Formal Discipline from the Stand- point of Experimental Psychology A THESIS presented to the Department op Psychology Leland Stanford Junior University IN PARTIAL fulfillment OF TliK requirements for the DEGREE oi Doctor of Philosophy By JOHN EDGAR COOVER April 8, 1912 [Published as No. 87 of the Psychological Monograph.] Formal Discipline from the Stand- point of Experimental Psychology A THESIS presented to the Department of Psychology Leland Stanford Junior University IN partial fulfillment of the requirements for the degree of Doctor of Philosophy By JOHN EDGAR COOVER April 8, 1912 [Published as No. 8: of the Psychological Monographs.] $^ \^ \(^ PREFATORY NOTE The writer was driven into this problem some years ago through reaction to instruction he was then receiving in educa- tion. He owes a debt of gratitude to those men who so faithfully and painstakingly retailed the newest notions culled from hereti- cal theory or hatched from the early and somewhat hasty labora- tory work. Interest was more fundamental in school training, in those days, than attention, suggestibility than obedience, and vo- cational training than mental training. Sherrington's 'spinal dog'^ had not yet removed by the skillful use of its scratch-reflex, the *reflex-arc' from the vertebrate nervous system, and it was supposed that we had been endowed with a mechanism by which reactions to stimuli could be literally 'specific'; school training consisted in 'specific discipline,' which was hysterically opposed to 'formal discipline.' Nor had the "All or None" principle of the terrapin heart,^ yet suggested for mental develop- ment the efficacy of strenuous innervation; training needed no spur but interest. Generous acknowledgment of all those incen- tives is hereby recorded; without them this investigation would not have been made. The writer is under many other obligations : First and fore- most to the Head of the Department of Psychology, Dr. F. An- gell, whose courteous extension of laboratory facilities and whose encouragement, advice, and assistance, have put the writer hope- lessly in debt ; to Professor Lillien J. Martin, for kindly interest and helpful suggestions ; to Miss Ruth Adele Sampson for valued assistance in the correction of the manuscript ; and to the follow- ing groups, almost wholly of fellow-students, who cheerfully and faithfully performed the onerous yet indispensable offices of reagents; those whose names are starred underwent periods of arduous training besides taking the tests : * Sherrington : Integrative action of the nervous system. 1906. Ch. IV. * Stiles : Am. Phys. Ed., 1910, 15 : i-S- iv PREFATORY NOTE In the experiments on Word-Marking and Weight- Lifting: *Mr. Geo. Snow-Gibbs. In the experiment on Discrimination : *Mr. G. Altnow, *Mr. T. Nakamura, *Mr. K. Yasuda, In the experiment on Reaction with *Miss E. Crandall, *Dr. F. Angell, *Miss Evelyn Brooks, Miss Reba Bland, Miss Rowena C. Bush, Miss Luella E. Behrens, In the experiment on Attention : *Miss Christine Madison, *Miss Elizabeth F. Lanktree, *Miss Susan M. Looney, *Mr. William T. Root, *Mr. Joel A. Snell, *Mr. Arthur Heche, *Dr. F. Angell, Mrs. Margaret E. Coover, Mr. Franklin F. Wolff, Miss Helen M. Clift. Miss Emma G. Gill, Miss Beatrice E. Freuler, Miss Ruth A. Sampson, Miss Rose F. Emery, Miss Bertha Van Zwalenburg, In the experiment on Reproduction: *Miss Alma Holmes, ♦Miss Gertrude D. Workman. *Mr. William T. Root, Miss Susan M. Toy, Miss Mabel I. Clark, Miss M. Dorothy Parkinson, Miss Nellie Allen, Miss Bernice C. Rowell, Miss Bonnye Anderson, Miss Gertrude B. Weaver. Discrimination and Choice : Miss Mabel McKibben, Mr. H. George, Mr. Geo. Snow-Gibbs. Miss Mila L. Coffin, Miss Ruth Cain, Miss Anita E. Sudden, Miss Ray Weaver, Miss Helen E. Evans, Miss Edith G. Engelhard, Miss Maud H. Bassett, Miss Nina E. Moise, Mr, Geo. L. Righter, Mr. Charles L. Allen, Mr. Asa C. Dimon, Mr. Roscal L. Draper, Mr. Stanley M. Reeve, Mr. Lloyd I. Tilton, Mr. Alexander Davis, Mr. Merton J. Price, Mr. Zeb. A. Terry, Mr. Robert W. Wilcox. Miss Mabel B. Angell, Miss Ruth Hutchinson, Mr. Harold A. Hughes, Mr. William H. Bloeser, Mr. E. Leib, Mr. Earl T. Dutton, Mr. Herbert L. Hubbard. CONTENTS PART I. ORIENTATION I. Introduction i II. Evidence from the literature of relationship between mental processes 5 1. Habituation to Distraction 5 2. Sensitivity 5 3. Discrimination 6 4. Association 7 5. Reaction 9 a. Interference 10 6. Memory 14 7. Voluntary control 30 8. Summary 31 PART II. EXPERIMENTAL I. Principally Quantitative 34 1 . Experiments on the more complex processes 34 a. Marking out words 34 ( I ) Conclusion 36 b. Estimating weights 39 ( I ) Conclusion 41 2. Experiments on the more simple processes 42 a. Sensible discrimination 42 ( 1 ) Control experiment 44 (2) Results 44 (3) Conclusion 46 b. Reaction with discrimination and choice .... 50 (i ) Control experiment 52 (2 ) Results 53 (a) Errors 53 (b) Time 56 (3) Conclusion 61 CONTENTS II. Principally Qualitative 64 I. Experiment on Attention (Variability in mental e. J- processes) 70 The tests 72 The training 73 Apparatus and procedure 74 The training results 82 (i) Extensive threshold of attention 82 (2) Learning 12-letter-rectangles 94 (3) Reaction to sound loi (4) Memory training 103 The test results 106 ( 1 ) Reaction to sound 108 (2) Marking out a's 113 (3) Marking out o's 116 (4) Card-sorting 118 (5) Typewriter-reaction 126 (6) Controlled reaction 131 (7) Sound discrimination 134 (8) Memory of sounds 138 (9) Memory of consonants 142 10) Memory of numerals 145 11) Memory of visual signs 148 12) Memory of associated pairs 149 13) Learning 12-letter-rectangles. Free. . 150 14) Same (With distraction) 154 15) Word-completion 157 16) Trains of ideas 158 17) Extensive threshold of visual atten- tion. Free 160 18) Same (With distraction) '. . 163 19) Tapping 164 Extent of variability 167 Causes of variability 171 The practice curve 173 General effect of special practice 178 Conclusion 181 Experiment on Reproduction 184 a. The tests 186 b. The traininar • • 188 CONTENTS vii c. Training results 190 ( 1 ) Processes 190 (2) Quantitative 196 d. Test results 199 (i) Recognition or choice of one of two letters 199 (a) Introspective analysis 199 (b) Quantitative analysis 200 (2) Reproduction and recognition of letters 205 (3) Sound discrimination 206 (a) Processes 206 (b) Scores 211 (4) Memory for visual symbols 211 (a) Processes 211 1 ) Immediate memory 212 2) Complete learning 212 (b) Scores 214 i) Immediate memory 214 2) Complete learning 215 (c) Summary 217 e. Conclusion 217 3. Critique of the Test-training-test type of experi- ment 219 4. Analysis of the 'Common Factor' 224 a. Factors involved in a single experiment .... 226 b. Practice-efifect upon these factors 227 c. 'Spread of Training' 230 d. 'Transference' 230 e. Both kinds of General Effect selected from former experience 231 f. Provisional classification of Common Factors 232 III. Conclusion 234 IV. Bibliography 243 APPENDIX A. Marking out words (Text pp. 34ff), Tables I-II, Plates I-VII 259, 261 Estimating weights (Text pp. 39ff), Tables III-X, Plates VIII-XII 260, 268 Sensible discrimination (Text pp. 42ff), Tables XI-XIX 269 Reaction with discrimination and choice (Text pp. 5off), Tables XX-XXVIII, Plates XIII-XVIII. . . 274 viii CONTENTS APPENDIX B. Data relevant to the Experiment on Attention, (Text pp. 7off) Fig. I. Symbol on cards (Text p. 76) 288 Fig. 2. Arrangement of compartments (Text p. 76) . 288 Fig. 3. Instruction to reagents (Text p. 76) 288 Fig. 4. Sample line of stimuli for typewriter-reaction (Text pp. 52, 76) 289 Fig. 5. Classes of scientists, etc. (Text p. 77) 289 Fig. 6. Series of stimuli in test on Controlled reaction (Text p. yy) 289 Fig. 7. Series of intervals in Sound discrimination (Text pp. 77, 78, 188) 290 Fig. 8. Symbols used in recording judgment (Text p. 78) 290 Fig. 9. Series in memory of sounds (Text p. 78) ... . 290 Fig. 10. Series for Memory of consonants, digits, paired associates (Text p. 79) 290 Fig. II. Visual signs (Text p. 79) 291 Fig. 12. Size and style of letter used in Tests 13 and 17 (Text pp. 80, 82) 291 Fig. 13. Word-Completion list of letters (Text p. 80). 291 Fig, 14. Practice curves of Training on Extensive threshold of visual attention (like Test 17) (Text pp. 83, 93) 291 Fig. 15. Practice curves of Training on Learning 12- letter-rectangles (like Test 13) (Text p. 94) 292 Fig. 16. Curves showing increase in complexity of pro- cess as effect of practice in Learning 12- letter-rectangles (Text pp. 95, 176) 292 Fig. 17. Curves showing change in kind of product as effect of practice in Learning 12-letter- rectangles (Text p. 174) 293 Fig. 18 Curves showing change in processes in Train- ing on Simple Reaction to sound (Text p. loi) 294 APPENDIX C. Scoring the 12-letter-rectangle (Text pp. 80, 187) 295 APPENDIX D. Influence of subliminal differences upon judgment in stimulus comparison (discrim- ination) (Text p. 185) 299 INDEX 301 PARTI ORIENTATION I. Introduction The conception of formal discipline^ belongs to the philosophy of education, and has dominated the educational theory of the greater institutions of higher learning from their inception to the present day. Twenty years ago it was vigorously assailed, however, by some of our eminent educators^ who claimed that our schools are suffering under the bonds of this tradition. It was charged that the conception is a myth; that the sole end of study is the information acquired by it; that no subject of study is of greater value than another except upon grounds of adjust- ing the student directly to his social, economic, industrial, en- vironment. It was explained that the 'Dogma' rests upon the 'faculty psychology' which lies respectably buried under the dust * Formal Culture: (formale Bildung; education formelle). The doctrine of the applicability of mental power, however gained, to any department of human activity. This doctrine is used as a standing argument for so-called disciplinary education, especially that in pure mathematics and classical languages. The assumption is that if the student masters these, he will thereby acquire a mental power that can be applied almost equally well to any kind of practical or professional life. This gymnastic theory of education involves the idea that it does not matter upon what the mind is exercised, provided only the exercise be rigorous and long-continued. — DeGarmo in Baldwin's "Dictionary of Philosophy and Psychology." This expression has been used to indicate the general reaction upon the ability of a student that is by many -supposed to spring from the method of study rather than from the content which is learned. — Ernest N. Henderson, in "A Cyclopedia of Education," edited by Paul Monroe. N. Y. : Macm. 1911. P. 642. ' Vid. Hinsdale, B. A. : The dogma of formal discipline. Proc. N. E. A., 1894, P- 625; also his book "Studies in Education," Ch. 2. DeGarmo : "On the report of the Committee of Ten." Ed. Rev. 1894, 7 -^y?. Rein: Outlines of Pedagogics (tr. Van Liew) 1895. P. 61. DeGarmo: Herbart and the Herbatians. P. 26. 2 JOHN EDGAR CO OVER of a century, and that its continued vitality is derived from its momentum in a field in which the 'resistance' of modern experi- mental psychology is wanting. It is true that the challenge was made by men of 'Herbartian' training, that the formal defini- tions of the conception were made by them and would not be entirely satisfactory to proponents, and that the Herbartian psy- chology is equally dead. Nevertheless, the point in dispute has shifted into psycho- logical territory, and it is for this reason that Experimental Psychology has come in contact with the venerable doctrine. The psychological question at issue concerns the relations which exist between the various mental processes. According to the 'faculty psychology,' the improvement of observation, or of reasoning, or of memory, or of any of the thirty-five faculties of the mind, in one field of experience was applicable to all other fields of experience; but improvement of one 'faculty' did not aflfect others. According to the Herbartian psychology, these 'faculties' are abstractions, adding nothing to fact and affording no explanation of mental phenomena;^ the Idea"* is the sole real content of the mind and through it alone are mental phenomena unified. Training of the mind is effected through the appropria- tion of ideas and is limited in its applications to the field of ex- perience to which the ideas belong. Both of these psychologies ^ Herbart's criticism of the 'faculty theory' lies in the two following ob- jections: (i) 'Faculties' are mere possibilities; there is no sensibility before sensation ; (2) they are class-concepts, obtained by a provisional abstraction from the inner experience, and then raised to the rank of fundamental forces of the mind and used for the explanation of our internal processes. Both criticisms are as telling against the established sciences of physics and chemistry as against the 'faculty-theory.' (i) The forces of physics do not exist apart by themselves any more than do the 'faculties,' but only in the phenomena called their effects; (2) they are abstracted from the con- crete phenomena and are class-concepts used for the explanation of the phenomena themselves. According to the criticism, there is no gravitation before the falling of the apple, nor can it as a class-concept, be used to explain the fall of the apple. * This Idea, as well as its relations to the emotions, feelings, and im- pulses, is hypothetical and does not at all square with the facts of inner experience. Vid. Wundt: Prin. of Phys. Psychology. (From Ger. Ed. 1893.) Pp. i8ff. FORMAL DISCIPLINE 3 have performed their service in the development of the science and are now of but historical interest. The question, therefore, so far as their authority goes, remains open. Special observation and investigation have sought to throw some light on the problem : (i) The contribution of formal school training to success or eminence in practical life has been estimated.^ (2) The relation- ship between undergraduate scholarship and preeminence in the graduate schools of law and medicine has been reported,® and the standing of our Rhodes scholars has been compared with that of their fellows in Oxford who had followed similar (classical) courses but with more rigorous training.^ (3) Experimental pedagogy has already contributed a large amount of data con- cerning the relationship between capacities employed in school work or between school subjects.^ (4) And the data of Cross- ''Halleck: Proc. N. E. A., Dep't of Superintendence. 1906. Pp. 34-41. Pritchett : Fifth annual report of the President and the Treasurer of the Carnegie Foundation for the advancement of teaching. 1910. P. 56. Dexter : High-Grade Men in College and out. Pop. Sci. Mo., 1903. 62 :429. Schuster : The promise of youth and the performance of manhood, a statistical inquiry. Univ. London. Galton Eugenics Laboratory Memoirs. 1907. in. Lowell : Appendix to report of the President and the Treasurer of Har- vard College. 1908-1909. Davis: An afternoon view of college life. Stanford Alumnus, 1912, 13:231. E.g., Cases as extreme as the following can scarcely be ignored : Every- one knows how formal and how unrelated to occidental affairs is the dis- cipline Chinese officials have undergone, yet "it is not on record in Washing- ton that the other foreign legations are in the habit of making allowance for any lack of acumen on the part of the Chinese legation; on the con- trary, the Chinese legation is regarded as one of the ablest accredited to this country." F. Angell, in a Commencement Address at Castellejo School, Palo Alto, California, 191 1. ' Lowell : College studies and the professional schools. Harvard Grad- uate Mag., 1910, 19:205. 'Pritchett: Ibid. p. 65. 'Catherine Aiken: Methods of mind training. Am. Bk. 1895. Harper's Mag. Editorial by Charles Dudley Warner. March, 1895. Winch : Accuracy in school children. Does improvement in numerical accuracy transfer [to accuracy in arithmetical reasoning] ? Jr. Ed. Psych., 1910, 1:557. 4 JOHN EDGAR COOVER Education have been found to contribute something toward an understanding of the 'general' nature of voluntary control in its various phases.^ It rests, how^ever, with experimental psychology to determine precisely the kind and extent of relationship that exists between the various mental processes, and this task appears to be well begun. There are already at hand some results pertinent to the question to be found scattered throughout the literature; and as a result of direct experimental investigation, the extent and causes of positive and negative influences upon one set of mental processes by reason of training upon another are, within the limits of the respective investigations, revealed. To these results we shall now turn, in the next chapter, before detailing our own experimentation in this laboratory. Winch : Further work on Numerical accuracy in school children. Idem. 2:262. Starch : Transfer of training in Arithmetical operations. Jr. Ed. Psych., 191 1, 2:306. Wallin : Spelling efficiency in relation to age, grade and sex, and the question of transfer. Ed. Psy. Monograph. Warwick, 191 1. Winch : Transfer of improvement in memory in school children. Br. Jr. Psych., 1908, 2:284; 3:386. Winch : Some relations between substance memory and productive im- agination in school children. Idem. 191 1, 4:95. Bagley : Educative process. 1907, Ch. XIII, p. 208. Ruediger: Indirect improvement of mental function through ideals. Ed. Rev., 1908, 36:364. 'Scripture, Smith, and Brown: On the education of muscular control and power. Yale Psych. Studies, 1894, 2:115. Scripture : Recent investigations at the Yale Laboratory. Psych. Rev., 1898, 6:246. Scripture: Cross-Education. Pop. Sci. Mo., 1900, 56:589. Davis: Researches in Cross-education. Yale Psych. Studies, 1898, 6:6; 1900, 8:64. Woodworth : Accuracy of voluntary movement. Psych. Rev. Mon., No. 13. 1899. Dresslar : Some influences which affect rapidity of voluntary movements. Am. Jr. Psych., 1892, 4:514- Raif: Ueber Fingerfertigkeit beim Clavierspiel. Zeits. f. Psychol., 1900, 24 :352-5. FORMAL DISCIPLINE 5 II. Evidence From the Literature of Relationship Between Mental Processes The results of experimentation in the psychological laboratory that have come before the writer's notice and that have a direct bearing upon the question of the kind and extent of relationship between mental processes, are portrayed here in topical fashion, which takes them out of their chronological order, with the hope that the grouping of evidence about the various types of mental activity will prove more satisfactory to the reader. The year of the published results can be seen in the foot-note references. I. Habituation to Distraction Vogt^" found that reacting on every stroke of a mentronome decreased the amount of continuous adding 47.7% ; that seven days' practice reduced this to 14% ; that the effect of distraction upon adding caused by synchronously reciting series of letters was decreased by practicing adding alone ; that the adaptation to the distraction of reciting series of letters while adding reduced the distraction of reacting to the stroke of the metronome while learning series of numbers by heart. Among his conclusions are : (i) An adaptation (Gewohnung) carries over from one special process to another; (2) Becoming habituated to a distraction while exercising one function, habituates to that distraction while exercising other functions. 2. Sensitivity Urbantschitsch^^ sought to determine whether there is a cross- effect (Wechselwirkung) between the senses. "While a uniform excitation was present to one sense a sensation was occasioned through another, from which I perceived accurately during the functioning of the new sense if there were any changes in the sensation of the originally stimulated sense." Listening to a tone lowered the limen for light ; it also affected olfactory, gustatory, "Vogt: Ueber Ablenkbarkeit und Gewohnungsfahigkeit. Psy. Arbeit., 1899-1900, 3 :62. " Urbantschitsch : Uber den Einfluss einer Sinneserregung auf die ubrigen Sinnesempfindungen. Archiv f. d. ges. Physiologie, 42:1 54- 6 JOHN EDGAR COOVER tactual, and thermal sensations. "The influence of one sense- excitation upon the sensations from other senses appears clearly to be a valid physiological law." Epstein^^ set himself to either controvert or support Urbant- schitsch's results by more carefully controlling the conditions of the experiment. He sought to find the influence of a sound- sensation upon (a) acuteness of vision, (b) acuteness of color- perception. The observer sat in a dark room and placed his eye to a telescope ; the cap was removed and he reported the number of concentric rings on a rotating disc ; the second stimulus was given during this fixation and he reported changes in the field of vision occasioned by it. 164 Experiments were made. Upon 60% of the reagents the sound impression increased both acuteness of vision and acuteness of color-perception; upon the other 40% it increased only the former. Dunlap and Wells^^ gave simultaneously visual and auditory stimuli to four reagents in reaction-time experiments, and found that when the auditory stimulus was reacted to, the reaction time was about 10 sigma longer than to the auditory stimulus alone; if the visual stimulus was reacted to, the reaction-time was about 20 sigma longer than to the auditory alone, but 40 sigma shorter than to the visual alone. 3. Discrimination "Volkmann*"* found that by practice of the left arm in dis- crimination until an initial ability of 23.6 improved to 11.2, the right arm without any practice showed an improvement from 26.4 to 15.7. Similar results were found for other cases of cross- education and for the spread of improvement in discrimination of touch at certain spots on the skin to neighboring spots." Bennett^ ^ trained 16 children (average age 11 years) of the " Epstein : Ueber Modification der Gesichtswahrnemung unter dem Ein- flusse von gleichseitigen Toneindriicken. Zeits. f. Biol, 1896, 33 :28. " Dunlap and Wells : Some experiments with reactions to visual and audi- tory stimili. Psych. Rev., 1910, 17:319. " Volkmann : Ueber den Einfluss der Uebung auf das Erkennen raumlichen Distanzen. Ber. der Kgl.-Sachs. ges. d. Wiss. (Math. Phys. Col.) 1858, 10:38. Quoted by Thorndike: Educational Psych. (First Ed.) P. 86. "Bennett: Formal Discipline. Teachers College, 1907, (pp. 59ff)- FORMAL DISCIPLINE 7 Speyer school in New York City, twice a week from October to March on discrimination of shades of blue, using a Milton-Brad- ley color-wheel. Judgments were made on whether the inner disc or the outer ring was the deeper blue (greater saturation). The stimuli were thus presented simultaneously. Tests were taken, before and after this training, on discrimination of shades formed by mixtures of (i) red+ white, (2) yellow+green, (3) orange-]- black, and of (4) pitch. For the latter test a Gilbert tone-tester was used, F sharp being taken for the norm. Effi- ciency was calculated in degrees of change of the sectors, or points of tone-change, corresponding to the range of the judg- ment "same." The average efifect of training was a reduction of this range: Boys 2.7° to 0.8°; girls 4.5° to 0.7°; or boys 65%, girls 80%. The range was reduced in the four tests: I 2 3 4 Boys 79% 60% 65% 28% Girls 84% 57% 56% 23% Since the First and Final tests were separated by five months, the practice efifect of the first test could not have been considera- ble. Clear transfer is thus shown from improvement in discrim- ination of shades of blue to shades of other colors, and in a less degree to discrimination of pitch. 4. Assotciation Thorndike and Woodworth^^ trained reagents in estimating areas, weights, and lengths : Six reagents were trained in estimating areas of rectangles from 10-100 sq. cm. in size, "until a very marked improvement was made." Tests were taken before and after this training on estimating areas of : a. Same shape and same size b. " " but 140-200 sq. cm. in size c. " " " 200-300 " " " d. Different shape and same size e. " " " 100-140 sq. cm. in size f. " " " 140-200 " " " g. " " " 200-240 " " h. " " " 240-over " " "Thorndike and Woodworth: The influence of improvement in one mental function upon the efficiency of other mental functions. Psych. Rev., 1901, 8:247ff, 384ff. 8 JOHN EDGAR COOVER Efficiency was calculated in amount of average error, and the final results were given in per cent showing the proportion of late to early error. In the whole table of 44 final figures 13 show lack of improvement. And of the 14 totals of the two tables, only one shows lack of improvement. The average results of four reagents (from Table IV, p. 384) show that training reduced error to 48.4% of the initial amount; that improvement was transferred to estimating the other sizes and forms to the fol- lowing per cents of their original error: a. 39, b. 67, c. 102, d. 62, e. 50, f . 78, g. 86, h. yy ; averages for two other reagents, who omitted c. and d., were a. 66, b. 61, e. 61, f. 84, g. 84, h. 76. Counting the complements of these per cents of error as per cent of gain, the gain in the training was 51.6%, the average gain in the tests within the field, same form, was 61 % ; different form, 38%; above the field, same form, 16%; different form, 51%; for all the tested abilities it was 29%. The extent to which the special practice shows general effect is, therefore, 55%. Two reagents were trained on a set of seventeen weights ranging from 40-120 grams inclusive, 5-gram intervals, and similar except in weight. Tests were taken before and after training, on estimating (a) 8 common objects averaging 95.8 grams, the weights of which fell within the field of 40120 grams; (b) 12 common objects averaging 736 grams, the weights of which fell above the field of 40-120 grams. Efficiency was calculated in deviations in grams. W. made in the last test in the training series 51% of the error made in the first test, T. 59.3%. W. showed in the after-training test on objects within the field over 100% of the improvement made in the training series; T. 32% ; W. on the objects above the field gained 6y% ; T. showed no improvement. Averaging the gains of both training and test series, 44.8% improvement was made in the former and 27% in the latter. As a result of the whole experiment the special practice showed general effect to the extent of 60%. The failure of transfer of improvement to diminish as the material estimated becomes less like the training material should be noted, for its bearing on the proposed explanation for trans- ference through 'identical elements.' FORMAL DISCIPLINE 9. 5. Reaction Angell and Moore^'^ carried out a long series of experiments with three reagents in reaction-time, in which the responses were made with the hand, the foot, or the Hps, to auditory and visual stimuh, in both 'sensorial' and 'motor' forms. Most of the visual series were not begun until after the auditory series (of from 700 to 1500 reactions) had been completed, and they show- ed a much shorter time than is usual for reaction to visual stimuli ; the authors observed that the form of reaction was the same in both groups and that the decrease in time in the latter must be referred to the practice-effect of the former. Gilbert and Fracker^^ tested three reagents on time of simple reaction, and time of reaction with discrimination, to sound stimuli, light, electrical, and tactual, stimuli ; then trained two of them on reaction with discrimination, and one on simple reaction, to sound, for 12 days. Of the 27 records in the re-test with the other stimuli, 25 showed transference of improvement to both simple reaction and reaction with discrimination. Thorndike and Woodworth^^ trained five reagents in reacting to words, in reading matter, containing both the letters e and s, ' by marking out the words. Before and after the training, tests were made in marking out ( i ) words containing, i-t, s-p, c-a, e-r, on similar pages; (2) a-n, l-o, e-r, on pages differing from the training material in length of line, size of type, and style of matter; (3) misspelled words and (4) the letter A from pre- pared sheets. Average gain in the training was 38%. In the tests the average per cent of gain was for (i) 21, 22, 10, 14; for (2) 28, 33, 31; for (3) 16; for (4) 10. The transfer of practice-effect was larger for (2) (dissimilar material) than for (i) (similar material). Altogether, the special training showed general effect to the extent of 44%. " Angell and Moore : Reaction time ; a study in attention and habit. Psych. Rev., 1896, 3:24s. " Gilbert and Fracker : Effect of practice in reaction and discrimination for sound upon the time of reaction and discrimination for other forms of stimuli. Univ. Iowa Studies in Psych., 1897, i ^62. "0/'. cxt. 10 JOHN EDGAR COOVER Carrie W. Liddle^*^ sought to determine whether practice in discrimination and sorting of one set of cards bearing colors or geometric signs would assist in discriminating and sorting another set with different colors or signs. Each set of 102 cards contained six colors, or six designs, was shuffled so that no color or device repeated itself, and was sorted into six compartments. The first six cards of the pack determined the order of colors in the compartments according to which the rest of the pack was to be sorted. Nine reagents took part and the experiment con- tinued two semesters. There was transference of practice-effect from one set of colors to the other set of colors, and to the geometric forms; and from one set of geometric forms to the otiier and to the colors. Increased powers of discrimination and attention were thought to be the causes of transference. (a) Interference Some contributions to transference of practice effect point out the fact that the effect is not always positive. It nevertheless indicates functional relationship of processes and is therefore important. Most of the material used in these contributions consists of discriminative reactions, which justifies general treat- ment in this place. Jastrow and Cairnes^^ found that when two mental processes, as (i) finger-movements involving rhythm and counting, and (2) adding or reading, are carried on at the same time, the fol- lowing effects are produced: (a) Simple movements are not interfered with; (b) maximum beating or beating in groups is interfered with: (c) beating in two's and three's (alternately) seriously interferes with reading aloud : (d) the maximum rate of beating hurries the mental process. Bergstrom has reported experiments showing interference clearly : ^'Liddle: Unpublished thesis for the degree of M.L. at the University of California. May, 1904. ^Jastrow and Cairnes: Interference of mental processes. Am. Jr. Psych., 1891-2, 4:219. FORMAL DISCIPLINE ii First.22 (i) A pack of 80 cards, 10 kinds in each pack, was distributed upon 10 piles. Another pack bearing the same ab- stract words at the top was distributed immediately after upon re-arranged places ; the results of the six reagents showed inter- ference in longer time for the second pack. (2) Cards with pictures of common objects sketched upon them were sorted as before; but with the difference that the time between the packs was varied from 3 to 960 seconds. Series were taken by four reagents for 20 days. Interference was shown for the following intervals (in seconds) between packs: 3, 15, 30, 120, 480; it decreased regularly with the increase of the interval; two-thirds dropped away in the first minute. Time of sorting decreased greatly with daily practice but the amount of the interference did not. In the shorter intervals the reagent was acutely conscious of it; in the longer, not at all. If the interval between the packs is increased to 24 hours, the second pack is sorted in less time; practice effect has overcome interference. Second.^^ Miinsterberg's "Gedachtnisstudien" (Beihefte, Heft 4) suggested the question, "Can a given association func- tion automatically while some effect of a previous and different association with the sense stimulus remains?" Packs of 80 cards, as before, were sorted on ten piles; orders of piles were changed and cards were changed. The answer to the question is affirmative: under certain "simple conditions, interference ef- fect of an association bears a constant relation to the practice effect and is equivalent to it." Bair^^ performed two experiments in which the practice effect much outweighed interference: (i) "Six keys of a typewriter are labeled with six symbols (letters or figures). Fifty-five of these letters or figures, in chance order, are now shown one by one, and the subject on seeing one taps the corresponding key. ** Bergstrom : Experiments upon physiological memory by means of the interference of associations. Am. Jr. Psych., 1892-3, 5 1256. ^ Bergstrom : Relation of the interference to the practice effect of an association. Am. Jr. Psych., 1894, 6:433. ** Bair : The Practice curve ; A study in the formation of habits. Psych. Rev. Mon. Supp., 1902, No. 19. 12 JOHN EDGAR C DOVER The time taken to tap out the series is recorded. Six different symbols are then used with a new series composed of them, and the subject's time record is taken as before. This is continued until twenty different sets of symbols have been used. Although the symbols have been changed each time, there is a steady im- provement, ranging for the four subjects in the following de- crease in time: 62 to 52, 95 to 85, 71.5 to 58, 65 to 56. The major part of this gain could not have been due to merely getting used to the machine or to the general features of the experiments, for the fourth subject was already used to these and still gained about nine-tenths as much as the other three.^^ "(2) The other experiment consisted in taking daily records, for twenty days, by means of a stop-watch, of the time required "The relation between interference and practice has also been just re- cently shown by Dr. Warner Brown (Habit interference in card-sorting. Univ. Calif. Pub. in Psychol., 1914, 1:269-321). Fifty-two playing-cards were sorted, according to suit, into a row of four boxes. In the First Series, 26 reagents worked 13 days, sorting 8 packs per day (except on the 3d, 5th, and 7th days when 4 packs were sorted), during the first 8 days with the "original" order of lebels ( D C H S), during the 9th to the 12th days, in- clusive, with a new order each day for the 3d to the 6th packs and the original order for the first and last two packs. Interference in passing from the original order to one of the new orders ranged from 1% to 23%, averag- ing about 12% (op. cit., p. 294). But in spite of the intrusion of practice in antagonistic reactions, practice on the original order resulted in considerable improvement: The actual amount of improvement from the 8th to the 13th days was considerably greater than from the 4th to the 8th days. The constant interference with practice did not prevent the steady improvement of skill with the original order (p. 307). In the Second Series, 18 students practiced card-sorting (twice a week) for 8 days, according to the same procedure as above except that 24 hours later they took an equal practice on the order C S D H. It was ascertained, after the elimination from the two series of all but 14 pairs of scores showing equal initial ability, (i) That antagonistic tendencies were carried over from one order to the other result- ing in loss of speed at the beginning of each day's work, (2) that the inter- ference phenomenon was confined to the first few trials of the day's sorting, (3) that it did "not affect the increase of skill in performing the action in the accustomed manner" (313), and (4) that practice on the second order helped in the learning of the first order. From the data of the First Series the author observed that "Four trials of any new order suffice to acquire a speed that it required six days, or 40 trials, to acquire in the original prac- tice (307). FORMAL DISCIPLINE 13 to repeat the alphabet from memory. Each day's experiment was as follows : First, the alphabet was repeated as rapidly as possible forward; second, the letter n was interpolated between each of the letters; third, the alphabet was repeated as rapidly as possible backward; and last, the alphabet was repeated back- ward interpolating n between each two of the letters. At the end of twenty practices in each order the subject repeated the alphabet first forward interpolating instead of n the letter x and repeating three times; secondly, interpolating r and repeating three times; then lastly, repeating backward and in like manner interpolating x and r and repeating three times. There was im- provement in the test series, the effect of the twenty days' training with the training series being to put the abilities in the test series as far ahead as three days of the direct training would have done."26 He concludes, concerning the first experiment, that "continued practice in one order increases proportionately the ability to make quickly and accurately a new and antagonistic order."^''' And that "any bit of special training helps us to find ourselves. It gives us a method of orientation which leaves us in our reactions not entirely at the mercy of chance even in unfamiliar situations. The experience which we get from special training gives us a general power to meet any entirely new situation with a more favorable response than had we not had this special training.^^ Louise E. Ordahl,^^ as a result of her work under Sanford's direction at Clark, concludes : "What Bair says in regard to the general ability given by special training, e.g., *to a new situation we react by a general discriminative reaction and are more likely to hit on a favorable response than without this special training,' is true of all learning. For no matter what new acquisition is undertaken if it is possible to master it, some previous general ^'Quoted from Thorndike : Educational Psychology (ist Ed.), p. 92. " Psych. Rev., 1903, 10 :s8o. ^ Bair : Contributions to Phil. Psych, and Ed., Columbia University, 1902. Vol. IX. ^'Ordahl: Consciousness in relation to learning. Am. Jr. Psych., 191 1; 22:158. 14 JOHN EDGAR COOVER training has either been developed by the individual or through the inherited co-ordinations of his ancestors." McMein and Washburn^*^ performed experiments with card- sorting to determine "Whether two relatively complex habits in- terfered with each other in a less or greater degree than two relatively simple habits." The more complex processes showed less interference. The relation between interference and practice effect then de- pends upon the time between practices and the complexity of the processes. The work of Liddle and Blair seems to indicate that improvement can be made in overcoming interference by frequent changes of the associations. This would be a very important general effect of practice. 6. Memory Some of the first work to show transference of practice effect in memorizing was done by Bergstrom^^ in connection with his investigation of interference in mental activity. He found that upon memorizing four series of non-sense syllables in suc- cession, with but 10 seconds between series, interference was progressive; i.e., each successive series took a longer time to memorize. Similar progressive interference occurred in learning 3 series of 30 digits in succession. This result conforms with that of Ebbinghaus.^^ To determine whether interference under these conditions was caused by recurring materials in re-arrange- ment, another experiment was performed in which the first series of non-sense syllables were made up of the letters of the first half of the alphabet, and the first series of numbers of the first half of the digits, while the second series were made up of the remaining elements. Since no materials recurred in re-arrangement and the time of the second series was greatly reduced, interference did not take place. Interference was considered an "after-image" of '"McMein and Washburn; Effect of mental type on the interference of motor habits. Am. Jr. Psych., 1909, 20:282. "Bergstrom: Influence of interference upon mental activity. Am. Jr Psych., 1894, 6 -.267 ff. *■ Ebbinghaus : Ueber das Gedachtniss, 1885, S. 95- FORMAL DISCIPLINE 15 central activity, and since it occurs with associations formed from impressions from disparate senses, it opposes the dismemberment of memory for facts into different sensory types, where by this distinction it is meant that the same thing can be learned more easily by one sense than by another. The principal fact to note here is that the practice effect of the first series on the second was marked. Bennett^^ reports transference of improvement in an experi- ment on memory. Two reagents took part; K. trained 28 con- secutive days, learning 16 lines a day of "In Memoriam"; B. trained 35 consecutive days, learning two stanzas of "Faerie Queene" daily. K. was tested on learning a row of 30 digits each day for five days, before and after the training; B. was tested on learning a list of 15 names of places each day for five days, before and after training. K. gained in tests 58% in the reduc- tion of time; B., 22%. Perhaps the most important investigation of the general effect of special training in memorizing, because of both the length and the rigor of the training, is that of Ebert and Meumann.^"* Al- though no control experiment was carried along, to determine the effect of the tests upon themselves in the repeated 'cross-sec- tions,' and the quantitative results, therefore, cannot be accepted as conclusive, the amounts of gain shown in the repeated tests appear sufficiently great to create a presumption in favor of the authors' conclusions which seem more fully warranted by the qualitative part of their study. This research continued from November, 1902, to August, 1903. Six reagents were trained on memorizing non-sense syl- lables, three taking 64 series, and three 48 series, of 12 syllables each. Usually the work of one day's training consisted in learn- ing two new series and relearning two series that had been learn- ed the preceding day. The number of the days of training for three reagents would thus be 32, and for the other three 24. ^'Bennett: op. cit. (pp. 45 f). ^ Ebert and Meumann : Ueber einige Grundf ragen der Psychologic der Uebungsphanomene in Bereiche des Gedachtnisses. Archiv f. d. ges. Psych, 1904, 4:1- i6 JOHN EDGAR C DOVER ' Before the training, after the six reagents had learned 32 series, after the training, and after a three-month interval of no practice, tests were given. There w^ere thus four 'cross-sections' of memory capacities taken : One for the purpose of determining initial efficiency, two for the purpose of testing the influence upon those capacities of the training on the non-sense syllables, and a final test given after a long interval without practice, to determine the durability of that influence. The tests involved (a) 'imme- diate memory' or the capacity of reproducing as much as possi- ble of a series of stimuli after a single presentation; and (b) 'complete learning' or the capacity of reproducing a series per- fectly after the fewest possible number of presentations. Reten- tion was also tested by relearning after 24 hours the series that had been 'completely learned.' The series for testing both sorts of memorizing consisted of both sense and non-sense material, and were presented auditorially for the tests on 'im- mediate memory' and visually, by means of a revolving drum, for the tests on 'complete learning.' The tests and results of the first three cross-sections follow : Table I. Effect of training on 'immediate memory.' (a) Number of units correctly reproduced after one presentation. ' Cross-sections Per cent gain f s ' A I 2 3 2 over I 3 over 2 3 over Numbers 7 8.8 11.2 29 26 59 Letters 7.2 9-5 1 1.3 36 19 58 Non-sense syllables 5.2 6.2 7-3 20 19 42 *Italian words 5 5-5 6.5 10 18 30 *Lines of poetry 15 17 19 13 12 27 *Lines of prose 17 19 22 12 16 29 * Only two reagents. (b) Number of units reproduced of which a third were errors. Numbers 10.3 15.8 17.7 56 12 71 Letters 11.2 14-3 16 28 12 43 Non-sense syllables 7.7 11.2 12.2 49 9 59 *Italian words 7.5 11.5 12.S 53 9 67 * Only 2 reagents. FORMAL DISCIPLINE 17 Table 11. Effect of training, on (a) 'Complete Learning,' and (b) Retention for 24 hours. Number of Presentations per unit, — per line for poetry and prose. Cross- sections Per cent gain r I 2 3 r 2 over I 3 over 2 3 over Non-sense syllables (a) 2.11 .83 48 61 43 n (b) •49 •27 20 45 35 59 Optical symbols (a) 3.83 2.23 90 42 60 11 (b) .68 .35 30 49 14 56 Italian words (a) •273 • 175 108 36 38 60 (b) .056 .040 036 29 10 36 Lines of poetry (a) ■75 .60 47 20 22 Z7 (b) .14 .08 07 43 13 50 Lines of prose (a) 145 .82 50 43 39 66 (b) •30 .10 09 67 10 70 From the above tables we observe ( i ) that the total improve- ment in the untrained 'special memories' compares very favora- bly with, indeed sometimes exceeds, the improvement in memory for non-sense syllables; (2) that the gain of the 3d over the 2d cross-section compares favorably both with the gains of the 2d over the ist, and with the improvement made in the second period of training; (3) that the first observation above is applicable to "retention after 24 hours," but that the second is less so; the improvement in retention was not so great for the second period of training (35% as against 45%), and segregation of the scores shows that the group of reagents who took but 16 series in this part of the training is more responsible for the lack of great improvement than is the group who took 32, to the extent of the ratio 19:38 (%'s), and is more responsible for the lack of great improvement in the other tests to the extent of the ratio 8:11. It may be remarked, however, that this group had training equal in extent to that of the other group before the 2d cross- section, and yet in the test on the training material it showed less improvement in retention to the extent of the ratio 28 153 al- though it showed equal average improvement in retention in the other tests (48%). And it may also be noted that this group made less improvement on Non-sense Syllables in the 2d c-s, and i8 JOHN EDGAR COOVER more than the other group in the 3d c-s, while it equalled the other in average per cent gain on the other tests of both 2d and 3d cross-sections. (4) In connection with the preceding observation it may be noted that although great improvement was shown in the wider capacity of 'immediate memory' (permitting 33% er- rors — see Table Lb), the second part of the training did not contribute much toward it. (5) Amount of improvement does not seem to depend upon similarity of the test to the training material; e.g., Numbers in Table I. and Optical Symbols in Table II. were more dissimilar to the training material than any of the other non-sense tests, yet improvement in the former ex- ceeds and in the latter equals that shown on the training material : and the gain shown on lines of Prose, in Table II., which differs greatly from the training material in the form of connection be- tween the units of the series, is higher than that shown on lines of Poetry which in the respect noted is more similar to the train- ing material. The permanence of the improvement in memorizing, as shown by the fourth cross-section,^^ would seem to support the claim that it was not effected by the test practice. Unless we assume that all the gain on the tests, other than on non-sense syllables, is the result of the practice effect of the pre- ceding tests, we must refer some of the gain in these tests to the training on non-sense syllables ; how much, cannot be safely esti- mated until the tests are repeated, without training, under the same conditions.^® ^Five of the reagents were here given tests on non-sense syllables, and two of them on verses of Poetry, after an interval without practice. This interval was 75, 85, 91, 146, 156, days respectively. There was no decrease in efficiency; some of the reagents showed an increase. '^ This appears to have been done by Dearborn (Psychol. Bull., 1909, 6:44) whose results "indicate that a considerable part of the improvement found must be attributed to direct practice in the test series, and not to any 'spread' of improvement from the practice series proper." G. E. Mviller (Zeits. f. Psychol., 190S, 39:111-125), Wessely (Neue Jahrb. f. Pad., 1905, 8:379-380), Sleight (Br. Jr. Psychol., 191 1, 4:39ofif), among others, have also criticised the quantitative evidence in the original research. A pupil of Meumann's, however, who took the precaution of performing control experiments, pre- sents, in a Dissertation published from Ziirich, evidence that special training FORMAL DISCIPLINE 19 The introspective evidence, as noted above, seems more con- clusive than the quantitative. The training effected ( i ) a change in the whole psychical habit of memorizing, which was applica- ble to the varied test material: Distaste for the exacting work changed to zest ; muscular tensions decreased in intensity and ex- tent ; innervation for the work became strong and constant ; atten- tion became more economically directed over the parts of the presented material; concentration became more constant; etc. And it effected (2) a change in the method of memorizing : The progress in 'complete learning' became methodical, and learning a series which soon after the beginning of the training was carried on in the various successive stages by definite and distinct mo- tives, as (a) orientating, (b) apperceiving, (c) combining units into a rhythm, (d) anticipating syllables, (e) proving memory, became more of a continuous process in which the various mo- tives were economically combined; the mnenonic aids at first grasped at by all reagents were advantageously discarded, etc. The experience of the reagents thus attests the fact of the general effect of their special practice, and reveals in some manner the cause of this general effect. Fracker^'^ makes another important contribution proving gen- eral effect of training when the materials memorized are of a simple nature. A group of four untrained reagents took the tes:s with the trained reagents, permitting determination of practice effect of the first test upon the second, and the plan and control of the experimentation are excellent. Eight reagents were given four weeks' training (two or three days a week) in memorizing series of 9 sounds made up of four intensities. These were produced by an electro-magnetic fork and were delivered through a telephone. The four different in- tensities were produced by switching into the circuit different resistances. Improvement was made by all but one reagent (Fs.). in memory is also general training in memory. (Radossawljewitsch: Das Fortschreiten des Vergessens mit der Zeit, 1907, S. 182). " Fracker : On the transfer of training in memory. Psych. Rev. Mon. Sup., 1908, No. 38:56-102. 20 JOHN EDGAR COOVER Before and after the training the eight trained and the four untrained reagents took the following memory tests : 1. Poetry (Two stanzas of "Eve of St. Agnes") 2. Order of Four Grays (40 groups) 3. " " Nine Tones (20 groups) 4. " " Nine Grays (20 groups) 5. " " Four Pitches (40 groups) 6. " " Nine Geometrical Figures (S Trials) 7. " " Nine Numbers (10 series of 9 double numbers) 8. Extent of arm-movement (10 trials for each of 3 standards) No's 2, 3, 4, 5, were given in "double fatigue order" to equal- ize fatigue and practice effects upon them. The relation of the tests to the training series was as follows : No. 2 different in content, same in method. No. 3 same in content, different in method. All others, different in both content and method. No. 2. A group consisted of 4 grays (No's 2, 7, 30, 45, — Hering) exposed at the rate of one second, remaining exposed one-half a second, with an interval of 4 seconds between the groups. In this interval the reagent responded aloud in num- bers I to 4, 4 being the darkest gray, reproducing the order of the 2d preceding group. No. 3. A group consisted of 9 intensities of sound, delivered at the rate of one second, each continuing one-half second. In a nine-second interval between groups the reagent responded aloud in numbers i to 4, 4 being the loudest sound, reproducing the order of the preceding series. No. 4. Same as No. 2, except that 9 grays were given in 1 group. The reagent responded aloud in numbers, during a nine- second interval between groups, as in No. 3. No. 5. Same as No. 2 in method, except as to the response., which was made by naming Do, Mi, Sol, D0-2. The stimuli were the notes of the major chord struck upon a piano. No. 6. The geometric figures were drawn by joining three straight lines (two long, one half-length) so that they joined only at the ends or in the middle, the long lines always adjoined, and formed right angles, none crossing. The nine symbols were FORMAL DISCIPLINE 21 exposed simultaneously for 10 seconds. The reagent responded by drawing the figures, within a time limit of one minute. No. 7. Nine two-place numbers were read aloud at the rate of one and one-half seconds. The reagent responded by record- ing within the time limit of 15 seconds. No. 8, The reagent, with eyes closed, moved his finger with free arm movement along a glass rod from a stationary piece of tubing to one adjusted by the experimenter; he moved his fin- ger out and back twice ; then moved it out to the position he esti- mated to be the same (the adjustable tube being removed) ; three standards were used. The tables show that of the 8 trained reagents six made their greatest gains, and the others made large gains, in No. 2, where the material was grays, and the method was the same as in the training on sound ; the other two made their greatest gains, and three others made good gains, in No. 3, in which the content was the same as in the training, but the method was different. In No. 4, four made large gains. In No. 5, in which the method was the same as in No. 2, but in which the material was series of pitches, responded to by name, four made large gains. Of the tests which differed widest from the training in material and method, three made large gains in No. 6 (Geometrical figures) ; three made fair gains in No. 7 (Nine numbers), and one in No. 8 (Movement). In No. i (Poetry) four made fair gains. By grouping the tests in the order of similarity to dissimilarity *as compared with the training, and averaging the per cent gain^^ of the trained and of the untrained reagents, we get : Similar Dissimilar Tests 234 5 6781 Trained reagents 36% 22 19 10 13% 407 Untrained reagents 4 II 10 -2 80-12 Difference 32 11 9 12 54-15 The average gain in training was 21 %. ^ The per cents gain are the differences between the scores which were expressed in per cent of a perfect score; they are not reckoned upon initial efficiency. 22 JOHN EDGAR COOVER And if we average the per cent gain for the respective re- agents, we may compare the (a) test gains with the (b) training gains, and the (a) gains of the trained with the (c) gains of the untrained reagents : Reagents i 2 3 4 5 67 8 9 10 11 12 (b) 32 25 -9 41 27 29 17 (9) (a) 10 15 6 15 30 15 6 14 (c)s 8 10 13 The average gain for the trained reagents on the similar tests was 22%, on the dissimilar 6% ; for the untrained reagents 6% and 3%. Which indicates that gain in the dissimilar tests was harder to make; that there was greater transfer of improvement in the training to the similar material than to the dissimilar mate- rial, and that there was transfer to the dissimilar. The greatest transference to the similar material, however, appears to have been made on No. 2, in which the method and not the content was similar to the training. But Fracker's emphasis is rightly placed on introspective an- alysis rather than on quantitative results, and he gives us a good account of the processes involved in the work of his reagents. The training in memorizing the order of four intensities of sound developed for nearly all the reagents individual systems of visual, visual-auditory, or visual-motor, imagery, involving four steps or four positions by which the sequence of presented intensities was remembered as imaged movement among these positions. These systems of imagery were carried over to the tests involving sequence of four graded units (No's 2, 3, 4, 5) and replaced, for the trained reagents, the changeable and tenta- tive methods employed by all reagents, in the first series of tests. Where the tests were more favorable to the application of this developed imagery, as with the four grays ( No. 2 ) , it contributed most effectively to the increase in the score ; where conditions did not permit its application in toto, as with series of nine units be- fore response (No's 3, 4) and with a series demanding a different form of response (No. 5), there was interference and its efficien- cy was decreased. Apparently, the dependence upon the system of imagery where interference is great, as in the changed form of FORMAL DISCIPLINE 23 response (No. 5), results in lower scores than were made wher»f the imagery was seen not to apply and reliance was placed upon other and more general effects of practice: more improvement was shown with the simultaneously presented geometric symbols (No. 6) than was made with the pitches (No. 5). Those other effects of training responsible for improvement here, and also in the tests on numbers (No. 7) and poetry (No. i) appear from the introspection of the reagents to have been (a) systematic grouping of material, (b) freer use of imagery in connection with this grouping, (c) more economic distribution of attention, (d) better concentration of attention, (e) more confidence in power to master the situation, etc. Fracker explicitly states three factors of improvement in training and of transference to the tests, be- sides the development of these systems of imagery: (a) attention to the essentials, (b) association responsible for building the sys- tems of imagery, (c) automatic use of the imagery where applicable. It is the chief virtue of this research that it has made clear the importance of individual systems of imagery as factors in the general effect of special practice in memorizing. So far, repre- sentative imagery had not been given, in researches connected with the theory of formal discipline, the attention it merits, al- though its place in mental life is well known. ^^ The ancients, according to Cicero, based their systems of mem- ory training upon spatial position or location, and it is not unusual todav for university students to fix points in a lecture in mind by ranging them along the windows and doors of the room. The psychological laboratory has revealed the prominence of this kind of imagery in the mental organization of not a few re- '^ The writer desires in this connection to allay the misapprehension on the part of the author {op. cit. QSflf.) that his results do not accord with those published by Coover and Angell (Am. Jr. Psych., 1907, 18:327) with respect to the relation between imagery and improvement in practice and between imagery and transference of practice-eflfect. Representative imagery may be of high value in processes of memory and at the same time be one of the chief distractions in processes of discrimination and of reaction with discrimination and choice. 24 JOHN EDGAR COOVER agents. As an example, Miss Gamble'*^ uses systems of represen- tative imagery extensively in memorizing series of stimuli, whether they are odors, colors, or non-sense syllables. The in- crease in facility in memorizing smells was due to a newly ac- quired system of associating members of a series with spots on the table-top, which she had previously developed while working with colors. She has worked in smell experiments for eleven years and learns large series of smells with phenomenal facility, yet she has no true smell imagery ; her olfactory impressions are replaced primarily by color-images suggested by the materials. With respect to the influence of practice on associations, she says that representative associations remain but that auxiliary asso- ciations (as, black-pink, being held by "red eyes of a mourner"; or green-yellow, by "green grass a hunting dog runs over and the corn-meal mush he is fed on";) drop away. Sleight^ ^ has recently made a substantial contribution to the question of transference of improvement in memory. He takes his departure from a criticism of the work of Ebert and Meumann, Fracker, and Winch. His objections to the first are (i) too few reagents for statistical treatment of results, (2) no control seriei- to determine the practice-effect of the tests themselves, (3) no evi- dence that the tests in the various cross-sections were of equal difficulty, (4) inadequacy of the per cent form of statement of results, for one reason because per cents are not equivalent when they are not reckoned upon equivalent stages of practice. His objections to the second are confined to the first of the four just quoted. These criticisms will receive some attention in theoretical discussion later. His first series of experiments were performed with children of three girls' schools, numbering 21, 28, 35, respectively (aver- age age, 12 yrs. 8 mo.). Ten different kinds of tests were given to ascertain the mem- orizing power of each child; these constitute the first 'cross-sec- tion,' upon the combined results of which the children of each ^"Gamble: Study in memorizing various materials by the reconstruction method. Psych. Rev. Mon. Sup., 1909, No. 43:1. *^ Sleight: Memory and Formal training. Br. Jr. Psych., 1911, 4:386. FORMAL DISCIPLINE 25 school were arranged in order of merit and divided into four approximately equal sections of equal merit ; one group was prac- ticed on learning poetry by heart, one on tables, one on prose substance, and one had no practice but spent the time on arithme- tical problems or some other task not involving memory work. The practice period lasted six weeks, four days a week, 30 minutes a day. A second cross-section of tests was taken in mid-practice, a third at the close, and a fourth after six months. In practice ( i ) the poetry group repeated line by line after the experimenter until the average child could repeat the whole with- out help. Meter and Hues varied. ( 2 ) The group on tables learned multiplication, denominations, squares, fractions, etc. (3) The substance-prose group heard read twice selections from scien- tific, geographical, historical, narrative material, and reproduced the substance of the narratives. The ten tests were intended to appear to develop out of ordin- ary school work and were representative of many different pro- cesses regularly involved in memory work; they included verbal and logical associations, in couplets and continuous, of letters syllables, and names; spatial associations with one presentation and with several. Averages were tabulated ; and the differences between the aver- ages of cross-sections one and two, two and three, and one and three, for each group of children for each test, were divided by the average of the test standard deviations of the three respective schools. A table of significant values was formed by subtracting the difference-score of the unpracticed group from the difference: scores of the three practiced groups, in each test for the three test-comparisons. Where these values (the difference between the difference-scores of the practiced and unpracticed children) were over three times as large as the probable errors, they were accepted as significant. Since we are limited to a consideration of the 'significant' values, of which there are a few, it is interest- ing to note that ( i ) the group trained on poetry showed transfer effect to non-sense syllables and the map test having lost in the test on poetry by an insignificant amount. (2) The group 26 JOHN EDGAR COOVER trained on tables showed transference to points (spatial memory) and to non-sense syllables, having lost on dates by an insignificant amount. (3) The group practiced on prose-substance showed transference to prose-substance and to names. (4) Of the three trained groups only one shows transference to the test most similar in material (Prose-substance) to their training. (5) If the indications of transference are limited to those values which "present really strong evidence" (five times their probable error), but three cases remain: Prose-substance to prose-sub- sfance, poetry to non-sense syllables, and tables to non-sense syllables, in the order named, the last being the greatest. (6) The only negative transference occurred from tables to dates, in which arbitrary 'paired' associations were common, but in the latter the sequence of the pairs was disturbed. The following causes of transference were conjectured : ( i ) From poetry to non-sense syllables, the common element of rhythm; (2) from tables to non-sense syllables, rhythm and arbitrary associations; from tables to points (spatial memory), visual imagery; (3) from prose-substance to prose-substance, identical material. Since the results of this experimental series with children were of an unexpected nature and their interpretation had to rest wholly upon statistical analysis, another series was carried out in order that they might be confirmed or contradicted by the new results and introspective analysis. The second series of experiments were conducted upon two classes of young women (average age. 18-19 years), first-year students, in the Training College (London). The method and tests were approximately the same as in the first series, except that but six tests constituted a cross-section. The training of three groups of poetry, tables, and prose-sub- stance respectively, continued twelve consecutive days (a Sunday excepted), 30 minutes daily, and differed from the training of the children in that poetry and tables were learned from manuscript instead of from oral repetition. The statistical treatment of the results yields six 'significant' values, and two of these are nega- FORMAL DISCIPLINE 27 tive : ( I ) The group practiced on poetry showed transference to nonsense syllables and to poetry ; (2) the group practiced on tables showed transference to dates; (3) the group practiced on prose- substance showed transference to prose-substance, and negative transference (interference) to non-sense syllables and consonants. Causes for transference, based largely upon introspective an- alysis, were presumed to be : ( i ) From poetry to poetry, identical material and rhythm; from poetry to non-sense syllables, rhythm; (2) from tables to dates, visualization ; (3) from prose- substance, to prose-substance, identical material; negative trans- ference to non-sense syllables and to consonants, repugnance for mechanical learning after training on easier logical material. This second series seems to be quite independent of the first, the significant values of the respective tables of general results agreeing in but one case. The introspections of the adults throw no light upon the results from the children, and fail, in their naivete, to furnish satisfactory explanations for their own re- sults; they are not to be considered comparable to the introspec- tions of the reagents of Ebert and Meumann, and of Fracker. The results of both series must stand on statistical analysis alone. In his analysis of results, Sleight finds Thorndike's early law of 'Identical elements' too simple. Cases of transference shown in the above paragraphs do not follow it: (a) The group prac- ticed on tables did not show improvement, but loss, in their 3d test on dates ; due to the fact that in the tests they were told the number of repetitions that would be given, whereas in the training they were not, which resulted in a change in the direction of the attention; and (b) the group trained in poetry lost in their third test on poetry, probably for the same reason. Trans- ference to the tests more remotely related to the training material was occasionally large; the greatest transference being to tests on non-sense syllables from training on poetry, and on tables. "The relation which produced transference is not necessarily (a) an external relation perceivable by an observer, nor (b) a relation perceivable by the learner; but (c) a common factor, of which the individual mind makes use, consciously or uncon- 28 JOHN EDGAR COOVER sciously. The individual's awareness of the usable common element may produce an earlier and greater effect." The com- mon element "must be separable from the complexes" in which it occurs. This disintegration resists transference. "The fac- tors which make for transference are similarities of a funda- mental nature, such as specific forms of attention, imagery, rhythm; in short, similarities of procedure;"^- changes in these affect transference more than changes in material. The general conclusion to which the author is led is that "Specific memory training is specific in its effects."^^ It appears to the writer that there are apparent reasons which mitigate this conclusion: (i) The unpracticed group was not unpracticed, since these tests involved processes largely exercised in the ordinary work of school. This influence tends to erase the difference between the practiced and the unpracticed groups. (2) There is sufficient indication by significant positive and negative values, of relationship of processes to count against merely specific effect of practice. (3) The assumption that "the mental processes have probably been independent" when the influence of the training is not revealed by "significant" dif- ference-scores, is opposed to the general introspective evidence of other investigators. (4) The criticism of irregularity of results, applied to Winch by the experimenter, seems applicable here, since, as was noticed above, the two tables agree in but one entry. Miss Gamble ^^ in the course of memory investigation with odors, colors, and non-sense syllables, found a great increase in facility in memorizing smells after practice in memorizing colors with which she developed a spatial system of representative imagery. The new system was carried over. And "The results of G. seemed to show that practice gained in the earlier ex- periments with smells and colors was transfered to the learning of non-sense syllables." FORMAL DISCIPLINE 29 Rall^^ tested two groups of students, upon each of three days, in memorizing poetry (Evangehne) and in memorizing non-sense syllables. One group (44) trained on memorizing for four weeks, 20 minutes a day, individuals choosing various material, such as poetry, prose in English and in foreign lan- guages, irregular verbs and vocabularies in foreign languages. Improvement in training was measured by comparing the last three with the first three days. The second group of 28 took no training. Results showed wide variations: Of the 44 trained reagents^ 4 lost in the training and one lost in the Evangehne tests ; 6 out of 34 lost on the non-sense syllables. Of the untrained group, 4 out of 28 lost on Evangeline, 3 out of 16 lost on non- sense syllables. In the training 22 improved more, 20 less, than in the Evangeline test, and 2 improved the same amount; 23 out of 34 improved more (or lost less) and 11 improved less in the training than on the non-sense syllables. The following table gives the average improvement in per cent : No. Training Evangeline Non-sense Syllables Trained 25 32-5 26.9 24.5 Untrained 20 17-8 12.0 Difference 9-1 12.5 In this experiment, then, 25 students gained in training 32.5%, of which 28% was transferred to the poetry test, and 38% to the test on non-sense syllables. A second investigation was made which agreed with the first in showing the general effect of special practice, but the amount of the transference was smaller. Radossawljewitsch^^ in conducting experimentation directed by Meumann found that special practice in memorizing improves memory in general. ^Rall: Some experimental evidence on the transfer of training in mem- ory. Psych. Bull., 1912, 9:88. "' Radossawljewitsch : Das Behalten und Vergessen bei Kindern und Er- wachsenen nach experimentellen Untersuchungen. Pad. Mon. Von Meumann. 1907. 30 JOHN EDGAR C DOVER Miss Talbot ^'^ in training her visual memory, not in power but in frequency of use, found general effect in its making memory more sure. 7, Voluntary Control Book ^^ found, in his investigation of learning typewriting, that in learning to 'short-circuit' to a higher order of habits, besides habits of manipulation, there were involved 'habits of control.' These with other mental habits "when developed in the sight method of learning, were carried over to the touch method of learning and were used to good effect." Judd and Cowling ^^ report that improvement in drawing an imaged form with the eyes open was transferred to efficiency in drawing with the eyes closed, Wallin ^^ trained two observers in nonocular control of re- versions in a number of reversible perspective outlines, such as a book or a pyramid. "Practice consisted in the attempt to uniformly envisage the infrequent or non-predominant perspec- tive." Improvement in 9246 trials, expressed in per cent of suc- cessful control, amounted to an average of 42% between the averages of the first and last 20 days (40% and 82%). Tests showed that the improvement was shared by the unpracticed eye. "The effects of practice are central; the training of the one eye established certain cortical tendencies and mental attitudes. The unused retina therefore tended to respond in harmony with the central disposition." Also, earlier in the experiment, reversions occurred about two and one-half times faster in direct vision than in peripheral vision; after practice, reversion occasionally occur- red most readily when the figure was in peripheral vision ; there was "transference of fixation motives attaching to the fovea to the peripheral retina. . . . The foveal tendency was transmuted into a 'generalized retinal habit.' " " Talbott : Attempt to train the visual memory. Am. Jr. Psych., 1897, 8 :4I4. *' Book : Psychology of Skill with special reference to its acquisition in Typewriting. Univ. Mont. Bull., 1908, 53:75. *'Judd and Cowling: Studies in perceptual development. Psych. Rev. Mon., 1907, No. 34:349. *" Wallin: Doctrine of formal discipline: Two neglected instances of transfer of training. Jr. Ed. Psych., 1910, 1:168. FORMAL DISCIPLINE 31 8. Summary The following summary gives briefly the results of the psycho- logical investigations, coming before the notice of the writer, and reviewed in the preceding pages, which have a direct or in- direct bearing upon the question of functional relationship be- tween various mental processes : (i) Improvement in habituation to distraction is general (Vogt) ; (2) sensitivity of one sense is increased when impres- sions from another are simultaneously received (Uibantschitsch, Epstein) ; (3) reaction-time to one of two simultaneous stimuli is different from the time to either alone, being shorter to a visual stimulus when a sound stimulus is also given than to a visual stimulus alone, and longer to a sound stimulus when both are given than to a sound stimulus alone (Dunlap and Wells) ; (4) increase in sensible discrimination of two points is shared by surrounding areas (Volkmann) ; (5) improvement in discrim- ination of shades of blue (for school children) was transferred to facility in discrimination of shades of other colors, and of pitch (Bennett) ; (6) improvement in estimating areas, weights, and lengths, was transferred to capacity to estimate areas, weights, and lengths, different from those used in the training (Thorndike and Woodworth) ; (7) time of simple reaction, and time of reaction with discrimination, to light, electrical, and tactual, stimuli, were reduced through training on simple reaction or reaction with discrimination, to sound stimuli (Gilbert and Fracker) ; (8) reaction time to visual stimuli was lowered by pre- vious practice in reaction to sound stimuli (Angell and Moore) ; (9) improvement in marking out words containing each of two given letters, on pages similar to and different from those used in training, and in marking out capital A's from a sheet contain- ing 500 capitals, resulted from training in marked out words containing the letters e and ^ (Thorndike and Woodworth) ; (10) there was transfer of practice-effect in sorting cards, from one set of colors to another set of colors, and to geometric forms, and from one set of geometric forms to another, and to the colors (Liddle) ; (11) functional relationship and interdepend- ence of mental processes are shown by the phenomena of inter- 32 JOHN EDGAR COOVER ference: (a) simultaneous processes may interfere with each other or augment one of them (Jastrow and Cairnes) ; (b) sort- ing a pack of cards in re-arranged compartments immediately after sorting another, takes longer time, and in learning rows of non-sense syllables successively, the time becomes progressively longer if they possess recurring elements (Bergstrom) ; (c) fre- quent changes in typewriter-reaction, however, result in capacity to make new and antagonistic series of reaction in less time, and practice in repeating the letters of the alphabet with the interpola- tion of a given letter between each two, increases capacity to repeat the same series with the interpolation of a different letter (Bair) ; (d) and there is less interference between complex than between simple processes (McMein and Washburn); (12) training in memorizing poetry improved memory for digits and for names of places (Bennett); (13) memorizing non-sense syllables im- proved memory for letters, numbers, words, meaningless sylla- bles, Italian words, verses of poetry, lines of philosophic prose, and optical symbols (Ebert and Meumann) ; (14) memorizing series of sounds improved memory for series of grays, tones, pitches, a square of geometrical figures, and verses of poetry (Fracker) ; (15) training of memory (for children and students) upon matter similar to schoolwork, resulted in a few significant gains in memory for material similar to, and different from, that used in the training, and deteriorated memory for some similar material (Sleight) ; (16) memorizing colors gave greater facility in memorizing odors, and practice gained in memorizing odors and colors was transferred to memorizing non-sense sylla- bles (Gamble) ; (17) memorizing poetry or prose in English or in a foreign language, or irregular verbs or vocabularies, improved memory for poetry and- for non-sense syllables (Rail); (18) improvement in voluntary control was carried over from sight to touch typewriting (Book), from drawing with the eyes open to drawing with the eyes closed (Judd and Cowling), from use in connection with one eye to use with the other, and from the foveal to the peripheral vision (Wallin). An examination of the data reveals the fact that special prac- tice is not wholly general in its effects ; is often not largely gen- FORMAL DISCIPLINE 33 eral, but probably is always somewhat general. Under the conditions of the experiments it usually ranges in amount from a fourth to three-fourths of the gain shown in the training.^ ^ Of greater importance than the fact of the general effect of 'specific training' however, is the theory that will account for all the facts that have come to notice through the course of inves- tigation and that point to the relationship of all mental processes, but to a relationship that is by no means a simple one. Progress toward such a theory is made by the discussions of investigators, quoted above, who have subjected their numerical and introspec- tive results to a more or less thorough analysis. But since there is, on the surface at least, some conflicting evidence, further work is necessary to unravel the tangled skein of positive and negative relationships, and to account for some apparently anomalous re- sults^^ which occasionally occur. "It is interesting and perhaps significant that the distinction between gen- eral and special effects of practice is so generally recognized by the Danes that their language provides separate terms for them (Faerdighed, general; Udenadlaeren, special), vid. Meumann : Beitrage zur Psychologie des Zeit- sinns. Phil. Stud., 1893, 8:435- "Of which further notice is taken on pp. 64ff'. PART II EXPERIMENTAL I. PRINCIPALLY QUANTITATIVE The following four experiments (performed 1903-5) took their departure from the work of Thorndike and Woodworth.^ The amount of transference was still in question, and the theory was still that of "identical elements" of a relatively simple nature^ largely motor. Although the principal contribution is quantita- tive, the analysis of processes is not neglected — all experiments were accompanied by introspections. 1. Experiments on the More Complex Processes In order that we could start on an even footing with the inves- tigation referred to, it was necessary to repeat some of the experi- ments ; the following two were chosen : a. Marking out words containing two given letters, and b. Estimating weights. a. Marking Out Words Two reagents were trained for 1 1 days in marking out words containing e and .$■ in selected columns of the "Outlook" Maga- zine. Each reagent looked over 12,000 words in each day's practice. Tests were taken before and after training, in marking out (i) Words in "Outlook" columns containing e-s, i-t, s-p, c-a, e-r. (2) Words on manuscript pages containing a-n, l-o, e-r. (3) Common nouns in "Outlook" columns. (4) Words in "Outlook" columns containing e-s. The manuscript pages were prepared with a script 'type-wheel' * op. cit. FORMAL DISCIPLINE 35 on a Blickensderfer typewriter and were typed in purple ink. They differed from the printed columns in (a) length of line; (b) color, (c) size, (d) form of letters; (e) letter-spacing; (f) line spacing. This provision was made in order that we might determine if marking out words consisted simply of re- actions to visual symbols. If so, then training the e-s function on printed matter would show much less effect upon the functions employed on the manuscript pages than on the printed columns. And if improvement in the trained function consisted to an ap- preciable degree in eye-movements or other habituation to the printed material, there would also be less improvement carried over to the functions employed on the manuscript pages. The results in time and accuracy are given in Table I (Appen- dix A, p. 259) and also in per cents in Table II (p. 260). From Table I are drawn the two curves on Plates I and II (p. 261) which show in absolute amounts the acquired efficiency (in speed) in the e-s function, and the amount efficiency in the other functions was increased thereby, for the two reagents. From Table II are drawn the curves on Plates III and IV (p. 261) which show the per cent of acquired efficiency in the trained e-s function and the per cent of improvement made in the tested abihties of other functions. Plate VII (p. 262) shows the course of efficiency with e-s words during training. Calculating gain by finding the complement of the per cent of time, the two reagents gained in the training series 57% and 31%, in the test series on "Outlook" columns 50% and 20%, on the manuscript pages 30% and 30% ; on nouns 14% and 6%, and on all 42% and 24%. The special practice shows gen- eral effect, therefore, to the extent of 74% and yy%. That the improvement in the final tests is not due to the practice-effect of the first tests may be learned from Table I, in which are shown the results on e-s words with which the first test began and ended; For Gs. this effect was 12% , for Cr., -3%. The accuracy is quite uniform for the tests, although in the training considerable improvement was made. (Table II, Plates V and VI, pp. 260, 262.) 36 JOHN EDGAR COOVER Our results in comparison with those of Thorndike and Wood- worth are given in the following table. They show about the same improvement in training, but more general practice-effect. 75% instead of 49%. Table Comparing Data Per cent Improvement Extent In In Tests of general Rea igents Training On Similar On Dissimilar Practice-ef- Data Data Avg. fect in % Thorndike Ber. 19 12.S 4 10 53 and Wood- Br. 51 21 35 19 2,7 worth's Be. 37.5 16.5 30.3 21 57 Reagents Wh. 44 5 25 10 23 EMT, . 36 24 22> 25 69 Our Gs. 57 50 30 42 74 Reagents Cr. 31 20 30 24 77 (The per cents are reckoned on the time of the first tests; in last column, on the amount of improvement in the training.) The table shows more improvement in tests on the dissimilar data than on the similar data, by 4 reagents in Thorndike and Woodworth's experiment, and by one in ours, (i) Conclusion Our results seem to indicate that habituation to the training data was not a great factor in transference of the practice-effect. Some slight improvement may have been due to habituation of eye-movement and eye-adjustment to the word-hunting process, to facility in pen-manipulation and to the dropping away of gross motor accompaniments of mental effort.^ A little may have been due to learning a method of looking for words ; as, keeping more prominent in mind an image of the least frequent letter and using it as a cue. But the transferred improvement seemed principally due to reducing the recognition of a word as contain- ing given letters to its essential process. Introspections note the ease and automatism of the process in the after-training tests: '^ Lindley (Am. Jr. Psych., 7:49ifif) finds a considerable list and points out that some of them are detrimental. FORMAL DISCIPLINE 37 Cr. remarked, after test on c-a, "seemed almost as easy as e-s, though the words as recognized were not of that familiar cast which some e-s words had acquired." Gs. remarked, after test on e-r, "Seemed to me as if I were marking almost as mechani- cally as I had done the e-s." Gs. remarked, after test on a-n words in manuscript, "marked with great facility of recognition and very mechanically." The process of recognizing words as containing given letters had been relieved of the unnecessary and retarding accompaniments {kinaesthetic , motor, and acoustic images) noticed in the introspections of the first tests and in the early training. These retarding accompaniments consisted in (a) repeating over and over again, in inner speech, the names of one or both of the letters sought, or their sounds^; and in (b) repeating the separate words, or actually reading the text, in order to determine whether the words contained the sounds which were held in mind as auditory images. Even when the method was adopted to recognize by visual images, these were strongly supported by kinaesthetic-auditory imagery; which indicated that the most difficult feature of the exercise lay in recognizing words as con- taining the critical letters. Training reduced this process to an almost automatic visual- motor act, and greatly reduced the time of a given performance. There was no kinaesthetic or auditory tendency in the process, in the final tests, except where the process was felt to be specially difficult (Gs. could scarcely distinguish the a's from the o's in the manuscript, and consequently had a tendency to pronounce the sound of a at times; Cr. had some difficulty in recognizing i-t in the printed columns). This chief factor of improvement in the training is identical with the factor of improvement in the tests. The processes for both reagents, in the different experiments in the first test, were various: (i) reading, (2) kinaesthetic- auditory cue of letter-names, (3) of letter-sounds, (4) visual ' Secor (Am. Jr. Psych., 11:236) found auditory and articulatory factors to be aids, especially with difficult material, but not necessary elements in visual reading. 38 JOHN EDGAR C DOVER image cue of letter forms, (5) imagery of the one letter only as the cue, (6) lines, or words as units for search, (7) mechani- cally searching for projections of letters, etc., sometimes varying within the single experiment, but usually playing the dominant role throughout an experiment. In the final tests the process was almost uniform, for each reagent, throughout the different experiments. That this change was not merely a change in method, is indicated by the fact that it took five days of training (12,000 words per day) to get rid of the retarding accompaniments of the recognition process. (See practice curve in Plate VII, p. 262.) Some special factors were noticed in the training which would not contribute to general effect: certain common words often recurring were reacted to by the word-marking impulse without engaging the word-inspecting process ; and the familiarity of the images of the capital forms of the letters rendered capitalized words easily recognizable. In the first test Gs., in marking out s-p words, took p for the cue and ran his eye along under the line for the projecting stem; and l-o words in the manuscript, above the line for the loop of the /. In the final tests, although p and / were the visual cues, respectively, the words were searched for the complete letters. (This accounts for the smaller gain on s-p and l-o words as shown by the tables and charts, in the Appendix A, pp. 259!?.) Cr. used the same trick with s-p words in the first test, but not in the final, which yielded him also a very small score — about half of the score for the e-r words. The observation that marking out words containing different pairs of letters engages essentially different processes cannot be taken as a criticism of the claim that the main factor of improve- ment is also a factor of transference. Marking e-s words is not a simple but a various process, even with a single reagent. It does not only change as a result of training, but at any one sitting it varies with words of different length and of different distribution of the critical letters. These variations may be so regular as to show in a large amount of marking (by a given reagent), various constant errors. E.g., Gs., in his training on c-s words, omit- FORMAL DISCIPLINE 39 ted a much greater proportion of 6-letter words than any others, and of words in which e and ^ were separated by three or four letters than any others, of words also containing x than of those also containing ch, €, g, or z. Cr., in his training on e-s words, omitted a much greater proportion of 4-letter words than any others ; of words in which e and ^ were separated by four letters than any others ; of words also containing z than also containing ch, €, g, or x. Gs. recognized e-s words principally through kin- aesthetic or sound image of s, x was his great distraction, and half the words he marked in error contained € interpreted as g. Cr, recognized e-s words principally through their visual images, using ^ as a cue, and z was his great distraction. Although the process varies not only with various words and with various reagents, general effect of the special practice is shown quantitatively by the tables and qualitatively by the intro- spections, to be the rule for both similar and dissimilar data. And we take the principal factor of improvement to be essentially general in nature. b. Estimating Weights Two reagents were trained for 14 days on a set of 17 Chicago suggestion blocks ranging from 40-120 grams, all similar except in weight. Each reagent took 100 series, making 1700 judg- ments. (See Tables V and VI for data of the first and last series, pp. 263-4.) Tests were taken before and after training on estimating (a) Ten common objects, averaging 67.5 grams in weight, and all falling within the 40-120 gram field; and (b) Ten common ob- jects, averaging 552.7 grams, and all falling above the 40-120 gram field. (Detailed results are given in Tables VII to X, pp. 264ff ; averages, in Tables III and IV, pp. 260, 263.) All weights were lifted from a cushioned surface through a distance (about six inches) limited by a taught cord, during a given time (one-half second) controlled by a metronome, and were replaced in the same time. Efficiency was measured in per cent — the ratio of the amount of deviations in the final to the first tests. The following table gives the averages for each reagent : 40 JOHN EDGAR COOVER Reagents Training Test Test Series Within Field Above Field 40-120 g. 40-120 g. Gs 74 71-6 56.7 Cr 80 70.8 244 If per cent of improvement is found by taking the comple- ments of the per cent of error, we get the following comparison of averages with the results of Thorndike and Woodworth : Reagents Training Test Test Series Within Field Above Field 40-120 g. Thorndike & Wood- worth's Re- W 49 62 2>Z-2 agents T 40.7 I3 -i-2 Our Gs 26 28.4 43.3 Reagents Cr 20 29.2 -244 From the tables it is noticeable that our results are more uniform for the two reagents than are those of Thorndike and Woodworth's. In each case one reagent shows no improvement in the tests on objects above the field, Cr. making 244% more deviation in his final than in his first test. With this exception our reagents showed in the test series more improvement in every case than was made in the training, and thus show more general practice-effect than was shown in the original experiment. Cr.'s loss of 244% in the test on objects above the field is easily explained by his introspection that his estimates "were mere calculations upon the old reproduced kinaesthetic image of the 1000-gram scale weight" which was accidentally handled just before the first test was taken, and by the fact that his devi- ations in the first test were remarkably small, totaling only 15% of the total weight lifted; (his deviation on objects inside the field in the first test was 26.1% of the total weight lifted; the corresponding figures for Gs. were, inside the field 29.7%, out- side the field 52.8%). (Besides the tables referred to, see also Plates VIII to XII, pp. 266fTf for curves.) FORMAL DISCIPLINE 41 (i) Conclusion The process during the training in Cr.'s case was a building up of a definite idea of the 40-120 gram field, and of acquiring facility in estimating the relative position of any weight within that field ; in the case of Gs., a deepening of the impressions of the 40-gram and 120-gram blocks, which were used as bases of judg- ment. No hard and fast associations between the "heft" of a weight and the idea or expression of its weight in grams could be detected. No definite associations were built up for probably two reasons : (i) Each series consisted of 17 blocks, which number was proba- bly too great to permit identity^; and (2) the "heft" member of the association is exceedingly variable, for besides the influence of the preceding weight, of the preceding estimate, and of the preceding error, it was affected by (i) the height at which the weight is grasped with the fingers, for if it is grasped at the top the block will "swing" and if it is grasped at the bottom it will "topple," in either case being "active" in comparison to the equilibrium of the grasp at the center of gravity, it is likely to be over-estimated; (2) the tightness with which it is grasped, for a loose grasp (a) permits a "pull" on the skin of the finger and thumb, tending to cause over-estimation, and (b) does not permit "coldness" of the weight to be so well sensed, tending to cause under-estimation ; (3) whether or not the reagent has been holding his pen tightly in writing introspections; (4) whether the hand is cold or warm, upon which depends the seeming tem- perature of the block, a cold block appearing heavier than a temperate one; (5) whether the weight is lifted with a jerk, which increases its "heft"; (6) whether it is stopped with a jerk before setting it down, which also increases its "heft"; (7) whether reagent is in "good tone," else all weights seem heavier; (8) and whether one's attention is distracted by counting metro- nome beats by which to gauge the movement. This process of estimating weights is therefore a very complex one, and for that reason not best fitted to throw light upon our * That recognition of stimuli diminishes with increase of members in the series, was shown by Lehmann (Ueber Wiedererkennen. Phil. Stud., 5:138). 42 JOHN EDGAR COOVER problem, too many irrelevant factors are possible in both training and tests, to leave our results unambiguous. Witness the fact that more improvement was made in the test than in the training series. Both of the activities engaged in these two repeated experi- ments were rather complex, and both included motor elements that come in for their share of the general effect of special prac- tice. These experiments had been devised to learn the relation- ship of activities with reference to the product of their application rather than to their kind, — they were word-marking and weight- estimating activities among which we have found general practice effect, and have determined the principal factors to be general in nature. An ideal experiment for determining functional relationship of mental activities, it would seem, should be devised with reference to some well-known and simpler kind of mental activity, with conditions such that identical motor elements do not share re- sponsibility with mental factors for general effect of practice. 2. Experiments on the More Simple Processes The following two experiments^ were devised to meet the re- quirements just mentioned. They employ the same kind activity upon dissimilar stimuli and engage different motor elements. Any transference here of acquired efficiency from a trained to an un- trained mental activity must be explained upon psychical grounds alone. The first experiment was devised to determine the effect of practice upon efficiency in an unpracticed activity; the second, to determine the effect of practice upon facility of improvement in another practiced activity. a. Sensible Discrimination. Four reagents were trained in Sensible Discrimination of the intensities of sound for 17 days during an interval of 57 days. Each reagent took 40 judgments per day's training. " A brief report of these was made by Coover and Angell in the Am. Jr. Psych., 1907, 18:327, under the title of "General Practice Effect of Special Exercise." FORMAL DISCIPLINE 43 Before and after training they were tested in sensible dis- crimination of intensities of brightness, each test consisting of 3 days' tests of 35 judgments each. In the training on sound the stimuli were given with a Wundtian sound-pendulum.^ The method was that of constant changes, Right and Wrong Cases, procedure without knowledge. The variable succeeded the norm in 4 seconds. Some D = o, many D J S and the larger D > S. The number of V > N = V, <, |||, and? (greater, less, like, and undecided). In a couple of series each day introspections were noted down by the reagent after each judgment; in the remaining series noting introspections was reserved until after each series. Reagents sat in marked posi- tions with backs to the apparatus. This training consisted of very careful work, the data being designed for use in another investigation also. Efficiencies at the beginning and the end of the training were calculated in per cents of Right cases in the. first ten and last ten judgments made upon six values of D. Judgments on D^o were not included, and 'like' judgments on other values for D were counted an 'undecided. '''^ Table XIV (p. 270) gives the number of R, W, and U judg- ments, per cent of R judgments made on each variable, and the per cent of R and U judgments made in each series, at the beginning, and Table XV, at the end of the training on sound. In the tests on brightness a Marbe color-mixer^ was used to present the stimuli. Artificial light was used, and the apparatus, including the disc when not exposed, was securely screened in black. Apparatus was also mounted on noiseless bearings. Values for D were chosen after a preliminary series so that there would be some D>S=V>N, and some D>S=V " 67.9 " " " 4.7 « a 72.0 " " ' 9-9 (( 73.0 " " " 1-4 « 65.6 " " " 5-9 « 68.4 " " " 2.3 (( 77.8 " " " 7-9 « 81.6 " " " FORMAL DISCIPLINE 55 by the above data as to the relation between errors and speed: Cr. said, "Made mistake and was bothered thereby," (March 3), "Errors result in confusion and pauses," (March 15). Al. re- marked, "the large number of mistakes impedes rapidity, as one is troubled by them," (April 27). And Bs. noted, "mistakes were noticed and caused confusion at the time," (March 20). For these three reagents, errors reduced the speed by confusing or vexing them; but for CI., by inducing introspection: "I do not care for the making of a mistake in and of itself, but I always pause to think of it and wonder why, and that lessens my attention to the business in hand," (March i) ; "The time I spent thinking of a mistake caused a delay," (Feb. 24). Our errors in reacting are thus causes of decrease rather than of increase in speed. As a matter of fact the errors are not greatly increased as may be seen from the daily averages in Table XXII : for CI., Cr., and Bs,, the increase is less than 2% in the second training in typewriter-reaction; for Al. they increased but 6%, and much of this may be accounted for by the fact that Al. fell more heavily under the unfavorable conditions of the second training, to be noticed later, than did the other reagents. Some of the increase in errors for all reagents is undoubtedly the result of the same unfavorable conditions of the second training, and some of it is due to the inability to inhibit anticipa- tory reactions during the rapid rythm of the later practice, which had been inhibited in the slower rythm of the earlier practice. This latter fact is also supported by evidence from introspection, e.g., when reactions were running 73.5 sec. per 100, CI. said, "Several mistakes inhibited," (March i), and after practice had reduced the time to 63.3 sec. per 100, she said, "Was running quickly; reaction simply came before I could inhibit making mistake of an a for an n," (March 10). The increase of errors may not mean that the efficiency of reaction to letters is less therefore, but that facility of reaction has increased to such a speed that anticipatory reactions previously inhibited are now made before they can be inhibited. And the error once made, the speed is retarded. 56 JOHN EDGAR COOVER In comparing efficiencies in the typewriter-reaction, therefore, we may look upon the time record as a satisfactory measure. For the purpose of learning whether the more unstable asso- ciations between the various letters and their proper reactions of the first training were also those of the second training, the errors of the whole period of training in typewriter-reaction were collected and classified. The weaker associations were the same for both periods. For CI. e and t were interchanged more than any other two letters, and a and t were the second pair in instability; for Al. a and t, and e and t; for Cr. e and t, and a and t; for Bs. e and a, and e and t. The card-sorting did not change the peculiarities in the errors of the respective reagents. There were probably six causes for the errors : ( i ) Lack of coordination between the letter and its proper reaction, (2) an- ticipation of a letter in which the reaction took place before the letter was cognized, (3) false cognition of the letter, (4) reac- tion incited by rhythm without recognition of the letter, (5) misplacement of the fingers on the keys, and (6) raising a fairly automatic process into clearer consciousness and giving the control for the moment to voluntary attention. The 4th and 6th causes enumerated are probably the chief factors in the increase of errors in the second training. (b) Time The gradual improvement made in the card-sorting may be seen in Table XXV (p. 282) and on Plate XVI (p. 284). If the first and last 400 reactions are taken for calculating initial and final efficiencies, the training resulted in the following prac- tice-effect, in decrease of time: CI. 23.5%, Al. 28%, Cr. 22%, Bs. 30%. The decrease in time in the second training in typewriter- reaction is noticeable in both Tables (XXI, p. 275, XXII, p. 276) and Curves (Plates XIII, p. 279, XV, p. 281). But inspection of the Tables XXIII and XXIV (pp. 276-8), and Plates XIV (p. 280) and XV (p. 281) of the control reagents shows that they also improved in speed after their unpracticed interval, which suggests that the improvement of the regular FORMAL DISCIPLINE 57 reagents may not have been the result of the training in card- sorting, and invites an examination into the relative facility of improvement.^^ The curves of the regular reagents, after training in card- sorting, show more improvement in regularity than do those of the control reagents, after the unpracticed interval; they also shov^ the drop in time to be greater (except in the case of Ge., whose curve is quite abnormal). (The line through the curves shows the average of the last day before card-sorting.) To express quantitatively the relative improvement in speed, data were selected from Tables XXI and XXIII and arranged as is found in Table XXVII (p. 283). The average reaction time per 100 was found for the second and third days of training for both regular and control reagents; with this was compared the average time per 100 for the 4th and 5th days of training, which followed the 2d and 3d in continuous training in the case of the regular reagents, but which came after the unpracticed " To indicate the pitfalls for the statistician who is so completely absorbed in the tables of quantitative results as to neglect the processes involved in producing the results, a recent reviewer may be aptly quoted at this point: "Nothing whatever is needed to account for the improvement in typewriting save the special practice in it" (Thorndike: Educational Psychology, vol. II 407) : the average "improvement from the first three days before training to the second three before training is greater than the improvement from th2 three before, to the three after, training" {op. cit., 406) ; as though one might expect considerable improvement in this typewriter-reaction after the practice-effect of 3000 reactions which for half of the trained reagents had produced about maximal efficiency. And it is claimed that the control reagents "show no inferiority" to the trained reagents {ibid. 407) ; whereas, as is shown later, the gain during the interval by the only control reagent whose results are comparable to the others is exceeded by the gains of the trained reagents Al. and Bs. and about equaled by the gains of CI. and Cr. for whom, the averages of the table might have suggested, the reactions had become automatic. Apart from the unwarrantable aggregating of results so disparate in ef- ficiency (or their place on the practice-curve), and a disregard of the analysis in the text, the reviewer's error is minimized by the fact that by chance only the daily averages were accessible to him. The curves (pp. 279!?) plotted by successive 100 reactions show the facts more clearly, and, if examined with the cautions noted in the text, will, in all probability, clear the authors of the charge that they "endeavor to extract evidence" {ibid. 408) of transfer- ence. The facts would then seem to prove the conclusions. S8 JOHN EDGAR COOVER interval in the case of the control reagents. The gain made by the control reagent Mn. (7%) is about the same as that made by the regular reagents (CI. 6%, Al. 0.8 fo, Cr. 3%, Bs. 8%) in continuous practice. (We disregard Ge.'s data — 16% — here as abnormal, for reasons given later, and because they represent series of reactions in which frequent and often long lapses of at- tention played a great part). Mn. shows gain as a result of the interval. But comparison of the average time per 100 reactions for the last two days before training in card-sorting with the average for the first two days after training shows the regular reagents to have gained more than Mn. after the unpracticed in- terval (CI. 6%, Al. 16%, Cr. 5%, Bs. 12%), except in the cases of CI. and Cr., to be noticed later, in which the gain was about as much. A comparison of the results in Table XXVII (p. 283) as well as of the curves is misleading unless several things are borne in mind, the principal of which are ( i ) that the control reagents had trained but three days (Gs. one day) before their unpracticed interval, while the regular reagents trained five days before their training in card-sorting, and for that reason the former would be expected to show more improvement in their second training as a result of practice in that training; and (2) that the control reagents in their training before the unpracticed interval showed comparatively long times from which great improvement in both absolute amount and per cent could be made before much skill would be shown: the last daily average before the interval, for Mn., is as high as the first daily average of two of the regular reagents (CI., Cr.), and for Ge. almost as high as the first daily averages of the slowest regular reagents (AL, Bs.). With these facts in mind the comparison of averages in Table XXVII (p. 283) of the curves on Plates XIII-XV (pp. 2yg^) discloses a noticeable improvement in regularity in the cases of CI. and Cr., and a marked improvement in speed in the cases of Al. and Bs., which is attributable to the training in card-sorting.^"^ "The curves of the second group of control reagents (Bd., Bh., Bs. 2, and Cf.) compare favorably, in time, with those of the regular reagents CI. and Cr., for whom the processes had become automatic, but they contrast mark- FORMAL DISCIPLINE 59 Some explanation might be offered for the fact that the im- provement in speed was not more marked. First, we had sought to determine the effect of training in card-sorting upon the practice-curve of typewriter-reactions. But the typewriter- reaction curve had already attained to records of approximately the maximum speed for both CI. and Cr. CI. is skilled in piano-playing, and Cr. in typewriting. They trained for five days before the point was reached where the influence of another training should be determined. On the second day of practice CI. remarked, "The fingers begin to react at the mere sight of the letters now"; on the third day, "Reacting today is growing more mechanical ;" and on the edlj' with those of AI. and Bs., indicating more certainly than was done be- fore the influence of the card-sorting practice. The particularly significant comparable curves are those of the regular reagents Al. and Bs. and the control reagents Cf. and Mn. Some slight additional evidence in support of the hypothesis of a function- al relationship between the two kinds of reaction we have employed is afforded by an auxiliary experiment in which the influence of practice in typewriter-reaction upon card-sorting was tested. The four reagents, men- tioned above and on pp S2f as the second group of control reagents, served in this experiment as the regular reagents, practiced by 4000 typewriter-re- actions each. Immediately before and on the last day of that practice they sorted at one sitting four packs of 50 playing-cards into four compartments labeled Diamonds, Spades, Clubs, and Hearts : D S C H The intervals between sortings were 58. days for Bd., Bh., and Bs. 2, and 44 days for Cf. Two control reagents, Cn., and Sn., sorted four packs each, before and after intervals of 56 days and 53 days respectively. Both groups of reagents had had extended practice throughout the pre- ceding semester in sorting playing cards by suit according to a different method. They sorted in piles upon a table instead of into compartments, and sorted to the place of the preceding card instead of to the place of the card held. All were therefore nearer practiced condition in sorting cards before the interval than after it, which may account for the loss of the control reagents. One of the regular reagents (Bs. 2) was suffering from a severe cold on the day of her final sorting and shows no improvement; the others seem to have carried over some practice-effect from typewriter-reaction to card-sort- ing, as may be seen from Table XXVIII (p. 286) and Plate XVIII (p. 287). 6o JOHN EDGAR COOVER fifth day, "Reaction to the sight of the letter is so automatic that I can think of other things as I operate. . . . During the reaction attention was not well concentrated at times because I was think- ing of the influence of thinking about your rate of speed upon your speed." Reacting had obviously become automatic. And Cr. on the third day of training remarked, "reacting is becoming very mechanical." Hence in these two cases we have determined the effect of training in card-sorting upon typewriter-reaction after it had already been trained to a high degree of automatiza- tion. The cases of Al. and Bs. are different. Neither is skilled in typewriting, although each had used a typewriter, and Al. remarked, "No application of former practice with the type- writer, — motion a new one," (Feb. 24). They did not speak of the process as being automatic before the card-sorting began, — and they show greater practice-effect. Another reason may be given for the reduction in time not being more marked. The second training on the typewriter was taken by the Regular reagents during April 26-28 while everybody was rushed with work preparing for the closing of the semester, and meeting the many social duties coming at that time, all of which occasioned general fatigue in at least three of the reagents. Cr. took the first day's work in the second training so late that the room was dusk and his atten- tion was almost wholly directed to the perception of the letters : "Had to strain to perceive the letters; this part of the process demanded all the attention; some errors are due to mistaking at first the letter outline" (April 26). Al. was doing heavy work day and night: "Reagent very tired, cannot hold atten- tion; up till I o'clock preceding night busied with exhausting and harassing work" (April 27). And CI. noted, "Have been up late for six out of seven nights past and feel somewhat below normal." But the most marked effect of card-sorting on the typewriter- reaction is shown by introspections. On the first day of the second training CI. remarks, "Sight of letter produced the reac- tion movement without my thinking of my fingers and not at all of the sight of the keys" (April 26). Al. remarked, "No FORMAL DISCIPLINE 6i headache, no nausea, as before card-sorting," "Much easier than at first" ; "General background of f eehng is probably not unlike that of card-sorting, but I did not think of the card-sorting during the trial" (April 26). Cr. said, "Process was sur- prisingly automatic and was accompanied with ease" (April 26) ; "seems more automatic than ever before, and even more so than the card-sorting. I do not pay slightest attention to the fingers or the keyboard when the process is going best ... it appears that the old associations have not only not been interfered with by forming new ones in card-sorting but that they have become firmer and action upon them more ready and automatic than it was before or than it was in card-sorting," (April 27). On the last day of typewriter-training before card-sorting began, Bs. remarked, "Pauses between letters caused by having to think which finger I should put down," and on the first day after card-sorting, "Seemed more natural than I thought it would," and the next day, "Seemed more natural to react today, demands less attention, tendency to become automatic." The introspections of the control reagents, none of whom was familiar with typewriting, show that while some ease and facility were experienced before the interval, reaction afterwards seemed unexpectedly "difficult" and "unhandy:" Mn. stated in the second training that "Reacting seemed difficult," "Seemed to have to stop to think which finger was to react to the dif- ferent letters." Ge. in the training before the interval said, "The reactions are becoming more automatic," and afterward, that he "looked at letters not knowing what to do" and that "combination of letters here was particularly unhandy," The interval without practice resulting in a feeling of discomfort and difficulty, strengthens the point of the preceding paragraph that the training in card-sorting is the cause of the increased ease and facility experienced by the regular reagents in the second training in the typewriter-reaction. (3) Conclusion We conclude, therefore, from the results of this experiment, that training the activity of reaction with discrimination and 62 JOHN EDGAR COOVER choice by sorting cards into compartments has increased the facihty of the same activity in both speed and regularity in typewriter-reaction (a) noticeably in regularity in two cases after the latter had become automatic, and (b) markedly in speed in two others in the course of practice. The cause of transferred facility could not have been identical motor elements. In the typewriter-reaction the eyes rested sharply fixated upon one spot on the screen while the fingers, with coordinating wrist and forearm movements, tapped down the keys, accessory muscles being mainly used; in the card-sort- ing reaction the eyes moved from the cognized colors on the cards about over the compartments of the cabinet or followed the movement of the hand and fixation was nowhere so intense, while the right arm contributed most of the movement in grasp- ing the cards, turning up the under color-surface, and casting them into the compartments, using mainly fundamental muscles. Neither could the cause have been habituation to the stimuli, for they were different in character: one being a small symbol (form) and the other a comparatively large rectangular surface of color. Nor was the cause identical associations between stimuli and reactions. According to the introspections of the regular reagents on their card-sorting training, the process of reaction is variable. At the beginning of training they matched the color of the cards with the labels on the compartments then to increase speed a system of mnemonics is employed designed to form associations in the mind between a compartment and its color; this system then goes through a process of mutation — becoming abbreviated, changed in part, or supplemented, — or is superseded by another; finally, through repetition reactions to particular compartments become coordinated with their respective colors and are made directly, free of the system except in cases of emergency. Synchronously with the growth of these coordinations, adven- titious processes, as pronouncing the color when cognized, move- ment and strain of the whole body, and disadvantageous move- ments of the hand, decrease to a minimum. According to the introspections of both the regular and control FORMAL DISCIPLINE 63 reagents on the typewriter-reaction, the process of reaction goes through precisely the same stages, except that it starts with a system of mnemonics, and systems are superseded more often. No two systems are aHke. But all finally give way to the direct reaction which has been coordinated with its letter. Adven- titious processes, as pronouncing letters upon cognizing them, visualizing keys, fingers, on their order, likewise decrease to a minimum. These two processes so summarily described are necessarily related to their particular stimuli and their appropriate reac- tions, and seem therefore fairly independent of each other. But there is one common factor already apparent: The habit of stripping the essential process of its adventitioits accessories. This is one of the causes of transferred facility, and we shall notice two others. Introspections also state that in any one series several systems of mnemonics may be operative ; there may be some direct reac- tions due to coordination of stimulus and reaction, some due to a kinaesthetic image of recent reaction; that in one part of the series one stimulus causes difficulty and in another part another stimulus; that some stimuli are harder than others throughout the whole series. Improvement here seems to consist in re- solving the reaction process to a single type, except in so far as reactions become direct, and to attend somewhat more closely to the difficult stimuli until their reactions become as ready as those to the other stimuli ; but also, not to confine attention so closely to the troublesome stimuli that their reactions anticipate other stimuli and cause errors thereby. Again, introspections and records show that even after a mnemonic system has been successfully applied and has served to bring stimulus and reaction to a fair degree of coordination, lapses of attention occur during which the "mind is a blank/' and the drum records abnormally long reactions. Improvement here consists in keeping attention upon the matter in hand so constantly that irrelevant stimuli are unnoticed. We find, therefore, the causes of the transference of facility to be (a) the habit of reacting to a stimulus without being delayed by prominent kinaesthetic^ acoustic, and motor accom- 64 JOHN EDGAR COOVER paniments of recognition, through gradually dropping these out of the process, or reducing them to a minimum; (b) an equitable distribution of the attention to the various possible reactions so as to be about equally prepared for all; and (c) the power of concentrating upon the process through a whole series to such a degree as to eliminate distraction. II. PRINCIPALLY QUALITATIVE Experimentation thus far had pretty well established the fact of the general effect of special practice, but there is not yet a general agreement in the interpretation of the results, as to extent and causes of transference of practice-effect; and in almost all investigations there stand out anomalous cases which remain unexplained. Most of the introspective analyses of practice effect show it to be extensively general in both positive and negative influence, but Sleight argues from the paucity of 'significant' values in his tables that it is narrowly specific. Ebert and Meumann found training-effect to be permanent for at least three months ; Sleight interprets his evidence to indicate that "the effects of both 'direct' and 'indirect' practice are usually not permanent."^^ Fracker ^^ Op. cit., p. 451. The effect of practice is found to be more permanent than the opposing factors of interference (Bair: The Practice Curve. Psych. Rev. Mon. 1902. 5:No.2, and Bergstrom : An experimental study of some of the conditions of mental activity. Am. Jr. Psych., 1893-4, 6:272) and fatigue (Kraepelin: Die Arbeitscurve. Phil. Studien., 1902, 19:476; also, A measure of mental capa- city. Pop. Sci. Mo., 1896, 49:760), and although some studies show that the greater part of it is soon lost (Kraepelin: loc. cit.), it may endure for a long time : Kraepelin found that two hours of practice in adding left noticeable traces three months afterward, and that the effect of 17 presentations of associations was perceptible i^ years later (op. cit. S. 476). Ebbinghaus (Grundziige der Psychologie, 2te Auf. Bd. I. S. 633) saved 7% in number of repetitions upon relearning stanzas from Byron's Don Juan after an interval of 22 years ; and 20% after an interval of 17 years. Bourdon (Recherches sur I'habitude. Annee Psych. 1901. 8:327-340) found that skill acquired in marking out letters in a page of print dropped FORMAL DISCIPLINE 65 found the greatest cause of improvement and transference in memory of simple sounds or shades, to be the development and carrying over of representative imagery of a very simple and definite type; while Ebert and Meumann found that in training on non-sense syllables the use of, and dependence upon, repre- sentative imagery of all kinds dropped away. And Fracker is inclined to oppose his results to those of Coover and Angell because the latter pointed out the function of representative imagery in the processes of discrimination, and of reaction with discrimination and choice, to be respectively a distraction, and a temporary but probably a necessary means of effecting auto- matic coordination between stimulus and reaction. Almost all investigators conclude that when transference of practice-effect takes place, it varies in amount with the similarity of either the material or the method of the tests to the practice, and there seems no reason why one should not hope for this law to be finally established. But, so far, either the ways in which practice and training may be similar must be multiplied and further defined or the exceptions to the law appear, upon inspection of the results, to be the rule. A few of the anomalies, for the purpose of illustration, follow : From the results of Thorndike and Woodworth, we note that the average gain in training on estimating areas was about 52%, that tests on the same form within the field of change of area in training yielded an average of 61%; but that above the field, same form, 16% ; while above the field, different form, 51%. Again, one reagent (W) in training on estimating very little during a rest of 236 days ; and that most of it remained at the end of a rest of seven years. Book (Psychology of skill, with special reference to its acquisition in typewriting. Univ. Mont. Bull. 1908. 53 75) found effect of practice in typewriting to persist after a rest of one and a half years. Swift (Relearning a skillful act: An experimental study in neuro-muscu- lar memory. Psych. Bull. 1910. 7:17-19) found skill in tossing balls to be retained 275 days ; and that it could be rapidly regained after an interval of 6 years. Skill that had required 42 days' practice to attain was regained in II days. In the last "try" of his test he exceeded his former skill, in number of catches, in the ratio of 1600:1051. 66 JOHN EDGAR COOVER weights made 49% improvement, while estimating objects within the field of training he made 62%. In Ebert and Meumann's research we found improvement in 'immediate memory' for non-sense syllables (the training material) to be 20%; while for numbers it was 29%, and for letters, 36%; if the grade of memory at "two-thirds correct" is considered, for non-sense syllables it was 49%, but for num- bers 56%. In 'complete learning,' for non-sense syllables it was 61%, for poetry 20%, but for prose it was 43%. One would surmise that the more arbitrary sequence of words in poetry would make it more similar than prose to the training. Retention after 24 hours, of non-sense syllables, was improved 45%, but of optical symbols 49% and of prose 67%. From Fracker's research we found the average gain in train- ing upon memory of sounds 21%, and the average gain upon the test on grays 36%, seven of the eight trained reagents gain- ing more in this test than in the training. Reagents exceeded their training gain in other tests also : on nine tones, 4 reagents ; on nine grays, 4; on four pitches of a tone, 2; on geometrical figures, 3. From the monograph, we learn that F. S., "a grad- uate student in psychology, who was trained in many forms of experiment" (p. 64), trained on sound 8 days, taking four sets of 75 sounds each, per day; he started at 70% reproduced elements, and finished with 40%, while three of the other reagents finished near or above 90%. In spite of his inverted practice curve he shows gains in his tests, however. If his most efficient imagery was auditory, as it seems from intro- spections upon training, and his test on four pitches (p. 82), to have been, how did he by a change of method make his greatest gain (26%) upon the four grays, and how make so little gain upon nine tones (i %), and upon nine numbers (4% ) ? Another reagent (H. C. E.) who developed during training a very definite visual system of imagery involving four posi- tions, made more gain upon the square of geometrical figures (35%) than he did in training (27%), and less upon the four pitches (23%) than upon the figures. From Sleight's research we noted that of the three trained FORMAL DISCIPLINE 67 groups of children, but one group showed improvement in the tests more similar in material to their training and that accord- ing to the two sets of data there are scarcely any correspondences between the 'significant' values of the children and the adults. It is probable that the conflicting nature of results rests largely upon differences in detail of experimental procedure: the kind and length of training, the kind, length, and number of tests, and the manner of scoring and comparing results. The paucity of 'significant' values in Sleight's tables may indicate merely that (a) the training-effect of practice on the trained groups is but slightly greater than the training-effect of other school work upon all the reagents; (b) statistical treat- ment smothers the facts: e.g., with a probable error of 16, and a difference-score of 53, a 'significant' value would necessitate a gain of 30 points in the second test over the first when the average score in the first was but 66.8 (Group 2, on points ),^^ which means that if the gain in the second test is not about half the initial capacity it cannot be considered really significant ; (c) the probable error is unduly large, by reason of widely differing initial efficiencies of individuals (in the test on points initial efficiencies ranged: for School X 27-93, for School Y 32-129, for School Z 23-132).^''^ Where individuals differ so greatly in capacity, we learn in following experiments, they are not doing the same thing, not applying themselves to the same kind of work.^^ The permanence of training-effect found by Ebert and Meu- mann may, as Sleight thinks, be due to the inadequacy of the per cent method of expressing results, or to greater ease of the later tests ; again, it may be due to the rigor of the training, which greatly exceeded that of Sleight's reagents. The usefulness of representative imagery found by Fracker may be largely due to the small number of separate units of ^' Op. cit., Tables, pp. 412-417. ^'' Idem, Table T, pp. 410-411. " Hollingworth (Individual differences before, during, and after practice. Psych. Rev., 1914, 21 :8) upon the basis of low correlation in early practice changing to high correlation in later practice, concluded that in the early trials he was not "measuring the same thing with all performers." 68 JOHN EDGAR C DOVER which his series were made up, and to the simplicity of the stimuli, both of which conditions were foreign to the training material of the reagents of Ebert and Meumann. The anomalous cases may be due to inadequate methods of comparing results or to a radical change in the reagent's processes. If the initial capacities of two reagents are not about equal, and the improvement is made through increasing the number of units of work within a given time, not only will the difference- score of the lower capacity have the advantage of the higher, because of more room for improvement, but expression in per cent of gain will exaggerate this advantage. (When the score is reckoned in decrease of time taken to perform a given amount of work, the expression of the score-differences in the form of per cent, of course, compensates in a measure the advantage of the low initial capacity). That some anomalies are the result of radical changes in the reagent's processes may be illustrated by a few cases re- ported in experiments some pages back (pp. 38ff.). In the word-marking experiment Gs. made his lowest gain on s-p words (15%)^^ when the material was similar to that used in the training, while on material different from the training he averaged 30% — in spite of the fact that the per cent form of expression tended to reduce the inequalities among initial capacities. Upon inspecting the table^*^ we note that his initial capacity was about twice as high as in the other tests. Intro- spections state that his method of marking s-p words in the first test was radically different from his method of marking other words :^^ he took p for the cue and ran his eye along under the line for the projecting stem; he took advantage of the form of the letter to turn from hunting letters to hunting a certain rare but easily recognizable projection. In the second test he hunted for letters. His low gain in marking out l-o words from manuscript pages was due to a similar change in process: ^^Vid., Table I, (Appendix A., p. 259). ^Vid., Table II, Plate I, (Appendix A., pp. 260, 261). " P. 38. FORMAL DISCIPLINE 69 he sought above the hne for the loop of the /, to the first test. Cr.'s low gain on s-p words was owing to the same cause. In the weight-estimating experiment, Cr.'s great loss on ob- jects outside the field was owing to greatly over-estimatmg the weights of two of the ten objects— a bottle of muscilage and a volume of the Psych. Rev.-^ He used in both tests on this series of objects, as a basis of judgment, a kinaesthetic image of a 1000-gram weight accidently handled some days before the first test, and it served more poorly for the second test since it was more vague, — a process very different from his estimation of the other series of objects. Large differences in an individual's results appear to be due to a radical change in processes and for that reason do not seem properly comparable with his other difference-scores — the scores in question do not measure the same kind of work. Whether great variation in initial scores of different reagents make them, for the same reason, non-comparable, remains for analyses of processes to determine; if they are non-comparable, no other statistical device is likely to prove more satisfactory in comparing them than the absolute or the per cent form. Evidently we need to know more about (a) the extent of variability in processes, with a single reagent and between dif- ferent reagents,23 who set themselves to the same objective task; =^ Table X, (Appendix A, p. 265). ^Contributions to individual psychology show us that individuals may vary tremendously in their mental processes, vid. Binet et Jassey: fitude de psychologic sur les auteurs dramatiques. Annee Psych., 1895, 1:60-118. Binet et Henri: La psychologic individuelle. Annee Psych., 1895, 2:411-465. Toulouse : Enquete medico-psychologique sur les rapports de la superiorite intellectuelle avec la nevropathie. 1896. Oehrn: Experimentelle Studien zur Individualpsychologie. Psych. Arbeit- en., 1896, I :92-iSi. Binet: L'fitude experimentale de I'lntelligence. Paris. 1903. Sharp: Individual psychology: A study in psychological method. Am. Jr. Psych., 1899, 10:329-391- Whitley: An empirical study of certain tests for individual diflferences. Archives of Psych., 191 1, 3' No. 19. Also the studies which attack directly the problem of the Diagnosis of mental imagery (like those of Secor, Segal, and Fernald) or which make 70 JOHN EDGAR COOVER (b) the causes of this variability; and (c) its effect upon the scores. We might then be able to determine more precisely how training affects the processes engaged in the training, how training-effect affects the test-capacities, and how these facts may be properly expressed in quantitative terms. To this task of qualitative analysis we set ourselves in the fol- lowing two experiments, and at the risk of tedium we report on both tests and training as fully as our introspective data permit. 1. Experiment on Attention. (Variability in Mental Processes).' In order to give greater opportunity for variability and for afunctional relationship of processes to appear, it seemed desirable to extend more than is usual the variety of tests and training; the variability for the large number of tests and different kinds of training could be worked out in detail and thus furnish a basis for a critique of this type of experiment, while functional relationship, at the risk of not being found at all, might be detected where it is scarcely expected, thus revealing some new causes for general effect of training, both positive and negative. Fortunately there is an excellent opportunity for organizing a number of variously related tests into this standard type of experiment. In many of the discussions psychologists have at- tributed the general effect of special practice, in part, to training in attention. ^^ In the want, as yet, of any direct method of an analytical inspection of the memory process (like those of Kuhlmann, Gamble, Cohn, Sybel, Wreschner, and Miiller), as well as the early work in which the concept of "imagery types" was originated and developed (Fechner, Galton, and Charcot). " Performed during the year 1910-1911. ^ Vid. Angell, J. R. : The doctrine of formal discipline in the light of the principles of general psychology. Ed. Rev., 1908, 36 :8. Aall : in review of Coover and Angell. Zeit. f. Psych., 1908, 48:303. Mijller : Zur Analyse der Gedachtnistatigkeit u.d. Vorstellungsverlaufes. Zeits. f. Psych., 1911. Erg. Bd. S. 244. Ebert und Meumann {op. cit. 205), Fracher {op. cit. 95), Sleight {op. cit. 442-3). Stumpf (Tonpsychologie, 1:81). FORMAL DISCIPLINE 7^ measuring attention,^^ lies our opportunity. If attention can be measured at all,^^ and, in view of the fact that it is often practically gauged by casual observation, it seems reasonable to hope that it can, it must be measured by the product of such mental activities as are known to depend in the highest degree upon the attentive state. Such activities are sensitivity, sensible discrimination, reproduction, and voluntary activity ;^^ and a large number of fairly standard tests are at hand which may be used to measure them. If we get the initial capacity of a number of reagents in a number of these tests, we may be said to have obtained indirectly a cross-section of their initial capacity in attention. Then, if special training is given to some of them, and the tests are given to all of them again for a final measure of attention, our experiment will conform to the type usually employed in our field of investigation. Although we are for the moment assuming that attention is a simple and uniform state of consciousness, we may, perhaps, have to recognize (a) that there are phases or 'moments' of at- tention^^ such as degree of concentration, quickness of adapta- tion, and duration of concentration; and (b) that individuals might possess (i) dififerent types of attention,^^ such as broad or narrow spanned, alert or sluggish, visual or auditory or motor, or • (2) different types of consciousness,^*' such as dual or multi-level. We may then, provisionally, call this an experirnent on atten- tion; our chief aim being to learn the extent and causes of variability in processes engaged in tests and training of widely different kinds, incidentally, to trace through introspective ^^In 1893 Kiilpe wrote: "The discovery of a reliable measure of the at- tention would appear to be one of the most important problems that await solution by the experimental psychology of the future." (Outlines of Psy- chology. 19CI. Sec. 73.2). And in 1908 Titchener said: "The discovery has not yet been made" (Lectures on the Elementary psychology of feeling and attention. 1908, p. 276) which still remains true. -"For present methods of measuring attention see Titchener (ibid. 276ff), and Pillsbury (Attention, ch. VI.). =" Vid. Kiilpe : op cit. Sec. 72,. ^* Chase: Some aspects of the attention problem. Fed. Sem., 1909, 16:281. ^McComas: Some types of attention. Fsych. Rev. Mon., 191 1, No. 55 '-SS- ^"Geissler: The measurement of attention. Am. Jr. Psych., 1909, 20:473fi. 72 JOHN EDGAR COOVER analysis the functional relationship between sets of processes varying more or less from the training in kind, when applied to tests varying more or less from the training in material; and in case we get sufficient data which may be justly compared, we may be able to determine whether improved attention is a cause of general effect of special practice. a. The Tests The following tests^^ were chosen because of their dependence upon attention : I. Reaction 1. Simple sensory to sound (50) i 2. Compound a. With discrimination ( 1 ) Marking out small a's ( 100) 2 (2) Marking out o's ( 100) 3 b. With discrimination and choice ( 1 ) Card-sorting (200) 4 (2) Typewriter-reaction (200) 5 (3) Controlled reaction (50) 6 II. Sensible discrimination of sounds (90) 7 III. Reproduction I. Unequivocal (Rote memory) a. Successive presentation (i) Memory of sound intensities (50) 8 (2) Memory of consonants (50) 9 (3) Memory of Arabic numerals (50) 10 (4) Memory of visual signs (10) 11 (5) Memory of associated pairs (50) 12 b. Simultaneous presentation (i) Learning 12-letter-rectangles (a) Free (10) 13 (b) With distraction (10) 14 2. Equivocal — ^Word-completion (10) 15 3. Free — 2-minute trains of ideas (3) 16 IV. Extensive threshold of visual attention 1. Free (15) 17 2. With distraction (10) 18 V. Maximum voluntary activity — tapping (5 30") 19 (The figures in parenthesis indicate the number of reactions, memory units, or experiments, in the test.) ^ How these tests are related, in method or material, both to laboratory work and to standard mental tests, is indicated by some representative bibli- ography given with the discussion of Test Results (pp. io6ff.). FORMAL DISCIPLINE 73 These tests were taken by lo reagents, 8 of whom took train- ing between the first and final series which were separated by an interval of 55 days. The first series of tests occupied 12 days during a period of 36 days; the second, or final, 10 days during a period of 21 days. Each pair of tests was separated by an interval of about 66 days. The pairs of tests were also taken by two reagents of a group of 21 control reagents. There were thus two sets of control reagents: The two who took all the tests, and the 21 each df whom took only one pair or a few pairs of tests. The conditions of the final tests were as nearly identical with those of the first as circumstances permitted. The material was also identical with the exception of the Tests 4 and 5, in which the orders of the symbols were changed to avoid learn- ing sequences of reactions. Of the reagents who took training, 7 had training in advanced psychological laboratory work; i had no training. Of the first group of control reagents (2), i had training. Of the second group (21), one was taking elementary laboratory work; the others were in a class of general psychology. The students of this second group represented nine departments in the university ; there were two graduates, eight seniors, five juniors, and six sophomores. On the whole they ranked younger than the other two groups, b. The Training During the 55-day interval between the tests, two reagents (Mn., Le.) took training 18 days on Test 17; 25 12-letter-rect- angles were presented daily, making in all 450 experiments each. Two reagents (Rt. and SI.) took training 18 days on Test 13; 20 i2-letter-rectangles were presented daily, aggregating 360 experiments each. One reagent (Ly.) took training in simple reaction to sound for 11 days, iioo reactions in all. (Le., who took training on test 17, also took training in this simple reaction to the extent of about 500 reactions). Two reagents (He., Cr.) took training on memory schemes for about 14 days. 74 JOHN EDGAR COOVER And one reagent (Al.) took training on Test 17 for 8 days, almost consecutive, to the extent of 200 experiments. In the training as in the tests introspections were written or dictated. c. Apparatus and Procedure (i) Reaction to Sound For simple reaction to sound, the Morse key (set at a tension of 100 grams, and an amplitude at the button of 3 mm.) was released by raising the index finger of the right hand. The reagent was instructed to lay the arm and hand on the table forming a hand-rest by folding in the fingers and extending the thumb and to give a stimulus-direction to attention. No ap- paratus except the key was visible to the reagent, the Wundtian sound-hammer^^ for presenting the stimulus being located a meter to the right and back of his head. The whirr of the chronoscope could be heard from the adjoining room and acted as a second signal. A "ready" signal was called at irregular intervals from two to five seconds before the stimulus was given. After reaction the reagent noted introspections and called "ready" for the next experiment; he was practiced before the experiment and the first three reaction-times of the first daily series, and the first reaction-time of the other series, were discarded. The Hipp chronoscope was tested daily with a fall-hammer^^ and its MV. found to be less than 1.5 sigma. Twenty valid reactions were taken on the first day; 30 on the second. For each series of 10 reactions were found the (a) arithme- tical mean, (b) probable error, (c) standard deviation, (d) PE of the standard deviation, and (e) coefficient of variation,^^ be- sides the (f) MV. and the (g) relative variation.^^ These were averaged for the five series of the test. Inspection showed, however, that owing to the increased sensitiveness of those ^Illustrated in Wundt : Grundziige der Psysiol. Psych (5 te Auf.), 111:503. ^ Illustrated, ibid., S. 397. ^Vid. Titchener: Experimental Psych., Vol. II, Pt. II, p. 361. '^Vid. ibid. Pt. I, p. 181. r.v.=MV.xioo/M. FORMAL DISCIPLINE 75 measures which rest upon the square of the deviations from the mean, over the MV. and the r.v., for occasional large deviations (yet not sufficiently large to discard), ^^ the former possessed no advantage over the latter which alone were consequently retained. (2) Marking OtU a's The reagent was instructed to mark out all the small a's be- tween the pen marks across a printed page. The material consisted of page 12 of Horace Davis' Phi Beta Kappa address on Vocational Training, delivered at Stanford University, May 8, 1909. Between lines 2 and 30, inclusive, there are 100 small a's. The page was placed before the reagent right side up, covered with a screen which was removed at the moment the stop-watch was started for timing the test. Introspections were written after the test. The Efficiency Index was computed according to Whipple's formula. ^'^ Since no letters were wrongly crossed out, the Efficiency Index became the "time per 100 reactions." (3) Marking Out o's Page 13 of the same material as above was used for marking out small o's, of which there are 100 between lines i and 30 in- clusive. The page was placed before the reagent inverted. Further procedure as above. Tests 2. and 3. were taken in a single sitting. (4) Card-Sorting A Jastrow^* cabinet of six compartments, and four packs of 50 cards (6 x8.5 cm.) were used for card-sorting. The cards were not glazed but were smooth and fairly stiff. In a central position at one end they bore a printed device (14 mm. in ■^ Times with deviations greater than 4xMV. were discarded. "Whipple's Manual (ist ed.). 260!. E=T/A, A=c-w/c+o, where T=time, A=accuracy, c=letters crossed out, w=letters wrongly crossed, o=a's omitted. ^Jastrow: A sorting apparatus for the study of reaction times. Psych. Rev., 1898, 5 :279ff. yd JOHN EDGAR COOVER diameter), ^^ chosen with the design of prohibiting verbal classi- fication, which was turned thru o°, 90°, 135°, 180°, 270°, and 315°, to make the six classes of cards to be discriminated. The labels on the compartments were so arranged as to avoid easy classification. After reading typewritten instructions^*^ the reagent took a standing position at the cabinet, informed the ex- perimenter that he was ready, and started with the experimenter's signal "go." The time of each pack was taken with a stop-watch. Introspections were written after each pack. Two packs were sorted on one day, the remaining two on another. Efficiency consisted of the average time for the four packs in each test. The errors were very few and were fairly constant in each reagent's work. ( 5 ) Typewriter-Reaction Reactions were made with the first two fingers of the two hands to the four letters a, t, e, n, as they appeared automatically through a screen on the (Blickensderfer) typewriter. Reaction to one letter brought the next into view. The four series were com- posed of 50 letters each, so arranged that each letter preceded and succeeded itself and each other about equally often.^^ Opportunity was given for writing introspections after the 2d and 4th series. The typewriter was connected in circuit with a marker which recorded the reactions, upon a smoked drum, beside a synchronous seconds-record furnished electrically by the laboratory clock. Efficiency was calculated from the kymograph records in seconds per series of 50 reactions. The average of the four records constituted the score of the test. Accuracy was checked by the record of reactions made by the typewriter, and since it was found to be uniformly high, it was not used to modify the time-score.^^ The four series of reactions were taken at a single sitting. ^* See Appendix B, Figs, i and 2 (p. 288) for reproduction of device and arrangement of compartments in cabinet. ** Appendix B, Fig. 3 (p. 288). "Appendix B, Fig. 4 (p. 289). "No reagent made more than 3 errors; the general average was 1.6 per series. FORMAL DISCIPLINE 77 (6) Controlled Reaction Reaction was made with the first two fingers of the left hand the first three of the right hand on the typewriter. The respective fingers represented, from left to right, Poets, Phil- osophers, Statesmen, Scientists, and Musicians. Ten familiar names of each class^^ ^^j-e arranged into lo favorable series^* and were pronounced singly ^^ Time was taken with a stop- watch which was selected from a collection of lo for its favor- able action. A day or two before the experiment the complete list of names was read by classes to the reagent to determine whether any were unfamiliar, and, if so, to learn them; and before the experiment he was drilled upon the class-finger coordinations until he was sure that he knew them. Introspec- tions were written after the fifth and the last series. After the test, free reactions were taken to one-, two-, and three-syllabled names, in which all the five fingers were used, not in regular order, to determine how much variation in time might be owing to length of name.'*^" The test was taken in a single sitting. Efficiency was calculated in terms of time alone; the mean of the 50 reaction-times, and the mean for each class. (7) Sound Discrimination Nine intervals of intensity^'' (including D:=o) were given with a Wundtian sound-pendulum.^^ The upper and lower intervals could almost always be clearly cognized. The time interval between norm and variable was about three seconds, between pairs about 12 seconds, and between series of nine *• Appendix B, Fig. s (p. 289). ** Appendix B, Fig. 6 (p. 289). *" With one exception : Cr. reacted as rapidly as possible to each half of the 10 series, which was presented simultaneously to his vision. **The general averages for all reagents were: .69, .73, .81; making a difference of 0.12" between the shortest and longest names. Since the long names were distributed throughout all series, no reference to this is made in the discussion of results. *" Appendix B, Fig. 7a (p. 290). ^'Illustrated in Wundt: Grundzuge d. Physiol. Psych., I:5ii. Ours is the single pendulum type. 78 JOHN EDGAR COOVER judgments 2.5 minutes. Ten series^^ constituted the test. Procedure was without knowledge and by the method of Right and Wrong cases. The reagent sat facing dark screens, with his back to the sound-pendulum which was about 5 meters distant; he recorded his judgment upon the intensity of the second sound (always the variable) by the use of symbols^^ signifying "greater," "less," "like," and "doubtful." Intro- spections were reached between series and after the test all of which occupied the hour. Efficiency was calculated in terms of Right judgments and the 'Difference Limen.' (8) Memory of Sounds Five series^ ^ of 10 sounds each were constructed with four easily distinguishable sounds (relative intensities: 0.12, 0.8, 2.1, 5.)^^ produced on the sound pendulum. Members of a series succeeded each other at the rate of one second; an interval of 2.5 minutes was given between series. The reagent sat 5 meters from the sound-pendulum with his back toward it, and recorded the sounds in terms of i, 2, 3, 4, in the order of their intensity. He was familiar with the sounds of the instrument since this test followed the test on Sensible discrimination, and in a pre- liminary practice he was drilled until he could name instantly any of the sounds produced in irregular order. The reagents recorded in ruled forms ; and wrote introspections between series and after the test.^^ Efficiency was calculated with Spearman's "Footrule for scoring the memory test. "53 ** Appendix B, Fig. 7b, (p. 290). ^'Appendix B, Fig. 8, (p. 290). ™ Appendix B, Fig. 9, (p. 290). °^ From Fechner: Psychophysik., I. S. 181. "^ This holds true for all the succeeding tests on memory. ''Whipple's Manual (ist ed.), p. 367. d R = which gives a perfect (n=-i)/3 score of i ; the score of a correctly placed letter in a series of 10 = 0.1. FORMAL DISCIPLINE 79 (9) Memory of Consonants The consonants were printed by hand in capitals (12-14 "im. in height, strokes 5 mm. in breadth) and were clearly seen from the station of the reagent 5m. distant. The light entered the room through northern windows behind and above the reagents, and the Jastrow tachistoscope^^ was surrounded by black screens which concealed the experimenter and his manipulation of the apparatus. The test was composed of 5 series of 10 letters,^^ The letters were presented at the rate of one second, the series at the rate of two minutes. Spearman's "Foot-rule" was used for scoring. (10) Memory of Numerals The procedure was the same as above except that digits^^ were presented instead of consonants. (11) Memory of Visual Signs Procedure was the same as in No. 9 except that meaningless optical signs^'^ (suggested by and similar to those illustrated by Stratton)^^ were used instead of consonants, and the test consisted of a single series. The reagent reproduced by drawing. (12) Memory of Associated Pairs The letters of No. 9 and the digits of No. 10 were presented in pairs^^ at the rate of a pair a second, with the Jastrow tachistoscope ; after an interval of 60 seconds the series of letters was shown at the rate of a letter in three seconds during which time the reagent was required to record the digit asso- ciated with the letter. The five series were separated by intervals of 2.5 minutes. The reagents were especially instructed to depend upon association alone for reproduction. The scoring was made on the basis of o.i points for each correctly recorded digit. Tests No. 8 to No. 12 were usually given in two sittings, "Illustrated in Whipple's Manual (ist ed.), p. 365. ''^Appendix B, Fig. loa (p. 290). * °* Appendix B, Fig. lob (p. 290). "Appendix B, Fig. 11 (p. 291). °' Stratton : Experimental psychology and its bearing upon culture, p. 29. ■^ Appendix B, Fig. loc (p. 290). 8o JOHN EDGAR COOVER (13-14) Learning 12-Consonant-Rectangles Through a tachistoscope of the Wundtian type^° rectangular cards (10.2 x 15.3 cm.) bearing three horizontal rows of 4 con- sonants were exposed. The letters were printed in black ink with rubber type (21 mm. high), broad Roman rather than Gothic in style^^ (lightest strokes were 1.5 mm.; broadest 4mm. in width). The disposition of the reagents and apparatus, the condition of the light, etc., were the same as in the preceding memory tests. The letters were clear and distinct from the reagent's position. The time of the exposure was 10 seconds. It was preceded by a pre-signal, "Get ready for No. — ," given 7 seconds, and a signal "ready" given 2 seconds before the screen was removed; it was succeeded by a free interval of 10 seconds which was terminated by the signal "Now," when the reagents recorded in ruled forms. The test consisted of 10 experiments, which came at the rate of 2.5 minutes. Intro- spections were written between experiments and after the test. In the test with distraction, the reagents were required to add eight digits called out by the experimenter at the rate of one a second, during the 10'' interval after the exposure, and to record the sum before beginning to record the letters. Records were scored according to the following values for a reproduced letter : 2 points for the correct line, i point for the correct column, making 3 points for a correctly placed letter. ^^ Tests No. 13, and 14 were given in a single sitting. (15) Word-Completion The reagent was furnished ruled paper, and a blotter to use as a screen. Then three separate columns of 10 consonants^^ *" Illustrated in Wundt : Grundziige der physiologischen Psychologic (ste Auf.) 1902, Bd. III., S. 334. ®^ Appendix B, Fig. 12, (p. 291). *^ Although this method is not satisfactory, and Cohn {op. cit. S. i6iff.) and Segal (op. cit. S. 136) justly claim that errors demand separate treatment, if a single score is to be derived for a measure of reproduction, some arbitrary method must be accepted and this one combines favorable features in the methods of Winch (Br. Jr. Psych., 1:129) and Smith (Mind, N.S. 5:52). See Appendix C, (p. 295). "Appendix B, Fig. 13, (p. 291). FORMAL DISCIPLINE 8i each were dictated to him which he recorded and covered with his screen, leaving a space of about 4 cm. between the columns. He was then told that under his screen were ten lines of three letters each, and that upon signal he was required to remove his screen and to fill in letters anywhere in order to complete a word on each line. Time was taken with a stop-watch. In case the ten words were not completed within 5 minutes, the unfinished work was taken. (16) Trains of Ideas Reagents were instructed to begin immediately from the stimulus-word and write by word or phrase as many ideas as possible within the two minutes allowed them. Three stimulus words were given: horse, potato, flute. Connections were then explained and the ideas counted. Tests No. 15, and 16 were usually given in the same sitting with No. 2 and 3. (17-18) Extensive Threshold of Visual Attention The same apparatus and like material and procedure as in No. 13 were used except the exposure was about o.i seconds (0.085''), the free interval between exposure and reproduction was 5 seconds, and the rate of experiments was two minutes. The whole card fell within the angle of acute vision. In the test with distraction the 5'' interval after the presenta- tion of the card was used by the reagent for adding 4 digits pronounced at the rate of one a second. The sum was recorded before the letters. Tests 17 and 18 were taken within the hour. (19) Tapping The reagent tapped five series of 30" each, with intervening rests of 2.5 minutes. He used his right hand in a manner chosen after a preliminary practice; he was instructed to keep his method constant and to tap as rapidly as possible. The Morse key was sent at a tension of 50 grams, and an amplitude (at the button) of i mm. (approximately the adjustment pre- ferred by telegraph operators). Records of the tapping were 82 JOHN EDGAR COOVER taken on a kymograph beside a seconds-line furnished by the laboratory clock. The kymograph records were scored from the time-line and the taps counted for every interval of 5 seconds. Total efficiency was found by the average of the taps per series of 30 seconds; the "Fatigue Index" was derived by dividing the average of the last five intervals of the first series by the num- ber of taps in the first interval. All procedure was according to Wells^^ except in the use of the right hand only. d. The Training Results According to the design of our investigation our chief con- cern is with analyses of processes, their changes during training, and the factors responsible for improvement. The amount of improvement in training, the probability of the attainment of maximal efficiency, what general conditions correlate with poor and good daily averages, and what special causes of variability affected the scores, are also noticed. (i) Extensive Threshold of Visual Attention These tachistoscopic practices consisted of 25 i/io-second exposures per day of cards (4"x6'') bearing 12 capital con- sonants (21 mm. high) (See Appendix B. Fig. 12, p. 291) printed with rubber stamps evenly in three rows. Reagents sat 5 m. distant ; experimenter was concealed behind the apparatus which was screened in black. The record of the reagent was made in a ruled form and was begun five seconds after the exposure was made; in scoring, a letter was evaluated three points if correctly placed, two points if misplaced in the line, one point if it appeared anywhere else. Experiments took place at the rate of i 1/2 minutes. Introspections followed each experiment. Three reagents, Mn., Le., and AL, took the training, the first two upon three days of the week, the latter upon successive days, at the same hour of the day, and under as nearly constant " Wells : Normal performance in the tapping test. Am. Jr. Psych., 1908, 19 :437ff. FORMAL DISCIPLINE 83 conditions as possible. The extent of the training was re- spectively 18, 18, and 8, days; and the gain over the first day's average was respectively, 29%, 26%, and 29%. The practice-curves indicate that maximal efficiency had probably not been reached {vid. Appendix B. Fig. 14, p. 291). The day's work was not long enough to produce fatigue. Days of poor scores were days of poor control of attention, of discouragement, of thinking that maximum efficiency had been attained, of being bored by the experiment. Days of good scores were days of good attention, and the best days usually came after a week-end or a short vacation. The process varies considerably, even during a sitting, and, although many of the factors of variability will be found in the following descriptions of changes in processes during train- ing, some of them may be noticed here. The degree of attention may vary from 'slack' to "wide-eyed and breathless" (Al.). The presentation may occur just before, at, or just after, the crest of the rhythm of attention. The ex- tent of attention may be confined to a few letters, when they will be clear and the rest of the card may not be seen at all (Mn. 2 : 24) ; or it may cover the whole card in which case all imagery may be so vague as to effect a blank score, (Mn. 2: 13). In the intermediate cases the attitude toward the more vague content determines whether the interval after per- ception is given to rote repetition of the 'clear' letters or is given in part to 'maturing' vague letters. Variability arises in the effort to coordinate the latter two processes. The material for any reagent is not uniform : Some letters, such as form familiar initials or abbreviations, are 'significant' and may challenge apperceptive elaboration at the sacrifice of further perception, (Mn. 6: 11) ; or they may prove more facile in apprehension and more ready for recall, (Le. 15 : 16). Some letters are more difficult to name (Le. 2:16, 11:8) and some groups are particularly non-euphonious (Mn. 2,:y). Among other subtle causes of variation may be mentioned an inducted effect through rapport between the experimenter and reagent, of the experimenter's variable state of alertness. 84 JOHN EDGAR COOVER During Mn.'s nth and Le.'s 12th day the experimenter inten- tionally varied the manner in which he gave the two signals ("Ready for No, — ," given seven seconds before, and "Ready," given two seconds before each exposure) according to three types, (a) sharp, alert, (b) indifferent, (c) low, decisive. The averages of both reagents were lowest for the 'indifferent' type; Mn. averaged highest for the 'sharp,' and Le. for the 'low, decisive' type. The experimenter chose the last type and was thereafter careful to keep his own condition of attention more constant. Since introspections, under the limitation of time, could not be complete, and the respective reagents differed in their selec- tion of factors, the manner in which processes changed through the training will be noted for each of the three reagents separately. Mn., during the habituation period of the first few days, changed from visual to a kinaesthetic retention because of the higher reproductive value of the latter. Only letters repeated by name were readily reproducible, and the period of exposure or the entertainment of the after-image was so brief as to limit naming to but a few letters: "I seemed to have the time in this case to repeat the (4) letters. I can remember the letters more readily by this process, but usually the time does not seem long enough for this" (ist day, 3d experiment). At the beginning, when scores were relatively low, attention was usually confined to the first line, or to a part of it, often leaving all the rest of the card unnoticed: ("All the (4) letters were clear, but I didn't seem to see the rest of the card," (2 :24) ). At this time a whole line produced a good score, but when the score was to become larger, the extent of the attention had to be increased, and consequently its direction changed. This change was facilitated by occasional good scores from relatively indistinct impressions resulting from chance failure to direct attention (i: 10); but a too diffuse attention was checked by an occasional blank score due to the imagery being too vague : "Attention on the whole card — can't recall a letter," (2:13). Attention by the 7th day came to be directed upon some other FORMAL DISCIPLINE 8s than the top Hne (7:3), and by the 12th day, since it was found that scattered letters appeared most frequently above those in clearest vision, the lower part of the card was chosen, the 3d line being favored : "Find I cannot do well at the middle line," (12: 16). But these conscious changes in method corre- spond to 'rests' in the practice-curve, and were not for any length of time strictly adhered to. Not only did the spread of attention now include letters that at first could not be seen, but visual imagery was required to reproduce the additional letters, since the interval was taken up by the verbal repetition (13:7,15:3). This coordination of the two kinds of imagery holding their respective content was not at first very good and never did become efficient; the letters held in the visual imagery were prone to escape before they could be recorded (3,7:17,15:5) although they some- times returned (5:14). Associations of letters with familiar abbreviations, initials, names, words, etc. were not much used. Introspection notes no more than one case (2 or 3 letters) on each of the following days: 2, 6, 12, 13, 14, 15, 16. But the last two days they promised to become prominent; on the 17th 11 letters were reproduced from them, and on the i8th 15 letters. During these days they merely supplanted verbal repetition of letter- names and did not contribute to the larger scores. But had training continued, they probably would have been coordinated with the verbal and visual imagery to the end of fixing more of the impression before it got away. The attitude toward the vague impressions changed and un- doubtedly contributed toward the rise in the curve. It was not until the second day that the five-second free interval before recording was used in part for developing or defining impres- sions; it was customarily used in merely repeating the names of the letters that had been clearly seen. On the 3d day vague imagery began to yield letters : "The first two letters were very distinct, but I do not know how I saw the last ones: When I was recording them, I felt them rather than had an image," (3:1); "During the interval I tried to recall the letters; and 86 JOHN EDGAR COOVER those recorded are the letters that came, although I have no visual image to prove that they are correct" (3:6). But on the 5th day uncertified material was accepted with caution: "J (a clear letter) always seems associated with L for some reason, and the L forced itself [properly] into the record" (5: 20). By the 9th day, however, this material was regularly given a chance, which was almost always warranted by the score. The following will show the character of the material: 7 : 8 "The impression of the three letters seemed very indis- tinct, but during the interval they seemed to take a definite form." 8:6 "P was particularly clear ; I do not know how I got S for I cannot remember seeing it especially." 9:18"! remember only seeing V, but during the interval the rest of the line (3 letters) was completed." 10:6 "When I repeated the letters, K and Z were the only decided ones ; but during the interval the others seemed to fall into place (2 letters)." 11:4 "L and T were the only letters that were distinct. The others seemed to fall into place during the interval." 12:8 "I do not think I saw the last two letters, but in some way they came to my mind during the interval while I was repeating the others." 14: 10 'T did not see the D as the third letter, but something put it there during the interval." 15:3 'T am not sure what the third letter was. When I repeated them I could get no sound in the third space; but visually it seemed to be F, although it does not seem to fit when I repeat them." (The first, second and fourth letters were clear. The third is often omitted by this reagent, and indeed also by others.) 16:5 "I had no idea of the third letter. K came when I repeated them, so I recorded it." (Correct). 16: II "C came as a feeling; seemed to see a rounded form." (Correct). 17:8 'T got a vague idea of a number of letters, but R was the only clear one:" 4 others were correctly recalled and a P recorded in error for a B. FORMAL DISCIPLINE 87 That this material was related to vague visual impression was indicated by an occasional error: "C and N came during the interval; N was especially insistent," (14:24). (Wrong — the two letters were G and V; similar forms). There is a fairly good correspondence between a favorable attitude toward the indefinite impressions and the rise in the curve. Improvement was largely due to the development of vague unrecognized material into correctly recorded letters.^^ While this development was in progress there were some ten- dencies working against good scores : Lines distinctly seen some- times failed to mature into letters, (8:25), and sometimes letters plainly seen failed to be named and could not be recalled (7:17)- Certain peculiarities of the individual consciousness or of the process had some effect upon the scores: Among them are: (a) the unfavorable effect of favorite or significant letters (M, X, C, Q,) (6: II, 7: 9, 10: 20, 4:6), through narrowing conscious- *°The "maturing" of a latent or subliminal impression, in tachistoscopic experimentation, has been observed by others : The distinction between the "recalled" and the "means" of recalling, insisted upon by Cohn (Beitrage zur Kenntniss der individuellen Verschiedenheiten des Gedachtnisses. Dritter Internationaller Kongress fiir Psychologie, in Miinchen, 1896:456-458), involves a maturing of imagery; he says that the retained letter may be a visual image while the means to its recollection is an involun- tary image in an abstract connection (p. 457), and that when it is schematic, or incomplete, it often "matures" (erganzt sich) through reflection. Hylan (The distribution of attention. Psych. Rev. 1903, 10:398) found that "a special efifort to recognize an indistinct character would frequently cause it to mature into complete recognition before others w^hich were at first more distinct," and that sometimes the indistinct letter would "come floating into the mind as an afterthought when all had been given that could at first be re- membered." The impression of letters exposed serially (rapidly) could be held an appreciable length of time without recognizing a single letter, until each was recognized one at a time. Bergstrom (Effect of changes in the time variables in memorizing, together with some discussion of the technique of memory experimentation. Am. Jr. Psych. 1907, 18:236) found that the rapid displacement of one impression by the next interferes greatly with the memorizing, and inferred that an uncon- scious organizing process (apperceiving or fixing) continues some time after the impression has been received, which is necessary for its permanence and revivability. 88 JOHN EDGAR COOVER ness to them alone; (b) the persistence of impressions from former experiments; (c) the difficulty of cognizing letters of similar form, or of naming and of reproducing non-euphonious groups of letters, (3:7); and (d) the shifting of letters to fill a space occupied by an unseen intermediate (usually the third) letter. The last was largely corrected during the training through increased power to develop letters from obscure imagery. The facts that the coordination of the two kinds of imagery (visual and kinaesthetic) was not yet perfected; that the co- ordination of associations with the imagery was just on the point of beginning ; and that the coordination of the reproducing and the recording processes was still defective, supports the objective indication of the practice-curve that training had not yet reached maximum efficiency. Le. changed direction of attention from the first line, when increasing score demanded more than four letters, to the middle line (7th day), and then to the blank space just above the middle line (13th day). These changes were made on the basis of the adequacy of chance variations from the usual method. Extent of attention changed from the first two letters of the first line to four; then this tendency to limit the range to clear impres- sions changed to include impressions of various lower grades of clearness. Intensity of attention varied from day to day and from experiment to experiment, but effort was made to keep it at maximum: "Find that fair (as distinguished from good) attention makes considerable difference; I see the card as well, but the letters are less clear, and fewer are noted"; and good scores are correlated with introspective report of good attention. Although extent and intensity of attention, or distribution and clearness, are supposed to vary inversely, there was during the training a development of the field of maximal clearness as well as the lower levels of clearness ; and there was a gradual transition from a one-level to at least a clearly eight-level or multi-level clearness of tachistoscopic impressions; the table below, which is made from introspective data, will illustrate: FORMAL DISCIPLINE 89 Table showing eight-level clearness The figures in the table represent the number of letters re- produced from a single exposure. The figures in the first line stand for the following headings : I. Day and Experiment 4. Fair 7. Strongly suggested A little vague 8. Suggested 9. Unrecognized content 2. Very clear 5- A 1 3. Clear 6. Vaj I 2 3 4 1:3 2 1:5 I 2 3:12 2 I 14:2 I 3 14:5 2 I 14:16 14:24 2 15:2 5 16:2 5 16:3 3 16:21 2 17:2 4 18:9 I 2 18:11 5 18:13 3 I 6 2 I I I 1 I 2 I I 2 I 2 I In this selection of introspections it is seen that each con- secutive level is distinguished from one or more other levels in some individual experiment. The fringe material gives rise to imagery all the way from (a) no recognition to (b) visual recall, recognition, and re- production of letters in their proper places; that is, apart from gradual extension of some degree of clearness into its field, there is on the van of this conquest an occasional 'maturing' of its material into correctly recorded letters : "Had idea of other lines but not enough for recognition." "Had strong idea of other lines but not enough for recog- nition." "Had idea of other letters but could not recall." "Saw lines but couldn't recognize letters." "Saw whole line and noted others as containing letters — in past experiments when one line was seen nothing of other lines was noted." "Other lines noted, a little vague to recall." "Other letters noticed, almost recognized." "Saw no letter clearly; all suggested by lines and made defi- 90 JOHN EDGAR COOVER nite by repeating during the interval; not sure of any." (Good score). "Saw lines, later recognized K." 'Tartly seen and later recognized as S." "In interval named letters in line but did not decide that I had seen R (a fifth letter) until after recording line; then had vague visual image of R." "Added C after writing line, from image." "Image not determined until after line was recorded." Again, the suggestion may be definite without recognizing any material as imaged ; "No image, letter came with feeling of familiarity." The way in which the boundary of clearness is extended into the territory of the unrecognizable, may be illustrated by the advance beyond the capacity of the eleventh day to apprehend clearly the four letters in the line fixated : "Idea of other lines, but not enough for recognition." "Stronger idea of other lines; not enough for recognition." "Line clear, extra letter a little vague." "Line clear, extra letter clear." "Line clear, two extra letters vague." "Line clear, two extra letters clear." At first the extra letter comes as a maturing visual image, vaguely, then more clearly; then it is clear enough to be named and fixed kinaesthetically with the others. In the very first experiments Le. held her visual image during the 5-second interval after which she named and recorded the letters. Then she named the letters during the exposure, re- taining by repetition and recording from kinaesthetic imagery, occasionally receiving an extra letter from visual imagery later. But by the 6th day she gave up the moment of exposure to the vivid life of the after-image, to "Einpragung," and named the letters in the interval afterward; "Named from image one by one as would from card, but slowly." She customarily recorded from kinaesthetic imagery. During this "Einpragung" she 'felt' clear letters in her mouth, the vague letters, suggested by lines, being named for recognition afterward. If letters were not so clear, or formed non-euphonious groups the names were FORMAL DISCIPLINE 9i repeated two or three times during the interval. The visual image of an outlying letter sometimes escaped during the pro- cess of naming the others; and when the letter was recalled more vividly for retention during the naming of the others the latter became doubtful owing to faulty coordination of the two processes. Certain peculiarities of the reagent's process affected the score: (i) She often, especially in dull mental condition, found it difficult to name the letters (2:11, 13, 14, 15); sometimes she was balked for several seconds (4:15, 5:11, 8:7); and sometimes she would make the best of the situation by miscall- ing: "Called Q 'H', holding a (corrective) visual image of Q," (4: 19). This difficulty was more prone to occur when Q, Y, or Z began the line fixated (11:8). An effect of the difficulty in naming was to confuse the order (5:8), in which case the letters were named in trial orders until the reagent was satisfied by a feeling of familiarity (4:11). (2) Certain combinations of letters flashed into significant groups which at first challenged attention to the exclusion of further perception, but later served readier apprehension; such were the initials of friends, silver-ware marks, etc. : J B, J D, T M, S K, L K, T B (Tuberculosis), SF (San Francisco), etc. W M were noticed to be similar in construction. (3) All through the training there was occasional self-con- sciousness in fixating (11:18) resulting in (a) more or less irregular phenomena, such as winking, shifting, or squinting the eyes, just at the moment of exposure, and (b) eye fatigue and strain (11 : 19, 14: 17). Al. used an essentially different method from the others; he endeavored to get the whole presentation, inhibiting the ten- dency to limit the area of attention to a smaller and clearer field. From the 'fringe' levels, letters appeared often through kinaesthetic-auditory imagery and were correctly recorded al- though they were not remembered as seen; they often matured in visual imagery, generally with doubtful position; and often ^ JOHN EDGAR COOVER letters insinuated themselves correctly into the record because they "seem to belong there," neither 'cue' nor verifying imagery being accessible to consciousness. Perhaps the general method of attending intensely to the whole card favored 'fringe' phenomena. The fixing of letters during exposure, or while the after- image persisted, was effected by naming the letters; and kinaes- thetic imagery was the customary 'cue' for reproducing and recording; sometimes the imagery was supported by visual imagery, and sometimes additional letters were visually repro- duced and converted into kinaesthetic before recording. Sometimes a "curious rivalry" between the two kinds of imagery occurred with respect to what letter belonged to a certain position, in which case records customarily followed the stronger kinaesthetic-auditory. But in some cases the position was medial between the two letters. Again, a kinaesthetic-au- ditory C was corrected visually to Z. Certain assimilative effects evidently resulted from the rivalry of the imagery : J seems G when the latter was just diagonally below it. Again, this assimilative effect may involve only visual factors : M seems X when the latter was just above it. Mnemonics were avoided and very few associations involun- tarily occurred. Development consisted in more adequate control of the atten- tion during (a) apprehension, (b) fixing, and (c) reproducing periods, (a) On the second day the reagent recorded, "Have learned to regard the card as a whole; distinguish many more letters than at first, but cannot fix them;" and of the letters distinctly seen, three introspections of the second, third, and sixth day's record: "Saw more letters distinctly than ever be- fore," "Saw more letters, say six," "Saw more letters, say eight or nine." Particularly for the earlier part of the training, apprehension in its advance far out-distanced fixing and repro- duction, (b) More visual images were named and fixed, and (c) in reproduction the reagent ceased to distrust insistent, but not perceived, letters, as persisting from some former apprehen- sion, and became willing to give them a chance in the score. FORMAL DISCIPLINE 93 Many letters, no doubt, through this change of attitude toward the 'fringe,' 'matured,' as is indicated above in connection with attention; at any rate the scores generally justified the record. Further and more direct evidence of the more efficient control of the attention, noticeable through lack of strain, is given in an introspection on the 7th day: "Attention as usual, 'open-eyed,' to take in whole card, but not breathless as formerly." Al.'s maximum daily average occurred on the 3d day (see Curve, Appendix B. Fig. 14, p. 291), and the question arises as to whether all the improvement of his training was made during the first three days of work, and therefore likely to lie in an habituation which lacks the general character requisite for transference. If there was further improvement it was such as failed to affect the scores. Were the same question to be put concerning Mn.'s training for the same length of time (8 days) the same hypothetical conclusion would result, for the first daily average to exceed her third was the ninth. Yet in the face of the gradual ascent of her practice curve up to the 1 8th day, the probability of improvement on the days between the 3d and the 9th can scarcely be denied. In what then would Al.'s improvement beyond the 3d day consist ? Tabulation of letters recorded as doubtful indicates that the averages of the days following the 3d were made up more largely of 'fringe' letters and less clear letters; if we express in per cent of the average scores the amount contributed by 'fringe' letters during the course of training, we get : for the 1st and 26. days, 8% ; 3d and 4th days, 4.4% ; 5th and 6th days, 20%; 7th and 8th days, 17%. The averages from which we have inferred corresponding capacity therefore represent, in part, disparate processes. And since the conquest of the 'fringe' content in consciousness was seen to be an essential part of progress in the cases of Mn. and Le. and since it has been taking place with Al. since the 3d day, it seems more than probable that improvement has con- tinued throughout the training. This view is strengthened by the introspective notes clearly indicating increase of sensitivity, 94 ' JOHN EDGAR COOVER and by the experience, common to all reagents, of confusion in reproduction of the distinct content through effort to mature further 'fringe' content. Advance has been made in appre- hension; training had not continued long enough to coordinate the part-processes sufficiently for reproduction to show the gain. For all of the reagents the effect of training was in the follow- ing points the same: (For Al., omit a, d, f, h) : (a) More letters were clearly seen, (b) More letters were distinctly seen in less clear imagery, (c) More 'fringe' material developed into letters, (d) There was more efficient cooperation of kinaesthetic and visual imagery for the purpose of recall, (e) There was less variability of attention, (f) There was more adequate distribution of the attention to the part-elements of the process, (g) There was greater ease, less strain, in perception, re- tention and reproduction, (h) There was more apperceptive and associative process in apprehension of letters, serving a fuller content and surer recall. (2) Learning 12-letter-Rectangles Rt. and SI. gained upon their jfiirst day's average 53% and 42%, respectively, during their 18 days of training. Maximal efficiency had not yet been reached (see practice curves, Ap- pendix B. Fig. 15, p. 292). The daily work was not sufficient to show the influence of fatigue. Days of low scores were : after Easter vacation, when efficient coordination of part-processes had to be built up again, for there was a reversion to earlier processes; and days of poor attention. Rt. began learning regularly with the first line and proceeded as in reading. He got a strong visual impression and converted it into kinaesthetic-auditory imagery for retention and reproduc- tion, repeating it during the lo-second interval. Upon the first day he began to convert the stimulus immediately into kinaes- thetic-auditory imagery without reinforcing the visual impres- sion, and when visual imagery of letters revived after they were retained and reproduced by the Kin.-aud. imagery, it was deemed FORMAL DISCIPLINE 95 a hindrance. In case some letters were held visually after the screen had fallen they too were named and included in the repetition of kinaesthetic-auditory imagery. But when such a visually held letter was remote, in the third line, it was retained visually while repeating the other letters in kinaesthetic-auditory imagery (which by the end of the first day decreased in its auditory support) and was converted only upon recording. Up to near the end of the 2d day the few associations that occurred were also deemed a hindrance, for the letters were, like the earher visual letters, retained kinaesthetically : "Associations annoy me." But in a few experiments an association came in a convenient place, after six kinaesthetically retained letters, and its letters were not repeated during the interval but were recorded from the visually held association. This is, in brief, the way in which visual imagery in direct form, and association, grew into value as supplementary devices to the kinaesthetic rote method of learning, the coordination of which played so great a role in raising scores as training proceeded. The first appearance of the developed method was on the 2d day (i6th experiment) when the first six letters were held kinaesthetically, the next two by associations (CP, chemically pure), and the next two visually; only the first six letters being repeated by rote during the lo-second interval before recording. In the first experiment of the 4th day this method recurred, the association being LB (pound). But it took the training of the 4th, 5th, and 6th days to make this the predominant method, which with favorable modifications gained in elasticity and value until the maximum scores were made. During the period of emergence of this method, of the three- fold content, a two-fold content was dominant (see Analysis Curves, Appendix B. Fig. 16, p. 292). After the third day it was a rare score that contained only kinaesthetically held letters, although 12 out of the 3d day's 20 were such. The independent visual images of letters were held side by side with the kinaesthetic, except that the kinaesthetic were being constantly repeated during the free interval, while the visual persisted from the first reinforced impression. During this 96 JOHN EDGAR COOVER development of the coordination of the two processes of reten- tion and recall, a secondary visual imagery emerged, weaker than the other and often accompanied by doubt as to its value. It was imagery persisting apparently in its own strength, not having been reinforced during perception; and it often came late, — after all other letters had been recorded. It was this last kind of visual imagery, strengthened and made more reproducible through training, which later became avail- able when associations could not be readily formed, to prevent the learning process from lapsing into the early two-fold form. Under favorable conditions six letters were as many as could be safely held kinaesthetically, and four visually (primary), which would give a score of 30 points, and some of the daily scores exceeded this. Since coordination of the three-fold pro- cess involving associations, which became dominant on the 7th day, had resulted in carrying out the two-fold process, in ex- periments in which associations were not found, before the interval of exposure closed, time for strengthening the secondary visual imagery was provided. In successful work therefore, the process became at least three-fold. On the loth day a four-fold process appeared several times, (letters held in kinaesthetic imagery, visual associations, primary reinforced visual imagery, and secondary visual imagery). The general method became elastic in recognizing favorable associations in other positions than in the two spaces following the 6th, to which they had heretofore been confined; i.e., the six letters to be learned by rote changed from the first six to any not available for associations. Three times on the i6th day associations fixed the first two letters, and through the later period of practice all spaces presented favorable associations. The character of the associations seems to be largely visual; the stimulus yielding a word in visual form. It seems doubtful if many are accompanied by meaning, or at least if the meaning is prominent as it would be if the associations were apperceptive ; the reagent calls them "visual" associations throughout. FORMAL DISCIPLINE 97 Letters held in each kind of imagery sometimes escaped, and occasionally returned later in time for recording.^ The method of recording passed through a development. At first it followed the order of the letters on the card. Later, the visual imagery, which under low conditions of attention faded rapidly, was recorded first, then the kinaesthetic, then associations, and last the secondary visual. The order of re- cording followed the increasing reproductivity of the material, except for the secondary visual imagery. Some variable influences upon the score were apparent: (i) If attention was not good, (a) Naming for kinaesthetic images was retarded, (b) Visual imagery was weak, and (c) Possible associations escaped notice; (d) All imagery, especially visual, faded rapidly, and some- times escaped during the interval. (e) In recording, some of the imagery escaped, and the order of the letters recorded was confused. (2) Some material was harder than others to pronounce and retarded the process. (3) Unusual associations, especially at first, retarded the process. (4) Since forming associations became a prominent motive in the method, material difficult to associate decreased, and material furnishing easy associations increased the score. Improvement seems to consist in (a) Higher sensitivity for the visual material, since the secondary visual images developed, and moreover, into two grades of clearness. (b) Higher reproductivity of letter-names. (c) Coordinating kinaesthetic and visual processes, so as to develop from a two-fold to a four-fold content. (d) Greater facility in forming visual associations. (e) Greater facility in apportioning the letters on the card to the appropriate part-processes. (f) Better coordination of the recording process with the retaining process. (g) Better method in recording with respect to the vivid- ness of part-contents. 98 JOHN EDGAR COOVER (h) Possibly an increase in liability of reproduction of all imagery, (i) Decrease in variability. SI. began learning by repeating the letters on the card in reading order. Upon the first day he attempted to visualize the lower row, middle row, or the top row, while he repeated by rote the other two rows which he intended to hold kinaesthetically ; occasionally he sought by tense staring while naming to have visual imagery support the kinaesthetic. In the 15th experiment of that day he hit upon an association: BROM (Laxative Bromo Quinine) and recorded LBRQ, and noted in his introspections that "It probably did not help me to remember." The next day he tried to fix all the letters by naming, and found it too much, losing, in that experiment, all but one line. He then began his later method by confining his rote mem- orizing (Kinaesthetic) to the eight letters of the first two lines and held as much as possible of the third in visual imagery. In a few cases the kinaesthetic imagery of the first two lines was supported by associations: JPN (Japan) and BHSR (Belshazzar) ; and again WMBR (Wamba). Upon the 3d day his visual imagery of the third row was assisted by associations (Z, last letter of the alphabet, and J his initial). Thus appeared the alternate form of his method. But it was not yet a conscious method, and was not much used until the 9th day from which time it was used to fix about as many letters in the third line as were reproduced visually. "Can repeat only names of first two rows so as to remember them; either have mnemonic for third row or visualize." A curious thing appears to have occurred during its emergence at that time, which seems rather an effect than a cause of its use. The visual imagery used during the preceding four days seemed to change to what was called in the discussion of Rt.'s results a 'secondary' form {i.e., not reinforced during perception) or an apperceptive form involving alphabetic position, which re- lated it to a class of the associations available and used at first FORMAL DISCIPLINE 99 particularly for X and Z, as is shown by the prevalence of substitutions of near-lying letters, as H for J or K. The reagent repeatedly said he did not visualize, neither did he repeat, nor did he form those letters into associations — he "just remem- bered them." "Last letters written in lower row are remembered by gazing steadily at them and names are not pronounced neither are they visualized." He also records some letters from this field which he cannot account for: these are what I have called the 'secondary' visual form, when they came in the visual mode. Apparently one cause for this change in the character of his visual imagery is that as he gazes at them he is not so much intent upon sharp visualization as upon seeking mnemonic devices to fix them. Up to this time he had made three perfect scores by holding the lower row in visual imagery. From now on, all his perfect scores (22) with one exception were assisted by associations, the first three occurring on the 9th day. The following table shows the relative use of his three kinds of retention for the third row, the 'visualized' and the 'secondary' visual letters combined : Day No. of letters recorded from each Kin. Vis. Ass'ns I 4 I 2 2 3 I 3 II 2 4 7 8 S 5 22 6 6 10 4 7 28 8 18 8 9 9 9 18 10 18 17 II 29 17 12 15 17 13 10 18 14 12 8 IS 20 32 16 34 II 17 16 32 18 14 34 loo JOHN EDGAR COOVER By the ninth day his method was to gaze at the 3d row during the exposure seeking a mnemonic association or visuaHzing while he was repeating by rote the eight letters above held in peripheral vision. Conflict between the visual and kinaesthetic processes en- couraged the search for mnemonic aids. His visual imagery was very unstable and vanished upon slight provocation. If it was to persist, it had to be made as vivid as possible during the impression; but that took attention from the rote process with the result of losing some of the kinaesthetic letters, or of losing the rhythm and sequence, which led to transposed letters in the record; then again, if effort was put disproportionately upon the kinaesthetic process, the lower row would be lost. The associations were not used advantageously above the 3d line; in fact they were disturbing there. They often resulted in substitutions of other letters, indicating that the form of association was not visual, as in the case of Rt., but more apperceptive. Sometimes they were of a complex nature : MDKS was held by "M.D., Mark Keppel, Snell;" ZGJX, "between special letters ZX, George Jones." Sometimes letters in a well known combination would be absorbed in another more special but earlier one: PBDV gave Peabody, although since the com- bination had occurred before, "deo volente" had been common for DV. The visual imagery of letters seemed to be more easily held if they were accompanied by associated letters, and the process was often thus a three-part process : Kinaesthetic for 8 letters, associative for 2, visual for 2. A peculiarity of Sl.'s imagery was that as long as he held his eyes on the screen, after the exposure, the letters remained there; but as soon as he glanced down to the paper, they either vanished or took a position on his forehead where they were very unstable and were apt to escape while he was recording the kinaesthetic letters. It took special efifort to retain them. His other imagery was also quite unstable, for any inner distraction was fatal to it; upon one occasion he began record- FORMAL DISCIPLINE loi ing with the wrong letter and almost lost the whole score. He tried to hold the imagery by keeping his body rigidly in the same position. Recording was an especial distraction. A development of the kinaesthetic process took place: At first he merely named letters "in his mind," then in a whisper, finally with very slight throat and tongue movement. The first was not vivid enough and the second was too slow. The variable influences upon the score recorded in the dis- cussion of Rt.'s training were present here also; except that No. 3 may be omitted and No. 4 made applicable to the 3d line only. Improvement seems to consist in (a) Coordination of visual, kinaesthetic, and associative processes. (b) Using incipient pronunciation with rote process. (c) Better at)perception of visual images. (d) Coordinating the recording and retention processes. (e) Making association a method. (f) Decrease of variability. It does not seem that sensitivity to visual imagery, or that reproductivity except as better effected through (b), was in- creased; and the method remained fairly mechanical. (3) Reaction to Sound Ly. did not greatly reduce her reaction time by her training of II days, 100 reactions per day, (1.4%). That there was a change in the process is indicated by the character of the practice curve : There was a drop down to the 6th day, showing at that point over initial capacity a gain of 19% ; but from then on there was a gradual rise to almost initial efficiency. If the ten reactions each of the 27 series, on the 100-gram tension of the key, are distributed and plotted (5 sigma to the plot) in three distribution curves of 90 cases each (correspond- ing curves, with 10 sigma to the plot, are shown in Appendix B., Fig. 18, p. 294), it is seen that the motor reactions with times around 100 sigma have fallen away, after the middle period; 102 JOHN EDGAR COOVER that the highest mode has shifted from 158 in the first period to 178 in the middle and last periods; that the chief block of reactions which came, in the first, between 130 and 170 sigma, has broadened in the middle to 180; and that the last curve is cut by a cleft at the point where the first has the highest mode, giving rise to two blocks: 130-155, 160-195. The median has moved from 159 to 160 and 165. The introspections throughout are headed "Sensorial"; at the same time, individual introspections note pressures, strains, and tensions in the fingers, hand and arm, and premonitory reactions indicate a motor direction of attention. Although the introspections are not full enough to determine definitely, yet there is some evidence that the process of reaction became less motor as practice continued, after the sixth day. It may be that as training proceeded the sensorial form was more nearly approached, yet the first considerable mode of the first curve, from 130-145, continued and broadened toward longer time, in the middle curve, has split in the last curve with the narrower mode at 130-135 and the broader at 140-155, whereas the former mode of 130-135 would seem to be the place for a practiced sensory time. The widening of this mode in the second curve, and the growth of a wide mode from 140-150 in the last, seem to indicate the development of factors in the process of reaction which (a) shifted most of the 'sensorial' reactions to 140-150, and (b) gave rise to a more frequent type of reaction at 160-195, where the last curve has its largest block of reactions and the first curve is most serrated. Intro- spections do not make clear what those factors are; but they must be such as cause reactions to fall central upon 148 and 178 sigma. The process at the beginning of practice was sometimes pure motor (90-110 sigma), often sensori-motor (i 10-120), very often sensory (130-140), and more often a complicated sensory (145-165). During training, the motor disappeared, the sensori- motor diminished, and the practiced sensory narrowed (130- 135), the sensory (135-150) became more frequent, and the FORMAL DISCIPLINE 103 complicated sensory (160-195) most frequent. This develop- ment into longer time conforms with Bergemann's results on influence of practice on sensorial time.*^® There was a decrease in variability, which is usually taken to indicate improvement in attention, but it is still large (MV., 13%), about 5% above practiced form (8% of the mean). (4) Memory Training He. and Cr. tested mnemonic devices for memory of figures, dates of events, and lists of words. Although the modern*''^ systems of mnemonics have the ad- vantage of the older^^ in that they rely upon the congruence rather than the incongruence of the supplied mnemonic word or phrase, they are open to two chief criticisms: (a) They add to the material to be memorized, and (b) the supplied mnemonic is likely to be replaced by a false one (He. lost the key-word "pass-key" in the mnemonic for Homer, by substituting "latch- key", which changed the date from 907 to 567). The chief advantage of mnemonics appears to lie in the better apperception of the data to be remembered, by reason of subjecting them to especial attention in order to determine that they are correctly represented in the key-word of the mnemonic. Since the natural relations in the data are obviously the more profitable to be scrutinized, the advantages outweigh the disadvantages, if anywhere, only in the learning of dates, telephone numbers, street numbers, or other disconnected numbers. That great advantage is gained by inspecting and grouping numbers, not for the purpose of transposing them into mnemonic phrases, but for noting the relations subsisting between the re- spective groups, was shown by the correct reproduction of series of 24 digits after grouping them into three's and inspecting them just once, ^'Wundt: Physiologische Psychologic, III:42i, (ste Auf.). '"E.g., Harvard College — Teach much — t-ch-m-ch — 1636. {Vid. James: Briefer Course in Psychology, p. 74; or Prin. of Psychology, vol. I, p. 669). ^^ E.g., Xenophon — a-zang-for-fun might do damage with matches — m-tch-s —360. 104 JOHN EDGAR COOVER The principal object of the training, however, lay in the repe- tition of lists of words, in conformance with Loisette's^^ in- structions for the training of attention by "compelling the intellect into some particular channel and keeping it there," "by compelling the intellect to stay with the senses," through analysis of connections between words in a series and frequent rapid repetition with full concentration. He claimed to have made weak memories strong, and good ones better, by this method. The method involved (a) an attentive apprehension (auditory or visual) of a list of words, having a more or less obvious sequence, with the view of noting the relations between each two words, and (b) an immediate recall of the whole Hst after the single apprehension. Later, on succeeding days, the lists were to be repeated orally from memory, forwards and back- wards, as rapidly as possible, recalling clearly the relations between the words. Notes were taken of the time and errors. The reagents set for themselves three subsidiary aims: (a) to determine the advantage of scrutinizing the relations between the successive words, for reproduction after a single impres- sion, (b) to learn whether the initial stability of the particular connections remains relatively the same in successive repro- ductions, and (c) to learn if the liability'''" of recall remains the same in successive reproductions, as shown by the time to reproduce the series. The second list of 30 words began with : Building, dwelling- house, parlor, partridge, feathers, light, lighterman; the rela- tions between the words are : genus and species, species and genus, whole and part, partial identity in sound, whole and part, substance and attribute, partial identity in sound. The list was dictated to He. by Cr. at the rate of one word per second, and was reproduced orally by He. in 160 sec. and by Cr. in 58 sec. Three lists were learned and reproductions were made on nine days after March 3d, to April 14th. ^ Loisette : Assimilative Memory, 1896, p. 20. '" For the special meanings of "liability" of reproduction, and "fidelity" of reproduction, vid. Kiilpe : Outlines of Psychology, p. 197. FORMAL DISCIPLINE 105 The chief subsidiary results may be stated as follows : (a) Scrutinizing the words for relationships results in suc- cessive associations of continguity reinforced by a unity in mean- ing: i.e., in the "Pointer, oak, ax, steel, ore, mine," series, Pointer, oak, are not only together in consciousness as two words, but the pointer is visually or kinaesthetically imaged as an oak pointer; and oak, ax, are unified in an image appropriat- ing any suggested relation between them, as of cutting down the oak tree, or of fashioning the pointer; Building, dwelling, house, are imaged as separate buildings in a hamlet, bearing spatial relations to each other, and parlor and partridge are in the dwelling — they all constitute a simple 'complex'. In- centives for recall are thus doubly strong. The reagents were surprised that the process of reproduction after a single im- pression went off with so much facility and with so few errors. (b) Although certain parts of a series visualized or imaged in a 'complex', fitted into a story built up through the imagery suggested by the words at the first impression, seemed to be more difficult and to demand more attention, in subsequent re- productions, than other parts, this difficulty resulted in halting the rhythm of repetition rather than in errors. The errors were almost wholly omissions, occurring in almost any part of the series, but seldom recurring in subsequent reproductions. They are caused almost wholly by incentives of recall reaching beyond the next member in the series. Introspection indicates that incentives reach from one 'complex' to another so that the most important member in the next 'complex' comes into consciousess before all the members of the last and the beginning members of the next are recalled, and that this consciousness of direction gives confidence and facilitates repetition. In this way ground is covered from one 'complex' to another by neg- lecting some of the intervening members. The incentives of recall of the next word constantly vary in successive repetitions. To attain accuracy intense concentration seems necessary, unless the process is to become merely rote-repetition by the vocal organs, which throughout was guarded against under the in- junction of the training to "keep the intellect with the senses." io6 JOHN EDGAR COOVER (c) The liability of recall, as shown by the time taken to reproduce the whole series, fluctuates constantly from day to day, yet considerable practice-effect was shown by both reagents, which indicates that in general liability or recall increases under conditions of relatively few repetitions during a long interval of time. If a free interval before a subsequent reproduction is but two days, the liabiHty of recall is considerably increased, as shown by decrease in time; if four or five days, it is still increased, if 14 or 15 days, it is slightly decreased — about as much as it is increased after four or five days. The advantage of the use of mnemonic connections in learn- ing German-English vocabulary {e.g., mistrauen^ — miss — shy girl — shy — diffidence) was not definitely determined; although a list of vocabulary could be repeated with certainty after one perusal, that perusal took about as much time as equally efficient direct learning; and although the former has the advantage of the use of the logical memory, while the latter relies more upon sensory memory, its connections seem to drop away with the flight of time about as rapidly as sensory memory falls away. Concerning the principal object of the training, introspection says : "Takes intense concentration ; afterwards feel tense in the frontal regions." But beyond the feeling that attention was keenly experienced in the training, and therefore presumably improved, there is no quantitative evidence at hand to show that improvement was made. This training was not as regular nor as systematically con- trolled as the foregoing. e. The Test Results In the following discussion of test results we shall notice chiefly ( i ) the extent of variability in mental processes between different reagents who apply themselves to the same objective task,'^^ (2) the extent of variability in a single " Some of this variability will undoubtedly result from differences in 'type,' whatever that proves ultimately to be, but the following discussion takes no account of it : owing to the difficulty of diagnosing and adequately describing an individual's 'mental type,' since all individuals are probably mixed types FORMAL DISCIPLINE 107 reagent's processes while applying himself to the same objective task, and (3) the conditions which must be met for the scores to be comparable. (4) The causes of variability, so far as they appear, and (5) the effect of variability upon the scores, are not neglected. According to the form into which we have cast our tests and training, it must be remembered, the particular influence of training upon these tests which we are seeking is that of improved attention. The initial efficiencies of the first series of tests constitute a cross-section of the initial capacity of at- tention; the final series of tests gives the final capacity of attention. If the processes in the final test have changed essen- tially from those employed in the first test in some other way than may be attributable to better attention, if, for example, the method of work is different, then the difference between the two scores would represent something besides the change in attention, and could not be used. At best, such a difference- score could only indicate the advantage or disadvantage of doing the task in another way. Scores of the same reagent to be comparable must consequently represent similar processes; scores of different reagents to be comparable, we must assume, should also represent similar or at least equivalent processes, and score-differences, or per cent of change in eflSciency, to be comparable should be based on something near equal initial efficiency. When comparable, the difference-scores of the trained and control reagents will be inspected for transference of improved attention. The degree of analysis in the various tests is not uniform for the reason that the processes engaged in the tests varied which vary according to means of diagnosis (vid. Segal, op. cit. and Fernald op. cit.), only the following general and perhaps untrustworthy characteriza- tion of our reagents, based largely upon questionnaire replies (Wissler: op. cit. 8-9) supplemented by oral report, may be offered : Mn. Le. Rt. SI. Ly. He. Cr. Al. Ms. Wf. Visual Strong Strong Strong Weak Strong Fair Weak Fair Strong Weak Auditory Good Strong Weak(?) Weak Good Good Good Good Strong Weak Kinaesth. Good Strong Strong Weak Good Good Good Good Strong Weak io8 JOHN EDGAR COOVER greatly in complexity, and the introspections of the reagents consequently varied correspondingly in completeness; but effort is made to analyze the tests employing the less complex pro- cesses fully enough to serve our primary purpose and to con- tribute to the popular notion of the nature of 'mental tests.'^^ That the extent of variability may be adequately indicated, all the tests are subjected to analysis, — an analysis that is not merely a logical schematism, but an empirical construction built up from the introspections. ( I ) Reaction to Sound Simple Reaction Time has been used in the study of Attention (Angell and Moore/ Binet^), in determining mental and physical correlations with children (Gilbert/) and with university students (Cattell/* Cattell and Far- rand/ Wissler"), in determining psychological norms of men and women (Thompson®), and in the study of individual psychology (Binet et Henri/ Henri*). The latter'" emphasize the value of the M.V., which has been suggester by Titchener" and Pillsbury^" as a possible measure of attention, and has been denominated by Buccola "the dynamometer of the attention."" ^ Angell and Moore : Reaction Time : A study in Attention and Habit. Psych. Rev. 1896. 3 :245-3S8. ^ Binet : Attention et Adaptation. Annee Psych. 1899. 6 -.276^. * Gilbert : Researches on the mental and physical development of school children. Studies from Yale Psych. Lab. 1894. 2 :8i. '"Cattell: Mental Tests and Measurements. Mind. 1890. N.S. 15:376. * Wissler : Correlation of mental and physical tests. Psych. Rev. Mon. No. 16. 1901. P. 7. ^Cattell and Farrand: Physical and mental measurements of the students of Columbia University. Psych. Rev. 1896. 3 :639flf. ^ Thompson : Psychological Norms in men and women. Univ. Chicago Contrib. to Phil. 1903. 4:8ff. 'Binet et Henri: La psychologie individuelle. Annee Psych. 1895. 2:445. * Henri : fitude sur le travail psychique et physique. Annee Psychol. 1896. 3 :24s. *" Also, Binet : A propos de la mesure de I'intelligence. Annee Psychol. 1905. 11:69-82. 'Titchener: Simple Reactions. Mind. 1895. N.S. 4:79; also. Lectures on the elementary psych, of feeling and attention. 1908. P. 280. ^^ Pillsbury : Attention 1908. P. 89. " Euccola : La legge del tempo nei f enomeni del pensiero. Milano. 1883. P- 155; (Quoted by Oehrn : Experimentelle Studien zur Individualpsychologie. Psych. Arbeiten. 1896. 1:113)- " In order to allay any suspicion that our tests are not representative and are pecuHarly subject to variability in processes, a paragraph, with references, showing the relation they bear to other tests reported in the literature, is added to the discussion of the results of each test. FORMAL DISCIPLINE I09 It is well known that intensity of attention lowers both the tirne^^ and the relative variation in simple reaction. But for averages to show this, they must not be influenced by other causes of variability : Health, 'Anlage,' habituation to external distraction, and the direction of the attention must remain uniform. The introspections indicate extraneous causes of variation : (a) Conditions of health, through good, dull, tired, nervous, etc. (b) 'Anlage', from interest to indifference, calm to anxious, natural to muscular or sensorial set of consciousness. (c) Process, as regards habituation to external distraction (such as noise, unaccustomed finger reaction, pressure on the key, temperature of the hands) ; as regards fluctuation of the attention, within the series, between muscular and sensory reaction. Another variable factor lies in the movement: In raising the finger, the extensor muscle must overcome the flexor, and owing to the balance between the tensions of these antagonistic muscles, the reaction movement is not simple but varies from a simple extensor reaction retarded by flexor tension, to an extensor reaction preceded by antagonistic flexor reaction which delays the reaction movement 40-50 sigma,"^^ and three types of this variable factor have been observed. ^^ Examination of the test averages and the distribution curves confirms and supplements the evidence of introspections. Clearly, the test averages cannot be handled recklessly. They cannot be compared at random for at least four good reasons : (a) The change in the direction of the attention within the series, as shown by a bifurcated distribution curve, and by a large mean variation, is compatible with good attention. The average of such a series might lie above or below that of a series obtained with an equal degree of attention, but where the direction of the attention remained constant : e.g.. The atten- tion of Le. may have been quite as good in the series giving an "Cattell: Phil. Stud. 3:329ff.; Pillsbury: Attention, 82; Kiilpe: Outlines, 432. ''* Smith, W. G. ; Antagonistic Reactions. Mind, 1903, 12 :47-58. "Judd, McAllister, and Steele: Mon. Supp. Psych. Rev., No. 29, pp. I4iflf. no JOHN EDGAR COOVER average of 117.2 with a mean variation of 16.6, as in the series giving 136.4 with a mean variation of 4.6; and it may have been no better. (b) Even when the direction of the attention is constant and the psychical process is about the same for the individual reagent, his results cannot be compared with those of another reagent whose process, as shown by a widely different average time, is essentially different, for the 'abbreviated' and the 'com- plete' forms are not merely different forms of the same act but are different acts,''^® and practice-effect is greater upon the 'complete' than upon the 'abbreviated' type.'^''' (c) Test averages which include great practice-effect are not comparable with those which do not, for they are not so reliable a measure of efficiency. (d) Test averages of reagents showing widely different facility also are not comparable because the reagents cannot be assumed to be doing the same work. Were all of the variable influences to remain about the same in the final tests as they were in the first, and were the essential processes also to remain the same, for each reagent, then all the averages might be used in determining influence of the training interval. In so far as our results vary from this requirement, they have to be put aside. Distribution curves show changes in the essential processes in the final test, for the majority of the reagents : Le. from an automatic (95 sigma) and a muscular mode (i 10-135) to a sensory (140) ; SI. from a sensory (145) to an automatic (95) and two muscular modes (130, 115); Ly. from sensory (145- 155) to automatic (100-125); Cr. from muscular (111-120) to sensori-motor (130) ; Ms. from less to more automatic (94-100) and muscular (125-130); Ct. to less automatic (90-105) and to more motor-sensory (144-150). The reagents differed greatly in initial efficiency (ranging from 1 16.0 to 186.1), and fall, in that respect, into four groups about the following averages: 115, 130, 155, 185. " Angell and Moore : Reaction Time. Psych. Rev., 1896, 3 1245. "Wundt: Physiologische Psychologic, 1903 (ste Auf), 3:419. FORMAL DISCIPLINE iii The per cents of change in the final test from the first, for the reagents whose scores are fairly comparable, are: Trained ist ControP 2d Control Group 2 Mn. -4.3 PE 4.38 Wf. -0.15 PE 3.40 He. 5-8 PE 5.62 3 Rt. -17.0 PE 6.05 4 Rr. -6.7 PE 20. And these figures are not entirely free from other causes of variation besides change in processes : Those of Rt. and Rr. are too great because of the large practice-effect in their first test (Rt. 80 sigma, Rr. 40). The loss of He. resulted from the fact that his first test was taken in practiced form, just after long practice in another experiment. Wf.'s decrease of time should have been more, for at the beginning of his final test he was "rather fatigued," and for the latter part "somewhat nervous," and his first test was taken while still in practiced form from preceding experimentation. The absolute difference for Mn. is a third larger than the probable error, for Rt. three times as large, for He. a half larger, for Wf. and Rr. much less. In concluding our comi)arison we can only consider it possible that Mn. and Rt. have transferred some improvement to this test: Mn. from tachistoscopic training, and Rt. from a slight practice in simple reaction to visual stimuli if not from his train- ing in Learning 12-letter-rectangles. Nor with the averages for variability are the results more decisive; yet, as was noted above, they are recommended as measures of attention. This would be true, no doubt, if the measure is to include steadiness of direction as well as of intensity of attention, as was suggested by Whipple.'^*^ Yet, were the direction of the attention constant, it is not obvious that a given amount of variability from a 'sensorial' average is just equivalent to the same amount of variability from a 'muscular' '^The ist Control reagents are those who took all the tests; the 2d Control, those who took but one pair or a few pairs of tests. "Whipple: Reaction Times as a test of Mental Ability. Am. Jr. Psych., 1904, 15:496. 112 JOHN EDGAR COOVER average, even when both averages are made by the same reagent f^ much less when made by different reagents. On these grounds then, the variation averages of all those who changed in the form of their reaction, in the final test, must be disregarded, which leaves the following: 2d Control Trained 1st Control MV r.v. MV r.v. Group 2 Mn. -4.7 -5-1 Wf. 2.7 1.9 He. -0.9 Group 3 Rt. -1.7 0.9 Group 4 Rr. -10.8 -3.8 The r.v. = MV/M x 100,^^ and makes the figures somewhat more comparable than without the reduction. Variation in initial relative variability ranged from 8.0 to 18.8. The increase of Wf. is due in part to his low variation in the first test, taken when he was in practiced form, and in part to nervousness and exhaustion in the final; the decrease of Mn. and Rr. is principally due to better habituation to ex- perimental conditions, Mn. possibly bringing some advantage from her training with the tachistoscope. Part of Rr.'s great decrease must be attributed to his extreme nervousness in the first test (his first Avg. MV was 45.9, while Mn.'s was 17.5). It is evident that in the reaction-time experiment the reagent must, as Wundt claims, be "thoroughly practiced in the tech- nique," or "there can be no hope of obtaining reliable results f^^ and that lack of expertness in introspection makes it difficult to group the processes according to kind, so that they may be measured and their measurements justly compared. It is claimed that some reagents are so incapable of control of the direction of their attention that they cannot be used in the ex- periment.^^ As to the effect of practice on variability, it has been shown under certain conditions to increase it.^* '" Vid. Alechsieff : Phil. Stud., 1900, 16 :24. *^ Titchener : Experimental Psychology, 1905, II, I:i82. ** Wundt: Vorlesungen, 1911 (ste Auf.), S 312. *" Lange : Phil. Stud., 1888, 4 1479. ** Angell and Moore : 'Reaction Time, Psych. Rev., 1896, 3 :24S-2S8. FORMAL DISCIPLINE 113 (2) Marking Out Small 2,'s The discriminative reaction of "Cancellation" was commended by Pillsbury* as probably the best test of the positive type for measuring attention. It has been recommended and used for this purpose in almost all of the important studies in individual psychology (Binet et Henri,^ Henri,* Toulouse/ Oehrn,° Binet," Sharp,' Whitley,*) ; it has been used as a mental test in correlational studies (Cattell and Farrand,' Wissler,^" Brown,"), as a means of studying the* processes of recognition and discrimination (Bourdon," who originated the test), attention and adaptation (Binet"), fatigue (Ritter"), habit (Bourdon^^), distraction and habituation (Vogt'"), general practice effect upon like or re- lated processes (Thorndike and Woodworth," our own experiment on mark- ing out words, pp. 34ff), practice effect upon individual differences (Wells,^* HoUingworth") ; and it is included in Whipple's ManuaP with tests for "At- tention and Perception," where an historical and descriptive account of the test may be found. One letter or character may be crossed out, as the small a, (Bourdon," Binet et Henri,'' Toulouse," Sharp,' Whitley*) or more than one letter, as a, e, I, t, or a, e, d, r, s, etc., wherever they occur (Binet* " Ritter," Bourdon," Vogt," Brown,"), or words which contain given letters, as both e and r, may be crossed out (Thorndike and Woodworth" and our own experiment, pp. 34ff). The matter containing the letters to be cancelled may be ordinary printed text or printed mixed words, in a known or in an unknown language, printed pages from printer's "pi," small letters or capitals, pages of spaced or unspaced digits, etc. ' Pillsbury : Attention. 1908. Pp. 84ff. ' Binet et Henri : La psychologie individuelle. Annee Psych. 1895. 2 :446. ^ Henri : fitude sur le Travail psychique et physique. Annee Psych. 1896. 3:239. * Toulouse : Enquete medico-psychologique sur les rapports de la superiority intellectuelle avec la nevropathie. (Zola) 1896. P. 226. ° Oehrn : Experimentelle studien zur Individualpsychologie. Psych. Arbei- ten. 1896. I :98. ' Binet : L'fitude experimentale de Tintelligence. Paris. 1903. Pp. 236ff. ' Sharp : Individual psychology : A study in psychological method. Am. Jr. Psych. 1899. 10 :356. * Whitley: An empirical study of certain tests for individual differences. Archives of Psychol. 1911. No. 19. 3:ii4- * Cattell and Farrand : Physical and mental measurements of the students of Columbia University. Psych. Rev. 1896. 3:641. ^"Wissler: Correlation of mental and physical tests. Psych. Rev. Mon., No. 16. 1901. 3:7. "Brown: Some experimental results in correlation of mental abilities. Br. Jr. Psych. 1910. 3:297. " Bourdon : Observations comparatives sur la reconnaissance, la discrimina- tion, et I'association. Rev. Philos. 1895. 40:167. ^* Binet: Attention et adaptation. Annee Psych. 1899. 6:364. " Ritter : Ermiidungsmessungen. Zeits. f . Psychol. 1900. 24 :424. "Bourdon: Recherches sur I'habitude. Annee Psychol. 1901. 8:330. 114 JOHN EDGAR C DOVER Marking out small as from lines of English print, 93 mm. long, on a page containing 100 of them (see p. 75), is a fairly- simple task, but, as introspections show, may involve quite dissimilar processes : (a) Movement along the line may be like that in reading, may alternate in direction, may embrace more than one line at a time, may be interfered with or facilitated by following a pencil-point. (b) The essential process may be (i) a search for the form among all the letters, without a unit of material to search through, or with the word or the line as a unit, (2) a search for the sound-image of the letter by pronouncing the words in inner speech, (3) an incipient pronunciation of all the words with the reliance mainly upon the kinaesthetic image of the sounded letter, (4) a reading of the text and reacting upon the words known to contain a. (c) The main process may not be pure, and may be sup- plemented by (i) elimination of words and suffixes known not to contain a; (2) by activity of attention in peripheral vision so as to command a larger field, leading to inaccuracy in reach- ing too far forward and to accuracy in catching omitted letters in the lines above. (d) The process may be retarded by distractions such as (i) difficulty with the pen, (2) appeal of the context, (3) look- ing back to catch possible omissions, etc. These various processes may be employed singly, in combina- tion, or in succession by a single reagent in a single test; which is sufficient warning that the averages of such tests may not be used for comparison unless similarity of processes yielding them is assured.®^ "Vogt: Ueber Ablenkbarkeit und Gewohnungsfahigkeit. Psych. Arbeiten, 1899-1901. 2--72,- " Thorndike and Woodworth : The influence of improvement in one mental function upon the efficiency of other functions. Psych. Rev. lOOi. 8:553. ^ Wells : The relation of practice to individual differences. Am. Jr. Psych. 1912. 22 -.77. " Hollingworth : Individual differences before, during, and after practice. Psych. Rev. 1914. 21 :3. "" Whipple : Manual of mental and physical tests. 1910. Test, 26, pp. 254ff. *^ Peters (Auf merksamkeit und iReizschwelle : Versuch zur Messung der Aufmerksamkeitskonzentration. Archiv. f. ges. Psych. 1906. 8:391) thinks FORMAL DISCIPLINE "5 Ly. used the visual cue; Mn., SI., and An. reacted to a by an auditory image of the word; Al., Le., He., and Wf., to the word as known to contain a, supplemented by the visual cue. Methods changed some, or were of mixed type, for all reagents during the test. For reagents who differed in kind of performance, the averages are not strictly comparable ; and for those whose process changed in its dominant elements, in their final test, the average difference-scores express something more than the effect of the training or the interval upon this test, and must be disregarded : Rt., in his final test, added to his method of attending to words as units and searching for the visual form, direct reaction to the word without search ; Cr. changed from the kinaesthetic- auditory cue to the visual cue ; Ms. from the visual cue, supported by a kinaesthetic-auditory image, to the word cue; CI., from marking words known to contain a, without notice of context, to reading the text; and Ly., from the word unit to reading the text. These changes in method are independent of the in- fluence of training in sustained attention, and were as often disadvantageous as advantageous. Initial efficiencies ranged from 113-282 seconds, and fall into six groups: 113, 129, 145, 165, 190, 280. The more comparable scores yield the following per cent of improvement (decrease in time of 100 reactions) in the final test: Trained 1st Control 2d Contr( Group I Group 3 He. 9 Wf. 10 Group 4 Le. 14 SI. 9 Group 5 Mn. 29 Al. 15 Group 6 An. 1 1.3 It seems probable that the practice-effect of the test on itself is about 10% ; that it was benefited by the tachistoscopic training about 10% more, but that it was not benefited by training in reaction to sound. Learning 12-letter-rectangles, or by Memory the process so complex that it must be analyzed before any part of it can be subjected to measurement; for this reason he did not use this test. ii6 JOHN EDGAR COOVER Training, or by the other tests in the series. The quick per- ception of capitals in the tachistoscopic training presumably lowered sensitivity for visual impressions, or shortened cog- nition-time, which was available for more ready perception of small as in the text. (3) Marking Out Small o's Marking out small o's from an inverted page of English print containing 100 (see p. 75) of them, was intended to be a process somewhat free from the distraction of the context,^^ which attended the preceding test, and to offer opportunity for a still more simple process, not so capable of change. But introspections proved it to be also quite variable. (a) (As above). (b) The essential process may be i) a search through all the letters for the visual form, without a unit or with the word or the line unit, and with a purely visual image of the or the visual image strongly supported by an auditory or a kinaesthetic image of it; and the kinaesthetic image may be that of a) re- peated pronunciation of the name of the letter, b) breathing its sound continuously, or c) pen-movement in tracing its form; 2) a search along the line for the only natural letter, since its form alone is not altered by inverting the page; 3) a blocking of the inverted page into units of a line, or a part of a line, and a 'spotting' of the letter that 'stands out' from its sur- roundings, (in which case, at moments when the central prepara- tion for its cognition was not perfected by the attention, the *° Other methods of avoiding the distraction of the text were noted above; such as, the use of text in unfamiliar language, of unspaced or "pied" ma- terial, etc. Woodworth and Wells (x'X.ssociation Tests. Psych. Rev. Mon. 1911. V. 13, No. 5, pp. 24fif.), Wells {op. cit.), and Hollingworth {op. cit.), used digits, which appear to be much the best material yet proposed. The first two researches involved the cancellation of zero's, — a close approach to the present test. Woodworth and Wells (p. 28) note some of the irrelevant individual differences obvious to the experimenter upon examination of the checked page, or upon observation of the performance : variation in the man- ner of making the cancelling stroke, reversal of direction in inspection of the line, misunderstanding instructions as to amount to be checked over, etc. FORMAL DISCIPLINE u? failed to "stick out in relief from the general blur of letters" and either had to be painstakingly searched for, or its omission risked, either case causing retardation or distraction). (c) The essential process may be supplemented by: i) re- acting to words recognized as containing o, 2) reading in- verted words in the search for the auditory or the kinaesthetic image, 3) the elimination, without search, of words or suffixes known not to contain 0. (d) The process may be retarded by peculiar distractions: i) Confusion with inverted c, 2) perseverence of tendency to react to a's (this test followed marking out a's),^'^ 3) looking back for omissions, 4) tendency to turn the head and eyes to read, 5 ) recognition of inverted words. The following reagents changed their processes essentially in the final test : Mn. changed from the visual cue to word-reaction supplemented by looking "for the most natural letter"; Le. changed from one or two words as a unit, to a whole line; He. changed in part to word-reaction; Ms. from inverted image of the word to visual cue and reaction to small words. Among the other reagents the methods differed somewhat: Wf., Ly., and Al. used the visual cue and reacted to small words; Rt. skipped words and endings known not to contain 0; Cr. and SI., and possibly Gl. and An., used the visual cue simply. The more comparable results, because the methods are more nearly equivalent, range in initial efficiency from 125-380 seconds, and fall into four groups: 125, 190, 225, 380. Changes in efficiency are as follows, in per cent of improvement : Trained ist Control 2d Control Group I Cr. o Group 2 Al. 13 Wf. 17 Rt. 13 SI. 3 Group 3 Ly. 4 GI. 4 Group 4 An. 22.4 Disregarding the results of the 2d Control reagent in Group 4, whose initial efficiency was but half that of Group 2, it appears *' Binet (Attention et adaptation, loc. cit.) found interference in changing from a, e, d, r, s to i, 0, I, j, t, (p. 370). ii8 JOHN EDGAR COOVER reasonable to expect about 5 fo improvement in practice-effect of the test itself; and perhaps 10% more as the result of training on the tachistoscope ; none from the other tests or training. Rt.'s gain is largely due to increased facility in recognizing small words and suffixes known not to contain and passing them by; there is no introspective evidence upon which to explain Wf.'s gain, but it is probably due, in part, to change in process. Those reagents who changed processes in the final test show loss more often than gain. (4) Card-Sorting Card-sorting, a series of reactions with the mental processes of discrimina- tion of the stimulus and the choice of the appropriate movement interpolated between stimulus and reaction, has been used to determine the influence of mental work upon rate of tapping (Dresslar^), to learn the conditions of mental activity (Bergstrom^), to investigate the influence of interference of associations upon memory (Bergstrom^) and upon the practice-effect in form- ing associations (Bergstrom'' and Brown **) to determine the effect of mental type on interference of motor habits (McMein and Washburn"), to test mental ability (Bagley,' Burt'), to test motor ability (Thompson*), and to study the learning process in relation to transference and interference (Kline and Owens'). In the last research ordinary playing cards were sorted into 52 compartments, but usually the stimuli consist of letters (Bergstrom^), non- sense syllables (McMein and Washburn), words (Bergstrom^) pictures (Berg- strom*), or colors (Burt, Thompson, our own experiment on pp. 5off.) ; the number of compartments, 4 (McMein and Washburn, Thompson), 5 (Burt), 6 (McMein and Washburn, and our own experiment, pp. soff.), 8 (AlcMein and Washburn), 10 (Bergstrom), 12 (McMein and Washburn); the packs con- tain 10 cards of each stimulus, except those of Bergstrom which contained 80 cards, and our own (see pp. 75 f.) which contained 50. * Dresslar : Some influences which affect rapidity of voluntary movements. Am. Jr. Psych. 1892. 4:5i4ff. * Bergstrom: Experimental study of some of the conditions of mental activ- ity. Am. Jr. Psych. 1893-4. 6:247. * Bergstrom : Experiments upon physiological memory by means of the in- terference of associations. Am. Jr. Psych. 1892-3. 5 :256ff. * Bergstrom: Relation of the interference to the practice effect of an asso- ciation. Am. Jr. Psych. 1893-4. 6:433ff. " Brown : Habit interference in sorting cards. Univ. Calif. Pub. in Psych. 1914. Vol. I, No. 4. ° McMein and Washburn : Effect of mental type on the interference of motor habits. Am. Jr. Psych. 1909. 20:282ff. •Bagley: On the correlation of mental and motor ability in school children. Am. Jr. Psych. 1900-1. 12:195. FORMAL DISCIPLINE II9 Reaction with discrimination and choice is probably more vari- able in its processes, in the work of the same reagent as well as in the work of different reagents, than is simple reaction, since there is more psychial process interpolated between the stimulus and the movement, and variability in reaction experi- ments varies directly with the amount of interpolated process.*^ But the variability may be such that two scores still belong to the same kind of act as is not the case in simple reaction when one of them is of the muscular and the other is of the sensorial type. Yet even here the scores may also be incomparable because the acts are different in kind. Each pack of fifty cards in this experiment was made up of six symbols shown in the accompany- ing plate of the compartments accord- ing to which arrangement they were distributed (see Appendix B. Figs, i, 2, p. 288). In the cognition of the smybol, the diameter was seen by Cr. and Wf. as classifying the cards into vertical, horizontal, and oblique, pairs; by others as merely forming the base of a pointer. The radius was accepted as a pointer by all, except Ly., who named the location of the filled half of the circle, and He., who named the location of the blank half, and was interpreted as pointing right or in, left or in, up or out, down or out, down-oblique or slant down, and up- oblique or slant up; Ly. and He. used the same terms of direc- tion. The cognition of the symbol was especially difficult for Ly. These terms of direction served to locate the three pairs on the cabinet, in the scheme shown above by the connecting lines, ' Burt : Experimental tests on general intelligence. Br. Jr. Psych. 1909. 3:136. * Thompson: Psychological norms in men and women. Univ. Chicago Contrib. to Phil. 1903. 4: No. i. 15. ' Kline and Owens : Preliminary report of a study in the learning process, involving feeling tone, transference, and interference. Psych. Rev. 1913. 20:2o6ff, ^ Vid. Kiilpe: Outlines, 422; Oehrn : Psychol. Arbeiten, 1:131-2; Alechsieff: Phil. Stud. 16:24. 120 JOHN EDGAR COOVER for most of the reagents (Mn. Le. Rt. Ly. He. Cr. Wf. Dn.) before the first test was over; SI., Ms., and Fr., differed, in that the latter two had formed no scheme in the first test, and the former merely formed a mnemonic device to hold the two rows ; as, upper row has two radii East and one North, lower row has two West and one South (map directions), without further placing of either pairs or single cards. Individual variation in the process of distributing was con- siderable.^^ As is indicated above there were some marked dif- ferences in the main outlines of recognizing the symbols and of forming a scheme of the compartments. But even where the main outlines were similar, great variation obtained in the detail of the development in the course of the tests, so that, strictly speaking, no two averages are measurements of precisely the same mental processes. This variation consists in (a) Differences in the predominant imagery (verbal, visual, kinaesthetic) used in cognition of the symbol and in locating its compartment; in consequent variation in the number of steps in placing a single card (as, i — visual impression of the card, 2' — verbal image accompanying its cognition, 3 — visual or verbal image of the compartment, 4 — kinaesthetic image of movement of arm to that compartent, 5 — visual impression of the compart- ment, 6 — impulse of movement; or, merely steps i, 3 and 6). (b) Varying dependency of the imagery used as a cue for the movement, upon a definite memorial scheme of the compart- ments, or upon random memorial elements of preceding sortings. (c) Relative adaptability of the scheme adopted (Sl.'s was particularly unwieldy). (d) Facility in perfecting the scheme. (e) Facility in superseding it with automatic coordinations. On account of these individual variations in the process of distributing, at any given point in the tests, not only did the schemes differ in detail, between the individual reagents, but they varied in degree of completion, resulting in varying amounts of primitive and mechanical matching of cards and labels, or in varying amounts of spontaneous sorting from perfected or '°C./. Individual variation reported by McMein and Washburn (op. cit p. 283). FORMAL DISCIPLINE 121 from automatic coordinations. Some reagents (Mn. Dn. He. Cr. Wf.) developed a scheme early, and had the process partly automatic before the fourth pack of the first test was sorted; others (SI. Ms.) did not get a scheme well developed at all, or only at the end of the first test (Fr.). Many other individual variations in processes could be noticed; such as, (a) effect of sequence upon placing a card — one reagent prefers the succeeding card to belong to the same pair (Cr.); another prefers any other (Rt.) ; (b) relative preference for the three pairs — to one the oblique is easiest (Rt.), to another it is the hardest (Ly.), etc. Besides variation in the process, there were variations among the influences upon it : (a) Emotional — The sorting was vexing and disliked (Mn.), or disagreeable (He.), caused nausea and trembling (Ly.), or was interesting (Le., Wf.), or agreeable (Cr. SL), or indif- ferent (Rt.), and each effect contributed toward the attitude with which the reagent came to the experiment. (b) The anxiety for speed varied. (c) Physical causes — such as, "sticking of cards" (Mn.), or stiffness of cold hands (Wf.). (d) Conditions of health — Ly., He., Cr., were fatigued for their final test, and Wf. was nervous; and Rt. and Al. were less alert than usual. Since it is our interest to compare averages of tests separated by a long interval of training or of rest, still other variations are of great importance : Those resulting from the different degrees of reproduction, in the final test, of the schemes or of the coordinations formed in the first test. Some of the reagents (Ly., Al., Fr., Wf.) had them well in mind and could use them early in the final test; others (Le., Rt., SI., He., Cr.) had practically to begin anew, recovering the effects of their former experience in varying degrees and at varying points in the final tests; and with both classes there was variation in respect to further development, particularly on the /part of the latter, whose developments occasionally conflicted with returning de- tails of earlier processes. The combined practice of both tests 122 JOHN EDGAR COOVER was not sufficient to lead to automatic coordinations to any considerable degree, and all averages include practice-effect. The development of processes through practice in the tests may be illustrated by the following record from the in- trospections of Cr. : First Test Pack I. Matched cards with labels of the c^)mpartments continually. No compartments placed; pairs of symbols becoming distinguished. Pack 2. Some cards distributed from memory of position from distributions recently made; some from a developing scheme of the compartments, which is now drawn from memory (correctly) with some hesitation: The compartments are grouped in pairs (according to the dotted lines in the plate a few pages back) determined by the direction of the diameters: Pair I. (positions i and 6) Are verticles and occupy the opposite corners; radii in. Pair 2. (positions 2 and 4) Horizontals; oblique pair to the left; radii out. Pair 3, (positions 3 and 5) Obliques; oblique pair to the right; radii difficult, but such that I may classify "middles up" (referring to positions 2 and 5). Continuity of attention to the process and deliberately hold- ing the parts of the developing scheme constituted a persevering 'anlage.' Pack. 3. Worked the scheme in about tenth the time; sorted some by exclusion; e.g., horizontals belong to compartments 2 and 4; have a horizontal in hand, match with 4 and throw it into 2 without looking at the latter's label ; but the process is almost wholly matching. Pack 4. Scheme comes easier; though instead of working- outright without matching, it simply facilitates matching. About an eighth of the time it works adequately alone. Final introspection. Chief hindrance to the sorting is losing calm control of the partial application of the scheme and of its extension. This does not seem to depend so much upon the intensity of the attention (which the process seems to keep high) as upon its distribution so that the memorial factors are kept sufficiently in process, which is very difficult; matching is easier. Some strain, mental and in left hand. Rather .agreeable. (3 errors in the 200) , FORMAL DISCIPLINE 123 Final Test Pack I. Recognize symbols as familiar, but had to work up scheme of cabinet; some old classification returns, but new organization begins to supplement it: Pair i (positions i, 6) "In"; Pair 2 (positions 2, 4) "Up" and "down"; Pair 3. Contrary, "Down" and "up." But must glance at the cabinet continually to assure myself of correctness of compartment. Pack 2. Scheme did not develop so satisfactorily as it prom- ised; for at first the process did not require glancing at the case; then had to glance to avoid having to wait on memory, which resulted in confusion to the memorial element and a resort to matching. (Best principle would have been to depend upon memory, even at occasional loss of time, thus avoiding confusion and retaining scheme). Application of scheme isn't simple : sometimes mechanical conception of direction, some- times incipient naming according to scheme, sometimes simply matching. (The change in Pair 2. from "Outs" to "up" and "down" resulted in erroneous application of "in" to them, causing 2 errors. ) Pack 3. (Tired; have had a wearing day.) Scheme in early part of series took on the following simplification: Pair i. (position I, 4) "in"; Pair 2. (positions 2, ^) "up"; Pair 3. (positions 3, 6) "down." But after working a short time it failed, and I was compelled to resort to matching. (Probably because of conflict with the old pairs, and lack of energ}^ in holding memorial elements firm). Pack 4. Scheme worked better; I depended upon it. Feel that I could make great headwaj^ in the next few packs, for the scheme is not fully mastered, but during this pack it de- veloped rapidly. Instead of the old incipient pronunciation of the class-word, the cards are placed by a more facile cue: kinaesthetic imagery of the meaning of the class-word. No consciousness of any influence of training (on memory schemes). In spite of all the degrees of variation between the dif- ferent reagents, the general experience was similar — reaction with discrimination and choice — and, since almost all reagents developed or used the same methods in the final test that they developed in the first, their results may, with reservations made obvious by the above discussion, be compared. The results of Regular 1st Control 2d Control He. 23 Ms. ID Dn. 13 Le. 12 Mn. 22 Rt. 15 Ly. 21 Cr. 22 Al. 27 SI. 4 124 JOHN EDGAR COOVER two reagents are disregarded: those of Wf. and Fr., control reagents because of Wf.'s "mental practice" on the test, and Fr.'s review of the compartments and the perfecting of a scheme, between tests, when they should have had no practice.^" Initial efficiencies ranged from 79.0-105.4 seconds per 50 reactions, and they fall into four groups around averages of 80, 87, 95, and 105 sec. The results of the reagents who maintained and developed their old methods, in per cent of decrease in time, are : Group I Group 2. Group 3. Group 4. The more important variations affecting these figures are*. Le. took her last half of the final test 14 days after the first half, which deprives it of the practice-effect shown by the other reagents who took their's two days after. The interval between the first and final tests was much shorter for Al, which no doubt contributed somewhat to his improvement; on the other hand he was less alert than in the first test. Rt. complained of a "muddy" attention in the final. Sl.'s scheme (map directions) was so unwieldy that in his final test his efficiency fell back to its initial position and he had to cover again the ground he had covered in his first test; this is one illustration of the fact that greater "room for improvement" cannot always be pre- dicated upon low initial capacity (his was the lowest in the series). ^*^° From the table it appears probable that 10% improvement may be expected in this test as a result of practice-effect upon itself ;^^ that 10% more was contributed by the training on other material. ** Mental practice was shown by Johnson (Experiments on Motor Education. Studies from the Yale Psych. Lab. 1902. 10:87) to facilitate simple reaction to sound. *"' On this matter, see footnote 139, pp. 222-3. "This supposition is supported by the evidence afforded by the former FORMAL DISCIPLINE 125 The principal factors of improvement seem to be : (a) A more automatic use of the memorial elements of the old method from the beginning (Al., Ly.) or soon after the beginning (Mn., Le., Rt, He., Cr.) of the final test. (b) Some further development of the old scheme (Al., Cr.), although in this there was conflict between the new and old elements. (c) A readier apprehension of the symbol (Ly., for whom it had been particularly difficult. ) The manner in which training affected the test is a matter of hypothesis and might be offered as follows : The training on the tachistoscope heightened sensitivity for visual impressions, applicable to cognition of the cards; all the training in unequal degrees heightened reproductivity of imagery, applicable to the memorial elements of the scheme; the tachistoscopic training and the learning of 12-letter-rect- angles exercised coordination of part-processes, applicable to coordinating discrimination and reaction; the training on the memory schemes exercised the continuous attention and the reproduction of memorial elements, demanded in this test, and all training involved habituation to distraction which would be applicable here. A few applications of former experience may be noted as ex- amples of transference and spread of training: Mn. and Wf. were students of higher mathematics; they were assisted im- mediately to the formation of a scheme by reason of the per- ceived relations of the radii; in contrast to this, Le. did not notice until the fourth pack of her first test that the radius formed a right-angle with the diameter on the oblique pair of symbols, which relation was then of assistance to her. SI. brought his unwieldly scheme from the class-room. training in Card-Sorting (see Table XXV in Appendix A) ; in which, al- though it was probably benefited by preceding training in Typewriter-Reaction, the practice-effect of the first four packs upon the second four was, in per cent of improvement, CI. o Al. 17 Cr. 12 Bs. 5 The second four packs were distributed the next day after the first four by CI. and AL, three days after by Cr., and in the same hour by Bs. AL, Cr., and Bs., are AL, Cr., and Ms. of the present experiment. 126 JOHN EDGAR COOVER (5) Typewriter Reaction Typewriter-reaction (see p. 76) engages the same essential processes as card-sorting, its chief difference lying in more apparatus and in the necessary changes in forms of stimuli and movements. This form of reaction was used by Bair^ in the study of the practice curve and interference, and in our own experiment on "Reaction with Discrimination and Choice" (pp. 5off) ; it was selected as a type of mental work by Seashore^ for the convenient measure- ment of which he devised a special apparatus — The Psychergograph. This test involved but four coordinations of stimuli and reac- tions, and thus permitted a more rapid reaction, a slower practice-effect,^^ and a greater degree of automatization, by the end of the final test, than was possible in the card-sorting test. The order of the letters, from left to right, was a-t-e-n. The cognition of the letter as it appeared was not merely a per- ception of the letter, but an apperception of it as belonging to a certain position, to a certain key, or to a certain finger. This process varied considerably in its dominant imagery: (a) Visual purely (Rt., He., Cr,). (b) Sound image i) of the phonetic value, or 2) of the name of the letter. (c) Kinaesthetic image of the letter's sound or name (Le. Ly. Wf.) ; or of the movement of the eye toward its position (Wf.). (d) Combination of the above, usually accompanied by actual incipient movement of the vocal organs in pronouncing the sound or the name of the letter, or movement of the eye or arm and fingers in locating the letter. The placing of the letter necessary for carrying out its appropriate reaction usually involved a mechanism of some sort serviceable for keeping the order of the letters clearly in mind and the reaction to each ready. This also varied greatly : (a) Continual repetition, in inner speech, of the letters in ^ Bair : The practice curve: A study in the formation of habits. Psych. Rev. Mon., 1902. 5 : No. 2. ^Seashore: A method of measuring mental work: The psychergograph. Univ. of Iowa Studies in Psych. 1902. 3:1-17. **The average time of the initial 50 reactions in card-sorting was 106 seconds; in the typewriter-reaction, 50 sec. The average practice-effect in the first test in the former was 32%; in the latter, 14%. FORMAL DISCIPLINE 127 their order (SI. Mn.), sometimes assisted by a mnemonic device, as "Aetna" (Ly.). (b) Classification of the letters, to hold their positions without repeating- them in their order : * i) Left, a-t; right, e-n (Rt). 2) I, 2, 3, 4; vowels odd; consonants even (Mn.) or vowels left, consonants, right, and alphabetic extremes on the left hand (Wf.). 3) Terminals a — n; t belongs to a; e to n (Le.). 4) Alphabetic order of a-t, of e-n, and of a-e (Wf.). The position referred to the keys (Le., Rt., SI., Ly., Wf., Dr.) or to the fingers (He., Cr., Wf. 2) ; in case it referred to the former, the finger was sometimes removed and the key glanced at to verify a judgment or to restore the order of the letters to the memory (Ly., SI., Wf.). Besides the foregoing causes for variation in the results, there were some others peculiar to the task : (a) Anticipation caused premonitory reactions which re- sulted in errors and confusion (Le., Ly., He., Sn.) ; the same eftect is caused by rapid rhythm when the process has become largely automatic (He. Cr. ). (b) Those who formed coordinations with the keys, often had difficulty in the control of their fingers (Le., Ly., Wf., Dr.) ; the difficulty also occurred when the scheme paired the letters into right and left groups and operated through kinaesthetic imagery in the arms, while the order of the letters in each group was determined by visual imagery (Rt.). (c) The mental tension involved, sometimes produced dizziness and nausea (Ly.). (d) The action and noise of the typewriter is so different from the type-bar machines that it was very distracting to some of the reagents who were typists (Rt., SI., He.). As a result of the influences of these factors of variation, the task was performed in many different ways by the respective reagents; and the essential method sometimes changed during the tests of the individual reagents. The former gave rise to, and may be said to be indicated by, great variation in initial ability, ranging from 36.4-67.7 seconds. 128 JOHN EDGAR COOVER which shows a preference for three types : one at 40 seconds per 50 reactions (Mn., Rt, He., Cr., Ms.); the second at 55 seconds (He., Wf., Sn., Dn.) ; and the third at 65 (SI., Ly.). If the chang-e in the individual reagent's work occurs in an early series of his first test, it lowers his average for that test and decreases the difference between the results of the two tests; if it occurs early in his final test or late in the first test it operates in the opposite direction. The change is sufficiently radical in the case of two reagents to throw out their results: Mn. changed from repetition of the series of letters, in the beginning of her final test, to the use of a classification; Wf., at the beginning of his final test, changed from "verbal image — key" to "verbal image — finger," made a radical change in method, and just before taking the final test took up practice in typewriting. The improvement, in per cent of decrease in time, made by the other reagents was : Group I. Group 2. Group 3. Since the reagents in groups 2 and 3 did not attain in their practice in the two tests an efficiency equal to that of the reagents in Group i at their beginning, inspection of results will have to be limited largely to the latter. As to the practice-effect of the test upon itself for Group 2, it must be about 14% ; we have, unfprtunately, no trained reagents whose results can be used in comparison to test training- effect. For Group i, however, we are able to make an esti- mate. From the data of the old investigation (see Appendix A, Tables XXI p. 275 and XXIII p. 276). It is calculated that continuous practice in work identical with this test, for reagents whose initial efficiencies place them in Group i, resulted in a loss of y% (CI.) and 10% (Cr.), and a gain of 0.4% Regular 1st Control 2d Control Le. 14.6 Rt. 12.4 He. -6.1 Ms. -07 Cr. 4-9 Sn. 13.9 Dr. 13.6 SI. 10.4 Ly. 4-7 FORMAL DISCIPLINE 129 (Mn.) ; and the effect of a free interval^^ of 45 days, after practice on 36 series, was a gain of 4.4% (Mn.). Since our practice in the first test amounted to but 4 series, and the interval was twice as long, it is possible that Le. Rt. and Cr. were benefited from 4% to 10% by their training. This benefit could be described, however, only in general terms: (a) The coordinations became more nearly equal in strength, as is shown by decrease of the smaller variations; (b) Steadiness of attention increased, as is shown by decrease in the number of 'balks,' (except for Cr., for whom there was increased distraction in operating the recording apparatus) ; and (c) Control of the memorial element increased, as is shown by the decrease in the length of the 'balks.' (a) is shared by Mn., SI., and Ly. ; (b) by SI.; and (c) by none of the other reagents. The training of the tachistoscope may have con- tributed higher sensitivity for the visual stimuli and better control in distributing the attention over part-processes; and the training on memory schemes may have contributed some- thing toward better control of the memorial element. What distinguishes the three types, as intimated above, are differences in the performance of the task: Group i begins with a fairly direct coordination of either visual impression, or the sound or kinaesthetic image into which it is converted, and the key or the finger; and reactions early begin to become more or less automatic (Rt., He., Cr., Ms.). The first effect of practice is likely to be loss in efficiency, as recorded in a preced- ing paragraph, due to evolving a scheme, more or less simple, which is to be instrumental, if practice is continued long enough, in effecting automatic coordination of visual impression and im- pulse of finger. Rt. began with (i) visual impression, (2) Kinaesthetic image of movement in the right or left arm, (3) visual image of key, (4) impulse of finger, and dropped (2) and (3). Cr. began with (i) visual impression, (2) verbal image, (3) impulse of finger, and dropped (2). Schemes were **The effect of the 80-day interval upon all our ireagents was an average loss between the last series of the first test and the first series of the final test, of 6% of the initial capacity. I30 JOHN EDGAR CO OVER very simple, relating the stimulus to kinaesthesis of arm or fingers; for Cr. this consisted in 'feeling' the letter in the finger. There was no alphabetic, logical, or phonetic classification. Group 2 begins with a more indirect coordination of stimulus and reaction, by interpolating retarding accompaniments of cognition and a more or less elaborate or complex classification of the letters, or an inefficient memorial representation of them, so that the reacting process is circuitous. Wf . began with ( i ) visual impression, (2) verbal image, (3) kinaesthetic image of eye-movement to key, (4) impulse of finger, and dropped (3). Schemes were complex, alphabetic, phonetic, and logical. Group 3 adds to these interpolations some especially inefficient process, probably usually of method. SI. and Ly. matched letter and key, for which the reacting finger had to be removed and returned before the movement could be carried out. The process is divided into two distinct acts : cognition of letters and searching for keys. The elements constituting the process of any reagent are at first selected from his stock of experience, and then are changed in accordance with the reagent's adaptability. No doubt Ly., who was not especially slow in card-sorting, came to the test with a strong sensory set of consciousness in accordance with her training in reaction to sound, for she complained that she gave her chief attention to the stimulus and could not get it directed upon the keys; the location of the key was either observed or remembered after its use was called for. Improve- ment was made, and many of the reactions were later carried out without involving a glance at the key, but the choice of the reaction remained a more or less distinct act following the discrimination of the stimulus. Although the course of practice varied with the type, the general efi;ect involved (a) the reduction of the interpolated process, (b) the simplification and finally transcendence of the scheme, (c) increase in partly or fully automatic reactions. The end cf continued practice would be the automatic coordina- FORMAL DISCIPLINE 131 tion of visual impression and finger impulse, into which all types would merge. .*»• {6) Controlled Reaction The Controlled Reaction involves further interpolated mental process be- tween stimulus and reaction, namely the referring of the stimulus to one of the classes which determines the choice of the reacting movement. It has been used in Jastrow's^ laboratory under the name of "Classification Time"; 10 monosyllablic words of each of the three grammatic classes, nouns, verbs, and adjectives, were used in irregular order as stimuli. Our test (see p. yy) follows Miinsterberg^ in using as stimuli the names of poets, philosophers, statesmen, scientists, and musicians. Both of these researches, however, used the cbronoscope for measuring the individual reactions, while we used the stop-watch (except in the case of Cr. who reacted as rapidly as possible to the successive stimuli). In this reaction there were five coordinations between class- name and fingers, but the class-name could be got only by classifying the stimulus; as, Newton, "scientist — 4th finger"; Mozart, "musician — fifth finger" ; etc. The variability of processes may be indicated by the following classification of methods : (a) The auditory impression of the name may be followed by a kinaesthetic image of the movement in the appropriate finger for the class to which the name belongs; and, as a result of practice, by a kinaesthetic impression from incipient move- ment in the finger, and later by the impulse of the finger in reaction. Attention is mainly upon classifying the name, and, at first, upon reviewing the class-finger coordinations. The coordination may be effected by conferring the professional air, disposition, or attitude of the class upon the finger (Cr.). (b) Or the auditory impression of the name may be followed by the pronunciation of the class-name, in the efifort of classifi- cation; then a judgment determining the class; then a choice of the appropriate finger; then the impulse of the finger (Ly.). (c) The coordinations may be more or less strong so that the attention may be given to the classification (Mn., Le., SI., He., Le., Re.) ; or the coordinations may be weak and demand most of the attention (Rt., He., Ms.) ; or both may be equally ^Jastrow, Morehouse, & Harper: Classification Time. Am. Jr. Psych. 1891. 4:411-415. * Miinsterberg : Beitrage fiir experimentelle Psychologic. 1889. Heft i :85. 132 JOHN EDGAR COOVER difficult so that the attention oscillates between more or less separate acts (Ly.); or both may be equally easy so that the attention by quick oscillations can carry them both along at the same time (Cr.). (d) The coordination may involve a scheme of the classes on the fingers (Le.) or in the positions the fingers occupy (Rt. ), or the classes may be simply held in sequential order (Wf.). The interval between series, or between reactions, may be used in reviewing the coordinations (Wf., He., Ey.), or in keeping the mind clear, in order to avoid anticipation of classes (Ey.)- Some further intimation of the variability of the processes, not only between different reagents, but with the same reagent, (as attention must oscillate from one weak part to another of the complex process) is given by the nature of the errors. They appear to have been owing to at least four principal causes : (a) False classification through their ignorance or confu- sion (Mn., Ly., He., SI., Ey.). (b) Coordination may be weak and the wrong finger be unconsciously used (Mn., SI., Ly., Ms.), or confused with another (Wf.). (c) Control of the fingers may be at fault; especially likely with the corresponding fingers of the two hands (Le., Rt., Ms., Ey.). (d) Automatic reactions may take place before classification is made, sometimes following the pronunciation of a class-name in trying to classify the man (Le.), sometimes as a result of strong expectation for a certain class (Ey.), or of a rhythm too fast to control (Mn., Ms.). (e) A few errors occurred through misunderstanding the name, as, Hayden for Hegel, Coe for Foe, Verdi for Virgil, (SI., Ms.). The greatest number of errors were made by those who put their attention mostly upon classification Among the reagents there was considerable variation in initial capacity; it ranged from 1.34-2.44 seconds per reaction, for 50 reactions, and falls into three groups: 1.4 (Mn., Ms., Wf.) ; 1.7 (Rt, SI, Ly., He., Cr., Ey.) ; 2.1 (Le., Re.). Analysis of results and examination of introspections, how- FORMAL DISCIPLINE i33 ever, do not reveal any simple characterization for the respective groups. That they are valid is attested by the 'reduced' time (total time minus free reaction, or cognition time), which would change but two reagents, by putting Rt. in group i and Ly. between Groups i and 2.^* The improvement in per cent of decrease in time was : Trained ist Control 2d Control Group I. Mn. 19 Ms. -7.0 Wf. 1.8 Group 2. Cr. 14 Ey. 6.7 He. ii.S Rt. II SI. 1.5 Ly. -5.6 Group 3. Le 26 Re. 20.8 Mn. and Wf. gained slightly, and Ey. considerably (7-16) in errors ; the latter owing to her classification of unknown men as scientists, the least known class. If the processes in the respective groups can be considered equivalent, and equally susceptible to improvement, some ad- vantage of training may be presumed from the tachistoscopic, and memory, training and from learning 12-letter-rectangles. In what way the training may be presumed to mfiuence the test is a matter of conjecture : The training on the tachisto- scope and learning 12-letter-rectangles demanded keen attention upon signal for a short time, and required its distribution over the part-processes involved; the former, and also the latter in the case of Rt., heightened reproductivity of imagery. The memory training may have contributed to steadiness in keeping the memorial element of the process ready to the end of forming automatic coordinations between class and finger. Improvement consisted in abridgement of the interpolated processes between stimulus and reaction. Some of the specific elements in this were: (a) reduction of the pronunciation of the class-word to a kinaesthetic image, or at least to incipient »*ThIs is due to the fact that they took a disproportionate time for the mechanical and perceptive part of the reaction; the average was .6 or .7 sec, and they took .97 and .99, respectively. 134 JOHN EDGAR COOVER movement (Mn.), or even to the professional air of the class the attitude of which is coordinated with the finger (Cr.) ; (b) effecting coordination between class and finger by use of a scheme; Rt. put poets and musicians on the ends, statesmen in the middle, and philosophers and scientists in the intervening spaces, holding the scheme visually; then the scheme became more kinaesthetic and less conscious; (c) the habit of using the intervals between series for reviewing the coordina- tions so as to be about equally prepared for all classes (He.), or particularly for those which have not been recently reacted to (Wf.); or for keeping the mind clear so as not to be in- fluenced by expectation (Ey.) ; (d) attention steadied over the whole process so as to keep it under better control (He., Cr.). Those who fared worse in the final test than the rest gave their attention to the classification mostly, which left the coor- dinations relatively free from practice-effect. Ms. lost some on all the classes, but most on poets, scientists and musicians; SI. and Ly. made gains on poets, scientists and musicians, but lost disproportionately on philosophers and statesmen, that is, their coordinations became more unequal in strength than they had been in the first test. None gained equally on all classes, and the greatest inequali- ties of gain were made by Le. who reduced her time almost wholly upon philosophers and scientists, classes that had taken on the average more than twice the time required for the other classes. But the effect of the interval on all those who gained much was to even up the strength of the various coordinations. (7) Sound Discrimination Sensible discrimination ranks with reaction time as a device or procedure in the psychological laboratory for the study of mental processes. Kiilpe^ notes its high dependence upon attention, Titchener* suggests it as a gross measure of attention, and Spearman* concluded, from correlational data, that it is most closely related to general intelligence. In the laboratory it has been used in the study of memory of lengths of lines (Hegelmaier*), of memory and recognition of tones (pitch) (Wolfe," Angell and Harwood,* Angeir), of intensities of sounds (Lehmann,' Angell'), and of shades of gray (Lehmann," Angell") ; and for the analytical study of mental imagery (Bent- FORMAL DISCIPLINE I35 ley," Whipple'') . It has heen used as a mental test in determining power to discriminate differences in pitch (Gilbert," Seashore," Wissler," Thomp- son," Spearman,' Krueger and Spearman," Burt"), shades of gray, (Thomp- son" and our own test on pp. 42ff.) and color (Bennett""). The test is standardized in Whipple's manual,"' which gives an account of its use and its results. Since qualitative variability in performance may depend in its con- crete aspects upon the nature of the differences to be discriminated, it must be remembered that our test (see pp. 77f-) involves differences in intensity. Since an analysis of the various processes of this test has .already been made in a former experiment on Discrimination,^^ and a characterization of individual performances is more 'Kiilpe: Outlines of Psychology. 1901. 38,4^9. ^ Titchener : Lectures on the elementary psychology of feeling and attention. 1908. 279. ^Spearman: "General Intelligence" objectively determmed and measured. Am. Jr. Psych. 1904. 15 :279. ^ , , „ *Hegelmaier, in Vierordt's laboratory about 1852 (Quoted by Bergstrom: Am. Jr. Psych. 18:211). "Wolfe: Untersuchungen iiber das Tongedachtniss. Phil. Stud. 1880. ' Angell and Harwood : Experiments on discrimination of clangs for dif- ferent ''intervals of time. I. Am. Jr. Psych. 1899-1900. ii:67ff. 'Angell: Discrimination of clangs for different intervals of time. II. Am. Jr. Psych. 1900-1901. I2:58ff. ' Lehmann : Kritische und experimenteile Studien iiber das Wiedererkennen. Phil. Stud. 1892. 7:204ff. . "Angell: On Judgments of "Like" in discrimination experiments. Am. Jr. Psych. 1907. 356ff. "* Lehmann : Ueber Wiedererkennen. Phil. Stud. 1889. 5 :96ff • "Angell: Discrimination of shades of gray for different intervals of time. Phil. Stud. 1902. 19:1-21. ,. . .J ,. . T r. u o ^ Bentley : Memory image and its qualitative fidelity. Am. Jr. Psych. 1899. 11:1-48. . . '^ Whipple : An analytical study of the memory image and the process ot judgment in the discrimination of clangs and tones. Am. Jr. Psych. 1901. I2:409ff, and 1902. I3:2i9ff. _ r v 1 1 -u " Gilbert : Experiments on the musical sensitiveness of school children. Studies from Yale Psych. Lab. 1893. i :8o-87. • , tt • c " Seashore : Hearing ability and discriminative abdity for pitch. Univ. ot Iowa Studies in Psych. 1899. 2:55-64. , . , -o , t^ Tv,r '"Wissler: Correlation of mental and physical tests. Psych. Rev. Mon. 1901. 3: No. 6. "Thompson: Psychological norms m men and women: Univ. Uhicago Contrib. to Phil. 1903. 4: No. i. Pp. 72, 81. " Krueger and Spearman : Die Korrelation zwischen verschiedenen geistigen Leistungs^fahigkeiten. Zeits. f. Psych. 1907. ,44:87. ^ ^ ^ , "Burt: Experimental tests of general intelligence. Br. Jr. Psych. 1909. 3 :98, 123. ^Bennett: Formal discipline. 1907. P- 59- ^. , • . =^ Whipple: Manual of mental and physical tests. 1910. Of brightness, IS9; of pitch, 180. " Pp. 47f. 136 JOHN EDGAR COOVER minutely made in a following experiment in which sensible dis- crimination constituted the training,^^ only sufficient analysis is made here to indicate the fact of variation in processes. (a) The sounds may be naively compared as external events (Mn., Vg.) sometimes accompanied by visual imagery of the experimenter's production of the sound; as, imaging him striking the desk with a pencil; and the weaker sound may be imaged as coming from a nearer source. (b) The image of the first may be compared with the sensa- tion of the second (Mn., Rt., He.). (c) The effect of the sounds on the body may be compared; as, the placing of the kinaesthetic auditory image higher or lower in the head (Le. ), or the blinking of the eyes (Ly. ). (d) The reactions to the sounds may be compared; as, move- ments of the hands, head, throat and tongue (Ly), nodding the head (Mn.), or breathing out forcibly like an axeman when striking the blow (He.).^^ (e) Or there may be no imagery, and the effects in the attitude compared (Rt., Wf.). (f ) And the sounds may be represented by verbal classification (Ly., Ms.). Certain method may be employed ; as, holding the breath and closing the eyes (Le.), holding the ear in a certain position to catch the sound ( SI. ) , making allowance for the second sound seeming relatively louder because of its getting closer attention (He.), or because it is accented by an iambic rhythm (Wr.). The process further varied because of certain distractions; as, external noises (noted by almost all reagents), distracting thought (Wf.), difference in quality of sounds (He.), expec- tation (Rt., Le.). (Many noticed the difference in quality of loud and weak sounds as appearing high or low in pitch, metallic or wooden, but did not consider it a distraction except in judging "like"— Mn., Rt, Ly., He., Ms.). ^ Pp. i99ff., 2o6ff. '^Kuhlmann (On the Analysis of Auditory Memory Consciousness. Am. Jr. Psych. 1909. 20: I94ff.) found the motor processes used in imitating the sounds to be the most frequent factor in the recall of details of sounds of familiar things. FORMAL DISCIPLINE 137 That the essential process itself varies with the individual reagent is indicated by introspective notes that the norm varied in intensity (Rt, Ms.). Some variation undoubtedly occurred by reason of un- familiarity with the (customary) symbols used in recording- the judgments (Le., Ms.), {vid. Appendix B. Fig. 8, p. 290). Although the processes varied with each individual, there is not sufficient introspective evidence at hand to indicate any radical change between the final and the first tests. Initial capacity ranged from 43%-73% of right judgments, and may be distributed in three groups :^^ 70% (He., Wf., SI.), 57% (Mn., Le., Rt., Ms.), and 45% (Ly-, Vg., Wr.). Improvement made in per cent of initial capacity was : Regular ist Control 2d Control Group I. *S1. 18 65 51 Wf. -4 -6 -46 He. 000 Group 2. Mn. 41 50 50 *Ms. 5 22 46 Le. -ID -5-58 *Rt. 9 14 35 Group 3. Ly. 2^ 25 26 Vg. 30 35 Z7 Wr. 25 21 21 * Foot-note 98 indicates that these reagents fall into lower classes, for the. 2d and 3d columns. The first column of per cents refers to judgments on all four intervals above and below the value of the norm, where D = o; the second column, to judg- ments on the first two intervals above and below the value of the norm only ; the third column to change in the "Difference Limen." °^ The initial capacity to judge differences between the stimuli of the first two intervals above and below the value of the norm, also varies, from 29% to 58% R cases, and would also classify the reagents into three groups around' the averages of 53%, 43%, and 34%. But it would change SI. to the 2d group, and Ms. and Rt. to the 3d group. Also the "Difference Limen" (that point between o and 4 intervals from the value of the norm at which 50% of the judgments are right) of the reagents ranges from 0.97 to 3.20; and classifies the reagents into three groups just as capacity of discrimination on the two intervals above and below does ; except that it places Ms. between Groups 2 and 3. (These three groups : 1.2, 1.7, 3-) 138 JOHN EDGAR COOVER About Group i nothing can be said except that initial capacity- was remarkably high and perhaps no improvement could be expected. Group 2 shows several consistent gains: Mn. in the three columns and SI. in the 2d and 3d columns. Abstracting from the results of Group 3, where gain should be relatively greater, Mn. and SI. seem to have been benefited by their train- ing. Mn. was exercised on minimal stimuli in the field of vision, and may be expected to show some profit from it in discrimina- tion; and both Mn. and SI. may have applied some training- effect in better attention for the moment, and in better reproduction of imagery, Le.'s results are peculiarly clouded. They do not show the improvement she made in discrimination, for, although half of her series in the first test yielded 4 R cases each, and half of her series in the final test yielded 6 R cases each, for some reason which her introspections fail to explain, a fifth of the latter yielded but 2 R cases each. It is possible that the erasures in her record indicate confusion of symbols; in which case her irregular results are accounted for. (8) Memory of Sounds The memory span, which is found by noting the longest series of digits or letters perfectly reproduced from a single presentation, usually auditory, is noted by Pillsbury^ as one of the customary measures of attention, and Kiilpe says <^hat "attention produces its maximal effect in the reproductory sphere."* Our tests in memory of serially presented members differ from the memory- span test in measuring the average number of members reproduced from a single presentation of a series too long to be completely retained (Methode der behaltenen Glieder, see p. 142). This test of memory of sounds (see p. 78) follows Fracker," and involves the perception and recall of four easily distinguishable intensities of sound. It is known that cognition of intensities of sensations occupies longer time than cognition of qualities, and that discrim- ination of intensities of sound is peculiarly difficult (Kiilpe*). These facts in addition to the novelty of the task and to the composite character of the auditory image for clangs, as found by Whipple," are calculated, in spite of the simplicity of the stimuli, to cause great qualitative variation in the processes used in the test. ^Pillsbury: Attention. 1908. 84ff. * Kiilpe: Outlines of psychology. 1901. 430. ' Fracker : On the transference of training in memory. Psych. Rev. Mon. 1908. Whole No. 38. Experiment III, pp. 59£. FORMAL DISCIPLINE 139 The four intensities making up the series of ten sounds were identified by the numbers from i to 4, in the order of intensity. Since this test closely resembles one of Fracker's,^^ the analysis of processes and the description of methods used by the various reagents will be of particular interest. His reagents used visual imagery mainly; some of ours used neither auditory nor visual imagery (Wf.). As the sounds came they had to be cognized, which was done by naming (giving them the numbers 1-4), or by comparing them with each other; the former needed a memory of absolute intensity, and the latter had to be done very quickly, in process, or it had to be left till the whole series had been received. These three methods of receiving the impressions would from the beginning of the experiment necessitate quite different pro- cesses, and they were all illustrated by the reagents in this test. The retention of the sounds was carried out in various ways : (a) Retained in auditory imagery (He.). i) The last few sounds of the series only (Le., Ly.). 2) In rhythm according to intensity, like chimes (Mn. Gl.). (b) Retained by kinaesthetic (Wf.), kinaesthetic-auditory (Al., SI., Ty.), auditory-visual (Ly., Rt.) imagery of the name (number) given to the sound when it came. i) Giving a rhythm to the naming (Le., Ms.). 2) Grouping the numbers (Le., Rt., He., Ms., Wf.). 3) Using mnemonic aids; as assigning the sounds to a spatial scheme, like numbered steps (Al.),ioo qj. associating their numbers with the visible numbers beside the blank spaces for the record (SI., Ms.). (c) Retained by kinaesthetic imagery of the response to the sounds (Ly.). ■* Kiilpe : op. cit. pp. 417-8. ° Whipple: An analytical study of the memory image and the process of judgment in the discrimination of clangs and tones. Am. Tr Psvch 1002 13 :259. J- • iA^ . °' Op. cit. Nine Tones, p. 59. ^°°A favorite method with Fracker's reagents {op. cit., 70-71); Kuhlmann (On the analysis of auditory memory consciousness. Am. Jr. Psych. 1909. 2o:i94ff.) found visual imagery that accompanied the auditory imagery a frequent means of recall of sounds of familiar things. 140 JOHN EDGAR COOVER The imagery of some of the reagents was of a mixed type (Rt.), and methods varied radically both between and with individual reagents. As a consequence of doing so many different tasks in taking the same test, the results varied greatly; initial capacity ranged from .3 to .7 average points per series, and fall into four groups of about .7, .5, .4, and .3 points. Improvement for the reagents who did not radically change their methods between tests, in per cent of initial capacity, was : Regular 1st Control 2A Control Group I. Gl. -17.6 Group 3. Le. Z7 Rt. 38 Group 4. Al. 52 Each of these reagents named the sounds as they came; for Le. the first group of four stood out in kinaesthetic imagery, the last two persisted in auditory imagery; but in the very last of the final test a new and more adequate method began to develop — retention of the rhythm of the intensities of the sounds in a series as in chimes. Rt. grouped the names in rhythms and recalled through kinaesthetic and auditory-visual imagery of the names. Al. recalled through kinaesthetic- auditory imagery ; his visual imagery is not strong or he probably would have made use of the vague spatial scheme that occurred to him near the end of the first test. Some of the other reagents made much more loss or gain (Group I. He. -11, SI. -3; Group 2. Ly. -16, Ms. 14; Group 3. Wf. 49, Group 4. Mn. 116%), but they all made radical changes in their methods in the final test. Mn., who made the greatest gain, grouped the sounds into three rhythms (4,4,2) in the first test ; but in the final followed the sounds as in a tune which gave the effect of chimes, the strains of which were remembered and transposed into numbers by comparison after the whole series was received. SI. simply named the sounds in the first test, but appealed to mnemonic aids in the final, and, for some series, associated the names of the sounds with the visible num- bers of their respective spaces on the blank record before him FORMAL DISCIPLINE 141 (the same method that Ms. used in her first and had to change from, when discovered, in her final test) or he associated them with his fingers. Ly., in the first test, reproduced from auditory and visual images of the names of the sounds, except for the last two or three sounds in the series which persisted in auditory imagery, but in the final test she used more kinaesthetic-auditory imagery for the early part of the series and kinaesthtic imagery of response to the sounds for the later members of the series. He., who found it extremely difficult in the first test to (a) name the sound, (b) place its order, and (c) hear the next one, in almost simultaneous process, and who combined kinaes- thetic imagery of names with auditory imagery of sounds, gave up, in the final test, the triple process and attempted to get an auditory impression of the whole series, making identification afterward by comparison; and while he stopped in the series with as many members as he thought he could retain, in the first test, in the final he tried for all, which resulted in indefinite grouping and fixing of the members. Wf. introduced in his final test rhythms of three. The changes that were made in the processes seem to be broadly adaptive, and in no way dependent upon the training during the interval. The reagents in the table above had practice in fixing and reproducing kinaesthetic and visual imagery of letters, and it is possible (a) that they did not change to other methods because their training improved them in the elements they had already used; and (b) that this improvement accounts largely for the improvement shown in the table. But since there are no com- parable records from control reagents at hand, the latter point must be left an open question. The changes in method were not necessarily beneficial; as is shown several paragraphs above, in parenthesis ; half of the reagents making them lost in efficiency. As to the effectiveness of the respective methods, those who associated the number assigned to the sounds with the numbers of the blank spaces for the record made the largest scores, 142 JOHN EDGAR COOVER Ms. producing a perfect score. The next efficient method was the auditory reception and retention of the whole series as chimes. Neither method involved the memory of sound in- tensity: The first being merely the memory of associated num- bers; the latter, of relations between intensities grouped in a very exceptional manner. It would seem that the visual forms of imagery employed with greatest success by Fracker's reagents were not necessary forms, but were preferences on the part of reagents who had good visual or kinaesthetic imagery, readily adaptable to the test, and encouraged by the training. (9) Memory of Consonants Memory for letters, usually consonants orally presented in a series too long to be reproduced from a single presentation (according to Ebbinghaus" Methode der behaltenen Glieder), has been frequently used in mental tests (Jacobs,'" Miinsterberg,^ Cattell,* Toulouse,* Pohlmann,° Winch,' Berg- strom'* Sleight'). Our test (see p. 79) is similar to Whipple's Test 38 A. (2), Variation (6).* The five series of 10 consonants were presented visually with a Jastrow Tachistoscope. The following is a classification of the more important methods followed by the reagents in this test: (a) Attention to the series as a whole with the view of retaining the visual imagery (Al., who recalled, however, from auditory imagery). (b) Naming the letters in rhythms of 4, 4, 2, (Le., Ms) ; of two's (Le.) ; of 4, 3, 3, etc. ' Ebbinghaus : Ueber neue Methode zur Priifung geistiger Fahigkeiten und ihre Anwendung bei Schulkindern. Zeits. f. Psych. 1897. I3:40iff. '* Jacobs : Experiments on "prehension." Mind. 1887 [o.s.] I2:75ff. ' Miinsterberg : Beitrage zur Experimentellen Psychologic. Heft 4. 1892. S. I2lff. ^ Cattell : Mental tests and measurements. Mind. 1890. 15 -.377. * Toulouse : Enquete medico-psychologique sur les rapports de la superio- rite intellectuelle avec la nevropathie. (Zola). 1896. 207. " Pohlmann : Experimentelle Beitrage zur Lehre vom Gedachtnis. Berlin. 1906. " Winch : Immediate memory in school children. II Auditory. Br. Jr. Psych. 1906. 2 :52ff. '"Bergstrom: Effect of changes in the time variables in memorizing, to- gether with some discussion of the technique of memory experimentation. Am. Jr. Psych. 1907. i8:2o6ff. 'Sleight: Memory and formal training. Br. Jr. Psych. 191 1. 4:430. ' Whipple : Manual of mental and physical tests. 1910. 366. FORMAL DISCIPLINE I43 i) "Impressing" the letter-names vividly while perceiving. 2) Strengthening the kinaesthetic impressions by serial repetition after each impression (Tn., Ds.), or in groups of four (Mn.)^^^ (c) Forming associations between the letters through familiar initials, abbreviations, cattle-brands, chemical compounds, etc. (Ms., Ly., He.). . m ^ (d) Making syllables by interpolating vowel sounds (Us.). Reproduction usually involved several kinds of imagery m each record ; usually the first part was reproduced by kinaesthetic or kinaesthetic-auditory imagery, or representative imagery (associations), and the last few letters by kinaesthetic or visual imagery; occasionally there was reproduction from auditory imagery (Ly., Al.). In case both parts of the series (begin- ning and end) were reproduced from kinaesthetic imagery, the former was the rote effect of repetitions and the latter the recently converted impression. Some of the reagents changed their methods in the final test : Le. changed from a rhythm of four to a rhythm of two in naming; and in her final recorded the last group, which was not accorded so much attention, first, leaving effort free to retain the earlier part of the series in kinaesthetic imagery; and it is impossible to determine whether the disadvantage of the former change is equivalent to the advantage of the latter. Ms., though not consistent in method in either test, made some use of associating letters with their preceding fellows, already re- corded, in the first test, but in the final made use of a rhythm of 4 in naming. Wf., who relied in the first test upon kinaes- thetic imagery for the main part of a series and visual imagery for the last few letters, made great use of associations for the former and kinaesthetic for the latter, in the final test. Ds. changed from forming syllables by interpolating vowels, in the first part of the series, and a visual retention of the last few letters, to rapid repetitions between exposures, of all the letters from the beginning of each half of the series; as, C, CV, CVJ, CVJB, etc. ^•^ Cf. Miiller, G. E. : Zur Analyse der Gedachtnistatigkeit und des Vor- stellungsverlaufes. Zeitschr. f. Psych. 191 1. Erg.-Bd. 5. S. 214. 144 JOHN EDGAR CO OVER There was some variation between the reagents who retained about the same methods for their final test that they used in their first. Mn. repeated in groups of four, reviewing after each group, recalling from kinaesthetic-auditory imagery, and occasionally recorded from the bottom up (perhaps from visual imagery). Rt. divided each series into two parts for rhythm, in the first test taking 5 members for the first group and in the final 6, which was recorded from kinaesthetic imagery, and recording the last few letters in the series from visual imagery. SI. recorded the last letter from visual, the others from kinaes- thetic imagery and associations. Ly. made use of associations and reproduced from kinaesthetic-auditory imagery and from visual imagery. He. occasionally recorded the last letter from visual imagery, but most of the reproduction was from kinaes- thetic imagery supported by associations. Cr. depended chiefly upon kinaesthetic-auditory imagery for all but the last two letters, which were retained visually; associations occurred occasionally. Al. tried to grasp the series as a whole, evidently a predominantly visual effort, and recorded largely from auditory imagery, as is shown by confusion between similarly sounding letters. Tn. repeated letters in intervals between exposures, each time from the beginning, until repetition excluded perception (at about the 7th consonant) ; then got the last few letters from kinaesthetic imagery from a single repetition. Initial capacity ranged from .3 to .7 points, per series, and classifies the reagents into four groups : .7, .6, .4, .3. Improvement in per cent of increase over initial efficiency, was made by those reagents retaining in the final test the methods used in the first, as follows : Regul ar 1st Control 2d Control Group I. He. Cr. 17 -13* Group 2. Rt. SI. 34 20 12 Tn. I Group 3. Ly. -3 Group 4. Al. 99 ''Irregular ; due to operator's error. FORMAL DISCIPLINE 145 Were one to consider Ly. as equivalent to a control reagent, since she was training in reaction time, it would give some slender basis for supposing that He., Mn., Rt., and Al were benefited by their training. The training of all, except He., was upon the reproduction of letters exposed as visual impres- sions. Rt. shows some carrying over of method in the 6-term grouping in his final test. The factors of improvement in this test are not so different from the factors of improvement in the training: (a) Better coordination of the part-processes; e.g., perceiv- ing and naming, grasping in rhythm, reviewing a rhythmic group hastily before fixing the next; recording and retaining, since recording involves kinaesthetic reproduction of different rhythm, also an accentuation of the group attended to, at the expense of other groups "retained kinaesthetically. This con- tributes toward the following classes, but does not account wholly for them. (b) Better fixing of imagery. (c) Better recall. The last two factors depend upon increased sensitivity and higher liability and fidelity of reproduction. Both are affected by. conditions of attention. The complete segregation of these factors is not possible with the present data, though each is occasionally clear, (a) is shown by Mn. in perfect records interspersed with low scores, due to kinaesthetic imagery disappearing while recording; (b) is shown by Rt. who extends the first group of letters from five or six to seven or eight, and Al. who extends the first group of a series from two or three, to three or four and the last group from two to three; and (c) is shown by Mn., Rt., and SI., in fewer substitutions and misplacings. Change of method was advantageous to those reagents who made it: Group 2. Wf. improved his score 11%; Group 3. Ds. 61.2%; Le. 39%; Group 4. Ms. 48%. (10) Memory of Numerals Memory for digits serially presented, usually in auditory form, has been used as a mental test (Jacobs,^ Bolton,'" Miinsterberg,^ Toulouse,' Cattell and Farrand,' Ebbinghaus,^ Wissler,' Binet,' Pohlmann,^ Krueger and Spearman,' 146 JOHN EDGAR CO OVER Sharp'") ; sometimes for the purpose of measuring attention (Bolton/" Binet^. Digits were preferred to non-sense syllables by Cattell and Farrand.^ Our test (see p. 79) is similar to Whipple's Test 38 A. (2), Variation (i) ;" series of 10 printed digits were presented visually with the Jastrow tachistoscope. Methods varied in their general characteristics as follows: (a) Repetition i) of single numerals (Tn., Le., SI, Ds., Cr.) 2) in rhythms (Le., Ms., Rt, Tn., He., Al.) 3) Naming once. 4) Naming over and over. 3) and 4) are both combined with both i) and 2). (b) Grouping (4-place Mn. ; 3-place Wf . ; 5-p Ds., Cr.).^"^ (c) Apperceptive grouping or relating (Ly., He., Wf.).^*^^ (d) Associating with numbers recorded in a preceding series (Ms.). Recall was through various imagery and was usually com- plex: Auditory (Mn.), auditory-visual (Ly.), kinaesthetic- visual (Rt, Le., SI, He.), kinaesthetic-auditory-visual (Cr.), and through associations (Ly., Wf.). Sometimes the disparate imagery supported each other on the same numerals recalled; 'Jacobs: Experiments on "prehension." Mind. 1887. [o.s.] 12 :75flf. '" Bolton : Growth of memory in school children. Am. Jr. Psych. 1891-2. 4:362ff. ^ Munsterberg : Beitrage zur experimentellen Psychologic. Heft 4. 1892. I2lff. ^ Toulouse : Enquete medico-psychologique sur les rapports de la superiority intellectuelle avec la nevropathie. (Zola). 1896. 222. * Cattell and Farrand: Physical and mental measurements of the students of Columbia University. Psych. Rev. 1896. 3 :644. ^ Ebbinghaus : Ueber cine neue Methode zur Priifung geistiger Fahigkeiten und ihre Anwendung bei Schulkindern. Zeits. f . Psych. 1897. 13 :4io. *Wissler: Correlation of mental and physical tests. Psych. Rev. Mon. 1901. 3 : No. 6. p. 9. ' Binet : L'fitude experimentale de I'lntelligence. 1903. Pp. 24off. ' Pohlmann : Experimentelle Beitrage zur Lehre vom Gedachtnis. 1906. * Krueger and Spearman : Die Korrelation zwischen verschiedenen geistigen Leistungsfahigkeiten. Zeits. f. Psych. 1907. 44:5off. '" Sharp : Individual psychology : A study in psychological method. Am. Jr. Psych. 1899. 10:351. "Whipple: Manual of mental and physical tests. 1910. P. 364. ^"^Cf., Miiller: Zeits. f. Psych., Erg. Band 5. S. 211. ^'^ Idem. S. 215; also Knors: Archiv f. d. ges. Psych., 17:340. FORMAL DISCIPLINE 147 sometimes cooperated in holding their respective numerals independently. The following reagents used different methods in the final test from those used in the first : He. added to his 2-place method of naming, an apperceptive element by relating the number to 100, and he dropped his visual reproduction of the last numbers in a series for kinaesthetic imagery. Cr. changed from kinaes- thetic-auditory, and visual for the last two letters, to a complex grouping involving 2-place and 3-place numbers, as, " '", " '". Le. changed from a rhythm of four to a mere repetition. Al. changed from a total impression to groups of 4, 6. Ms. changed from a rhythm of five to a single naming with effort to associate the numbers with the figures recorded in the preceding series, which essentially changed the process to memory of associated pairs. Wf., who had used kinaesthetic imagery supported by associations (perceived relationship between the figures in the series, such as products, squares, cubes, etc.), changed to group- ing into 3-place numbers. Tn. and Ds. changed respectively from a rhythm of 6, 4, and a grouping into 5 -place numbers, to serial repetition from the beginning, after each exposure, which they had just used with the consonants. Individual differences in processes between the reagents who did not change their methods radically were : Mn. recalled from auditory imagery of 4-place numbers. Rt. used a kinaesthetic rhythm for six digits, and visual imagery for the last four. SI. named in pairs, 2-place numbers, and recalled the last figure or two visually. Ly. used associations furnished by perceiving relations between the numbers, their sums and differences, re- tained the numbers expressing the relationships visually, and recalled the numbers themselves from auditory imagery supported by the association-meanings. Initial capacity ranged from .37 to .92 points, per series, and classifies the reagents into four groups : .8, .6, .5, .4. Improvement in per cent of increase over initial efficiency, for the reagents who did not radically change their methods, was : 148 JOHN EDGAR CO OVER Regular ist Control 2d Control Group I. Mn. 17 Rt. -7* Group 2. SI. 30 Group 3. Ly. 15 *This score illustrates the inadequacy of the Spearman "Footrule for scor- ing the memory test;" (vid. Whipple: Tests, p. 367). The loss was caused by four reproduced numerals in correct order being misplaced; omitting this score, the reagent shows a gain of 12%. These reagents presumably improved somewhat by reason of their training, the results comparing favorably with those of corresponding groups in the preceding test, but since there are no control averages to compare with them, the supposition rests unproven. Those who changed methods, with one exception gained more: Group i. Cr. 10%, He.-2%; Tn. 16.7%; Group 2. Wf. 52%, Ms, 33%; Group 3. Ds. 11%; Group 4. Le. 17%, Al. 70%. (11) Memory of Visual Signs The signs (see p. 79, and Appendix B. Fig. 11, p. 291) were so unfamiliar that, although they were clearly perceived in the one-second exposures, they were not apperceived and therefore were not usually subject to recall. However, some of the signs suggested to the reagents familiar things or con- ventional characters, and in this way made an impression that could be utilized for recall. This impression did not need to be largely visual, as the test presupposed, but could be converted into any other imagery that would serve as a carrier for the sug- gested thing, plus a modicum of visual or other imagery which suggested any necessary variation in the drawing. Thus, none of the reagents reproduced more than the last sign from purely visual memory; but gave the symbols names of the things they resembled, as, d, J, 8, 10, omega, phi, dutch cap, etc., and ap- proximated the signs in their reproductions through retention of these names, principally in verbal imagery, and secondary criteria of deviation from the outline of the conventional thing. FORMAL DISCIPLINE I49 The test shows the impossibiHty of forcing the use of visual imagery by such material and method. -^^^ (12) Memory of Associated Pairs Memory for "paired associates" is a variation of Miiller and Pilzecker's' "Trefifermetliode," and has been used with vocabulary material (Bourdon,^ Thorndike'). Our method (see p. 79; also Appendix B, Fig. 10, p. 290), differs from the original in the simultaneous presentation of the pairs to be associated. A printed consonant and a printed digit were presented together in series of 10; recall of the digit was required upon a second presentation of its associated consonant. The process was intended to be memory of contiguous asso- ciations, but since this was more difficult or unfamiliar than memory of serial order, the latter was not always inhibited, and introspections are not sufficiently full to be trustworthy in reporting this source of error. Processes, so far as introspections indicate them, varied greatly: The usual process was to name the letter and digit together so as to get a strong unitary kinaesthetic, auditory, or kinaesthetic-auditory, image of the pair; and when the letter appeared alone it was again named to recall its associate. It is a simple process, and if the reagents had been so instructed they probably could have adhered closely to it without great variation. But variation appears immediately in the 'aufgabe,' and intro- spections scarcely more than indicate that it occurred. Mn. said the numeral came as an unfinished syllable of a word, Mn., Le. and others found that familiar or significant letters held their associations best. He. and Ms. found visual (G6, C5) and other associations: H2 from chemistry; C3 from alphabetic order, etc. Rt. and Ly. supplemented kinaesthetic imagery with visual, and found some combinations easier to pronounce. Mn. and Ly. were caused some distraction by having to inhibit a tendency to notice serial order. ^Miiller und Pilzecker: Experimentelle Beitrage zur Lehre vom Gedachtniss. Zeits. f. Psych. Erg. 1900. i :2. ^Bourdon: Recherches sur I'habitude. Annee Psych. 1901. 8:327ff. 'Thorndike: Memory for paired associates. Psych. Rev. 1908. I5:i22ff. *"* Cf. Kuhlmann : On the analysis of the memory consciousness ; a study in the mental imagery and memory of meaningless visual forms. Psych. Rev., 1906, 13 :3i6ff. ISO JOHN EDGAR CO OVER He. found that some associates not repeated came up for recording. The attitude of the reagents toward these indefinite associations varied somewhat in the two tests and would be a source of error in the resuhs. This attitude is indicated by the relative number of recorded associations that proved correct : Per cent, of the numbers recorded, that were correct. First Final Mn. 39 43 Le. 56 83 Rt. S^ 42 SI. 54 41 Ly. 56 56 He. 91 80 Cr. 30 SO Al. 50 16 Ms. 31 66 Wf. 93 66 Es. 52 94 Pe. 42 46 Results show that there was a general tendency for associates of earlier series to persist and cause errors, especially if they were intensified by significant letters or by mnemonic connections. The initial capacities ranged from .08 to .42 points and re- sults are so irregular as to make their inspection unprofitable. (13) Learning 12-Letter-Rectangles Memory for consonants, simultaneously presented in a "letter-square" under conditions of number of letters and time of presentation that preclude a perfect score, has been used for two principal purposes: (i) to study the mental imagery (Binet et Henri,^ Toulouse,^ Cohn,^ Segal,*) ; and (2) to test mental ability (Binet,° Sharp,* Winch'). The test is standardized for the former purpose by Titchener* ; for the latter by Whipple.* It has also been used for the study of the relation of attention to memory (Smith*"). The letter-square, or, more precisely, the letter-rectangle usually contains 12 letters in three horizontal rows, and is presented from 10 to 20 seconds. Our test (see p. 80) is similar to Whipple's and follows Cohn, Smith, and Segal: i2-consonant-rectangles exposed 10" ; reproduction after a free interval of 10". * Binet et Henri : La psychologic individuelle. Annee Psych. 1895. 2 :436ff. * Toulouse : Enquete medico-psychologique sur les rapports de la superiorite intellectuelle avec la nevropathie. (Zola). 1896. P. 182. ^Cohn: Experimentelle Untersuchungen iiber das Zusammenwirken des akustisch-motorischen und des visuellen Gedachtnisses. Zeits. f. Psych. 1897. 15:162. * Segal : Ueber den Reproductionstypus und das Reproduzieren von Vor- stellungen. Archiv f . d. ges. Psych. 1908. I2:i33ff. FORMAL DISCIPLINE iSi This is a fairly simple exercise, like most of the other standard "Mental Tests," yet it, too, may evoke quite different processes in the various reagents, and may become several different tasks even to the same individual. The effect of practice on the processes involved was described some pages back (pp. 94ff.). ^n the discussion of training re- sults. Here v^e have to do with fairly initial abilities, except in the final tests of Rt. and SI., who trained on this work, in which variability shows itself freely. The reagents attempted the task by: (a) Reading in vertical columns of 3 letters. (b) Reading in lines of four letters (the general method). I ) Rote repetition, over and over, for kinaesthetic- auditory impression. 2 ) Fewer repetitions, apperceptive ; grouping of letters, rhythm, 3) Visual impression; intensified or casual. 4) Associating letters. ^°^ a) by sound, rhyme. b) by form (e.g., VWYM, CGQ). c) by alphabetic position (BC, KL, XZ). d) by signification; favorite form, initials, abbreviations, words, etc. (DV, deo volente; SFTR, San Francisco Teddie Roosevelt, or sifter). ^ Binet : Attention et adaptation. Annee Psych. 1899. 6 :324ff. " Sharp : Individual psychology : A study in psychological method. Am. Jr. Psych. 1899. 10 -.353. ' Winch : Immediate memory in school children. I. Visual. Br. Jr. Psych. 1904. I :i28. ^ Titchener : Experimental psychology. 1901. Vol. I. Part II. Instructor's Manual. 396ff. ''Whipple: Manual of mental and physical tests. 1910. Test 38, B. 368. ^^ Smith : Relation of attention to memory. Mind. 1895. N.S. 4:47ff. ^°^ For advantage of logical connections, vid. Balaban : Zeits. f. Psych. 56: 356-377, 379-400. For agreement of these results with others, vid. Smith: op. cit. p. 57; Cohn: op. cit. pp. I78ff. ; Segal: op. cit. S. 160; and Michotte et Ransy (Contribution a I'etude de la memoire logique. fitudes de Psychol. 1912. 1:72-81). Calkins (Psych. Rev. 5:460) found the tendency to combine unrelated concrete words in four-fifths of her 52 reagents; three-tenths of the words, chosen to prevent natural combinations, were forced through imagery into some sort of relation. 152 JOHN EDGAR COOVER To illustrate: (a) Continuous repetition of the first line and the P H Q K first column, on the accompanying 12- D S T N letter-rectangle, yielded 15 points (Le.) ; L G R F or of the first two lines, 22 points (Rt). (b) Repetition of letters in lines, by pairs, yielded 20 points (Le.). (c) Repetition of the first two lines; visual image of the last line on the forehead, gave 24 points (SI.). (d) Wholly representative (Phone quick, long distance, lograft) gave 36 points, a perfect score (Cr.). Clearly those four methods involve different kinds of work, besides yielding different scores. Reproduction was usually in combined imagery; the more common being, kinaesthetic-auditory, but often also auditory, visual, and representative (associations), upon the same letters. Then different letters were sometimes assigned to the different forms of imagery so that their cooperation yielded more letters than could be reprodued by combining on the same letters. Some- times only the imagery representative of the letters was prominent; i.e., but little imagery of the letters, when associations for them were used. Some of the reagents made radical changes in their methods in the final tests: Mn., Cr., Ms., and Bt., changed to much greater use of associations; Le. from more rapid repetition, over and over, to a single apperceptive repetition and the use of associations. Other reagents retained their old methods : Rt., who trained on this work, repeated the first six letters for reproduction from kinaesthetic imagery ; the next two letters were assigned to a visual association, or were reinforced for a visual recall, and the last line casually observed for visual recall to be recorded first while retaining the more vivid imagery to be recorded later. SI., who also trained on this work, repeated the first two lines for kinaesthetic recall, and fixed the last line by associations, with meaning, or by visual imagery, which was very unstable. He., Ly., and Wf. made forming associations a method. FORMAL DISCIPLINE iS3 Al, sought to get a kinaesthetic-auditory impression of the whole card. Initial efficiency ranged from 15.5 to 27.3 points, classifying the reagents into four groups: 2y, 22, 18, 16. Improvement upon initial efficiency, for the reagents whose methods remained practically the same, was : Regular 1st Control 2d Control Group I. He. 7 Group 2. Ly. 26 Wf. 8 Group 3. SI. 45 Rt. 44 Al. 4 Group 4. Ed. 12.3 Ly.'s is the only score that invites speculation. Her method was to form associations, and her improvement is due to doubling her facility in forming them ; this differs greatly from improve- ment in vivifying impression and heightening the power of recall of imagery, which contributed to the improvement of Rt. and SI., and seems in no way to depend upon her training in simple reaction. In her method variability is greater than might at first be supposed : In the final test, although the material presented was the same as that of the first test, (a) different letters were chosen for the associations, and (b) when the same letters were chosen the associations were sometimes different (DZ dizzy, Diaz; WSH wash, Washington). Facility in form- ing associations varied greatly, irrespective of material, even within the same test ; e.g., in the first test, Ly. began with form- ing associations as a method, and by the fourth experiment had made a perfect score by holding all the letters in associations, but when the 7th card, used in the above illustration, was ex- posed, she could find no associations for the letters and was reduced to repeating them in pairs, making a score of 10 points, (half her average). In her first test 48% of her recorded letters were held in associations, in her final, 90% were so held. Change in method was advantageous (improvements being: Group 2. Cr. 26%; Group 3. Mn. 64%, Bt. 27%; Group 4. Ms. 48%, Le. 14%), sometimes more than any other circum- stance; as is well illustrated in the case of Ms., a control reagent, who, although her initial ability was but a little below theirs, 154 /0//iV EDGAR COOVER showed greater improvement than was made by either of the two reagents who trained on this work i8 days. An interesting effect of the training on this work is found in the resuhs of Rt. and SI, in comparison with the results of others, obtained on additional experiments in this test, three in which numerals were exposed, and one in which commercial signs were exposed : These reagents carried over their method, as recorded above, to the numerals, changing from 4-place and 2-place grouping, respectively, and made 83% and 70% improve- ment (while the other reagents ranged from loss to 55% im- provement), and this improvement is greater than was made with the consonants upon which they trained. To the commercial signs, the method was absolutely unfitted, and they alone lost in the scores of the final test (Rt. 7%, SI. 17%), SI., whose method was the more mechanical, losing the most.^"^ The chief factors of improvement in this test were : (a) More adequate method. (b) Better coordination of the kinaesthetic, auditory, and visual modes of imagery, and assigning them to different material rather than permitting them to merely support each other in making vivid a smaller number of letters. (c) Better impression and recall through each kind of imagery. (d) More use of associations. (e) Associations more apperceptive. (14.) Same. With Distraction Distraction in learning consonants visually presented in letter-rectangles has usually been applied to the period of perception (Smith/ Cohn/ Segaf) and has consisted in counting, adding, or intoning a vowel sound; occasion- ally the distraction has been applied to the interval between perception and reproduction (Bingham,* Finzi^). In our test (see p. 80) the reagent was required to add digits pronounced by the experimenter during the interval between perception and reproduction, and to write down the sum before he began recording his letters. ' Smith : op. cit., 47ff. * Cohn : op. cit., 162. "Segal: op. cit., 133. ^ Bingham : Memory, II. Psych. Rev., 1894, i :453ff- 'Finzi: Zur Untersuchung der Auffassungsfahigkeit und Merkfahigkeit. Psych. Arbeiten, 1899-1901, 3:295. *"* Cf. Sleight, op. cit., pp. 440, et seq. FORMAL DISCIPLINE 155 Interpolation of processes, as usual, again increases variability in processes and results. In the preceding test attention could be given, during the lo-second interval between impression and recording, to the retention and fixing of the letters; in this, a variable division of the attention takes place between retention and adding, resulting in a further important source of variation beyond those of the preceding test. (a) The adding may be carried on either in visual imagery, or auditory imagery, partial sums being kinaesthetically imaged or repeated, while the attention shifts quickly^^'^ to the verbal or other imagery from which the recall of the letters is sought to be made. (b) The whole attention may be given to the adding in the interval while the numbers are pronounced. (c) Or, attention may be given to further fixing the imagery of the letters for surer recall, and turn with leisure and con- fidence to the auditory images of the numbers after they have all been pronounced. The first method is apt to lead to incorrect sums and to great decrease in the score ; the second to correct sums, but to decrease in the score; the third to correct sums, if auditory imagery is fair, and to a good score. Variability is further increased by the varying degrees of conflict between the part-processes. Difficulty in adding, using the first method above, dispersed clear imagery, because of confusion of mind it occasioned (Le.). Both kinaesthetic and visual imagery were interfered with, when the second method was used (Rt.). Associations escaped when they were merely verbal, because of the conflict with kinaesthetic impressions in pronouncing the partial sums (Ly. He.). Since recall is surer from associations than from imagery of the letters,!^^ almost all reagents resorted to them, but with "'Smith (Mind. N.S. 4:66) found that the attention of his reagents, who were set to learning the letters while adding, oscillated between the two tasks. *"In the preceding experiment, when letters had heen fixed in an associa- tion, no further attention was given to them, the interval being filled with 156 JOHN EDGAR COOVER varying facility. Not only the readiness to form associations was variable, but selection of such kinds of associations as involve more apperception was variable, and, consequently, retention of associations was variable. The whole association may be lost; or by a slight cue may yield its quota to the score. The following will illustrate how near a o score is to a score of loo^, when associations are used: A reagent gave full attention to adding during the interval, recorded a correct sum, and found that he could not recall a trace of his associations or even a letter; after a moment, weak kinaesthetic imagery brought back the association of the first two letters; then in recording these, associative connections grew stronger and the associations all returned (led by meaning rather than by verbal imagery), and a perfect score was produced (Cr.). An association may be recalled, however, and be either untrust- worthy or useless, because of the absence of further imagery relating the letters definitely to it. The influence of the imagery is seen in the record of VNWY from the associated word "Vanity"; Van(it)y; the substitution of "W" for "it" was held in visual imagery (Cr.). The adding was often fatal to this imagery as may be seen from the failure of "North Buckham" to produce NRT BKYM by producing NTH, thus yielding the lowest score of the series (Cr. ). None of the reagents were constant in their methods, and initial ability ranged from 8.3 points to 19.5. Distraction ranged from 15% to 60%. effort to retain by repeating letters held in less sure imagery. At the begin- ning of the first test, here, the reagent said : "I rested my attention on the last column more than on any other, but the second seemed to form a word; I retained the latter and could not remember the former." (Mn.). This con- forms with the findings of Balaban {op. cit.) that associative learning was eight times more effective than mechanical memorization, and of Arnold (The initial tendency in ideal revival. Am. Jr. Psych., 1907, 18:251) that recall is insured by a more closely organized disposition, a better developed meaning, a more complete organic whole. FORMAL DISCIPLINE I57 (15) Word-Completion This test (see p. 8of.) is an application of Ebbinghaus'^ "Completion Test" to words, and differs from Whipple's'' "Word-building" test in greatly limiting the possible combinations. It involves equivocal reproduction or controlled association, and was designed to measure a narrow type of inventive or im- aginative power. Some of the reagents completed the Hst in less than 100 seconds; others left it unfinished in 300. The process differed greatly, especially with respect to the 'Aufgabe.' All reagents sounded the consonants phonetically, interpolating between them indefinite vowel sounds, until a word was called up by the kinaesthetic-auditory cue. Upon a hitch in getting a word, some of the reagents (Le., Rt., Cr., Ms., Wf.) started systematically sounding definitely the various vowels, taking them in order, between the first pair of consonants. Only once did the visual form suggest a word (Rt.). Although instructions were clearly given to add letters any- where to complete a word, only twice did an added letter precede the first consonant (Ly., Ms.), and occasionally the reagents otherwise limited their task, e.g., Le. and Ms., after their experience in the first test, limited additions in the final test to the two spaces between consonants, and failed to com- plete the ten words within the 5-minute time-limit; He., in his first test, limited himself in the first few words to the space between the first pair of consonants, and after leaving an incomplete word and properly assigning himself to other spaces in completing following words, he returned to it, again limiting himself as before. Since the difficulty of the task depends upon the 'Aufgabe,' which the reagent holds in his mind, it is obvious that quan- titative results, without adequate introspection, are without value. Even with adequate introspection, there is no way to equate results when the processes are greatly different. The test is of value in showing how, under precisely the same * Ebbinghaus : Ueber eine neue Methode zur Priifung geistiger Fahigkeiten und ihre Anwendung bei Schulkindern. Zeits. f. Psych., 1897, I3:40iff. ; also Whipple's Test 48, p. 445ff. * Whipple: Manual of mental and physical tests, 1910, Test 27, pp. 44iff. 158 JOHN EDGAR COOVER instructions and external conditions, different reagents vary- widely (a) in the task to be performed, and (b) in the processes they use to carry out a similar task. (i6) Trains of Ideas The mental processes involved in free reproduction or uncontrolled asso- ciation made early contribution to experimental psychology. They furnished Galton/ as he walked along Pall Mall and noted the ideas that came into his mind as he scrutinized successive objects that caught his eyes, material which he learned to subject to measurement. They are, of course, employed for mental diagnosis in the famous "word-association" test; and they have been employed in unbroken series of from lo to lOO words* for the study of association (Cattell and Bryant'), of effect of immediate environment upon association (Flournoy^), of community of ideas of men and women (Jastrow, " ' Nevers," Calkins,^ Tanner,*), of individual differences in mental processes (Secor," Binet") ; but they have also been used in unbroken series for determining rate of mental activity (Wissler," Thompson," Brown"). In our test (see p. 8i) 2-minute series were written after the stimulus words : "horse," "potato," "flute." Since most of the individual scores run up to about the maximum speed in writing, it may be presumed that in this test, when the ideas are recorded by the reagent, reproduction is seriously limited by recording. Even were the phrases and ^Galton: Psychometric experiments. Brain. 1879-1800. 2:151. ' Vid. Whipple : Manual of mental and physical tests. 1910. Test 2>2>, PP- 3i3ff. 'Cattell and Bryant: Mental association investigated by experiment. Mmd. 1889. 14 :23off. * Flournoy : De Taction du milieu sur I'ideation. Annee Psych. 1894. I :i8off. 'Jastrow: A study in mental statistics. New Review. 1891. 5 :S59ff- * Nevers : Dr. Jastrow on community of ideas of men and women. Psych Rev. 1895. 2 :363ff. ^Jastrow: Community of ideas of men and women. Psych. Rev. 1896 3:68ff. * Calkins: Community of ideas of men and women. Psych. Rev. 1896 3 :426ff. ' Tanner : The community of ideas of men and women. Psych. Rev. 1896, 3 :548ff. '"Secor: Visual reading: A study in mental imagery. Am. Jr. Psych 1899-1900. ii:225ff. " Binet : L'fitude experimental de I'intelligence. 1903. P. 309- "Wissler: Correlation of mental and physical tests. Psych. Rev. Mon. 1901. 3 : No. 6, p. 8. "Thompson: Psychological norms in men and women. Univ. Chicago Contrib. to Phil. 1903. Pp. looff. " Brown : Some experimental results in correlation of mental abilities. Br. Jr. Psych. 1910. 3:306. FORMAL DISCIPLINE i59 words Spoken, the process would be similarly limited in speed. Scores, therefore, do not so much show rapidity of reproduction as rapidity of expression and regularity of reproduction. And since the whole field of the reagent's experience is available for reproduction, with all the grades of Hability for recall, variation in regularity is inevitable ; nor does it seem that variable attention rather than direction of 'leads' into various fields of experience is the greater cause of variation. Initial efficiency ranged from 55 to 100 ideas for the six minutes. Qualitatively, some interesting facts were presented : (a) The 'Aufgabe' was not the same for all reagents, although instructions and external conditions of the experiment were uniform. The usual method was to begin with the idea suggested by the stimulus-word and follow the course suggested by the last word until new 'leads' would develop from some prominent member of the last series, or 'story,' thus intro- ducing a new 'story,' etc. But SI. bound himself to record only ideas related directly to the stimulus-word, and gave second place to the suggestion of recorded ideas; and Wf. fol- lowed an intermediate course approaching closely Sl.'s method. (b) Most of the reagents find that of the flood of memories and images only a few, usually the most prominent, can be recorded. (c) The 'stories' from which the constellations of ideas are chosen for recording are related to various periods of the reagent's experience and tend to fall upon the same periods of experience, for all stimulus-words given in one sitting, but upon different periods for the same stimulus-word given upon different days. This merely indicates i) the organization of experience, in cross-section; and 2) the flux of experience, in longitudinal-section. Ms. in her first test drew mainly upon experience of her childhood; in her final test, upon those of recent years and days, (d) The remote and recent experiences are sometimes drawn upon in the same minute, due to the prominence of identical factors in their memories; emotional or cognitive. t6o JOHN EDGAR CO OVER ■(e) Anxiety about one's work, or great interest in an event, leads all trains of ideas to that center of interest, even when approaches are repeatedly inhibited (Cr., Le.). This shows how efficient the selective influences are, in the control of the central elements of consciousness. Monomania would seem to be but an exaggeration of this natural process. In this case the control seems to be given over to the emotional set of consciousness.^^^ (17) Extensive Threshold of Visual Attention. Free The number of letters that can be apprehended in a momentary exposure of a rectangle containing from 6 to 12 letters has been thought to constitute a measure of the "extensive threshold of attention" (Wundt^) or the number of elements that can be simultaneously grasped by consciousness. The time of presentation is usually about o.i seconds, less than the eye-reaction time, in order to limit apprehension to a single fixation of vision ; but owing to the persistence of the after-image for about 0.25 seconds (Schumann,^ Hylan,' Messmer*") and to the visual memory after-image that may appear a few seconds later (Schumann^), the attention may fluctuate successively over the elements so held, thus augmenting unduly the measure of the threshold of attention and relating it to the 'memory-span' for successively presented let- ters. Wundt' warns the reagent to avoid this error through introspectively distinguishing between the simultaneous and the successive activities. The test has been used to measure visual perception and attention (Griffing*) or the range of visual attention (Whipple^). In our test 12-consonant-rectangles were presented for about o.i sec. (85 sigma) ; reproduction was required after a free interval of 5 seconds (see p. 81). * Wundt: Grundriss der Psychologic. lote Auf. 1911. Sec. 15, Par. 6 (pp. 254ff.) ; also Grundziige der Physiologischen Psychologie. 5te Auf. 1902. 111:35 Iff. ^ Schumann : Die Erkennung von Buchstaben und Worten bei momentaren Beleuchtung. Bericht u.d. I. Kongress f. Exp. Psych. 1904. 34-40; also Psychologie des Lesens. Bericht u.d. II. Kongress f. Exp. Psych. 1906. S. 174- * Hylan : The distribution of the attention. Psych. Rev. 1903. 10:373, 498. , ^" Messmer : Zur Psychologie des Lesens bei Kindern und Erwachsenen. Archiv f. d. ges. Psych. 1903-4, Bd. 2, S. 206. *Griffing: On the development of visual perception and attention. Am. Jr. Psych. 1895. 7:277ff. "Whipple: Effect of practice upon range of visual attention and of visual apprehension. Jr. Ed. Psych. 1910. i :249ff. ^""This test furnishes a good diagnosis of the reagent's experience at the time it is taken. Repeated tests would make it quite possible to characterize the reagent's mental life. A person could not in a briefer or more efficient way keep a diary of his real interests than by this two-minute test, accom- panied by a commentary. FORMAL DISCIPLINE i6i Analysis of processes involved in the work of this test was made under the head of Training Results, some pages back (pp. 82ff.) ; only a general indication of the fact of variability of processes, therefore, need be made here. (a) The field of the attention may vary from the whole card to a line or to a few spaces. In the first case, vague imagery makes up the bulk of the material dealt with; in the second, both clear and vague are entertained; in the third, only clear imagery. (b) The type of imagery may determine the process: If the visual impression is immediately converted into kinaesthetic or kinaesthetic-auditory imagery, reproduction is largely limited to letters clearly seen; if the visual impression is held for a while, before being converted, more dimly seen letters become fixed. (c) If the attitude toward vague imagery is encouraging, some unrecognizable imagery 'matures' into letters, the record of which the score often justifies. (d) If, in recall, disparate imagery supplements itself on sepa- rate letters, more letters are recorded. (e) Some distraction seemed peculiar to the test: Readiness of the naming process was not at times satisfactory because of low reproductivity of the letter-names; to some reagents some combinations of letters were retarding because of the non- euphonious pronunciation of their names ; to some, special letters (significant or disagreeable) retarded the process; several times exposures were completely missed by winking. Introspections show that the process varied with each reagent in each test, and sometimes considerably between tests. There was a general effort to convert the visual impression or visual imagery into kinaesthetic imagery or impression (in- cipient pronunciation) ; this conversion reduces reproduction to a 'rote' process, which doubtless is the line of least resistance in memorizing. Introspection says : "First the imagery is very intense, visual, which I convert into kinaesthetic (verbal) ; then I repeat rapidly, almost as one word, and this is accompanied by weak visual imagery. Then I seem to feel sure of the letters — they seem to lie on my tongue and are less full of meaning than at first" (Rt.). "After repeating letters I felt confident that I had them; they would not fly away" (Wf.). i62 JOHN EDGAR COOVER In practice, the moment of this conversion was adjusted by the necessity of fixing many distinct impressions before they got away, and the necessity of permitting thern to mature more clearly in order to be either cognized or held in visual imagery in sufficiently stable form to remain while fixing others by naming. The shifting of this moment was necessitated by the varying grades of clearness of the imagery, and inexpert shifting was fatal to the score. Better coordination of processes here resulted through practice, and variation in methods as indicated above must be credited with considerable variation between individual reagents, resultant upon this particular practice effect. Initial capacities varied in average scores from 4.1 to 9.1 points (3 points to a correctly placed letter), and divide the reagents into four groups: 9, 8, 6, 5. Improvement in per cent upon initial capacity, for the reagents who retained in their final tests their old methods, was: Regular 1st Control 2d Control Group I. He. 12 Group 2. Le. Rt. 58 10 Wf. 21 Ty. 4 Group 3. Al. Cr. 45 42 Group 4. SI. Ly. 25 14 Al. differed from the other reagents in attending consistently to the whole card. Three reagents made radical changes in their processes in the final test: Mn., who trained 18 days on this work; and Ms. a control reagent, changed from attending to the whole card to attending to limited parts of it, usually the first line. Dn. changed from intentionally varied methods in the first test, involving attending to the whole card, limiting the field in various ways, and closing one eye and then the other to see if a clearer impression could not be obtained, to a regular method of restricting the field and of binocular vision. These changes were advantageous: Group 4. Mn. 147%, Ms. 47%; Group 5. Dn. 39%. FORMAL DISCIPLINE 163 That great difference in the score results from variation in the extent of the field attended to, is illustrated by individual scores made by Ms. in her first test. In the first six experiments, with attention upon the whole card, she made no score above 4 points; in the last four experiments, with the field restricted to the upper right-hand corner, she made no score below 8 points. Mn.'s great gain is based upon those six experiments, for the averages of the scores of the last four experiments in her first test is higher than the average of either the last four or all of the experiments of her final test. The results of this test offer several anomalies which indicate the futility of merely quantitative treatment of work of this kind. Ms., a control reagent, shows greater gain than does AL, who trained on this work for 8 days; besides (a) the difference in capacity, indicative of difference in kind of work, (b) the gain is exaggerated by the per cent being reckoned upon lower initial capacity. Wf., a control reagent, shows a greater per cent of gain than He. ; although their initial capacities are not so unlike, this excess of gain is illusory, since in absolute gain He. excells, and from He.'s greater initial capacity, it is conceivable, improvement is more difficult to make. Cr. and possibly He. are the only reagents who appear to have brought to the test any advantage from their training; both improved more than the other reagents who did not train in this work, in regular performance indicative of more constant conditions of attention. (18) Same. With Distraction Adding four digits during an interval of 5 seconds between perception and recording is a doubtful distraction, ranging from o to 59%. It was but negligible for Mn. in both tests; for SI., Ly., and Dn. in the first test, and for He. and Al. in the final test; and it varied between the two tests considerably for all the reagents except Mn., Cr., and Ty. The chief cause of variation in the scores (initial capacity ranged from 3.2 to 6.9 points, and final capacity from 2.8 to 1 1.4 points) was the conflict between the visual or kinaesthetic- i64 JOHN EDGAR COOVER auditory imagery in which the letters were being held and the kinaesthetic imagery of the partial sums in adding. Where dis- traction was overcome, the usual method was to quickly name the letters and thus fix them in kinaesthetic imagery, and add the numerals at leisure from auditory images after they had all been pronounced. (19) Tapping Tapping as rapidly as possible during an interval of from 5 seconds to 2 minutes, with a pencil, stilus, or telegraph key, has served as a test for vol- untary motor ability and, with the longer intervals, for fatigue (Bryan,* Dresslar,^ Gilbert,' Moore,* Binet et Vaschide,° Kirkpatrick,* Bagley,' Bolton,' Kelly,' Thompson," Burt," Wells"), and has been standardized by Wells** and Whipple." Our test follows Wells : 5 series of 30", with 2.5' rests. {Vid. p. 81 f.). Maximum rate of voluntary activity as expressed in tapping on a Morse key (adjusted as preferred by telegraph operators — I mm. amplitude, 50-gram tension) seems a very simple process. But our results show it to be variable and to depend upon other important factors besides attention and fatigue. Results gave the "Total Efficiency" (average number of taps * Bryan: On the development of voluntary motor ability. Am. Jr. Psych. 1892. 5:i23ff. ^ Dresslar : Some influences which affect rapidity of voluntary movements. Am. Jr. Psych. 1892. 4:5i4ff. ' Gilbert : Researches on the mental and physical development of school- children. Studies from Yale Psych. Lab. 1894. 2 148. * Moore : Studies of fatigue. Studies from Yale Psych. Lab. 1895. 3 :92ff. ° Binet et Vaschide : Experiences de vitesse chez les jeunes gens. Annee Psych. 1897. 4:20off; also fipreuves de vitesse chez les jeunes garqons. ibid. 64ff. * Kirkpatrick : Individual tests of school children. Psych. Rev. 1900. 7:274ff. 'Bagley: On the correlation of mental and motor ability in school children. Am. Jr. Psych. 1900-1. I2:i95ff. ' Bolton : Relation of motor power to intelligence. Am. Jr. Psych. 1903. 14 :354. ® Kelly : Psychophysical tests of normal and abnormal children ; a compara- tive study. Psych. Rev. 1903. io:345ff. '" Thompson : Psychological norms in men and women. Univ. Chicago Contrib. to Phil. 1903. 4: No. i. I2ff. " Burt : Experimental tests of general intelligence. Br. Jr. Psych. 1909. 3-132. *^ Wells: A neglected measure of fatigue. Am. Jr. Psych. 1908. I9:345ff. " Wells : Normal performance in the tapping test. ibid. 347-483. " Whipple : Manual of mental and physical tests. 1910. Test 10, pp. looff. FORMAL DISCIPLINE 165 per series of 30 seconds), and the "Fatigue Index" (found by dividing the average of the last five intervals of 5 seconds each, of the first series, by the number of taps in the first interval of 5 seconds). The manner of tapping varied between the different reagents, virith respect to the parts of the arm, forearm, wrist, hand, and fingers, put in vibration, with respect to the amplitude of vibration, the manner of grasping the key, the amount of innervation and bodily tension, determination to resist fatigue, and accompanying psychial processes, such as counting the taps in groups. And it varied, in a less degree, in an individual reagent's work. "Total Efficiency" ranged from 176.8 to 245.8, in the first test, and reagents fall into five groups: 246, 218, 205, 198, 180. Change in per cent of initial capacity, was as follows : Regular 1st Control 2d Control Group I. Rt. -1.5 Group 2. He. s-o Group 3. SI. 3.2 Cr. 3.2 Le. -0.2 Ly. -2.6 Mn. -3.3 Ms. -1.3 Group 4. Wf. -1.2 Group 5. Me. 0.6 Wx. 1.2 The changes in the table may indicate changes in fatigue, since objective fatigue is said to increase the time,^^^ and the test is recommended as a measure of fatigue ;^^^ they may indi- cate changes in the capacity of attention which opposes fatigue; or they may be due to both these causes as modified by others. That the last conjecture is sometimes true is shown by the results of Rt., Le., and SI. Rt.'s practice curves of the two tests are precisely the converse of each other; his first curve shows great practice-effect in the second series, reaching a maximum in the third series; his final curve begins with phenomenal speed (267), loses greatly in the second series, "" Moore : Yale Studies, 3 :95. *" Wells: Am. Jr. Psych., 19:344. i66 JOHN EDGAR COOVER reaching minimum in the third series, Le.'s curves are similar but show less practice-effect and loss. Sl.'s final 'fatigue curve' is inverted, differing from all others in showing marked practice- effect up to the fourth interval. Unfortunately, introspections are not sufficiently full to indicate the causes of these variations. Usually the two practice curves of a reagent are similar: Both show practice-effect in the cases of five reagents (Mn., SI., Cr., Ms., Wx.) ; no practice-effect in the case of four reagents (Le., Ly., He., Wf.) ; and loss in the case of one reagent (Me.). Susceptibility to fatigue, as is inversely indicated by the "Fatigue Index" (see p. 82), ranged from 87 to 97, placing the reagents into three groups: 95, 90, 85. Complete resistance to fatigue would be 100. Per cent of change in resistance to fatigue was: Regular 1st Control 2d Control Group I. SI. 18.5 Le. -9.5 Ms. -5.1 Wf. -1 1.7 Wx. -1.9 Group 2. Ly. 5.6 Cr. I.I He. -44 Rt. -54 Me. -i.o Group 3. Mn. -3.5 The agreement between the two tables showing the "state" of fatigue and the "susceptibility" to fatigue in the cases of Mn., Le., Rt., SI., Cr., Ms., and Wf. indicates a reliability for the test as a measure of fatigue which is supported, in part at least, by the fact that the final tests were taken at the end of the school-year when most of the reagents could be expected to be working under greater fatigue than when the first tests were taken. But there are disturbing influences : He. gained 5 % in "Total Efficiency" and lost 4.4% in resistance to fatigue; Ly. lost 2.6% in "Total Efficiency" and gained 5.6% in resistance to fatigue; SI. gains in both, but made in his final test the remarkable "Fatigue Index" of 115, which is supported by his remarkable 'fatigue curve' drawn from the averages of the respective intervals in all series of the final test, which shows a marked practice-effect. FORMAL DISCIPLINE 167 No relation between the results of this test and the training or improvement in attention is evident. f. Extent of Variability In the preceding analyses of processes it was shown (a) that in almost every test individual reagents differed from each other, often greatly, in the way in which they performed the work of the test;^^^ and (b) that it is the rule for the individual ^"That individual variation in kind of mental work performed, when re- agents set themselves to the same objective task, is not peculiar to the material or the method of our tests, may be seen by inspecting any report of investiga- tion in which the mental processes of the different reagents are subjected to analysis. The contributions to individual psychology and the studies of "imagery type," as has already been noted (foot-note to p. 69), reveal this qualitative variability in every class of mental activity. To select a few typical references for further explication, it is shown specifically in discrimination of clangs (Whipple: An analytical study of the memory image and the process of judgment in the discrimination of clangs and tones. Am. Jr. Psych. 1901. 12:425-433, 448-452) ; in memory for sounds of familiar things, presented by a graphophone (Kuhlmann: On the analysis of auditory mem- ory consciousness. Am. Jr. Psych. 1909. 20:i94ff), in memory for mean- ingless visual forms (Kuhlmann : On the analysis of the memory consciousness. Psych. Rev. 1906. I3:3i6ff), in memory for nonsense-sylla- bles (Miiller und Schumann: Experimentelle Beitrage zur Untersuchung des Gedachtnisses. Zeits. f. Psych. 1894. 6 :303-5 ; Pentschew : Untersuchungen zur Okonomie und Technik des Lernens. Archiv f. d. ges. Psych. 1903. 1 :4i7ff ; Ebert und Meumann : Ueber einige Grundfragen der Psychologie der Uebungsphanomene im Bereiche des Gedachtnisses. Archiv f. d. ges. Psych. 1904. 4:iff; von Sybel : Ueber das Zusammenwirken verschiedener Sinnesge- biete bei Gedachtnisleistungen. Zeits. f. Psych. 1909. 53:327ff), in memory for various material (Bingham: Memory. Psych. Rev. 1894 i:46iff; Whitehead: A study in visual and aural processes. Psych. Rev. 1896. 3 :258ff ; Gamble : Study in memorizing various materials by the reconstruction method. Psych. Rev. Mon. 1909. No. 43); in word-association (Galton: Psychometric ex- periments. Brain. 1879-80. 2:158; Calkins: Short studies in memory and in association. Psych. Rev. 1898. 5:460; Mayer und Orth : Zur qualitativen Untersuchung der Association. Zeits. f. Psych. 1901. 26: 1-13; Wreschner : Die reproduktion and Assoziation von Vorstellungen. Zeits. f. Psych. Erg. Bd. 3, 1907, S. 86ff; Koffka: Ueber Vorstellungen. 1911) ; in imagery of things (Philippe: Un recensement d'images mentales. Rev. Philos. 1897. 44:510; Lay: Mental imagery experimentally and subjectively considered. Psych. Rev. Mon. 1898. No. 7 ; Slaughter : A preliminary study of mental images. Am. Jr. Psych. 1902. 13 :526ff) ; in spelling (Abbott : On the analysis of the memory consciousness in orthography. Psych. Rev. Mon. 1909. No. i68 JOHN EDGAR C DOVER reagent to vary his processes while at work on a test and often to radically change, in the final test, the methods of work employed in the first.^^^ What the variations in processes were has been shown in some detail. Some indication of the extent of radical change in method may be indicated by the following table which lists the cases of such change when it occurred between the first and final tests : Tests I 2 3 4 5 8 9 10 13 17 Total No. of reagents II 12 12 12 II lO 12 12 12 12 ii6 No. Changed 6 5 4 O 2 6 4 8 5 3 43 The table shows that in the ten tests in which the results were the more regular, of ii6 difference-scores 43 (35%) "^ere affected by change in methods between the first and final tests. The changes are distributed over the reagents as follows: Regular— Mn. 5, Le. 5, Rt. i, SI. 2, Ly. 3, He. 3, Cr. 4, Al. i ; I St Control — Ms. 8, Wf. 4; 2d Control reagents — 3, and 4; which indicates that some reagents are more prone than others to radically change their methods of work. Introspections for some of the tests (No.'s 6, 7, 19) were not sufficiently full in detail to indicate what the changes were, although large variability among the scores of the series of which the tests were composed indicated that they occurred; 44:i27ff.). For especially good analyses the reader is referred to Binet (L'fitude Experimentale de I'intelligence. 1903. Pp. 282, 246, 306-7), Segal Ueber den Reproductions Typus und das Reproduzieren von Vorstellungen. Archiv f.d. ges. Psych. 1908. I2:i75ff), and particularly Miiller (Zur Analyse der Gedachtnistatigkeit u.d. Vorstellungsverlaufes. Zeits. f. Psych. Erg. Bd. 5, 1911 ; 8, 1913). The evidence in these references is based almost wholly upon analysis of adult introspections ; but it seems highly probable that functional variability is also extensive among children. Binet {op. cit.) found this true of his two daughters, and Winch (The faculty doctrine, correlation, and educational theory. Jr. of Phil. Psych, and Sci. Methods. 1911. 8:377) infers it from the fact that the coefficient of correlation was low for the early series of learning while it was high for the late series, a fact which when found by Hollingworth (Individual differences before, during and after practice. Psych. Rev., 1914, 21 :8) with naive adult reagents, was used to show "that we are not, in early trials, measuring the same thing with all performers." "* Change in processes during practice in discussed on pp. I76ff. FORMAL DISCIPLINE 169 and the results were so extremely irregular in some of the other tests (No.'s 11, 12, 14, 15, 16, 18) that any statistical treatment of them seemed useless. The extent of variability between reagents may be shown by the extent of the range of initial capacity for each test. If this is reckoned in per cent of the lowest average (the highest capacity for a time-unit; the lowest for a work-unit), we get the following table: I. 60 5. 86 10. 149 16. 82 * 135 6. 85 12. 425 17- 122 2. 150 7. 70 13. 76 18. 115 3- 204 8. 133 14. 13s 19. 39 4. 33 9- 133 IS- 452 ** II ^Relative variation of reaction time. **Fatigue Index. In Test I the range of initial capacity in reaction time was 60% of the highest capacity; in relative variabiHty 135%. In Test 8 the range, in memory of sounds, was 133% of the lowest capacity. (Uncertainty in scoring ruled out the results of Test II). The range in initial capacities is usually large. In only three cases is it below 50% of the lowest average; in six cases it is between 50% and 100%; in eight cases between 100% and 200%; in one case 204% and in two cases over 400%. Thus, in over half of the tests the highest initial capacity is more than double the lowest.^^'* "* Our tests are not peculiar in showing frequent wide ranges in initial capacity. On p. 67 we had occasion to refer to Sleight's (op. cit. 410-411) test on "points" ; the range in initial capacities for the three schools is shown by Table I to be 27-93, 32-129, 23-132, respectively. Binet's (L'fitude experi- mentale de I'intelligence. 236ff) two daughters upon the test of marking out the letters a, e, d, r, s, from French text made initial scores of 23.1 and 53.4 respectively; and his six dull and five bright boys upon the same test (At- tention et Adaptation. 364!?) ranged from 61-165 and 68-138 respectively,* Ebert und Meumann's (op. cit. 15, 47ff) six reagents ranged in memory span for numbers 5-9, in learning nonsense-syllables 13-41 presentations, note-form^ symbols 25-60, sickle-form symbols 33-75. Thorndike and Woodworth's (op. cit. 253ff, 385, 556) reagents ranged in average error in estimating areas,. 7.7-24.2, 10.5-28, 9-21.9, 5-47-8, 12.6-47.4, 4-5-I4-7, 20.1-37.8, 23.7-62.6, etc.; and in 170 JOHN EDGAR CO OVER Initial capacity, as the term has been used in these pages, denotes the average of the scores of the series of experiments constituting the first test. And since the results of all the tests, except No.'s 2, 3, 11, 15 and 16, were made up of averages of from four to ten scores of individual experiments or of series of experiments, the variation in initial capacity as here represented is not as great as would be shown had the initial scores been selected. The result of (a) change in processes between first and final tests and (b) the variability in initial capacity is a great reduc- tion in the strictly comparable difference-scores between the eight regular and the four control reagents. If the number of cases is selected from the tables reproduced in the discussion of The Test Results above (pp. io6ff.), we get the following table: Exp. Groups Group Regular 1st Control 2d Control Tot; 1. 4 2 2 I II a. 6 12 3. 4 4 4 I 12 4- 4 I 2 I I 12 5. 3 I 4 I II 6. 3 I I 2 II z 5 I 3 I I 7. 3 I 2 I II 2 3 I 3 r 2 8. 4 ID 9. 4 2 3 I 12 XO: 4 12 »3. 4 2 I I 12 17. 4 2 2 I I 12 19. 5 3 5 I II Total 36 12 9 149 To interpret the table, variation in initial capacity classified the reagents in Experiment i into 4 groups ; the only results of both regular and control reagents to fall within a single group were two of the regular and one of the 1st control reagents in Group 2; 11 reagents took the test. "time in marking out words, 170-232, 175-306. Culler's (Interference and adaptability. Archives of Psych. 1912. 3: No. 24:16) nine reagents in the FORMAL DISCIPLINE 171 The tables of relative variation in reaction time (of Test i) and the "Fatigue Index" (of Test 19) are omitted in this table. The aggregate of cases in which the difference-scores fall into some group in which comparison can be made is 57 (38%) out of a total of 149. And if the averages of tests in which the results were sufficiently irregular to be omitted, are combined with these, out of a total of 21Q difference-scores we have hut 57 {26%) which are comparable. g. Causes of Variability The more general causes of variability in a reagent's work were conditions of health, relative freshness or fatigue, emo- tional conditions, attitude toward the work of the test, etc. Incidental causes occasionally occurred such as, cold hands in reaction tests, winking at the moment of a rapid exposure, accidents in manipulation of keys, or in handling pen and paper in recording, unusual distraction, influence of preceding laboratory work (He. and Wf. in reaction to sound), mental practice (Wf, in typewriter reaction), etc. More specific causes varied in accordance with the nature of the work in the test, but may be described in general terms as, voluntary or undesigned shifts of the attention to various elements of the processes engaged, changes in the extent of the distribution of the attention over part-processes and their coordination, con- structing of more adequate methods, and practice-effect in dropping out of the process unessential factors, in heightening sensitivity, discrimination, reproduction, habituation to distrac- tion, and in building up habits of higher order. To these causes are due the serrated aspect of the practice curve of a reagent in any of the tests. The causes of variation in the results of different reagents must lie in individual variation with respect to the factors described above, and to others in addition to them. An impor- first group ranged in averages of the first five experiments in the typewriting practice, 20.8-49; the 7 reagents of the second group, 33.2-131 ; and the 8 re- agents of the third group, 28-60. These few samples show the tendency which is likely to be found in any table of initial capacities in mental tests. 172 JOHN EDGAR COOVER tant place among the latter must be given to the 'Aufgabe,' or the task held in the mind of the reagent: Simple reaction, for example, may be 'sensorial' or 'muscular' ; and word- completion may involve self-imposed limitations; etc. Radically different methods of work account for large dif- ferences in intial capacity or in per cent of improvement (even when based upon equal initial ability). The low initial capacity of SI. in card-sorting was due to marked variation in kind of work from the other reagents — his difficult map-scheme used for discriminating the cards and for locating the compartments was the prime element in the difference. The low initial capacities of Ly. and SI. in the typewriter- reaction were owing to different work, the prime difference being a difficult coordination of letter and key which interfered with a coordination of the two parts of the process — discrimination, and choice of reaction. The radical difference between kinds of work was often pointed out in the discussion of Test Results and the effect upon the averages shown. Memory of sounds was effected by kinaesthetic-auditory imagery of the names, by seizing the series as chimes, or by associating the number-names with a visible series; 12-letter-rectangles were learned through a combination of kinaesthetic and visual imagery, or through rep- resentative imagery (associations). In the distraction tests (No.'s 14, 18) the difference in the effect of the distraction depended upon variable method in avoiding conflict between the visual or kinaesthetic imagery of the retained letters and the kinaesthetic imagery of the partial sums in the adding process. And the groups into which the reagents were classified according to initial ability, in the typewriter-reaction test, were shown to be valid by analysis of the kinds of work. Great difference in per cent of improvement was also often shown to be due to difference in kind of work. Sl.'s map-scheme in card-sorting and Ly.'s difficult coordination of letter and key in the typewriter-reaction test, prohibited rapid improvement. The anomalies sometimes found belong here : such as the gain FORMAL DISCIPLINE i73 of Ms., a control reagent, in Test 17, which was greater than that of Al. who trained in the test material for 8 days, — they did not do the same kind of work; Al. consistently gave atten- tion to the whole card, Ms. to a limited part of it; and the gain of the same reagent (Ms.) in Test 13, which was greater than that of either of the two reagents who trained on the test material for 18 days,^ — she changed to the use of associations. As to the source of the more specific causes of variation, it lies in part in the practice on the tests themselves, in smaller part in the training (for the regular reagents), but in greater part in earlier experience. Change in the process seems to be effected through the selective function of the attention, by way of adaptation. The need of a discriminating mark for the cognition of the symbols on the cards was met by the selection of map-directions, on the part of SI., from his school- room experience and the relation between diameter and radius, on the oblique pair of cards, was selected from the geometrical experience of all reagents except Le. whom the pair confused; the schemes of classification of compartments in card-sorting, for Wf. and Mn., and of keys or fingers in typewriter-reaction, for Wf ., were adaptations of mathematical relations ; the grasp- ing of series of sounds as chimes, in memory of sounds, was an adaptation of musical experience; Ly.'s 'sensory set' of consciousness in typewriter-reaction was selected from her train- ing-effects in reaction to sound; and many other changes of method in the final tests were due to the selection of elements that had become prominent in experience since the first tests, often in the training of the regular reagents. h. The Practice Curve It was pointed out that the difference-scores of two reagents, even when initial ability had been the same, were strictly com- parable only upon the assumption that the respective ways of performing the task were equally susceptible to practice-effect. The fact that scores of different reagents measure different processes has its parallel in the fact that scores of the same reagent measured different processes; not only because the 174 JOHN EDGAR COOVER reagent changes these processes at random, by design and other- wise, in order to hit upon a more adequate method, but because practice-effect itself involves change in processes. This was particularly pointed out in the discussion of the Training Results (pp. 82ff.). The scores at the beginning of Al.'s training measured letters clearly seen; at the end, letters 'matured' from 'fringe' content of consciousness. The conquest of the 'fringe' content was the means of raising the scores of Mn. and Le. by supplement- ing the maximum perception and retention of clearly seen letters, and was the proper practice-effect in Al.'s training. Al., however, had not recovered, by the last day of training, from the disastrous effect of this direction of effort, upon the fixing and retaining of clearly seen letters which contributed his highest score on the third day. He was working under the disadvantage of striving to effect coordination of these part-processes and had not yet succeeded. The early scores in Rt.'s training in learning 12-letter-rect- angles measured letters recorded from kinaesthetic-auditory, or kinaesthetic imagery alone (see Analysis Curve, Appendix B. Fig. 17, p. 293). Then, additional letters from visual imagery began to contribute to the score and by the 7th day they reached their maximum. From the 5th day "visual asso- ciations" began to contribute as many letters as did the visual imagery, which had maintained its level. The scores at the beginning of the training measured letters from one kind of imagery — homogeneous letters, — at the end, letters from four distinct kinds of imagery — heterogeneous letters}'^^ How this change in the material measured was brought about, illustrates accurately what the practice-effect involves: Rt. at the beginning got a strong visual impression of the letters and then converted it, by pronouncing the letter-names, into kinaesthetic-auditory imagery for retention and reproduc- tion, repeating letter-names over and over during the lo-second ""The transition from simple imagery to mixed or complex imagery through practice in memory work was shown by von Sybel (Zeits. f. Psych. 53:338). FORMAL DISCIPLINE 175 interval between perception and recording. But on the first day he found that he could economize time and effort by con- verting the stimulus immediately into kinaesthetic-auditory imagery and by giving a rhythm to the repetition. The effort to make a strong visual impression was dropped out. This simplification of method was accompanied by a simplification of the kinaesthetic imagery itself through dropping away of the auditory component which formerly supported it. If visual imagery found place at all, it merely supported the kinaesthetic upon the same letters. The reagent found that the lo-second exposure limited the kinaesthetic method of fixing the letters adequately for recall to the first six places. But after repeti- tion had facilitated this process through a readier recall of letter-names in the naming process, and through a more definite rhythm in fixing the letters for recall, some time was gained which permitted attention to note other letters; these extra letters were recorded from visual imagery. This was the first coordination of different kinds of imagery from each of which different letters were recorded. This two-fold coordination then gave way to a three-fold coordination in which additional let- ters were recorded from "visual associations." The first oc- currence of the associations involved the letters CP, which as soon as seen stood for "Chemically Pure," When record- ing, these letters were found to be very vivid by reason of their association with the familiar phrase; and finding asso- ciations became a method. Another three-fold process devel- oped through the habit of intensifying the visual impression of some additional letters while repeating by rote the kinaesthetic group; when recording, it was found that other less vivid letters could be recalled from visual imagery besides those which had been intensified during perception. This three-fold process con- sisted in the coordination of kinaesthetic imagery, intensified visual imagery, and a secondary visual imagery which seemed to persist in its own strength. By the loth day a four-fold process was occasionally employed which coordinated all the kinds of imagery already mentioned. Single-fold imagery was dominant during the first three 176 JOHN EDGAR COOVER days; two-fold coordination was dominant during the 4th, 5th, and 6th days; three-fold coordination became dominant after the 7th day; and four-fold coordination appeared on the loth day after which it supplemented the three-fold imagery when associations were not found to hold sufficient letters. The single-fold method recurred on the 9th and 14th days; both days of reduced scores {vid. Analysis Curves, Appendix B. Fig. 16, p. 292). The greatest improvement as shown by the practice curve was coincident with the greatest use of the three-fold and four-fold methods; that is, with the highest coordination of the different kinds of imagery. And the rise of the curve is dependent throughout upon the increasing use of the auxiliary forms of imagery. On the nth and 12th days, when the practice-curve first reached its higher levels, as many letters were reproduced from the auxiliary imagery as from the kinaesthetic imagery, while on the first three days less than a fifth as many were so reproduced. Other effects of training were: (i) the method of assigning to associations favorable letters other than those occupying the usually preferred 7th and 8tli places on the card, (2) the method of recording the letters held in the weakest imagery first, and (3) a more adequate coordination of the recording and reproducing part-processes so that writing down some of the letters ceased to be a distraction on account of which other letters escaped recall. Per cent of improvement in this training evidently measures change in processes; not merely the more radical changes at the beginning, involved in adaptation to a new kind of work, hut those smaller and more orderly changes which constitute practice -effect. ^ ^® "*This fact of change in processes during practice is amply supported by the literature, typical references of which follow: Talbot (An attempt to train the visual memory. Am. Jr. Psych. 1897. 8:414-7) by exercising visual recall improved her memory which made more frequent use of visual elements than it had done before. Culverwell (The creation of a memory. Jr. Exp. Ped. 1911-2. 1:160-1) reports an interesting case of improvement through practice in changed mode. Change in processes during practice was shown in reaction-time to words by Berger (Ueber den Einfiuss der Uebung auf geis- FORMAL DISCIPLINE . 177 Practice-effect in the tests is a source of variability in the test-averages of different reagents. It was not operative in tige Vorgange. Phil. Stud. 1889. 5:170-178); in simple reaction-time by Angell and Moore (Reaction time: A study in attention and habit. Psych. Rev. 1896. 3:249-252); in discrimination of clangs by Whipple (An analy- tical study of the memory image and the process of judgment in the discrim- ination of clangs and tones. Am. Jr. Psych. 1901. 12:448); in learning paired associates of non-sense syllables by von Sybel (Ueber das Zusammen- wirken verschiedener Sinnesgebiete bei Gedachtnisleistungen. Zeits. f. Psych. 1909. 53 :338) ; and in memorizing non-sense syllables by Ebert and Meumann (Ueber einige Grundfragen der Psychologie der Uebungsphanomene in Ber- eiche des Gedachtnisses. Archiv f.d. ges. Psych. 1904. 4:202f¥, 2iofif, 228). Philippe (Sur les transformations de nos images mentales. Rev. Philos. 1897. 43 ^492) and Bentley (The memory image and its qualitative fidelity. Am. Jr. Psych. 1899. 11 :47-8) pointed out the characteristics of instability and mutability of imagery which would affect recurrent processes in which it plays a part, and Kuhlmann (On the analysis of auditory memory conscious- ness. Am. Jr. Psych. 1909. 20:i94ff) found that in later than immediate recall of sounds auxiliary visual imagery became more frequently the means of recall. With respect to changes in the processes of recall dependent upon degree of learning J. R. Angell (Determination of mental imagery. Psych. Rev. Mon. 1910. No. 53:70) wrote: "If I am obliged to repeat the words before they are completely learned, my recall is likely to be dominated by visual processes. On the other hand if I am allowed to proceed until the learning is quite perfect, the recall is likely to be mainly in auditory-motor terms, and the more perfectly automatized the act becomes, the more I lose the visual element. Judged at one stage of the process, I should then be set down as a visualizer; judged at another stage, I should be auditory-vocal- motor." But the most conspicuous cases are shown in studies of learning, in which well-defined stages are related to the curve of practice : Bryan and Harter : Studies in the telegraphic language. Psych. Rev. 1897. 4:27-53; 1899. 6:345-375. Swift: Beginning a language. A contribution to the psychology of learn- ing. Studies in Phil, and Psych. (Garman volume) 1906. 2^7-313, 304flF. Book : The psychology of skill, with special reference to its acquisition in typewriting. Univ. Mont. Bull. 1908. 53:1-188. Swift: Learning to telegraph. Psych. Bull. 1910. 7:149-153. Ordahl: Consciousness in relation to learning. Am. Jr. Psych. 191 1. 22:i58ff. Kline and Owens : Preliminary report of a study in the learning process, involving feeling tone, transference, and interference. Psych. Rev. 1913. 20:222-3. Cleveland : The psychology of chess and of learning to play it. Am. Jr. Psych. 1907. 18 :297. 178 JOHN EDGAR COOVER some of our tests, but it was clearly so in others. If it was much greater in the first than in the final test, initial capacity, as expressed by the average score of the first test, is lower relatively, than in the case of about equal practice-effect or no practice-effect in the first and final tests; the difference-score will be relatively too large. The extent to which this source of error was operative in our tests, as well as the frequency of practice-effect in both tests, is shown by the following table: Tests I 4 5 6 7 8 9 lo 13 17 19 Total a. 4641201124328 •b. 05100200221 13 a. Much greater practice-effect in the first test than in the final. b. Practice-effect in both tests. In these eleven tests there were 125 difference-scores; 28 (22%) of them are too large because of the greater practice- effect in the first test; and practice-effect occurred in both first and final tests in 13 (10%) cases. The distribution of the cases over the reagents is as follows: Regular Control a. b. a. b. Mn. 2 3 Le. 4 I Rt. 2 I SI. 2 Ly. 2 I He. 2 I Cr. I 2 Al. I 2 Ms. 4 Wf. 2 4 I — 2 I To be safe in the comparison of difference-scores, such tests should be chosen as are free from the rapid practice-effect of adaptation, or training should continue until initial efficiency is more stable and its quantitative expression more reliable.^^^ i. General Effect of Special Practice After seeking to avoid error by making the difference-scores as nearly as possible fairly comparable, our quantitative data "'For further discussion of this matter, see p. 22if. FORMAL DISCIPLINE 179 for generalization upon the general effect of special practice of the trained reagents are greatly reduced; and in view of the fact that even they are not decisive, our generalizations must remain statements of probability. With this caution in mind we may examine the cases in which training-effect seemed to show itself in improvement in the tests. These cases are collected in the following table : Training Reagents Tests 7,* 9, 10. Tachistoscopic Mn. Le. I7 2, 5,* 8. 4, 6, Al. 3, 4.* 8, 9. Learning 12-letter-rectangles Rt. I.* 5, 6, 8, SI. 7, 10. 'Reaction to sound Ly. 4,* 10, 13- Memory training He. 4,* 6, 9, 17- Cr. 4,* 5, 6, 17- *These cases were influenced by greater practice-effect in the first test. Common elements in part-processes, or in modes of attention, are not difficult to find: Training on the tachistoscope involved: quick perception (shared by tests 2, 3, 4, 5, 6, 9, 10), keen momentary attention (shared by tests i, 6, 7), reproduction of imagery (shared by 4, 5, 6, 7, 8, 9, 10), coordination of part-processes (shared by 1,4, 5, 6, 8, 9, 10). Training in learning 12-letter-rectangles engaged quick perception (shared by i, 5, 6, 7, 8, 9, 10), rapid kinaesthetic processes (shared by 8, 9, 10), apperceptive grouping (shared by 9, 10), reproduction of imagery (shared by 5, 6, 7, 8, 9, 10)! keen momentary attention (shared by i, 6, 7, 8, 9, 10). keen continuous attention (shared by 5, 8, 9, 10), coordination of part-processes (shared by i, 5, 6, 8, 9, 10). Training in reaction to sound involved quick perception (4, 10, 13), keen momentary attention (10, 13), quick reaction (4). Training in memory engaged keen perception (4, 5, 6, 9, 17), i8o JOHN EDGAR COOVER apperceptive grouping (9, 17), reproduction of imagery (4, 5, 6, 9, 17), continuous attention (4, 5, 6, 9). The probable cases of general effect listed here do not con- stitute a complete list for which there is quantitative evidence, for cases in v^hich great improvement was in part contributed by change in method were set aside; as may be illustrated by two cases of Mn. There is little doubt that she brought to the test on learning 12-letter-rectangles great advantage from her tachistoscopic training, yet her gain (64%) is set aside because of a radical change in her method induced by her training; there is no doubt at all that she improved in her training, yet her gain (147%) in the test identical with it was for the same reason set aside. The requirement of setting such cases aside, it will be remem- bered, is made by the form into which the investigation was cast: the ostensible purpose was to test for the influence of improved conditions of attention, and scores including the ad- vantage or disadvantage of change of method could not be used to measure difference in the capacity of attention. Apart from this requirement, however, there is another ground for discarding these difference-scores, which challenges their value; evidently their service is largely limited to determining the relative advantage of the different kinds of work. But we do not need to rely upon our quantitative data for proof of general effect of special practice. Introspections indi- cate it more clearly and more certainly than difference-scores can. Not only are methods of work, and systems of imagery, transferred from one kind of work to another where they are applicable, as the methods of fixing and reproducing 12-letter- rectangles were carried over to memory of consonants serially presented (Rt.), and to digits simultaneously presented (Rt., SI.), and as the better organization of associations in which letters from 12-letter rectangles were fixed and retained after training in memory (Cr.), but the process of adaptation to the strange work of the test consists in the selection and application FORMAL DISCIPLINE i8i of elements of former experience which, when acquired, must always have been 'specific' and which as applied are always 'general/ This principle is not only true of the exceptional and somewhat bisarre cases cited to illustrate unusual variations in the scores, as Sl.'s map-directions in card-sorting, and the grasping of series of sounds as chimes, but it is true of the usual and regidar processes of learning to do expertly the work of the test. The general effect was usually advantageous, as was often pointed out in connection with discussion of change of method, under Test Results, but it is not necessarily so; Ly.'s 'sensory set' of consciousness carried over to the typewriter-reaction from reaction to sound, the attempted transference of training method by Rt. and SI. to commercial signs presented in the method of training,^^^ were disadvantageous. Wf.'s complex scheme in typewriter-reaction, and Sl.'s unwieldy map-scheme in card-sorting, will serve to illustrate the fact of negative influence. Introspections indicate that the 'spread of training' need not be conscious,^^^ as may be illustrated by Cr.'s improvement in learning 12-letter-rectangles which was largely due to better organization of the associations used to represent the letters, a prime factor in the memory-training, yet he was not conscious of applying training-effect. Indeed, introspections in this type of experiment are not of value because they assert or deny the influence of training upon the tests, although such statements may often be true, but because they describe the processes en- gaged in both training and tests fully and accurately enough for the presence or absence of specific influences to be determined. j. Conclusion Variability in the mental processes engaged by the reagent on any test was found to be universal; radical change in the "'Agreeing with the results of Sleight {op. cit. p. 440) that elements may not be available where the whole form will not apply. The fact seems true particularly when the coordination of part-processes has become automatic or mechanical. "* Again in agreement with Sleight (op. cit. p. 440). i82 JOHN EDGAR COOVER processes between the first and final tests, for the ten tests in which the results were the more regular, occurred in over a third of the i66 difference-scores {vid., p. i68). Great variability in initial capacity, as represented by the average of the scores of the first test, was also universal; in twenty tables, there were but three cases in which the range of variability was less than half the lower average, 9 cases were between 100 and 200% of it, and two cases were over 400 9& of it {vid., p. 169). The effect of these two kinds of variability was to reduce the comparable difference-scores to a fourth {vid., p. 171). The causes of variability were general, such as health; inci- dental, such as accident in manipulation of a key or of cards; and specific, such as are involved in adaptation, or in practice {vid., p. 171). Great difference in the scores of an individual and great difference in initial capacity are signs of great difference in kind of work {vid., pp. 172). The source of the more specific causes of variation lies in part in the practice-effect of the tests, in less part in the practice- effect of the training, in greater part in earlier experience. Change in process is effected through the selective function of attention, by way of adaptation, and of acquiring skill through practice. Adaptation occasions the more sudden and radical changes, as abrupt change in method; skill through practice involves a more or less orderly sequence of changes depending in nature ujpon the task but always resulting in a marked difference in the kind of work performed in its initial and final stages {vid., p. 173). Practice-effect in the tests disturbs the quantitative results; if it is greater in the first than in the final test, the difference- score is too large. Any average which includes it and purports to measure efficiency is not strictly reliable. The extent of this source of error in 'mental tests' may be indicated by the fact that of 125 difference-scores, 22% were affected by greater practice-effect in the first test, and 10% by practice-effect in both first and final tests {vid., p. 177-178). FORMAL DISCIPLINE 183 Recognizing the inconclusive character of our quantitative results, even after the difference-scores were made as comparable as possible, we found 31 cases (7 of which were influenced by practice-effect in the first test) of probable transference of training-effect to the tests, some of which were greatly different in material and method from the training. In all, however, common elements in part-processes, and in modes of attention, were numerous {vid., p. i/Sff.). Introspective data are decisive in indicating the transference of methods of work and systems of imagery^^^ from one kind of work to another, where applicable, and that adaptation to the strange work of a test consists in the selection and applica- tion of elements of former experience which, when acquired, must always have been 'sjpecific,' and which as applied are always 'general.' This also appears to be a principle of the learning-process, in which skill is attained by smaller and more orderly changes in the processes {vid., p. i8of.). Transference need not be consciously effected. Introspec- tions are of value chiefly not because of asserting or denying it, but because they describe the processes fully and accurately enough for its presence or its obsence to be seen. And its effect is sometimes negative {vid., p. 181 ). As a measure of attention our tests are inadequate, and the question of transference of improved conditions of attention remains open. That our quantitative results were not more conclusive in showing the effects of training on attention may be owing, in addition to the disturbing influence of great varia- tion in processes and scores, to certain limiting conditions under which the experimentation was conducted: (a) Change in method between tests rules out the results as a measure of change in attention; (b) our reagents were university students whose habits of attention were pretty well established by former training; (c) our trained reagents were more mature than the control reagents and were experienced in laboratory work; (d) our period of training was relatively short (six weeks, three days per week) ; (e) our tests could not be long enough to give a reliable measure of initial ability, because of inter-test ^"'' In agreement with Fracker {op. cit, pp. goff.). i84 JOHN EDGAR COOVER practice-effect; and (f) adequate introspections were limited almost wholly to the trained reagents. The contribution of the experiment lies principally (i) in revealing the nature and the extent of variation in mental processes, both between different reagents and with an individual reagent, which may be expected when such 'tests' as ours are given under favorable laboratory conditions to intelligent young men and women; (2) in showing how such variation affects the scores; and (3) in offering suggestions toward improvement in this type of experiment. The interesting question of how statistical method is affected by these considerations is discussed later (pp. 219 ff.). 2. Experiment on Reproduction The purpose of this experiment" was (a) to supplement the Experiment on Attention by a more intimate knowledge of the factors of variation, through greater refinement of introspection and analysis, and (b) to set the conditions to test for a subtle but definite 'spread of training.'^-^ To meet the former aim, tests were chosen or devised which offer opportunity for introspection separately upon the successive 'moments' into which they naturally fall, or for fuller descrip- tion of the processes used in acquisition and reproduction by reason of the material used in them. To meet the latter aim, the tests and training were arranged to exclude the transference of methods of work or of systems of imagery, such as belong rather to the grosser and more radical changes in work, due to what was termed "adaptation," than to those smaller and more orderly changes which constitute practice-effect proper. It will be remembered that the most definite cases of the general effect of practice which came to notice in the preceding "Performed during the year 1911-1912. *^ For the distinction between 'Spread of Training' and 'Transference,' see footnote to p. 225, and the text on p. 230. FORMAL DISCIPLINE 185 experiment were of the nature of 'transference,' but that there was also some indication of 'spread of training' ; and that it was this latter sort of general practice-effect that was found in the earlier experiments on Marking-out Words, Discrimina- tion, and Reaction with Discrimination and Choice. The susceptibility to improvement of the capacity to 'mature' and reproduce weak imagery in the tachistoscopic training, in the Experiment on Attention, suggested reproduction of imagery as the work of this experiment. ^^^ Consequently, tachistoscopic and memory tests were devised and made to differ radically in method and material from the training, in order to avoid 'transference' and to invite 'spread of practice' ; a further test which differed from the training but slightly in material, and method, was included to determine if so slight a difference might cause interference. The training chosen was sound discrimination, and in order to insure practice in reproduction of imagery of some sort, the time-intervals between the two sounds to be discriminated were varied between 7 and 60 seconds, and series of sounds were made to vary widely, as a whole, in intensity. The tests were designed to measure the capacity to reproduce imagery occasioned by the presentation of materials of different "^ This process, or group of processes, so far as it lies above the threshold of consciousness, may be represented by a tension of attention under the influence of 'determining tendencies' analogous to the experience of recalling a name that seems close but delays in coming ; conceivably, how^ever, it also takes place wholly under the threshold, operates in part from "Unconscious psychical stimuli" (Lipps: Grundtatsachen des Seelenlebens, S. 125), involves as subtle processes as the "unconscious associations" reported by Scripture (Ueber den Associativen Verlauf der Vorstellungen. Phil. Stud., 1892, 7 78, 136), Jerusalem (Beispiel von Assoziation durch unbewusste Mittelglieder. Phil. Stud. 1894. 10:323-5), and Thomas (Ein weitres Beispiel. Zeits. f. Psychol., 1896, 12:60). Vid. Footnote on p. 87. Statistical evidence of the influence of subliminal impressions upon judgment is to be found in experi- ments in discrimination (where Right Cases fall off regularly with the magni- tude of D) {7nd. Appendix D, p. 299, and Peirce & Jastrow, quoted by Donaldson in The Growth of the Brain, p. 292), and has been reported from experiments on Guessing by Sidis (Psychology of Suggestion. 1898. 168-171) and Stroh, Shaw and Washburn (A study in Guessing. Am. Jr. Psych., 1908, 19:243-245). It is perhaps not necessary to point out that provision for introspection in this experiment is for other purposes than for a direct determination of the presence of these subtle processes or of increase in their efficiency. i86 JOHN EDGAR COOVER kinds, and the training was designed to develop greater power in reproduction of imagery of a certain simple kind. The method of experimentation was the same as the preceding. Three groups of reagents took the series of tests before and after an interval of from five to seven weeks. The first group took training during the seven weeks; the second group took all the tests before and after a free interval of five weeks; the third group took but one pair of tests each before and after a free interval of five weeks. Each pair of tests was therefore taken by three trained reagents, three control reagents who took the other tests also, and two control reagents who took no other tests. The trained reagents (Hs., Wn., Rt.) were seniors in Psy- chology, English and Education, respectively, who were doing advanced laboratory work in Psychology. The first group of control reagents (Al., Hhs., Br.) were a freshman and a sophomore in Psychology, both pursuing elementary laboratory work, and a freshman in Economics who had no laboratory experience. The second control group (Ck., Pn,, An., Ty,, Dn,, Hn., Hd.) included five seniors, one sophomore, and one fresh- man, representing the German, History, Education, Economics, and Pre-Legal departments of the university. a. The Tests Test I. Recognition or choice of one of two letters. The purpose of this test was to give a measure of the liability^^^ of reproduction. The apparatus was the same as that used in the Tachistoscopic test (No. 17) of the preceding experiment. The method was (a) to expose a 12-consonant-rectangle o.i Sec, (b) after a 3-sec. interval to expose two letters printed like those printed on the rectangle, one of which was to be chosen by the reagent as having been on the card and to be recorded in the section of a ruled form corresponding as nearly as possible to its position in the rectangle, (c) to time with a stop-watch the interval between the exposure of the two letters and the vocal choice of one of them. Rectangles were presented at the rate ^For use of this term, vid. Kiilpe: Outlines, p. 197. FORMAL DISCIPLINE 187 of one a minute. Fifty experiments constituted the test. All judgments carried an index of one of four grades of certainty as to the letter's existence on the card and as to the position in which it was recorded. Introspections were made upon four different intervals or moments of the experiment: (i) From "Ready" to the stimulus, (2) Perception, (3) On the interval of 3 seconds between perception and the appearance of the two letters, (4) The moment of recognition or choice. The two letters shown for choice were so selected that they were distributed evenly over the 12 spaces of the card, that each letter was on the card one-half times as often as it was exposed for choice, and that it occupied the second position as often as it did the first in the presentation for choice; thus it would be possible to learn if any part of the card is favoired in perception, if there were favorite letters, if the right or left letter presented for choice is favored, all of which would be sources of error, and under the conditions of the experiment would tend to bring the score of R cases toward the probability figure of 50% ; Reproductive tendencies would be shown by the excess over 50%. Fifty experiments constituted the test. Test 2. Reproduction and recognition of letters on 12-letter- rectangles. The apparatus was the same as that used in the pre- ceding test. The method was only slightly changed from that of Test 17 in the Experiment on Attention. 12-Consonant- rectangles were exposed o.i seconds; 10 seconds were given in which to record the letters perceived; the card was then re- exposed for 15 seconds while the reagents recorded in pencil other letters recognized as having been seen but which failed to be recalled. The rate of experiments was two minutes. In- trospections upon process and imagery were written and the following intervals of the experiment were treated separately: (i) From "Ready" to stimulus, (2) Moment of perception; (3) Upon the interval before recording, if there was one; (4) Interval of recording; (5) Interval of recognizing further letters. Twenty experiments constituted the test.^^^ "*See Appendix C, p. 295, Method "Text (2)," for values used in scoring the results. i88 JOHN EDGAR COOVER Test 3. Discrimination of sounds. The same as No. 7 in the Experiment on Attention; the same series of intervals were used {vid. Appendix B. Fig. 7, p. 290) and the stimuli were given with the same instrument — the sound pendulum. There were 10 series, of 9 judgments each, in the test. Intro- spections were taken on the process. The per cent of R cases was to constitute a measure of the sort of reproduction upon which training was taken; it being assumed that some sort of reproduction of the first stimulus, or of some function of it, was necessary to render judgment upon the second. Test 4. Memory for visual symbols. Reproductive ten- dencies could be measured here by the amount of correct repro- duction from a single presentation, and by the number of repetitions necessary to learn completely a series of 12 symbols; both measures .were attempted. The Miiller and Schumann memory apparatus was set at one revolution in 44.5 seconds, which presented the symbols at the rate of 1.07 seconds. And when the series was repeated, presentations came at the rate of 44.5 seconds, leaving an interval of about 33 seconds between the end of one and the beginning of another. The number of revolutions was recorded by the apparatus. The symbols were of the same kind as those used by Ebert and Meumann,^^^ one series of symbols resembling notes of music with the flags variously located; the other, of symbols resembling sickles, both elements changing in absolute positions and in relation to each other. The first part of the experiment consisted in reproduction of as many symbols as could be remembered from one presenta- tion each of two series of six symbols; the second part, of learning two series of 12 symbols by heart. b. The Training Three 4th year students (Hs., Wn., Rt.) took the training in Sensible Discrimination of intensities of sound, which ex- tended over a period of 48 days and consisted of 31, 59, 60 "' Archiv f .d. ges. Psychologic. 1905. 4 : 49 ; samples of these forms are also illustrated by Thorndike : Educational Psychology, vol. II, p. 369. FORMAL DISCIPLINE 189 series, 9 judgments per series, taken upon 7, 11, 13 days, aggregating 279, 450, 540 judgments respectively for Hs., Wn.. and Rt. The training was taken between 2 : 30 and 3 : 30 p.m., and continued about 45 to 55 minutes. The room during the experiment was kept closed and much darkened, and the reagents kept their positions constant, about 4 meters from the source of the sound, with their backs toward it. They took the training together and secured privacy at a long table by partitions of wooden screens. The sounds were produced by dropping steel balls from magnets upon a steel block, by the use of Krueger's "Fall- phonometer."^^^ Manipulation was noiseless. The weights of the pairs of balls were: I- 5-S6 grams 2. 8.33 " 3. 11.86 " and the heights from which they fell, and their relative intensities of sound were: Intervals cm. Intensity 9 96 1.78 8 84 1-55 7 72 1-33 6 63 1. 17 S 54 I. 4 47 0.87 3 40 0.74 2 35 0.63 I 30 0.56 These steps in intensity are about one-half a noticeable difference. The 54 cm. fall was used for the norm, and it with all the rest for variables; but by using the three pairs of balls, three norms of varying intensities (i.oo, 1.50, 3.13), with their corre- sponding series of variables, were obtained. Judgment was given upon the second sound, and the latter was the norm as often as it was the variable. The time intervals between the two stimuli were equally distributed over 7, 15, 30, 60 seconds, ^ Similar, in principle, to illustration in Wundt : Grundziige d. Physiol. Psych. (5te Auf.) I: 512. igo JOHN EDGAR COOVER These changes in intensity of the norms and in sequence of variable and norm were intended to prevent an easy classification of the first stimulus and to favor its being held in some sort of imagery; and the temporal intervals were made various lengths to make the retention or recall of that imagery necessary, and to facilitate a checking up of this fact by comparison of the number of the Right and Undecided cases for the various intervals. The method was that of constant difference, Right and Wrong Cases; procedure without knowledge. c. Training Results (i) Processes Hs. At the beginning and throughout the training Hs. held the auditory image of the first stimulus, or brought it back with more or less effort, and compared it with the auditory sensa- tion of the second stimulus (i, 32). But there were variations from this general method •}~'^ ^ In the deviations from the simple direct comparison of auditory imagery which resuhed in the training-series of these reagents, it is possible that reproductive processes of some sort found place. Whether the simple audi- tory impressions were supplemented or actually replaced by complicated auditory imagery or imagery from the other modalities (visual, tactual, kinaesthetic), or comparison was dispensed with in 'free' judgments, it would seem that imagery representing the first stimulus, or an organic reaction to it, mediated judgment. That these deviations are frequent, results largely from the fact, pointed out by Whipple (An analytical study of the memory image and the process of judgment in the discrimination of clangs and tones. Am. Jr. Psych., 1902, 13:259), Slaughter (A preliminary study of mental images. Am. Jr. Psych., 1902, 13: 526ff), Kuhlmann (Problems in the analysis of the memory consciousness. Jr. Philos. Psych. & Sci. Meth., 1907, 4:5ff), and others, that the auditory image is not simple and that it sometimes loses itself in its other-modal or organic constituents ; and partly from the impulse, insisted upon by Woodworth (Non-sensory components of sense perception. Jr. Phil. Psych. & Sci. Meth., 1907, 4: i69fTf), to clothe the sensory impression with ideational, perceptive, or emotional attributes. Bentley (The memory image and its qualitative fidelity. Am. Jr. Psych., 1899, 11:7). quoting Kiilpe, reminds us that reproductions are not weakened copies of sensations, that reproduced sensation is schematic, needs aids (words, movements, or- ganic sensations, feelings) to complete recollection, and adds that these aids may become the real vehicle of retention. FORMAL DISCIPLINE IQJ Sometimes the image of the first stimulus is overcome by- other imagery or by distraction: In one instance visual imagery of E manipulating imaginary apparatus crowded it out, after which it was recalled (i). When she was nervous or tired the auditory image was apt to get away, especially during the longer intervals (18,25); then she was particularly conscious of distraction — wind whistling by the windows (26), the stoi>- watch ticking (31), a throbbing noise down stairs (55) which rendered the renewal of the image very difficult or even im- possible. "Lost image entirely . . . was going to judge ' ?' when it returned and I was quite sure that my judgment 'greater' was correct" (22 15 , Norm-Variable 9, 60," Right). Instead of comparing the image of the first stimulus with the sensation of the second she sometimes had the feeling of converting her image of the first into sensation and of com- paring sensations (i), or in uncertainty she compared images of both (26:8); occasionally she interpreted intensities as qualities of pitch and translated tonal imagery into intensity (14, 46); she also based judgments on extraneous imagery as "the feeling in the hands of noise from wooden blocks" (29), and other kinaesthetic images in hands and arms (32:3, V3N — 15''). In some series she did not try to hold the image of the first, but often brought it back after the second stimulus was received (36, 41). In some cases she made 'free' judgments: "I lost the image entirely, yet when the second stimulus came it was very clearly 'greater' ... it was startlingly loud" (39 :2, — NV8 — 60" — R) ; "I could not bring back image of the first stimulus, but I was certain that the second stimulus was 'greater' ; I think I judged more by the way it startled me than by anything else" (42:1, — NV/' — 60'' — R). The judgment 'like' sometimes meant that (i) the "sensa- tion exactly fits in with the retained or renewed auditory image" (i), and is given with a feeling of certainty (21) ; and (2) the identity is doubtful (26). Introspective notes as to certainty in a series correlate with the number of R judgments (28, 29). Relief was felt when the intervals were short; the effort to 192 JOHN EDGAR COOVER renew and strengthen the image of the first stimulus is spared (27:9). Her training was taken under trying circumstances, occasioned by events not connected with her university work, which made her very nervous and given to abstraction. She omitted series 3-17 and 47-49 inclusive. Wn. Particularly at first and somewhat all through the training Wn. retained or revived images of the first stimulus in order to pass judgment upon the second; comparison however was difficult: "I seem to compare the first stimulus I have retained through images, with the second stimulus as soon as I hear the latter. When I have tried to compare them both as images I have found great trouble in keeping each clear in my mind; they have tended to overlap each other" (3). This difficulty in comparing images is mentioned several times (6, 8). Never- theless to pass judgment upon the sensation of the second stimulus, the image of the first seems necessary: "I find that if the first is not in my mind as one image or another when the second is sounded, I have great difficulty in comparing the two. It requires constant attention to keep it there," (6). The holding or reviving images is mentioned in series i, 2, 3, 14, 17, 20, 21, 22, 28, 38, 41, 48. It was resorted to in the heavily typed series (17, 38) upon inability to classify the first stimuli or to attribute to them a personal quality. The images are strong just after the first stimulus has sounded, but soon fade (2), are very hard to retain (6) or to revive after they have disappeared. Often they are recalled by the second stimulus (7, 12, 26). The imagery seems rarely purely auditory. It is usually auditory and kinaesthetic, the latter taking the dominant role, especially in reviving a waning or lost impression (2,3). In this the throat, head, and hands, are chiefly instrumental (i, 4, 7, II, 13, 14, 20, 28, 41). Sometimes the kinaesthetic image is based on a "singing" of the first stimulus, intensity being interpreted as pitch (15, 28, 41, 49); rarely is the pitch held FORMAL DISCIPLINE 193 in auditory imagery of another's voice (43,48). Occasionally the sounds are interpreted as 'stronger' instead of 'greater' (4) ; and value is given in terms of "sensations of being struck" (i). Judgments were sometimes based upon the "startling effect" (3,4a), or a "sensation of surprise," (5). Visual imagery played some part but was usually recognized as a dis- traction (4, 13) ; sometimes, however, as an aid (13, 14, 20, 27). It took various forms: "a piece of steel being snapped" (i, 13, 20) ; "E working with the apparatus" (4, 11, 14, 2y) ; "a piece of steel hitting a bar of iron" (4) ; and such as attended associa- tions involved in attributing personal quaHties to the stimuli (43). Other modes of passing judgment developed early and con- tinued throughout the training. The beginning of the method of classification resulted from a noted "familiar" quality of the first stimulus as a sound that was heard before and is now known (8, 9, 36), and a comparison of the first stimulus with preceding first stimuli (10, 11). This led to classification as "very loud," or "very soft" (10, 11, 14, 18), which developed into a method (20, 30, 31, 41, 49) : "I have developed quite a system of classification of the first stimulus as 'very large,' 'large,' 'small,' 'very small,' etc., so that I have no difficulty in recalling it upon hearing the second. I do not even have to think of it in the interval when the first stimulus is very decided one way or the other" (30). The first stimulus was then re- tained in verbal-motor imagery. Sometimes this method failed, however, and she had to fall back upon auditory-kinaesthetic imagery (33). Upon the break-down of this method, another was developed, suggested probably by an early instance (13:6) in which a loud stimulus brought to mind a very aggressive man of her acquaintance. Qualities of personality were con- ferred upon the first stimulus (13, 35, 36, 37) : "From classify- ing the stimulus in a general way as 'very great,' 'great,' etc., I have come to attributing qualities; as, aggressiveness, timidity, power, weakness, etc. I find that by doing this I can remember the stimulus easily when I want to compare it with the second stimulus and need not think of it during the interval, as is the case when I try to retain it by image" (35). Sometimes, how- 194 JOHN EDGAR COOVER ever, the stimulus failed to suggest personal qualities (38) ; "In such cases I fell back on the images, auditory-kinaesthetic, for a comparison, and was usually able to recall them" (38).i28 The obvious economy of these two methods in the longer intervals of 30'" and 60" is probably responsible for their use. Tree' judgments were seldom made (26.3). 'Like' judgments were usually based upon "similar effects — same kind of sensations" (i), or "the same sensation of surprise" (4). The short intervals {f\ 15'') were a relief (10), the long ones difficult (18, 23, 47), especially because the retained imagery may not belong to the last experiment. The heavier balls were also a relief (16, 22, 46) ; the weaker sounds were hard to retain and hard to classify (33). When the norm was the first stimulus it was often thought to be varying (17). Emotional factors probably played some part in determining judgment: Sometimes the second stimulus was anticipated (5); reagent was worried because there were so many 'less' (20, 43) or 'greater' (24) judgments occurring in the series; or upon giving several 'like' judgments she tried to make the rest either 'greater' or 'less' (32). Conferring personality was an "interest- ing" process (36) and may have urged attributes not appropriate. The 'type' as shown by an assembling of the data is verified by introspection: "I am surer of 'less' than of 'greater' judg- ments" (5). The R judgments for the cases in which the second stimulus was 'less' greatly exceed those of the cases in which it was 'greater.' Rt Rt. insists that no image of the first stimulus remains during the interval (i, 2, 4) : "In fact if I try to keep it in mind I am not sure that I succeed as well as when I don't try to do so" ( I ) ; but "an idea of the intensity of the first stimulus remains — though not in auditory imagery" (1:7). 'T do not carry the image over the interval; I do carry a sort of estimate of the "'That methods and standards of judgment alter in training was shown by Whipple's reagent O. (Am. Jr. Psych., 12:448). FORMAL DISCIPLINE I95 comparative loudness of the first sound: I cannot say just how I form the estimate" (5). At the beginning of the training, therefore, he began to classify the first stimulus which relieved him of carrying an auditory image. But there were times when the image of the first stimulus was distinctly recalled just after the second was received: "On the appearance of the second sound there must be a recall of the 'idea' of the loudness of the first sound, and I think the actual sound arouses an auditory image of the first" (36). As the training progressed, judgments were confidently given when no image or idea of classification was present (28), and it became difficult to introspect (31 ). The following is a descrip- tion of the process written after the last series: "I have a 'mental estimate' rather than an auditory image of the sound. There must be a vague fleeting imagery with this 'mental estimate' for if the interval is long this estimate is less definitely placed, is more elusive and fleeting and vague. A sort of 'stock pattern' of degree of loudness remains. This concerns only the interval. On the appearance of the second sound I am more able to judge than I expect to be" (49). There are many protests against the long intervals (i, 2, 3, 10) the reagent claiming that he loses attention; this would seem to indicate an effort to carry imagery along. Judgment was usually given quickly or upon the appearance of the second stimulus (2, 3). The most satisfactory state of the attention was not the highest (2), and the reagent com- plained that it was often poor (4, 6) and often that the sounds seemed vague (13, 14). Distraction was often noticed (5, 18, 26, 27, 35, 41). The judgment 'like' merely meant no perceptible difference (1:8,4:6). The reagent's 'type' is verified by his introspection upon the last day : "The long interval I feel makes the second sound seem louder. The first is fading in intensity and clearness and is becoming flitting in its recall. . . . There is either an unusual number of judgments 'greater' or else I have a decided prefer- ence for them in series 48 and 49." 196 JOHN EDGAR COOVER (2) Quantitative In reckoning Right and Undecided cases (a) judgments upon D=o are discarded,^^^ (b) 'Doubtful' and 'like' judgments are aggregated as Undecided; (c) per cent is reckoned upon the total number of judgments upon D>o. With the intermediate weight of ball a series of judgments yield a slightly greater per cent of R cases when the fall-phono- meter is used (as in the training) than when our test series is given on the sound-pendulum (as in Test 3), The following table gives the results of Rt. : (80 judgments each). Sound Pendulum Fall-Phonometer R U R U 62.5% 27.5% 65.0% 22.5% All other conditions remaining the same, a greater per cent of R cases occurred (a) when the absolute intensities were greater (heavier balls being used), (b) when the time-interval between the stimuli was 15'', (c) when the second stimulus is the less, for Hs. and Wn. ; when it is the greater, for Rt. From the following table, showing the number of R cases in 96 judgments on each time-interval, in seconds, it appears that "imagery" of some sort was more or less used : 7" 15" 30" 60" Hs. 46 57 48 43 Wn. 51 61 54 59 Rt. 63 68 61 54 The 7" interval was not sufiEicient for the imagery to become settled. The imagery seemed to dim with the flight of time; only for Wn., who used classification more consistently — an in- direct method, — was the longest interval as favorable as others.^^® '^ Vid. p. 43, footnote. *'° The indication of the dependence of judgment upon memorial factors becomes more definite if we separate the judgments according to the rela- tion of the intensity of the second stimulus to that of the first. Assuming, from the work of Ebbinghaus and others, some of whom are noticed below, that the memorial factor dims with time, one should expect (i) Overestima- tion of the second stimulus, (2) A falling off of Right Cases with time, in the gross averages (as is shown in the table above), and (3) upon segrega- FORMAL DISCIPLINE 197. If the R judgments of the training are aggregated for each of the reagents respectively, their relative efficiencies may be compared: Hs. 50.4%, Wn. 58.6%, Rt. 64.1%. A peculiarity of the training is that instead of improvement ting the data as suggested, a general restriction of this decrease in R cases to the group in which the second stimulus is less than the first. This third expectation is fulfilled in the following table (Per cent of judgments reck- oned upon all judgments upon D>o) : 7" 15" 30" 60" S,s, Hs. 69 56 57 34 Wn. 71 (^7 63 54 Rt. 69 71 50 40 Ave. % 70 65 57 43 The first expectation enumerated above, is met by all reagents when the stimuli were given 60" apart, and by Rt. in gross averages. It is opposed by 'Type' in the shorter intervals for Hs. and Wn. Another influence to veil in slight measure the fulfillment of these expectations is the presence of 'absolute' or 'free' judgments, shown by all reagents in excess of R cases in Order N — V over Order V — N ; it amounts to about 7% of all R cases on D>o. Our results conform pretty well with the literature. The over-estimation of the second stimulus (the Fechnerian Time-Error — vid. Psychophysik 1 :88) has been found with musical tone by Wolfe (Phil. Stud. 3:556), with inten- sities of sound by Starke (Phil. Stud., 3:27ofif), Merkel (Phil. Stud., 4:ii7flf) Lehmann (Phil. Stud., 7:205), and Tschisch (Zeits. d. Psychiatrie, 1896) with memory for passive movement by Schukowsky (Zeits. d. Psychiatrie 1899), with discrimination of shadows by Gerwert (Zeits. d. Psychiatrie, 1899, No. 8), and with discrimination of shades of gray by F. Angell (Phil. Stud. 19:5). A falling off of R cases with increase in time between the stimuli is shown by Wolfe {op. cit. p. 569), Tschisch (Dritter Int. Kong. f. Psychol., 1896, p. 103), Gerwert {op. cit.), and F. Angell {op. cit. From Table III, pp. 12-13). The general restriction of this falling off of the R cases with time to the group S,>S2 cannot be well illustrated from the literature for the reason that this group cannot be segregated from the other (Sis. 65.4 62.2 55.3 51.3 When intervals of from 5 seconds to 60 seconds have been used between the stimuli, optimal intervals greater than 5 seconds and less than 30 seconds have been shown by Gerwert (op. cit.), Tschisch (Dritter Int. Kong. f. Psychol., p. 103), Wolfe (op. cit.), and Angell (op. cit. p. 5). Influence of 'Type' was shown by Martin and Miiller (Zur analyse der Unterschiedsempfindlichkeit, pp. 128-134), Kampfe (Phil. Stud., 8:582), and F. Angell on Discrimination of Clangs (Am. Jr. Psychol., 12:72). A good critique of the variable influences upon R cases in this type of experimentation is to be found in Martin and Miiller (op. cit., pp. I7ff) and in Angell's review of the same in the Am. Jr. Psychol. (ii:266fif). FORMAL DISCIPLINE 199 Attention, there may have been some change made by the train- ing which the scores failed to show. Analysis rather than scores must be relied upon to determine this, and it was seen in the analysis of processes that change did take place. d. Test Results (i) Recognition or Choice of One of Two Letters The method of this test favors analysis of both introspective and quantitative results for determining many of the factors of variability in the processes involved in a task which seems quite simple and definite. (a) Introspective Analysis If one experiment be divided into four intervals upon which introspections were recorded, the method of the reagent and the accompanying processes may be stated as follows : 1. Interval between signal and stimulus {2 sec). A settling of the body into a comfortable (Al.) or a strained (Hhs.) atti- tude, a direction of the gaze upon the window of the screen, sometimes with the head at a slight angle (Wn.) ; Consciousness may (i) contain keenly the purpose of (a) seeing the whole card, or (b) a definite portion of it, or (2) the mind may be a blank (Hhs,). If perception of the whole card is intended, the center of the window may be keenly fixated or the gaze may be not so limited spatially. The trained reagents appear to have been more constant in holding a single purpose, though all reagents varied some in this respect during the 50 experiments of the test; the control reagents did more experimenting especially in regard to the extent and location of perception. 2. Moment of perception {o.i"). A more or less dim image of the whole card may result, but usually one line or a group of letters stand out more clearly, while the rest of the card may present some dim imagery or may appear blank. Some- times the presentation was read (verbally) from left to right or from top to bottom; sometimes just seized as a whole, vis- ually. The grades of clearness of the letters did not follow from the purpose, except in cases where perception was narrowed to a 200 JOHN EDGAR COOVER small part of the card. From 2 to 5 letters were the usual number that occurred in clear imagery. 3. Interval between the perception and the presentation of ike two letters {3"). Visual imagery was held and often read as far as clear letters appeared; effort to 'mature' dim imagery, and to locate the letters seen. Sometimes a vivid hope is enter- tained that certain of the clear letters will be exhibited. There was noted a feeling that more letters than those perceived would be recognized if shown (Rt.). 4. Moment of recognition or choice. If one of the two letters shown had been clearly perceived, it was instantly named; if neither had been clearly seen, effort was made to determine which had been on the card and one was chosen because it "seemed more familiar" either to sight or by pronunciation, because it brought back a vague image of the card, because its form was more pleasing, because of certainty that the other letter was not seen; or the choice was a "pure guess." Since the imagery was in all grades of clearness, certainty of judgment graded down from very certain to wholly uncertain. Four grades were re- corded. "Very certain" was given for very clear visual or kinaesthetic or combined imagery; "Certain" for less clearness; "Not just certain" either for such imagery as would support the choice of a similar letter, or for vague imagery; "Wholly uncertain" for very vague imagery or for none. Sometimes a letter was recognized as having been seen, while in process of pronunciation (Hs.). Other cases of 'maturing' imagery adequate for recognition, concurring with the feeling of famil- iarity, and efficiency in recalling a vague image of the card, referred to above, are indicated by changes in choice. That vague imagery often determined undecided judgments is shov.'n by the record of the influnce of the ist, 2d, and 3d preceding cards. (b) Quantitative Analysis Analysis of the quantitative results reveals some additional factors of variation in processes, indicating further complexity of the processes engaged in this simple task, and shows in what way training of the regular reagents affected these processes. FORMAL DISCIPLINE 201 The range of initial capacity was from 48% to 76% R cases, and its average of 61% was increased in the final test by 2%. But increase was not uniform, nor do the regular reagents show any advantage over the control reagents, and it will be made evident that these scores cannot be taken as a measure of re- production or, in fact, of any other capacity. The first patent indication that they measure greatly different processes is the fact that the results of two control reagents are directly opposed to each other : Al. made the highest score in the first test, and the lowest in the final, losing 18% ; Ck. made the lowest score in the first test and made the greatest gain (12%). Letters were chosen by all reagents with four degrees of certainty each of which involved different processes, as is shown not only by the introspections but by the regularity of their quantitative results when the latter are aggregated : VC C NC U % R cases 94 73.6 56.3 494 Avg. time of choice 1.22" 1.47" 1.68" 2.16" Avg. space error 0.79 i.iS 1.32 1. 41 The distribution of all judgments is shown by the following per cents : 14.6 18.4 21.6 45.4 Thus almost half of the aggregate of judgments was "Wholly Uncertain." The distribution of judgments over the various grades of certainty varied some between the first and final tests for all reagents. This was caused by variability in the direction and extent of the attention, owing to the difficulty of carrying out the instructions to attend to the whole card; only one reagent (Rt.) was able to do this consistently and he complained that the unnatural effort was a distraction. It has not been stated, perhaps, that all the letters on the card in all our tachistoscopic tests fall easily within the angle of acute vision. The effect of narrowing the attention to a portion of the card was to get a few letters in great clearness at the expense of any imagery' 202 JOHN EDGAR COOVER of the rest of the card however vague, to increase the number of "Certain" choices, and to decrease the influence of 'fringe' imagery in determining R cases in "Uncertain" choices. That the latter influence was present is indicated by the time- relations between the right and wrong cases in Uncertain choices : If the four averages containing the greatest increase in time of R over W cases, the four containing the least increase, and the five containing decrease in time, are selected and compared, we get the following table : Increase in time .54", Ratio iR:W 1.40 " " .125", " " .93 Decrease " " .14", " " -83 Which indicates that in "Uncertain" choices time was a func- tion of R cases. It should be noted that when R cases are not influenced by 'fringe' imagery, the ratio of R:W should be i., and that the table shows in the ratios of lower value an opposing influence which will be discussed later as the effect of preceding impressions. In "Certain" choices, of course, R cases correlate with less time. Variation between reagents in extent of the card attended to is indicated by the location of letters in "Certain" choices: ist line, Hhs. ; ist and 2d Hne, Hs., Rt, Wn.2, Ck. ; ist and 2d mainly, Al; 3 lines, Br., Wn.i. The averages emphasize three types : Lines I 2 3 Hhs. 24 2 I Hs., Wn., Rt., Ck. 14 10 3 Al., Br. 9 19 8 Other factors of variation in the processes involved especially in "Uncertain" choices, as shown by quantitative analysis, are the influences of favorite letters, of similar letters, of the position of the letters exhibited for choice, and of preceding letters. All of the reagents favored or ignored particular letters, al- though these letters rarely retained their special status throughout both tests, but rarely exchanged their status in the final test. This influence is verified by the fact that choice of the favorite FORMAL DISCIPLINE 203 letter is made in shorter time, besides being shown by greater frequency in a tabulation of all letters chosen. Partiality for the left or right letter exhibited for choice was shown by all reagents and is constant for both tests; in five cases it was as high as 1:2. Influence of preceding letters was suspected when it was noted that wrong cases would frequently have been right for preceding cards, and was found by comparing the number of identical letters preceding the recorded letter with the number preceding the true letter. The aggregate shows the influence clearly : Preceding cards 1st 2d 3d Certain choices 27.8% 5.3% 1.5% Uncertain choices 94% 45-0% -6.5% This influence upon "Certain" choices was shown only by space error, and must have been effective through facilitating perception ;^^^ upon "Uncertain" choices it was shown by both space error in R cases and by W cases, and must have been effective, in the former, by augmenting 'fringe' imagery into determining influence, in the latter, by either augmenting 'fringe' imagery of a similar letter, or by persisting to the exclusion of 'fringe' imagery from the last perception.^^^ A factor working for wrong cases, especially in "Not just certain" choices, and counteracting the influence of preceding letters toward R cases in "Uncertain" choices, was the influence of similar letters. All errors in "Certain" choices were attribut- "^ Schumann (Die Erkennung von Buchstaben und Worten bei momentaren Beleuchtung. Bericht u.d. I. Kongress f . Exp. Psychol., 1904 : 36) inferred from the fact that the recognized letters were not always those most clearly seen, that the recognition of letters must be assisted by residua of previous perceptions of them. *** In tachistoscopic presentation of mutilated typewritten words, Pillsbury (A study in apperception. Am. Jr. Psychol., 1896-7, 8 : 355, 357, 359) found an unconscious effect of preceding words upon the word-completion process ; introspective report of the absence of an association determining the comple- tion was untrustworthy. McComas (Some types of attention. Psych. Rev. Mon., 191 1, No. 55, p. zi) also found, in tachistoscopic work, a determining influence from presentations given 2 min. and 5 min. earlier; content appeared to lie dormant until reported as seen on a later card. 204 JOHN EDGAR COOVER able to this cause, and revealed as similar letters, TH, MW, DB, XK, NY, RB, PR, NK, CO, YV, BK. In testing for the influence of training in the reproduction of imagery upon this test we cannot use the gross scores of R cases; increase of reproduction is consistent with an increase of W cases in "Not just certain" choices because of the similarity of letters; we are consequently limited to the "Uncertain" choices, and, although they are influenced as shown above by many factors of variation, they furnish evidence that the training produced some effect. Since the range of initial capacity was but from 43% to 50%, results may be grouped and handled with a degree of confidence. The regular reagents show a greater increase in R cases : First Final Difference Regular 44.8% 58.1% +13.3 Control 45-9% 50.0% + 4-i Part of this increase must be due to the 'maturing' of 'fringe' imagery as is indicated by the greater increase in the time of the R over the W cases in the final test: First test .04" Final test .25" It was pointed out above that time here is a function of R cases. This increase is not shared by the control reagents nor by one of the regular reagents (Wn. ) who in her training avoided reproduction of simple imagery. Since there was some increase of influence of preceding letters, it is possible that this influence tended, more than in the case of the control reagents, to augment the 'fringe' imagery of the last presentation in a way similar to the effect of this influence in facilitating perception in cases of "Certain" choices. If the 'fringe' material from preceding cards is not 'matured' about a nucleus of 'fringe' imagery from the last presentation, its influence would be toward W cases; and the above table indicates by per cents below probability (50%) that this occurred in the work of both groups in the first test, and it may have occurred and have been overcome by its opposing factor in the FORMAL DISCIPLINE 205 final test. That this balance occurred in the case of the control reagents seems evident from the slight increase in the per cent of R cases, as is shown by the table above, in spite of great increase in the amount of influence from preceding cards: First Final Difference Regular 13.3% 16.7% + 3-4 Control 18.2% 32.5% +14.3 This table also shows that the control reagents were somewhat more influenced by preceding letters than were the regular reagents in the first test. The effect of the interval upon the control reagents was to shift this influence toward the more re- mote cards and to make it independent of imagery from the last presentation. The effect of the training upon Hs. and Rt. was to shift this influence to the more recent cards and to combine it with 'fringe' imagery of the last presentation. (2) Reproduction and Recognition of Letters^^^ The results of this test furnish another forceful illustration of the worthlessness of merely quantitative treatment of 'mental tests.' Initial capacity in recording letters from a 12-letter-rectangle exhibited o.i" ranged from 4.15 to 10.05 points. The three regular reagents occupy the three higher places. With one ex- ception gains in the final test vary inversely with initial capacity. The final capacity of but one control reagent exceeded the low- est initial capacity of the regular reagents, and that exceeded by 40% the final capacities of two reagents who trained, in the preceding experiment, on this work for 18 days. Intro- spections offer no explanation. The recognition part of the experiment, designed to test for reproductive tendencies too weak to reach the threshold, "' Experimental work justifying our selection of this test has been done by Robert Mac Dougall (Recognition and recall. Jr. Philos. Psych. & Sci. Meth. 1904, 1 : 229-233) who used words for his material. He found that where about a half were reproduced through recall, about three-fourths were recognized. "Characteristic differences between recall and recognition," have been more recently studied by Hollingworth (Am. Jr. Psych. 1913, 24:532- 544). 2o6 JOHN EDGAR COOVER was ill advised owing to the impossibility of checking reagents in their recognition and of equating their scores. Two important facts, however, may be extracted from the results. The regular reagent who avoided reproduction of simple imagery in her training and who was not benefited by the 'matur- ing' of the 'fringe' imagery in "Undecided" choices in the preceding test (Wn.) lost 6.2% while the other two regular reagents gained 7% and 5% respectively, yet her initial capacity was slightly below their's. And the introspective evidence of variability in processes recorded in Test 17 of the preceding experiment is supplemented by similar evidence showing that this variability appears in each 'moment' of the experiment; particularly in four of the five intervals into which the experiment naturally falls: (i) In the 'Anlage' between "Ready" and the stimulus; (2) In the direction and distribution of the attention during the per- ception, especially with reference to the treatment of the after-i image which persists after the screen has fallen; (3) In the process during the voluntary interval between perception and recording; (4) In the process of coordinating the maturing, retaining, and reproducing part-processes, during recording. (3) Sound Discrimination It will be remembered that the work of this test is the same in kind as the work of the training, the only difference being that here the sound-pendulum was used instead of the fall- phonometer {i.e., the sounds were produced with wood instead of with steel), and the time-intervals between stimuli were all short (about 3"). (a) Processes Reagents were not shown the sound-pendulum or the man- ner of producing the sounds, in order to avoid the 'stimulus error' and the complication of visual imagery. But usually curiosity was evinced as to how the sounds were produced, FORMAL DISCIPLINE 207 and in almost all cases visual imagery of imaginary apparatus was frequent in the process of discrimination. Hs. used auditory imagery for judging, and in both tests speaks of comparing the images; in the second test she speaks of ease of judging in one series because the auditory image of the first stimulus remained in her mind without effort. But in the beginning of the second test she also compared the image of the first with the sensation of the second stimulus, only bringing the latter into imagery for comparison when the judgment was at first doubtful. She was helped by kinaesthetic imagery into which the stimuli were converted by singing the tone, by striking or pressing down or weighing with her arms and hands, etc. When the kinaesthetic and auditory bases of judgment conflicted she speaks of following the latter. She had considerable visual imagery which was regarded as a hindrance : of the experimenter holding a hammer and knock- ing on the table, of a scale for distributing intensities, ^^^ and of steel balls falling. Another distraction was the anticipation of the second stimulus. An affective image, as of knocking in the back of the head, is spoken of at the beginning of the first test; and an auditory image of her own voice singing the stimulus occurred in the second. The interval of training seems to have diminished the visual imagery and other distractions, leaving her with the general method of holding the auditory image of the first stimulus as a basis of judgment upon the impression of the second or of comparing with the image of the second in case the difference seemed doubtful. Wn. also has varied imagery as bases of judgment; but her kinaesthetic imagery is perhaps predominant, affective imagery following, then visual and auditory. She speaks several times in her introspections in the first test, of comparing images; sometimes these are a combination of auditory and kinaesthetic ; and in case of doubt she reviews them over and over in her mind. Her kinaesthetic imagery began immediately and con- tinued throughout: She tried to remember by kinaesthetic '** Cf. Angell, F. : Phil. Studien. 1892. 7 : 414-468 . 2o8 JOHN EDGAR COOVER image in the mouth, head and arms ( i : i ) , pressed the teeth together (1:2), tried to sing the stimuH (1:3, 2:7), nodded hccK. i 1 .10) ; and several times the stimulus could be recalled only through kinaesthetic imagery (1:6, 7). She had affective imagery: felt knocks in the head (1:1), "feel the sound be- fore I hear it; if 'feel' and 'sound' do not agree I make both intensities with my arms; this aids my judgment more than auditory images" (1:5). Distracting influences were the visual imagery of woodpeckers and trees (1:1), of E with a little wooden peg (1:2) or hammer (1:4), of toy men hammering a block (though she thought this helped) (1:6); anticipation of the second stimulus (1:3); bothering about the instrument (i 13) ; bothered because of so many "greater" judgments (i :8) or "less" judgments (2:7). The interval of training seems to have qualified her to give judgment upon the second impres- sion; she says she compared auditory image with auditory impression (2:1). Her kinaesthetic imagery continues, mov- ing hand as if it held the hammer (2:7); and so does the affective : "feel the little hammer in my head and myself knocking with it" (2:2). The visual imagery does not seem quite so frequent. Although the interval was not over a few seconds, she felt that she makes the sounds more and more alike as she recalls them (2:6), which indicates that voluntary strengthening of the first stimulus may account for her type which was shown in her training results. Rt. from the first protested that he carried no image of the first stimulus over to the second (1:1,3); but that he "re- thinks" if not "re-images" it (1:1). When he thinks "just how loud was that sound?" and dwells upon its intensity, he be- comes confused in his judgment upon the second (1:4). An attentive attitude less than maximum was most satisfactory (i :3, 4, 5). The probability of a symbol of opposite meaning following several like symbols in succession in a series, he thinks, may have occasionally influenced his judgment in doubt- ful cases ( I :7). In both tests he was bothered by rhythm which would run through his head and tend to make the second FORMAL DISCIPLINE 209 Stimulus louder or weaker than it really was (1:8, 9; 2:4). The effect of the training was to make these 'wooden' sounds seem "novel"; he was very uncertain of his judgment in the first series of the second test, and his score was low. But in his 3d series he underscored four out of his nine judgments, showing that confidence soon returned. And near the end of the test he says: 'The whole of the work today is so much easier to follow, with the short [temporal] interval [between the stimuh] (2:8). Al. thinks that she compares impressions that seem to be in her head, in making judgment (1:1); she attentively fixes the first and waits a few moments after the second stimulus has been received before judging. But by the middle of the first test she remarks that it does not take so long to judge (1:5); even then, the first sound seemed to be an impression in her head. After the interval the clear differences were judged upon the receipt of the second stimulus; the smaller differences, after a few seconds of comparison (2:1-3). Later in the test she felt the effect of practice {2:4,6) when the differences were clearer and judgment was more readily given. The second test seemed much easier. Hhs. made judgment immediately upon hearing the second stimulus; seems to have classified his sounds as light, medium, harder, (as of striking a board) (1:7, 2:4) and also without using the class-names (2:8-10). He had visual imagery of a metronome (1:6), and of some one striking a board with a small mallet. Thinks the second test a little easier, and that there were more clear intervals in it. Shows considerable practice-effect in the first test : 3, 3, 1,4, 3, 5, 5, 5, 5, 6. Br. uses as a basis for judgment the effect the sound works upon him (i ; 2 :i). To this he added a sort of a classification. The second test seemed much easier than the first; more clear differences. Ty. thinks she compares auditory images, at first (1:1), which when clearly different are outside (1:4), but when not, they are brought inside the body and converted into a "tactual" 2IO JOHN EDGAR COOVER image in the eyes and nose (i :4). She has some kinaesthetic imagery in the arm as if striking {1:4,2 -.3). After the interval there is some change: she weights the after-images of the stimuH, if their difference is sHght, in her ear (2:2) ; and has kinaesthetic images in ear and throat (2 :6) and chest in breath- ing (2:6); sometimes of humming the stimulus (2:6) or tap- ping it; and feels that the after-image of the second stimulus is louder than it should be (2:5); the change in general was toward kinaesthetic and possibly affective (tension in the ear) conversion. Dn. says that in comparing sounds his "mind was concentrated on a point inside the ear" (1:2); he compared controlled ex- halations of his breath, like carrying the tone of a piano in the roof of the mouth; less intensity, smaller volume of breath breathed out. He also classified sometimes. In the second test the same method was continued, except that in breathing the tone the tension of the throat muscles was taken into account; and it was supplemented by kinaesthetic imagery of the head and arms as in nodding or striking; of the toes, as pressing down for a heavy tone, raising for light (perhaps transferred from piano-playing) . These analyses indicate clearly that for the unskilled reagent discrimination of intensities of sound is a very variable pro- cess. Practice may not lead to improvement, as was shown by our trained reagents, and is verified here by Ty. and Dn., and the chief causes of retrogression are indirect methods of dealing with the stimuli; Wn, Ty., and Dn. are the clearest examples of this. A point of interest is that the 'wooden' sounds caused the trained reagents some interference after their habituation to the 'metallic' sounds of the training: Rt. recorded that these sounds seemed "novel" and he was consequently very uncertain in his judgments, in the beginning of the second test; and the sum of R cases in the first two series in the final test was lower for each of the three reagents than in the corresponding series in the first test. By the third series, however, the interference had been overcome. FORMAL DISCIPLINE 211 (b) Scores Initial capacity ranged from 48.8% to 76.3% R cases. The three regular reagents occupy the three higher places, and their scores in the final test change in a way consistent with the results of their training: Hs. and Wn., who lost consistently in their training, lost here; Hs. owing to conditions of health, and to her extraordinary score in the first test; Wn. to the development of the inadequate method of classification and asso- ciation which prevented treatment of the stimuli in terms of simple auditory or kinaesthetic imagery. Rt., who, although he showed loss in the latter part of his training yet showed gain in the intermediate part, made a slight gain. Thus the training effect shows itself in this test. The control reagents whose initial capacity was nearly as high as that of Wn. and Rt. also lost, Dn. losing the greatest amount owing to the development of an indirect kinaesthetic method which was not so trustworthy as his earlier method of basing judgment upon 'affects' or more direct kinaesthetic imagery. The detrimental effect of an indirect method of discrimination has already been pointed out. (4) Memory for Visual Symbols (a) Processes In general all reagents were more or less helpless at the beginning of this test because the symbols were strange and evaded fixing. Almost all avoided the pure visual effort for which the test was designed ;^^^ they compared the symbols ^* In this respect they conform with the work of Ebert and Meumann's (op. cit., pp. 50, 116) reagents, who declared that the symbols could not be learned from the pure visual impressions. Kuhlmann (On the analysis of the memory consciousness, a study in the mental imagery and memory of meaningless forms. Psych. Rev., 13 : 342) also found indirect methods of memorizing largely employed : associations, verbal descriptions, motor ten- dencies of eye and hand. And in another place (Problems in the analysis of memory consciousness. Jr. Philos. Psych. & Sci. Meth. 1907, 4 : 5-6) he points out a more subtle mode of vicarious functioning; through the more or less extended organic reactions which are evoked by any sensory stimulus, 212 JOHN EDGAR C DOVER with conventional characters and things, gave them the names of these associated characters and things, and remembered the names, usually at first without the support of visual imagery necessary to record correctly. Owing to the inexperience of the control reagents in intro- spection but little knowledge is at hand of their processes; though that little points to a freer use of visual imagery and visual association of the forms of the symbols, than was true with the trained reagents. 1 ) Immediate Memory The process of introspecting after recording from the single presentation of six symbols was difficult, and the processes of perception and reproduction were very imperfectly revealed. There was sufficient evidence, however, to indicate that there was large individual variation. For the 'note' forms, the effort was a) to visualize the pre- sentation (Hs.), b) to transpose the content into a visual scheme using the musical staff (Wn.), or c) to classify the flags accord- ing to spatial relations (Rt.). For the 'sickle' forms, the effort was generally verbal, to name conventional characters to which the symbols were similar, and to reproduce from kinaesthetic-auditory imagery. 2) Complete Learning The trained reagents show clearly in their introspections that the processes employed in the 'complete learnmg' are quite complex and variable. For the 'note' forms there were three methods of learning employed : a) Naming the number and positions of the flags in the consecutive symbols, for recall in verbal imagery (Hs., Wn., Hhs.) ; b) classifying the symbols according to the num- ber of flags, and their relations of symmetry, counting the symbols to identify related symbols by number, for recall in mixed imagery largely visual (Rt, Br., Hd.); c) combination the lack of prominence of which in consciousness is no criterion of their utility. Bentley's statement that the aids to complete recollection may become the real vehicle of retention, has already been referred to (foot-note, p. 190). FORMAL DISCIPLINE 213 of naming and classifying (Al, Hn.). With methods a) and c) a few symbols at the beginning and the end of the series were usually held in pure visual imagery. The fact that the learning of this series may involve very different kinds of work is illus- trated by reagents employing methods a) and b) respectively: a) Hs. learned the series by naming, i. lower, middle, 2. middle top, 3. one middle, 4. two middle, 5. two bottom; etc. It was a straightforward 'rote' method, converting the presentation into verbal imagery; this was made vivid by verbal repetition, and was recalled through kinaesthetic-auditory imagery, sup- ported by visual memory. Wn. also named but used slightly different terms, adding left and right, and supported her verbal imagery with kinaesthetic imagery of direction in neck and arm. b) Rt. used the early presentations for inspecting the symbols and seeking relations in the position of the flags; then he began counting the symbols as they passed the window and fixing related pairs: No. 7 is No. i inverted, 6 is 12 reversed, 9 is I with flags at half-mast, etc. ; and by drawing the symbol in the air with his pen while cognizing it and repeating- the number, he formed associations between pen-movement for reproduction and the number of the symbol; reproduction was supported by visual imagery. This was a more logical method, involving the apperception of relations. For the 'sickle' forms, tha same three kinds of methods were used, and with two exceptions the reagents carried over the methods they had used with the 'note' forms: a) the symbols were named by calling them h, ?, ? inverted, wrong v, right v, y inverted, g, 5, etc., and in reproduction the kinaesthetic- auditory imagery was supported by visual imagery (Hs., Wn.) ; b) the symbols were classified according to form (Rt.) ; or c) were learned by classifying some and naming others (AL, Hd.). Hhs. used method a) in his first test and b) in his final; and Hd., who used method b) with the 'note' forms, used method c) here. In the final test methods usually continued without much change, except on the part of three control reagents : Br. changed 214 JOHN EDGAR COOVER to a more purely visual method, Hn. from more purely visual to naming and visual, and Hd. introduced a method of counting and fixing forms in their positions (the method Rt. used in both tests). The effect of the interval was to make the test easier for all reagents ; this was so marked that a third of them assumed that the series of the final test was identical with that of the first, and felt a strong recognition for some of the sequences. The prin- cipal factors contributing to this were: (a) Familiarity with the symbols, b) concrete knowledge of the demands of the test, c) possession of a method, d) in some cases an improvement in method. (b) Scores If increase in facility of reproduction of imagery can be shown at all, it must appear in increased scores. In some cases, however, the scores are affected by change in method and are unreliable as measures of reproductivity ; and in all cases, pre- sumably, there was some improvement in the application of method leading to more adequate impression of the stimulus, which cannot be separated from the factor of reproductivity in the scores, i) Immediate Memory Since the tests in 'immediate memory' were necessarily short, incidental causes of variation, including direction and distribu- tion of attention, must play a correspondingly large role. And since it happens that all of the control reagents, except possibly Hn., worked more freely with visual imagery than did the more mature regular reagents, they probably received more benefit from the practice-effect of the test, and this advantage might more than off-set a slight increase in reproductivity brought by the regular reagents from their training. Individual variation in initial efficiency, for the 'note' forms, ranged from o to .329 (i representing a perfect score), and for the 'sickle' forms, from .080 to .414. Per cent of improvement was reckoned upon the final score and is too irregular to make FORMAL DISCIPLINE 215 the examination of the tables profitable. It ranged, for the 'note' forms, from 25 per cent loss to 100 per cent gain, and the control reagents show the best gains; for the 'sickle' forms, from o to 81 per cent; neither group of reagents shows advantage over the other. Several scores are of interest : Wn.'s great improve- ment in the test on 'note' forms was due to a change of method from the use of the musical staff to naming; Hhs.'s o score in his first test was due to inverting each of the two symbols he retained (visually) ; Rt.'s large improvement on the 'sickle' forms was owing to greater facility in classification. The great- est and most consistent gain on both forms was made by Hd., shown later to be a special case, indicating that training in memorizing 'literal' ^^^ prose increased efficiency in this test. 2) Complete Learning The 'complete learning' enabled the reagent to collect himself and work out a method in the course of the first test; and the variation in the factors of attention and adaptation was not so potent a source of error. Initial efficiency ranged, for the 'note' forms, from 15 to 31.3 presentations, for the 'sickle' forms, from 12.3 to 42.8 pre- sentations. The trained reagents took, in general, fewer pre- sentations than did the control reagents. Improvement was reckoned in per cent, of initial efficiency: Regular 1st Control 2d Control 'Note' forms Hs. (15) 53-3 Wn. (16) 46.3 Al. (19.6) 53-1 Rt. (23) 52.2 Br. (24) 20.8 Hhs. (31.3) 42.S *Hd. (25.6) 45.2 Hn. (26.2) 39.9 'Sickle' forms Hs. (12.3) 50.4 Hhs. (24) 19.2 *Hd. (25.2) 64.3 Wn. (29.1) 52.9 Rt. (42.8) S0.9 The number in parenthesis shows initial capacity in number of presenta- tions ; reagents are ranged according to initial capacity. *Hd. is shown later to be a special case. "*In 'literal' prose the learning must be 'word perfect'; which contrasts with 'substance' prose, in which reproduction of ideas is tested. 2i6 JOHN EDGAR COOVER Since initial capacity of the regular reagents was higher than that of the control reagents, they could scarcely be expected to show as much improvement unless there was a relatively greater increase of capacity in reproduction; the fact that they show greater improvement than four control reagents out of five, therefore, seems good evidence of a definite advantage. Aver- ages indicate this advantage still more clearly: Note Sickle Trained 50.6% 51-4% Control 38.8% The trained reagents were somewhat more accurate in their reproductions : In the first test two out of six records were without error, as against one out of five by the control group; and in the final test they made four out of six without error, as against three out of six by the control group. One of the control reagents, Hd., was withdrawn from his group in the consideration of results, because he represents an exception; he had been given a rather severe course of training in 'literal' prose by his fraternity brothers just before the final tests. And although literal memorizing is hard for him, and he left the room after the first test tired out, he found the final test unexpectedly easy, made the two reproductions without error (both scores in the first test were imperfect: .937, .833), and reduced presentations 45.2% and 64.3%. There is no doubt about the influence of his prose training upon this test. A negative 'spread of training' seems to be indicated by Rt.'s results in the first test in 'complete learning.' His method was that of classification and grouping, and introspections show that his habit of disregarding the stem in the 'note' forms, which was invariable, was carried over to the 'sickle' forms, where both stem and curve were variable, and caused confusion; he learned the 'note' forms in 23 presentations, while it took 42.8 presentations to learn the 'sickle' forms. Not all of this differ- ence, however, should be attributed to negative spread of train- ing, for the 'sickle' forms are more difficult to classify and group than are the 'note' forms.^^^ ^^' Though not necessarily more difficult to learn by naming, as is shown by the initial capacities of Hs. FORMAL DISCIPLINE 217 (c) Summary The process of memorizing visual symbols did not engage much visual imagery; the usual method was to associate the symbol v^ith some familiar thing and remember it by verbal- motor imagery of the name of that thing, or to classify the symbols for reproduction from logical memory. Visual percep- tion served principally to analyze and interpret the symbol for conversion into verbal-motor terms or to effect classification. Where visual imagery was used, it supported the more promi- nent kinaesthetic-auditory. The complete process utilized a com- bination of verbal-motor, kinaesthetic, visual, and auditory imagery. The scores of the 'complete learning' show that although the control reagents began with less efficiency and therefore might be expected to make greater gain than the trained reagents, the latter, and one of the control reagents who had training in memorizing literal prose, made more improvement. The only explanation at hand is that training in retaining and recalling imagery occasioned by intensities of sound contributed to facility in retaining and recalling the combined imagery used in this test. e. Conclusion The results of this experiment supplement those of the Ex- periment on Attention in illustrating the variability in processes engaged upon a simple task and in locating the variability in each of the successive 'moments' of a test. Although the training, with its evident retention and repro- duction of auditory and other imagery, did not result in im- provement in sound discrimination, analysis of processes used in the training shows change in those processes, and training- effect shows itself in the tests. In Test I. (Recognition or Choice of One of Two Letters) the regular reagents increased R cases in 'uncertain' choices due to more 'maturing' of 'fringe' imagery of the last presentation and to strengthening this process by uniting to it the influence of preceding identical letters. In Test 2. (Reproduction and Recognition of Letters) the 2i8 JOHN EDGAR COOVER anomalous results emphasize the worthlessness of purely quan- titative treatment of the results of 'mental tests.' In Test 3. (Sound Discrimination) the results of the regular reagents are consistent with those of their training on similar material; but habituation to the 'metalic' sounds in the training caused interference with the transference of practice-effect in the early series with the 'wooden' sounds of the test. The great- est loss, by a control reagent, illustrates the detrimental effect of an indirect method of discrimination, thus agreeing with the training results of Wn. In Test 4. (Memory of Visual Symbols) the trained reagents show their advantage, in 'complete learning,' over the control reagents, except in the case of Hd. whose large and consistent gains in both 'immediate memory' and 'complete learning' indi- cate clearly that his inter-test training in memorizing literal prose contributed to his facility in memorizing these symbols. One of the corroborating facts in the results is that the regular reagent (Wn.) who avoided using simple imagery in her train- ing showed less gain in Test i. and a loss in Tests 2. and 3., where simple imagery was requisite, but was not handicapped in the 'complete learning' of Test 4., where the imagery could be more complex. It is noteworthy that all the gross quantitative results of Tests I, 2, 3, and of 'immediate memory' in Test 4, were worthless; only by careful analysis of scores and introspections was the fact of 'spread of training' made evident at all. The chief contribution of the experiment is : That the specific training did not result in gain in efficiency with the training ma- terial, but that it did result in changes in the processes it en- gaged and it showed its 'general' character in influencing the piocesses involved in Recognition or Choice of One of Two Let- ters (Test i) and in 'Complete Learning' of visual symbols (Test 4) both of which differed radically from the training in sense-mode of impressions and in methods of work. Facts of this character, already noticed, in the Experiment on Attention, in the failure of Al. to improve after his third day of training (p. 93), and in the failure of Rt. and SI. in their FORMAL DISCIPLINE 219 effort to apply the method of learning developed in their training to commercial signs (p. 154), indicate that not only is there gen- eral effect of special practice, both negative and positive, but there is an occasional negative special effect of special practice: Changes in the mental processes incidental (i) to the greater complexity of the processes involved in the next step of progress, (2) to the conscious application of a method already automatic, (3) to the development of a method which neglects the essential process, (4) to change of attitude toward the experiment, or (5) to chance variation in the processes, may cause retrogression in efficiency, indicated by a drop in the practice-curve. It is well known that in continuous practice the scores on the immediately following days suffer from this retrogression. Retrogressive effect of special practice must therefore be taken into account in testing for general effect. This negative special effect shows itself in negative general effect in tests similar to the training, and is not inconsistent with positive general effect in tests less similar to the training.^^^ 3. Critique of the Test-Training -Test Type of Experiment Much has already been said by way of a critique upon the Test-Training-Test type of experiment in the introduction to The Experiment on Attention (pp. 65-70) and under the head- ings of Extent and Variability (pp. i67ff) and The Practice Curve (pp. I73ff), in the same experiment, as well as through- out the text in the discussion of The Test Results ; but it seems desirable to bring the critique together into a plain statement unencumbered with the concrete evidences upon which it is based, and to illustrate with a clear hypothetical case. The prime requisites of this type of experiment are (i) that the tests shall afford a reliable measurement of a known process or group of processes, (2) that the training-effect shall not only be measurable but shall be capable of definite description in terms of changes in processes, and (3) that change in efficiency in the *^In this respect it resembles 'interference/ which, however, is more transient. 220 JOHN EDGAR COOVER final test shall be described with sufficient precision to indicate its dependence upon, or independence of, the training-effect. Only when these requirements are met can the functional re- lationship of mental processes of various kinds be determined with certainty. In order that the processes tested can be known, trustworthy introspection is essential. If it is dispensed with, or if it is faulty, the test-averages may not belong to the same denomina- tion, in which case they can be neither grouped nor compared for any intelligible purpose. Suppose that the capacity of the reagents K., L., M., N., and P. to progress from station A. to station B. is measured by a test ; that K. and L. are trained in progressing from R. to S. ; and all are again given the test from A. to B., for the purpose of determining the training-effect upon the capacity of K. and L. Suppose progression in this illustration is made by travel, and in the first test K., N., and P. walk, and L. and M. ride bicycles; in the training K. and L. both walk; and in the final test K. and P. walk, L. and M. ride bicycles, and N. takes an automobile. Let the following table show the results, in minutes : Regular Control K. L. M. N. P. Beginning 125 130 End 95 100 First 125 20 25 135 125 Final lOO 22 20 9 120 Training Tests Diflference — 25 -\-2 — 5 — 126 — 5 Average — 11.5 — 45 Conclusion : Results show that training in progression from R to S. has a negative influence upon capacity to progress from A. to B., since the practice-effect of the first test upon the final ("45) greatly exceeded the gain (-1 1.5) made by the trained men. Analysis of processes, however, shows that training-effect was transferred by K. ; he used the same processes, as affected by practice, in the final that he did in his first test and in his train- FORMAL DISCIPLINE 221 ing. A negative effect is shown by L., since his practice in walk- ing caused his bicychng muscles to retrograde. The anomalous gain of N, is the result of a radical change in processes, inde- pendent of the purpose of the test. For accurate knowledge of the reliability of the test-results other conditions affecting the progress should be known : whether the traveler had to turn out of the road into rough footing to pass vehicles, whether any part of the road was slippery, owing to the sprinkling cart or rains, whether any unusual conditions affected the work. In other words, the conditions must be under the control of the experimenter or must be known in order to be reckoned with; and in mental tests these conditions are both objective and subjective. This illustration represents the facts all too accurately for the comfort of those of us who have employed this type of experi- ment, and have struggled with anomalous results. Merely quan- titative treatment is worthless. Incidentally, it is difficult to see how any method of statistical correlation can improve upon the simple treatment shown above; and it is equally difficult to imagine a value for even the single test averages : e.g., what can we know by finding that the average initial efficiency in pro- gressing from A. to B. is 86 minutes, when no single ability or group of abilities approached that measurement? It seems patent that results from great numbers of reagents cannot be more reliable, when different kinds of work are mea- sured; this sort of error will not cancel itself out by multiplying it, nor will the 'probable error' have any meaning as an indica- tion of the reliability of the measurement. Practically, the requisite for introspective description of pro- cesses limits the number of reagents. The security of greater numbers may be obtained by repetition of the whole experiment. The requirement that the test shall afford a reliable measure- ment is not easily met, for the reason that it must be arbitrarily decided as to how much practice-effect to include. The usual preliminary series which gives an opportunity for adaptation to the work of the test does not remove the difficulty; practice- 222 JOHN EDGAR COOVER effect continues and at various rates with different reagents/^^ especially if different methods of work are employed. In almost any case a single test score is bound to be unreliable ; in some of our tests in the Experiment on Attention averages of five series, '^° Otis and Davidson (The reliability of standard scores in adding ability. Ele. School Teacher, 1912, 13 : 91-105) found different types of learning ability among 202 8th-grade children who were given 25 tests in adding; amount of improvement, moreover, bore no constant relation to initial effi- ciency. Wells (The relation of practice to individual differences. Am. Jr. Psych., 1912, 23 : 75-88) who trained 10 adult subjects 30 days on adding and marking out zeros, concluded that "A superior performance at the beginning of special practice is not necessarily, nor even probably, attained at the sacri- fice of prospects for further improvement. A high initial efficiency may carry with it as much or more prospect of improvement under special practice than a low one. . . . Not practice, but practiceability, is responsible for the su- perior position of such an individual" (p. 88) ; if the individual's high efficiency is the result of greater practice, his position may be near the "physiological limit" or end of his practice-curve, and practice-effect may be expected to be small ; if it is the product of greater ability (native endow- ment), his position is nearer the beginning of his practice-curve and practice- effect may be expected to be large (p. 75). That the reagent with the higher initial ability improves more in practice was found in tests in tapping by Bolton (Relation of motor power to intelli- gence. Am. Jr. Psych., 1903, 14:621), in discriminative reaction on a type- writer by Culler (Interference and adaptability. Archives of Psych., 1912, 3: No. 24, p. 57), in marking out zeros by Wells {op. cit. p. 79ff), in adding by Wells (op. cit.) and Hahn and Thorndike (Some results of practice in addition under school conditions. Jr. Ed. Psych., 1914, 5:79)> and in multi- plication by Thorndike (Effect of practice in a case of a purely intellectual function. Am. Jr. Psych., 1908, i9:374ff). That the reagent with the lower initial ability improves more in practice was found in training in learning non-sense syllables by Miiller und Schumann (Experimentelle Beitrage zur Untersuchung des Gedachtnisses. Zeits. f. Psych., 1894, 6 : 328) and von Sybel (Ueber das Zusammenwirken verschied- ener Sinnesgebiete bei Gedachtnisleistungen. Zeits. f. Psych., 1909, 53:356), in marking out letters by Binet (Attention et Adaptation. Annee Psych., 1899, 6:368), in free associations by Wells (Practice effect in free associa- tions. Am. Jr. Psych., 1911, 22:2-3), and in adding by Thorndike (Practice in the case of addition. Am. Jr. Psych., 1910, 21:485). Hollingworth (Individual differences before, during, and after practice. Psych. Rev., 1914. 21:1-8), who followed Whitley and Wells in attacking the problem directly, gave 175 repetitions of seven tests to 13 adult reagents, and calculated the coefficients of correlation between the orders of abilities in the 1st, 5th, 25th, 50th, 80th and 130th trials and the last trial. The averages FORMAL DISCIPLINE 223 taken on two days, were unreliable. The averages used in our Experiment on Sensible Discrimination were much more reliable, and they involved the work of three days, about forty-five minutes each. In reaction, discrimination, and tachistoscopic tests it should probably extend to at least 100 experiments. The number of tests should also be limited on account of inter-test practice-effect, although this source of error may be partly eliminated by arranging the series composing the tests in 'double fatigue order' ;^^® successive tests may be made equivalent in difficulty by using Miiller's 'cyclical changes' ("Zyklischen Wechsels").!^! The training should perhaps be longer than is usual, and as rigorous as circumstances will permit, say six months for such work as cannot be trained to maximal efficiency in less time. The control reagents should equal the trained reagents in of these coefficients were .41, .61, .-73, .77, .85, and .92, indicating a gradual approximation, through these various levels of practice, to the final order, and, consequently, the insecurity of the results of the early levels as indicative of the final relative capacities of the individuals tested. In adding and in cancellation of digits the coefficient of r =: + 75 was not reached until the 2Sth trial; in tapping, the 130th trial; but in naming colors and in coordina- tion (aiming) it was reached by the 5th trial. An indication of change in processes during practice, which had the effect of diminishing individual differences, is to be found in the coefficients of correlation between tests at varying stages of practice published by Winch (Jr. of Philos. Psych, and Sci. Methods, 1911, 8:377) and Hollingworth (Correlation of abilities as affected by practice. Jr. Ed. Psych., 1913, 4: 405ff) ; Hollingworth found the coefficients (averaged for his seven tests with 13 observers) for the ist, 5th, 25th, 80th, and 205th trials to be .065, .280, .320, .390, and .490, respectively. That differences in rate or amount of improvement in practice may be dependent upon differences in mental processes is shown by McMein and Washburn (Effect of mental type on the interference of motor habits. Am. Jr. Psych., 1909, 20:282ff) who found that reagents who used visual imagery freely made more rapid improvement in card-sorting, and overcame interfer- ence, produced by turning the compartment-scheme through 180°, more easily, than the other reagents. And von Sybel (Zeits. f. Psych., 53 1356) found that the practice-effect was greater for reagents with the less initial facility in method of learning. ^^* As was done by Fracker : op. cit. p. 63. ^*^Vid. Muller: Zeits. f. Psych., 1905, 39:116. 224 JOHN EDGAR COOVER number, in initial efficiency, and in facility in introspection, or their results may not be comparable with those of the trained reagents. If these conditions are met, comparison, both quali- tative and quantitative, may be intelligently made; and the simplest mathematical treatment w^ould seem the least objection- able, even the per cent form of expression being serviceable.^'*^ Finally, two more criticisms with reference to the interpreta- tion of difference-scores, are important: (i) An exceptionally large difference-score is a symptom of radical change in kind of work and calls for individual treatment; (2) A small difference- score does not necessarily mean that training-effect is absent; it may merely be the resultant of positive and negative factors in that effect. In either case introspective description of processes must, when possible, furnish the chief grounds for interpretation. 4. Analysis of the 'Common Factor' It is generally agreed that 'the Common Factor' is responsible for the general effect of special practice ; but as to what it is or may be, there is difference of opinion. It is interesting to note that its nature is growing more complex as experiment and discussion advance. In the early days it was regarded as largely physiological and related simply and immediately to the data of presentation, or, as mental habit, related definitely to the method of the experiment; in marking out words, for example, it was said to consist principally in eye-movement, pen-manipulation, and in the habit of looking for the least common letter. And "^The chief objection to the per cent form of expression involves the presumption of a direct comparison of measurements of improvement made by reagents who differ widely in initial efficiency, {vid. Whitley's hypothetical case showing how varying methods of portraying practice-effect lead to vary- ing conclusions, op. cit. 100-105) or who differ widely in their positions with reference to the end of their practice-curves (Wells : Am. Jr. Psych., 23:82-85). This objection appears to be removed if, upon the grounds of qualitative difference in work, capacities differing widely in initial status are declared to be not directly comparable, and if, upon grounds of individual variation in the form of practice-curves or of multiple types of practice- curves, the individual's position on the practice-curve cannot be determined and, consequently, no remedy is apparent. FORMAL DISCIPLINE 225 'general effect' was said to decrease directly with decrease in similarity of matter and method between the two tests. ^"^^ More recently, the 'Common Factor' has been found to be made up of subjective factors, not altogether, but largely, inde- pendent of the matter and method of the tests (admirably shown by individual variation), and to be anything but simple, although the illustrations used in discussion are still apt to give it the ap- pearance of a simplicity and definiteness which the authors do not really intend. ^^^ If one set of processes consists of elements a,h^c,d,e, and an- other of e,f,g,h,i, training in the one is said to improve the other through the common factor e; the effect is greater in case the latter set should consist of c,d,e,f,g. Fracker^^^ has suggested the term 'Transference' for this kind of influence, but he pro- vides for another kind of 'general effect' which he calls 'Spread of Training,' which is found when the second set of processes consists of f,g,h,i,j, none of the elements of which is common with any of the first set but some of which are 'consciously' or 'subconsciously' connected with some of the latter in the mind of the reagent.^"*^ And Sleight^"*^ claims that to be effective the 'common factor' need not be recognized by the reagent as com- mon ; and that it may not be effective at all because it is so firmly bound up with its associates that it cannot be lifted out and used where its associates are not also common. "' Vid. Thorndike and Woodworth, 1901, op. cit. "*Vid. Fracker, 1908, op. cit. Sleight, 191 1, op. cit. "' Fracker, op. cit. 85. "' The writer believes that Fracker has performed a real service in pointing out those distinctions and in offering apt technical terms for them ; and he has adopted the terms in the preceding pages, but has thought best to characterize them in a different way, suggested by the two widely different classes of ele- ments that contribute to improvement in training : The sudden rise of the curve in the beginning stages is due largely to 'adaptive' changes in which elements of former experience have been 'transferred' bodily to the new work; the later, more gradual, rise, broken by plateaus, is due to a more orderly development which was called "practice effect proper." The influence of this practice-effect, apart from transference, was termed "spread of train- ing (see p. 230). "' Sleight, op. cit. pp. 44of . 226 JOHN EDGAR CO OVER To locate and define the 'common factor,' a thorough analysis of a set of processes into its elements is necessary; and this is exceedingly difficult because introspections, even when expertly made, are scarcely ever complete descriptions and reveal only the elements selected from the content of consciousness which enjoy a considerable degree of clearness. No doubt this in- completeness tends to a portrayal of the elements as discrete units connected in simple mechanical relations somewhat con- gruent with the old associational psychology. Admittedly in- complete analysis, however, yet shows, when it is at its best, that the elements found are not of the same order and are not simply related. a. Factors involved in one experiment The difficulty of defining the common factor and of stating the laws under which it operates, may be best apprehended, per- haps, by following a single experiment through its three or more parts in an effort to locate some of the elements of the processes which may serve as the 'Common Factor' : (i) The first period, between the 'ready' signal and the ap- pearance of the stimulus, is characterized by a set of conscious- ness which includes certain predispositions or tendencies which affect the course of the processes to follow. It includes an emo- tional attitude toward the experiment and a corresponding 'will to succeed'; definite ideals of efficiency to be attained; ideas of method to be used. This 'anlage' comes to every experiment as an adaptation to a more or less novel situation, and is generalized therefore from past experience, is modified in recurring experi- ments on the basis of the resultant experiences from the fore- going; and engages at once the selective function of attention in the adaptation, and the function of control in carrying out the initial steps of the purpose, such as the readiness for a certain direction and distribution of the attention, and renewal of defi- nite imagery which is to be used in ready recognition or recep- tion of the material. (2) The period of dealing with the stimulus may be either short or long, (a) If short, as in simple reaction, sensible dis- FORMAL DISCIPLINE 227 crimination, or tachistoscopic work, quickness of adaptation or alertness of the attention, as well as degree of concentration, is a condition of good performance, since it affects directly sensi- tivity for the stimulus. The direction and distribution of the attention prepared for are carried out; part-processes are co- ordinated, external and internal distractions are inhibited, un- essential imagery is reduced to a minimum, (b) If long, as in repeated reactions (card-sorting, typewriter-reaction, marking out a's), or in memorizing, there are in addition to the fore- going, sustained attention, more complicated part-processes, greater use of categories of classification, more development of method in process, more use of individual forms of imagery. In any case the process varies more with the form (in reaction, or memorizing) than with the material of the experiment. (3) In tachistoscopic or memory work the period between the stimulus and recording may be used chiefly to make imagery definite for recording and may also serve for 'maturing' content into the field of clearness. In sensible discrimination it may serve for comparison of images. The control of the attention is therefore important, else the vague imagery will not be so di- rectly entertained, and distraction will cause a loss of some of the clear imagery. Coordination of retaining and defining proc- esses is demanded. (4) In the period of recording, coordination of the retentive and reproductive processes call for economic distribution of the attention. A single experiment thus contains many factors of a formal nature which cannot be treated as coordinate elements. ^^^ b. Practice-effect upon these factors Analyses of processes in training and tests in the preceding experiments show in what way these factors are changed through training : (i) The attitude becomes optimal in tone, the 'will to suc- "' And consequently cannot be illustrated by a, b, c, d, e, f, etc., as though simply and mechanically related either with each other or with 'content' elements. 228 JOHN EDGAR COOVER ceed' is firmer, the essential processes can be more definitely pre- pared for in the set of attention, strong expectation of particular content drops away. (2) (a) Where the period of dealing with the stimulus is short, as in reaction and sensible discrimination, intensity and di- rection of attention comes under better control, resulting in greater sensitivity for the stimulus, greater preparedness for reaction, or higher discriminability, and closer coordination of sensation and movement or of discrimination and report. Ad- ventitious imagery drops away, (b) Where the period is long and the processes are successive and complex, as in reaction with discrimination and choice (card-sorting, typewriter-reaction) there is first an elaboration of method involving analysis and classification of the stimuli and synthesis of like stimuli and simi- lar relations into a scheme which mediates between stimulus and reaction. Improvement here consists in adaptability in forming an adequate scheme, and in an economic distribution of the atten- tion to the various part-processes it involves. Then comes the dropping away of retarding factors, the stimulus becomes co- ordinated with its reaction, and improvement here consists in sustained attention and in a distribution of attention which equal- izes readiness for all reactions ; kinaesthetic accompaniments of recognition (as pronouncing, or movement of some part of the body) and of initiating the reaction, and those grosser bodily movements, tensions, and strains, accompanying effort, drop away. In memorizing cards of exposed letters, figures, or symbols, methods of analyzing, classifying, and grouping, and of co- ordinating part-processes involving disparate imagery and dif- ferent grades of clearness, grow up; systems of representative imagery (associations) develop; complexity of content increases from a two-fold to a four-fold system of imagery, and from a two-level to an eight-level grade of clearness. Improvement in this part of the experiment involves factors which are common to all processes of learning, and also factors which are specific to the material or the form of the experiment. But the relative importance of these two classes of factors may FORMAL DISCIPLINE 229 be illustrated from reaction with discrimination and choice. In no exercise, probably, are the special factors more important. The process of forming coordinations between specific stimuli and specific reactions is the end attained by the training. Yet those specific coordinations constitute the least part of the train- ing effect ; power of forming coordinations between other specific stimuli and other specific reactions has been so much increased that learning new systems is greatly facilitated. ^^^ Memorizing furnishes another illustration : Methods of apperceiving and rep- resenting the material are more permanent than the material learned, and have a wider application. ^^^ (3) In the period of fixing the imagery and of 'maturing' vague imagery into impressions definite enough for cognition, improvement consists in a peculiarly delicate control of attention in respect to direction, to distribution, and to degree; and this experience once realized probably makes it possible to entertain any kind of imagery in the same way. (It resembles in nature the hearing out of an 'upper partial' from the complex clang of a vibrating string, in that stimulation directs the attention; but it is not so simple a process.) Improvement through training has the effect of increasing sensitivity to central excitations, of increasing both the liability and the fidelity of reproduction. (4) Improvement in coordinating the retaining, reproductive, and recording processes effects an economic distribution of at- tention which results in a singular self-possession and poise, a satisfying feeling of fitness for the task. Economic distribution of the attention to such complex processes is a matter more of form than of content, for both stimuli and form of expression may change yet that complicated formal process is to some ex- tent available. ^^^ This statement has necessarily been made in general terms and has emphasized the more formal elements which are fitted to fill the office of the 'Common Factor.' The more concrete, and ^*^Vid. Bair, {op. cit.) and Liddle {op. cit.). ^^"Cf. Fracker {op. cit. gi), and Gamble {op. cit. 97, 149). '"C/. Fracker {op. cit. 95). 230 JOHN EDGAR CO OVER presumably more specific elements, dependent upon the particular work of the task, have been abundantly illustrated in the discus- sions of test and training results. The more formal elements appear at this juncture to need emphasis since they are so likely to be overlooked, especially when the analysis is expected to lead to simple elements mechanically related. c. 'Spread of Training When the more orderly changes in processes effected by prac- tice in one task are carried over to another we have what the writer thinks is best described as a 'Spread of Training.' This occurred more conspicuously in the Experiment on Marking out Words, the Experiment on Sensible Discrimination, and the Ex- periment on Reproduction. The 'Common Factor' appears to be formal and to be relatively detached from the data of presen- tation, and, in the last experiment, from method. No systems of imagery were carried over. The practice-effect consisted in stripping the essential process of unessential factors, in facility in developing automatic coordinations, in establishing habits of higher order, in dealing more effectively with vague imagery, etc. This kind of general effect seems to rest more directly upon modes of mental processes than upon the material of experience, and seems capable of description largely in terms of emotional and volitional attitudes and of the control of attention. d. 'Transference' The term 'Transference' seems particularly applicable to the carrying over, from one task to another, of the material of ex- perience. It takes place typically in the 'adaptation' to a novel exercise, and in the application of physiological processes and systems of imagery in an exercise similar to the exercise in which they were developed. These two kinds of general effect of special practice bear to each other a relation analogous to form and matter in logic, and, although inseparable, should be discriminated for the purpose of prohibiting the limit of search for the 'Common Factor' to the case of 'Transference' alone. Enumeration of the elements of FORMAL DISCIPLINE 231 the material of experience "will then not exhaust the processes in an exercise, to be considered for the purpose of locating the 'Common Factor.' In preceding pages when the causes of gen- eral effect of practice have been discussed, in the review of the results of experimentation conducted by others,^^^ and by our- selves,^"^ some elements properly falling under the head of 'Spread of Training' were always included in the consideration. If the reader should glance over the 'Summary' (pp. 3 iff) of the evidence for relationship between mental processes, from the literature, he will note that the eighteen numbered cases all in- volve 'Spread of Training,' all but eight involve it almost ex- clusively, and only two involve a large share of 'Transference.' e. Both kinds of General Effect selected from former experience It has already been pointed out^^^ that elements of former ex- perience contribute to the work of a test. This is applicable to both kinds of general effect, as may be seen by inspecting the analyses of processes in the last two experiments. To illustrate, with reference to 'Spread of Training,' the reagent comes to the experiment with emotional factors, voli- tional attitudes, and modes of mental processes already estab- lished : His work in tachistoscopic tests and in learning 12- letter-rectangles is influenced by favorite letters, preference given to right or left position, distribution of attention, etc., in dis- crimination tests, by concern for an equitable distribution of 'greater' and 'less' judgments and by other factors leading to constant errors ;^^^ he gives preference to imagery of one par- ticular sense-mode, as is illustrated by his representative imagery in tests in discrimination, and in memorizing, although he may vary this mode upon occasion, for varying content or method ;^^® he comes to tests in learning 12-letter-rectangles, and in memor- "^ Pp. 19, 22f, 23f, 2-/. ^"Pp. 36ff, 4ifif, 46ff, 6iff, 173, 2i7ff. '=^Pp. 173, i8of. "° Cf. F. Angell's Review of "Zur Analyze der Unterschiedsempfindlichkeit" von Martin u. Miiller, in Am. Jr. Psych., igoo, 11 : 266-7. "° Cf. Segal, J : Ueber den Reproductionstypus und das Reproduzieren von vorstellungen. Arch, f .d. ges. Psychol., 1908, 12 : 124-236. 232 JOHN EDGAR COOVER izing, with tendencies designed to facilitate a particular method of coordinating imagery of different modes, and of placing re- liance mainly upon methods of rote learning or of classification and grouping, and to tests in compound reaction and in memor- izing visual symbols with principles of classification; etc. To illustrate with reference to 'Transference,' the reagent ap- plies elements of the material of former experience, especially in the forms of particular imagery and systems of imagery. To the former belong the conventional things by which visual symbols are cognized and named for memorizing; groups of letters from common phrases and names, cattle-brands, wheat-sack initials, names from Biblical and classical literature, etc. used in learning i2-letter-rectangles; the various representative visual and kinaes- thetic imagery in discrimination of sound, as of flashing light, falling bodies, or raising the toes for louder sound ; etc. To the latter belong the musical staff for memorizing 'note' forms, map directions for cognizing symbols with radii variously turned, series of steps for memorizing sound intensities, mathe- matical relations for grouping numerals, logical schemes for classification of stimuli in compound reaction ; etc. This application of the elements of former experience is fa- miliar to every one who has considered at all critically the process of learning. But it should perhaps be pointed out that in the ap- plication the former elements are not merely picked up as stable units and mechanically inserted in a new process. The elements come with a certain mutation determined by the novelty of the situation, and their application involves a degree of invention. The elements of consciousness do not appear to behave in the manner of reflex-action, and it does not seem possible to get a purely 'specific' effect of practice when the practice is a conscious process. All elements may be the 'Common Factor' ; all effects of conscious practice are to some extent 'general effects.' f . Provisional classification of Common Factors In view of the complexity of the situation it is hazardous to attempt a complete classification of the 'Common Factors,' yet the writer feels inclined to offer the following provisional list : FORMAL DISCIPLINE 233 Common Factors I. Objective. 1. Likeness of material. 2. Likeness of form. IL Subjective. A. Form of Experience. 1. Aufgabe (interpretation of instructions). 2. Attitude (emotional, volitional). 3. Ideals (controlling concepts). 4. Purpose (definition of volitional attitude). 5. Intellectual processes. a. Elaboration of method. b. Short-circuiting of processes. c. Higher order of control. d. Elimination of accompaniments of ( 1 ) Discrimination, (2) Cognition, (3) Movement, (4) Reproduction. e. Growth in simplicity or complexity of imagery, f. Coordination of part-processes. 6. Control of attention. a. In degree, b. In direction, c. In distribution, d. In quickness of adaptation, e. In duration of concentration. B. Material of Experience. 1. Sirnple imagery (Visual, auditory, kinaesthetic, tac- tual, etc.) . a. Direct. b. Associative. (i) In the same sense-mode. (2) In different sense-mode. 2. Compound imagery (Of the above). a. Direct. b. Associative. 3. Complex imagery. (Spatial, temporal, causal, his- trionic). a. Direct. b. Associative. 4. Systems of Imagery (Representative schemes, mne- monic devices). III. CONCLUSION 1. Evidence from the literature of experimental psychology indicates a functional relationship between various mental proc- esses. This relationship is sometimes positive, sometimes nega- tive. Specially designed experiments show that 'specific' prac- tice is never wholly 'general' in its effects; is often largely 'gen- eral,' and is probably always somewhat 'general' Under the experimental conditions, the 'general' effect usually ranged, in amount, from one-fourth to three-fourths of the gain made in the specific practice (vid. Summary, pp. 3ifT).- 2. Results of the repeated experiments of (a) Marking Out Words, and (b) Estimating Weights, support the evidence for 'general' effect of 'specific' practice, and indicate through intro- spective analysis, what change in the practice is responsible for the 'general' effect. a. The training on Marking out Words, on printed pages, con- taining both letters e and s, increased efficiency by reducing the recognition of words as containing c and s to its essential process, through relieving it of unnecessary and retarding accompani- ments, chiefly kinaesthetic, motor, and auditory. This factor in the training-effect was responsible for the marked increase in efficiency shown in the tests on marking out words containing other pairs of letters, on printed pages and on manuscript sheets. Some 'specific' effects of practice (such as word-reaction) which would not contribute to 'general' effect, or, if so, only in a nega- tive form, were also found; but the principal factor of improve- ment in the practice was the main factor of improvement in the tests, — a general effect (34-39). b. The training on Estimating Weights resulted in building '^ The statements under this head will be found in more amplilied form in the Conclusions of the various experiments, which may be located by refer- ence to the Table of Contents. ^The page numbers in parenthesis refer to material upon which the state- ments are based, including conclusions to experiments. FORMAL DISCIPLINE 235 up a definite idea of the field represented by the training weights (40-120 grams), or in deepening impressions of the weights at the upper and lower limits of this field; and these ideas of value improved the capacity for estimating weights differing in kind, both within the field of training and above it. Both reagents showed more improvement in tests on objects dissimilar to the training-weights, but within the field, than they made in training ; and one reagent made his greatest gain in the test on objects above the field. These anomalous results, together with the in- trospective evidence of the complexity of the estimating process, suggest that simpler processes should be chosen for measuring and analyzing 'general' practice-effect (pp. 39-42). 3. The two experiments, (a) Sensible Discrimination, and (b) Reaction with Discrimination and Choice, designed to de- termine whether there is 'general' effect of 'special' practice when the processes involved are as simple as possible, and when the tests differ from the training (a) in sense-mode of reception of the stimuli, or (b) in the form of the stimuli and their motor- response, but when the tests and training involved the same kind of mental activity, contribute results indicative of the real com- plexity of the relatively simple processes, and of the 'general' nature of practice-effect. a. Training in Sensible Discrimination of intensities of Sound resulted in improvement in efficiency through divesting the dis- criminating-process of its unessential and complicating factors consisting in irrelevant or fantastic imagery, indirect sets of at- tention, vascillating attention, expectation, etc., all of which render judgment illusionable, and this improvement was trans- ferred to the tests in sensible discrimination of shades of gray (pp. 42-50). b. The training in sorting cards bearing distinctive colors into a cabinet with six compartments increased efficiency in Reaction with Discrimination and Choice, which was transferred to reac- tion to letters on a typewriter, (a) noticeably in regularity with two reagents for whom the latter had become automatic before the training in card-sorting was begun, and (b) markedly in speed 236 JOHN EDGAR CO OVER with two reagents for whom the latter was in course of practice. The practice-effect responsible for the improvement in the card- sorting and for the improvement transferred to the typewriter- reaction was ( I ) the habit of stripping the essential process of its adventitious accessories, consisting chiefly in (a) kinaesthetic and verbal elements accompanying and retarding the recognition of the stimulus, (b) mnemonic schemes which served the purpose of building up coordinations between stimulus and reaction, (c) false motions, and (d) bodily strain, and (2) such control of the atten- tion (a) that the various possible reactions were about equally prepared for and (b) that the series of continuous reactions were not so frequently broken by balks due to distraction (pp. 50-64). 4. Disagreement in the results of investigations as to the ex- tent of the general effect of special practice, in hypotheses pro- posed for the causes of the transference of practice-effect, and the frequency of cases of an anomalous character which defy any consistent hypothesis, are probably due to differences in technical procedure in experimentation, to differences in kind and length of training and in the relation of training to tests, and to dif- ferences in statistical treatment of results. They call for more qualitative investigation to the end of determining more precisely how training affects the processes engaged in the training, how training-effect affects the test-capacities, and how training-effect, both direct and 'general,' may be properly expressed in quanti- tative terms. A study of anomalous cases suggests the import- ance of determining (a) the extent of variability in processes, both with a single reagent and between different reagents who set themselves to the same objective task, (b) the causes of this variability, and (c) its effect upon the scores (pp. 64-69). 5. The Experiment on Attention, which provided a qualitative study of the kind, extent, and causes of variability in processes engaged by a single reagent in a single test and in a range of tests, and of variability between processes engaged by different reagents in identical tests, yielded results which show: a. That it is the rule for the individual reagent to vary his processes while at work on a single test and often to change FORMAL DISCIPLINE 237 radically, in the final test, the methods of work employed in the first. b. That the causes of this variability, beyond general condi- tions of health, interest, etc., and incidental occurrences, such as winking at the moment of the presentation of the stimuli, and accidents in manipulation of a key, etc., were of a fairly specific nature varying in accordance with the nature of the work of the test, such as voluntary or undesigned shifts of the attention to various elements of the processes engaged, changes in the extent of the distribution of the attention over part-processes and their coordination, the constructing of more adequate methods, and the practice-effect of dropping out of the process unessential factors, in heightening sensitivity, discrimination, reproduction, habituation to distraction, and in building up habits of higher order. c. That in almost every test individual reagents differ from each other, often greatly, in the way in which they performed the work of the test. d. That they differ in the 'aufgabe,' or their understanding of the instructions for the test, in their general experience from which elements are selected, by way of adaptation, to begin work, in the order and degree of the changes in processes due to the variability in the single reagent's work noticed in (b) above .(pp. 64-184, particularly 167-173). 6. The training results of this experiment brought into clear relief the fact that practice-effect itself involves changes in pro- cesses : At the beginning of training on Test 17, only letters that had been clearly seen were recorded ; at the end, letters were cor- rectly recorded that had not been "seen," but that 'matured' from the 'fringe' content of consciousness ; the conquest of the 'fringe' content was a practice-effect that extended the area of distinct perception whilst increasing efficiency on its own account. At the beginning of training on Test 13, only letters the visual im- pression of which was converted into kinaesthetic-auditory (ver- bal) imagery were recorded; at the end, letters were recorded from four distinct kinds of imagery ; at the beginning, the process was a simple 'rote' process ; at the end, it was a complex process 238 JOHN EDGAR COOVER involving coordination of many part-processes. The conception that repetition in the learning-process increases efficiency simi- larly to repetition of a skilled movement in fixing a habit is an absurdity against which the fact of the increasing keenness of consciousness accompanying the progress of the former ought to have been sufficient warning. Repetition in learning changes the process (pp. 173-176). 7. Altho the series of tests proved inadequate as a measure of attention, there was some quantitative indication of a practice- effect upon certain forms of attention (or attentive forms of consciousness) that were of general application: (i) The con- trol of attention to seize the stimuli of the moment; (2) equit- able distribution of the attenion over the various part-processes, leading to coordination of disparate imagery, of processes of re- tention and reproduction, or of imagery and movement; (3) sus- tained application, involving inhibition to external and internal distraction; (4) an intensity of application, effective in lowering thresholds of sensation and reproduction, and in fusing co- ordinations (pp. 178-180). 8. There was ample introspective evidence for general effect of special practice, both negative and positive. Not only were methods of work, forms of processes, and systems of imagery, transferred from one kind of work to another, but in the adapta- tion to the novel work of the tests elements of former experience were selected and applied, which when acquired must always have been 'specific' and as applied are always 'general' (pp. 180- 181). 9. The Experiment on Reproduction supplemented the preced- ing experiment in illustrating the great variability in processes engaged in a single test, locating the variability in each successive 'moment' of the single test, and in offering new evidence, both quantitative and qualitative, for the positive and negative general effect of special practice, when the tests and training differed so greatly in both material and method as to exclude a 'transference' of the elements of experience (184-219). 10. The training on discrimination of sound did not result in improvement in efficiency with the training material. But, ac- FORMAL DISCIPLINE 239 cording to introspective evidence, it effected changes in the processes employed. Quantitative analysis showed that the practice-effect of the evident exercise of retention and reproduc- tion of auditory and other imagery 'spread' to the tachistoscopic test of Recognition or Choice of One of Two Letters, and to the test on the Complete Learning of series of visual symbols, both of which involved retention and reproduction of imagery (pp. 184-219). Retrogression in efficiency in the course of practice, and failure to apply methods developed in training to material slightly different, etc., indicate a negative special effect of special practice, which should be taken into account when testing for general effect. The retrogressive effect shows itself in negative general effect in tests similar to the training, and is not incon- sistent with positive general effect in tests less similar to the training (p. 219). 11. Data from the latter two experiments indicate the im- portant bearing which variability in processes has upon the quan- titative treatment of results (pp. 167-176, 181 -184). Certain precautions need to be taken in order that the results of the test- training-test type of experiment may not be misleading (pp. 219- 224). It would seem that purely quantitative results are worthless. 12. Introspective description of the processes involved in the successive 'moments' of a single experiment, and of practice- effect upon them, indicates such a complex relationship of the elements or part-processes that are fitted to perform the office of the 'Common Factor' that it seems doubtful if there are any purely 'specific' elements in the conscious process, The 'Com- mon Factor' may be constituted of formal modes of conscious- ness (emotional and volitional attitudes, modes and habits of consciousness, control of attention) in which case it effects 'Spread of Training'; or it may be constituted of the material of consciousness (imagery, systems of imagery, direct and repre- sentative) in which case it effects 'Transference' of the elements of experience. Search for the common factor should not be so conducted as to locate only the latter form; experi- mental evidence indicates that the former is more largely 240 JOHN EDGAR CO OVER responsible for the general effect of special practice. In both forms it is selected^ from former experience and applied with some degree of invention to a more or less novel situation. It would seem that 'adaptability' consists in this 'general' use of the forms and elements of experience which had their origin in 'specific' reactions to specific stimuli,' and that the desideratum of the 'learning process' consists in extending the 'general' applications as widely as possible.* There is evidence that all modes and elements of experience may be the 'Common Factor' ; that all effects of conscious practice are to some extent 'general' effects (pp. 224-233). 13. Discussion throughout the text makes it clear that much further work is needed in this field. With improved method a complete survey of functional relationship between the mental processes should be made. Many outstanding questions await attack : . Conditions favoring and limiting 'general' effect, the scope and causes of negative influence, the duration of direct, and of the various 'general' applications of, trainir^-effect, and its dependence upon length and rigor of training and upon kind of work; relative dependence of individual variation in processes upon nature and nurture ; dependence of efficiency upon training in a 'best way' of performing a mental task, etc.^ ^ This selection and application is not necessarily purposeful or even evident to introspection. The value of introspection, as has been pointed out in the text (p. 181), does not rest upon its assertion or denial of relationship be- tween processes, although its testimony concerning this may often be true, but upon a sufficiently complete and accurate analysis of processes that their relationship may be evident. It is particularly competent to show in this way 'Transference' of the elements of experience; 'Spread of Training,' however, is likely to escape detection unless analyses are especially complete, for it is sometimes extremely subtle and evident only through quantitative analysis (as in the Experiment on Reproduction). * Colvin (Some facts in partial justification of the so-called Dogma of Formal Discipline. Univ. 111. Bull. 1910. 7 : No. 26, p. 31 ; also, The Learn- ing Process. 191 1. Pp. 24^2^) makes a plea for the application of this prin- ciple in school work, and gives "rules for securing transfer," or for securing a "general" training. 'The circuitous process of testing and abandoning various methods of work which often retards reagents in course of practice, and the arrest of an occa- FORMAL DISCIPLINE 241 After investigation has profited by adult introspective analysis and it has been carried out with adolescents, trained in introspec- tion, and has been quantitatively checked by results from children, we may perhaps by reason of our intimate knowledge of the functional relationship of mental processes, the integration of experience, the formal learning-process, be able to determine whether the limited time given to formal education should be spent primarily for discipline or for knowledge, whether infor- mation and aims belonging to adult vocations are as vital in the educative processes of children as such information and aims as are incidental to the successful functioning of experience through graded steps in a formal training. 14. Meanwhile, experimental research in the psychological laboratory has established certain functional relationships be- tween mental processes more or less simple, and has singled out factors responsible for the transference of improvement from one exercise to another. Its contribution to the general question of formal discipline is important, in that it shows the general effect of special practice; but it is limited in its scope to fairly simple processes, to periods of short training, and, for intro- spections, to reagents upon whom training does not produce great practice-effect. Were the processes as complex and various as those engaged in the study of Greek, Mathematics, or Science; were the training to extend over four school-years instead of over ten weeks under limiting laboratory conditions ; and were the subjects in the plastic period of the 'teens,' it is conceivable that the extent to which the training is general would be found to be greatly in excess of the laboratory figures. sional reagent upon a low plateau by reason of a peculiarly inept method of work, suggest that training in 'good form' might be the better part of prac- tice, especially in the early stages, but also in the later stages which involve transitions to new coordinations of part-processes. This phase of practice is considered important in athletics, and in the trades involving skilled move- ments; and Swift (Studies in the psychology and physiology of learning. Am. Jr. Psych., 1903, 14 : 224) suggests its application to the early stages of learning skill, while Bryan and Harter (Studies on the telegraphic language: The acquisition of a hierarchy of habits. Psych. iRev., 1899, 6:375), by pointing out the end-processes of training, suggest its place in the acquisition of skill in such types of learning as involve hierarchies of habits. ;<;42 JOHN EDGAR CO OVER 15. The contribution of Experimental Psychology to the edu- cational theory of training is timely, since intensity of work and drill^ is coming into vogue again. It is found that great effort and rigorous drill are necessary to pass through 'plateaus of growth' or, as Bagley^ calls them, "sloughs of despond." There seems to be a disposition to find the conditions of training that will "unlock reservoirs of higher power,"^ and those who find satisfaction in physiological explanations are turning to the theory of the 'synapse'^ and the "All or None"^^ principle. This increase of power is deemed necessary for the acquisition of such mental habits and such knowledge as, under the conditions of our civilization, constitute a reasonable preparation for complete living. ' Beckwith : Drill, a chapter in pedagogy. 1905. Book: Psychology of skill. Univ. Mont. Bull. 1908. P. 53. Bryan and Harter : Studies in telegraphic language. Psych. Rev. 1897. 4: 27. Swift : Learning to Telegraph. Psych. Bull. 1910. 7 : 149. Wallin : Has the Drill become obsolete? Jr. Ed. Psych. 1910. 1:200. ' Bagley : Craftsmanship in teaching. Macm. 191 1. "James: Energies of men. Science, 1907. 25:321; also Am. Mag. Nov. 1907. * Sherrington : Integrative action of the nervous system. Scribner's. 1906. " Stiles : "All or None" principle and its implications. Am. Phys. Ed. Rev.. 1910. 15:1. IV. BIBLIOGRAPHY (Numbers in parenthesis indicate the pages in the text upon which the reference is used). Aall, A. (Review of Coover and Angell). Zeits. f. Psychol., 1908, 48:303. (70). Abbott, Edwina E. On the analysis of the memory consciousness in orthography. Psychol. Rev. Mon., 1909, No. 44:127-158. (167). Aiken, Catherine. Methods of mind training concentrated attention and memory. Am. Bk., C1895, (3). Alechsieff, Nicholaus. Reactionszeiten bei Durchgangsbeobachtungen. Phil. Stud., 1900, 16:1-60. (112, 119). Angell, F. (Commencement address at Castilleja School, Palo Alto, California, 191 1). (3). Discrimination of clangs for different intervals of time. 11. Am. Jr. Psychol., 1900-1901, 12:58-79. (134, 198). Discrimination of shades of gray for different intervals of time. Phil. Stud., 1902, 19:1-21. (43, 135, 197, 198). On judgments of "like" in discrimination experiments. Am. Jr. Psychol., 1907, 18:253-260. (134). (Review of Martin und Miiller). Am. Jr. Psychol., 1900, 11:266-7. (198,231). Untersuchungen iiber die Schatzung von Schallintensitaten nach der Methode der mittleren Abstufungen. Phil. Stud., 1892, 7:414-468. (207). and Harwood, Henry. Experiments in discrimination of clangs for different intervals of time. I. Am. Jr. Psychol., 1899- 1900;11:67-79. (134). Angell, James R. The determination of mental imagery. Psychol. Rev. Mon., 1910, No. 53:61-108. (177). The doctrine of formal discipline in the light of the prin- ciples of general psychology. Ed. Rev., 1908, 36:9 ff. (70). and Moore, A. W. Reaction time: A study in attention and habit. Psychol. Rev., 1896, 3:245-258. (9, 108, no, 112, Arnold, Felix. The initial tendency in ideal revival. Am. Jr. Psychol., 1907, 18:239-252. (156). Bagley, William Chandler. Craftsmanship in teaching. Macm., 1911. (242). Bagley, William Chandler. The educative process. Macm., 1907. (4)- On the correlation of mental and motor ability in school children. Am. Jr. Psychol., 1900-1, 12:193-205. (118, 164). Bair, J. H. The practice curve. A study in the formation of habits. 244 JOHN EDGAR COOVER Psychol. Rev. Mon., 1902, No. 19; also review of the same in Psychol. Rev., 1903, 10:579 ^- (". I3. 64, 126, 229). Balaban, A, Ueber den Unterschied des logischen und des mechan- ischen Gedachtnisses. Zeits. f. Psychol., 1910, S^'-ZS^-Z77> 379" 400. (151, 156). Beckwith, Walter P. Drill: A chapter of pedagogy. 1905. (242). Bennett, Charles J. C. Formal discipline. Teachers College, 1907. (6, 15, 135)- Bentley, I. Madison. The memory image and its qualitative fidelity. Am. Jr. Psychol., 1899, 11:1-48. (135, 177, 190, 212). Bergemann, Robert. Reaktionen auf Schalleindriicke, nach der Methode der Haufigkeitskurven bearbeitet. Psychol. Stud., 1906,1:179-218. (103). Berger, G. O. Ueber den Einfluss der Uebung auf geistige Vorgange. Phil. Stud., 1889, 5:170-178. (176). Bergstrom, John A. Effect of changes in the time variables in memorizing, together with some discussion of the technique of memory experimentation. Am. Jr. Psychol., 1907, 18:206-238. (87, 142). An experimental study of some of the conditions of mental activity. Am. Jr. Phychol., 1893-4, 6:247-274. (14, 64, 118). Experiments upon physiological memory by means of the interference of associations. Am. Jx. Psychol., 1892-3, 5 :256- 369. (II, 118). Relation of the interference to the practice effect of an association. Am. Jr. Psychol., 1894,6:433-442. (11, 118). Binet, A. A propos de la mesure de I'intelligence. Annee Psychol., 1905, II :69-82. (108). Attention et adaptation. Annee Psychol., 1899, 6:248-404. (108, 113, 117, 150, 169,222). L'fitude experimentale de I'intelligence. Paris, 1903. (69, 113,145,146,158,168,169). et Henri, V. La psychologie individuelle. Annee Psychol., 1895,2:411-465. (69,108,113,150). et Jassey, J. £tude de psychologie sur les auteurs. dramatiques. Annee Psychol., 1895. 1:60-118. (69). et Vaschide, N. fipreuves de vitesse chez les jeunes gar- gons. Annee Psychol., 1897, 4:64-98. (164). Experiences de vitesse chez les jeunes gens. Annee Psychol., 1897, 4.200-224. (164). Bingham, John. Memory. II. Psychol. Rev., 1894, i :453-46i. (154, 167). Bolton, Thaddeus L. Growth of memory in school children. Am. Jr. Psychol., 1891-2, 4:362-380. (145, 146). Relation of motor power to intelligence. Am. Jr. Psychol., 1903, 14:615-631. (164, 222). BIBLIOGRAPHY 245 Book, Wm. Frederick. The psychology of skill, with special refer- ence to its acquisition in typewriting. Univ. Montana Bull., 1908, 53 :i-i88. (30, 65, 177, 242). Bourdon, B. Observations comparatives sur la reconnaissance, la discrimination, et I'association. Rev. Philos., 1895, 40:153-185. (113)- Recherches sur I'habitude. Annee Psychol., 1901, 8:327- 340. (64, 113, 149). Brown, Warner. Habit interference in sorting cards. Univ. Calif. Pub. in Psychol., 1914, i :269-32i. (12, 118). Brown, William. Some experimental results in correlation of mental abilities Br. Jr. Psychol., 1910, 3:296-322. (133, 158). Bryan, Wm. L. On the development of voluntary motor ability. Am. Jr. Psychol., 1892, 5:123-204. (164). and Harter, Noble. Studies in the physiology and psy- chology of the telegraphic language. Psychol. Rev., 1897, 4:27-53. (177,242). Studies in the telegraphic language. The acquisition of a hierarchy of habits. Psychol. Rev., 1899, 6:345-375. (241). Buccola, G. La legge del tempo nei fenomeni del pensiero. Bibl. scient. intern., 1883. Milano. (108). Burt, Cyril. Experimental tests of general intelligence. Br. Jr. Psychol., 1909, 3:94-177. (118, 135, 164). Calkins, Mary W. Community of ideas of men and women. Psychol. Rev., 1 896, 3 :426-430. (158). Short studies in memory and in association. Psychol. Rev., 1898, 5:451-462. (151, 167). Cattell, J. McK. Mental tests and measurements. Mind, 1890, 15:373-381. (108,142). Psychometrische Untersuchungen. Phil. Stud., 1886, 3:305-335- (109)- and Bryant, Sophie. Mental association investigated by experiment, Mind, 1889, 14:230-250. (158). and Farrand, Livingston. Physical and mental measure- ments of the students of Columbia University. Psychol. Rev., 1896, 3:618-648. (108, 113, 145, 146.) Charcot, J. M. Legons sur les maladies du systeme nerveux. Oeuvres Completes, 1886-1890, 3:154-193. (70). Chase, H. W. Some aspects of the attention problem. Ped. Sem., 1909, 16:281-300. (71). Cleveland, Alfred, A. Psychology of chess and of learning to play it. Am. Jr. Psychol., 1907, 18:269 fif. (177). Cohn, Jonas. Beitrage zur Kentniss der individuellen Verschieden- heiten des Gedachtnisses. Dritter Internationaler Congress fiir Psychologic, in Miinchen. 1896, 456-8. (70, 87). Experimentelle Untersuchungen iiber das Zusammen- wirken des akustischmotorischen und des visuellen Gedachtnis- 246 JOHN EDGAR CO OVER ses. Zeits. f. Psychol., 1897, 15:161-183. (70, 80, 150, 151, 154)- Colvin, Stephen Sheldon. The learning process. N. Y., Macm., 1911. (240). Some facts m partial justification of the so-called dogma of formal discipline. Univ. 111. Bull., 1910, 7:No.26. (240). Coover, J. E. and Angell, F. General Practice effect of special exer- cise. Am. Jr. Psychol., 1907, 18:327-340. (23, 42). Culler, Arthur Jerome. Interference and adaptability. Archives of Psychol., 1912, 3 :No.24 (170, 222). Culverwell, E. P. The creation of a memory. Jr. Exper. Ped., 1911-2, I :i6o-i. (176). Davis, Horace. An afternoon view of college life. Stanford Alum- nus, 1912, 13 :23i. (3). Davis, Walter W. Researches in Cross Education. First Series. Studies from the Yale Psychol. Lab., 1898, 6:6-50. Second Series, loc. cit., 1900, 8:64-108. (4). Dearborn, W. F. The general effect of special practice in memory. Psychol. Bull., 1909, 6:44-45. (18). De Garmo, Chas. "Formal Culture" (in the Dictionary of Philos- ophy and Psychology, ed. by Baldwin) . ( i ) . Herbart and the Herbartians. (i). The report of the Committee of Ten. Ed. Rev., 1894, 7:277. (i). Dexter, Edwin G. High grade men ; in college and out. Pop. Sci. Mo., 1903, 62:429-435. (3). Donaldson, Henry Herbert. The growth of the brain. (185). Dressier, F. B. Some influences which affect rapidity of voluntary movements. Am. Jr. Psychol., 1892, 4:524-527. (4, 118, 164). Dunlap, Knight and Wells, Geo. R. Some experiments with reac- tions to visual and auditory stimuli. Psychol. Rev., 19 10., .17:319-335- (6). Ebbinghaus, Hermann. Grundztige der Psychologic. Leipzig. 1902 (64). Ueber das Gedachtniss. 1885. (14). Ueber eine neue Methode zur Priifung geistiger Fahig- keiten und ihre Anwendung bei Schulkindern. Zeits. f . Psychol., 1897,13:401-459. (142,145,157). Ebert, Ernst und Meumann, E. Ueber einige Grundfragen der Psychologic der U^ebungsphanomene im Bereiche des Gedacht- nisses. Archiv f. d. ges. Psychol., 1904. 4:1-232. (15, 64, 66, 67, 70, 167, 169, 177, 188, 211). Epstein, S. S. Ueber Modification der Gesichtswahrnehmung unter dem Einflusse von gleichzeitigen Toneindriicken. Zeits. f. Biol, 1896, N. F. 15, 33 :28. (6). Fechner, Gustav Theodor. Psychophysik, 2te Auf., 1:181, (70, 78, 197; Appendix B, Fig. 7a). BIBLIOGRAPHY 247 Fernald, Mabel Ruth. The diagnosis of mental imagery. Psychol. Rev. Mon., 1912, No. 58. (69, 107). Finzi, Jacopo. Zur Untersuchung der Auffassungsfahigkeit und Merkfahigkeit. Psychol. Arbeiten, 1899-1901, 3:289-384. (154). Flonrnoy, Th. De Taction du milieu sur I'ideation. Annee Psychol., 1894, 1 : 180-190. (158). Fracker, Geo. Cutler. On the transference of training in memory. Psychol. Rev. Mon., 1908, No. 38, 5:56-102. (19, 23, 64, 66, 70, 138, 139, 183, 223, 225, 229). Galton, Francis. Inquiries into human faculty and its development. Lon., 1907. (70). Psychometric experiments. Brain, 1879-80, 2:149-162. (158, 167). Statistics of mental imagerv. Mind, 1880, O.S. 5 :30i-3i8. Gamble, Eleanor A. McC. Study in memorizing various materials by the reconstruction method. Psychol. Rev. Mon., 1909, No. 43, 1-211. (24, 28, 70, 167, 229). Geissler, L. R. The measurement of attention. Am. Jr. Psychol., 1909,20:473-529. (71). Gervvert, A. Experimentelle Studien iiber das Gedachtniss bei vis- uellen Empfindungen. Zeits. der Psychiatrie, 1899, No. 8. Re- viewed in Zeits. f. Psychol, 1900, 23:232-3. (197, 198). Gilbert, J. A. Experiments on the musical sensitiveness of school children. Studies from the Yale Psychol. Lab., 1893, i :8o-87. (135). Researches on the mental and physical development of school children. Studies from the Yale Psychol. Lab., 1894, 2:40-100. (108,164). and Fracker, G. C. Effect of practice in reaction and dis- crimination for sound upon the time of reaction and discrimina- tion for other forms of stimuli. Univ. Iowa Studies in Psychol., 1897, 1:62 fif. (9). Griffing, Harold. On the development of visual perception and at- tention. Am. Jr. Psychol., 1895, 7:227-236. (160). Hahn, H. H. and Thorndike, E. L. Some results of practice in addition under school conditions. Jr. Ed. Psychol., 1914, 5 :65- 84. (222). Halleck, R. P. What kind of education is best suited to boys ? Pro- ceedings N. E. A., 1906, 58-65. (3). Hegelmaier. (in Vierordt's Laboratory, about 1852. Quoted by Bergstrom, Am. Jr. Psychol., i8:2ti). (134). Henderson, Ernest N. Formal discipline. A Cyclopedia of Educa- tion, N. Y., Macm., 191 1. (i). Henri, V. fitude sur le travail psychique et physique. Annee Psychol., 1896, 3:232-278. (108, 113). Hinsdale, B. A. The dogma of formal discipline. Proceed. N. E. A., 1894, 625 E; also Ch. II, in Studies in Education, (i). 248 JOHN EDGAR COOVER Hollingworth, H. L. Characteristic differences between recall and recognition. Am. Jr. Psychol., 1913, 24:532-544. (205). Correlation of abilities as affected by practice. Jr. Ed. Psychol., 1913,4:405-414. (223). Individual differences before, during, and after practice. Psychol. Rev., 1914, 21 :i-8. (67, 113, 116, 168, 222). Hylan, J. P. The distribution of attention. Psychol. Rev., 1903,^ 10:373-403, 498-533- (87, 160). Jacobs, Joseph. Experiments in 'prehension.' Mind, 1887, 12:75-79. (142, 145). James, Wm. The energies of men. Science 1907, 25:321-332; also Am. Mag. Nov., 1907; also Philos. Rev., 1907, 16:1-20. (242). Principles of psychology. Lon., 1890. (103). Psychology; Briefer course, 1900. (103). Jastrow, Joseph. Community of ideas of men and women. Psychol. Rev., 1896, 3:68-71, 430-431. (158). A sorting apparatus for the study of reaction times. Psychol. Rev., 1898, 5 :279-285. (51, 75)- A study in mental statistics. New Rev., 1891, 5 :559-568. (158). and Cairnes, W. B. The interference of mental processes. Am. Jr. Psychol., 1891-2, 4:219 ff. (10). Moorehouse, Geo. W., and Harper, Mildred. Classifica- tion time. Am. Jr. Psychol, 1891, 4:411-415. (131). Jerusalem, W. Ein Beispiel von Association durch unbewusste Mit- telglieder. Philos. Stud., 1894, 10:323-5. (185). Johnson, W. Smythe. Experiments on motor education. Studies from the Yale Psychol. Lab., 1902, 10:81-92. (124). Judd, Chas. H., and Cowling, Donald J. Studies in perceptual de- velopment. Psychol. Rev. Mon., 1907, No. 34:349-369. (30). McAllister, C. N., and Steele, W. M. Series of studies of eye-movements by means of kinetoscope photos. Psychol. Rev. Mon., 1905, No. 29:141 ff. (109). Kampfe, Bruno. Beitrage zur experimentellen Priifung der Methode der richtigen und falschen Falle. Philos. Stud., 1892, 8:511-591. (198). Kelly, R. L. Psychological tests of normal and abnormal children r A comparative study. Psychol. Rev., 1903, 10:345-372. (164). Kirkpatrick, E. A. Individual tests of school children. Psychol. Rev., 1900, 7:274-280. (164). Kline, Linus W., and Owens, W. A. Preliminary report of a study in the learning process, involving feeling tone, transference and interference. Psychol. Rev., 191 3, 20:206-244. (118,177). Knors, Clemens. Experimentelle Untersuchungen iiber den Lern- BIBLIOGRAPHY 249 process. Archiv f. d. ges. Psychol., 1910, 17:297-361. (146). Koffka, Kurt. Ueber Vorstellungen. Leipzig, 191 1. (167). Kraepelin, Emil. Die Arbeitscurve. Phil. Stud., 1902, 19 :459-507. (64). A measure of mental capacity. Pop. Sci. Mo., 1896, 49:756-763. (64). Krueger, F., and Spearman, C. Die Korrelation zwischen ver- schiedenen geistigen Leistungsfahigkeiten. Zeits. f. Psychol., 1907,44:50-114. (135,145). Kiilpe, Oswald. Outlines of Psychology. Macm., 1901. (70, 71, 104, 109, 119, 134, 138, 186, 190, 299). Kuhlmann, F. Problems in the analysis of the memory consciousness. Jr. Philos. Psychol. & Sci. Meth., 1907, 4:5-14. (190, 211). On the analysis of auditory memory consciousness. Am. Jr. Psychol., 1909, 20:194-218. (70, 136, 139, 167, 177). On the analysis of the memory consciousness ; a study in the mental imagery and memory of meaningless visual forms. Psychol. Rev., 1906, 13:316-348. (149, 167, 211). Lange, Ludwig. Neue Experimente iiber den Vorgang der einfachen Reaction auf Sinneseindriicke. Phil. Stud., 1888, 4:479-510. Lay, Wilfrid. Mental imagery experimentally and subjectively con- sidered. Psychol. Rev. Mon., 1898, No. 7, 2:No.3. (167). Lehmann, Alfred. Kritische und experimentelle Studien iiber das Wiedererkennen. Philos. Stud., 1892, 7:169-212. (134, 197). Ueber Wiedererkennen. Philos. Stud., 1889, 5 :96-i56, (41, 135)- Liddle, Carrie Walker. Experiments on the transfer of mental faci- lity. (Unpublished Thesis for Degree of M.L. at Univ. Calif., May, 1904.) (10, 229). Lindley, Ernest H. Preliminary study of some of the motor phe- nomenona of mental effort. Am. Jr. Psychol., 1895-6, 7 :49i-5i7. (36). Lipps, Theodor. Grundtatsachen des Seelenlebens. Bonn, 1883. (185.) Loisette, A. Assimilative memory. N. Y., 1896. (104). Lowell, A. L. College rank and professional success ; in appendix to the Report of the President and Treasurer of Harvard Col- lege, 1908-1909. (3). College studies and professional schools. Harvard Grad. Mag., 1910, 19:205-211. (3). McComas, Jr., H. C. Some types of attention. Psychol. Rev. Mon., 1911, No. 55, vol. 13. (71, 203). MacDougall, Robert. Recognition and recall. Jr. Philos. Psychol. & Sci. Meth., 1904, i :229-233. (205). McMein, M. and Washburn, M. F. Efifect of mental type on the interference of motor habits. (In minor studies from the 250 JOHN EDGAR COOVER Psychol. Lab. of Vassar College). Am. Jr. Psychol., 1909, 20:282-284. (i4» 118, 120, 223). Martin, L. J., und Miiller, G. E. Zur Analyse der Unterschiedsem- pfindlichkeit. Leipzig, 1899. (198,231). Mayer, A., und Orth, J. Zur qualitativen Untersuchung der Associa- tion. Zeits f. Psychol., 1901, 26:1-13. (167). Merkel, Julius. Das psychophysische Grundgesetz im Bezug auf Schallstiirken. Phi. Stud., 1888, 4:117-160. (197). Messmer, Oskar. Zur Psychologic des Lesens bei Kindern und Erwachsenen. Archiv f. d. ges. Psychol. 1903-4, 2:190-298. (^60). Meumann, Ernst, Beitrage zur Psychologic des Zeitsinns. Phil. Stud., 1893, 8:431-509. (33). Michotte, A., et Ransy, C. Contribution a I'etude de la memoire logique. fitudes de Psychol., 1912, 1:1-95. (151)- Moore, John M. Studies of fatigue. Studies from the Yale Psychol. Lab., 1895, 3:68-95. (164, 165). Miiller, G. E. Zur Analyse der Gedachtnistatigkeit u. d. Vorstell- ungsverlaufes. Zeits. f. Psychol. Erg.-Bd., i. Teil, 191 1, 5; 3. Teil, 1913,8. (70,143,146,168). (Review of Ebert und Meumann). Zeits. f. Psychol., 1905. 39:111-125. (18, 223). und Pilzecker, A. Experimentelle Beitrage zur Lehre vom Gedachtniss. Zeits. f. Psychol., Erg. Bd., I, 1900. (149). und Schumann, F. Experimentelle Beitrage zur Unter- suchung des Gedachtnisses. Zeits. f. Psychol., 1894, 6:81-190, 257-339- (167,222). Miinsterberg, Hugo. Beitrage fiir experimentelle Psychologic. Heft I. Freiburg, 1889. (131). Heft 4, Freiburg, 1892. (11, 142, 145)- Nevers, Cordelia. Dr. Jastrow on community of ideas of men and women. Psychol. Rev., 1895, 2:363-367. (158). Oehrn, Axel. Experimentelle Studien zur Individualpsychologie. Psychol. Arbeiten, 1896, i :92-i5i. (69, 108, 113, 119). Ordall, Louise Ellison. Consciousness in relation to learning. Am. Jr. Psychol., 191 1, 22:158-213. (13, 177). Otis, Arthus S., and Davidson, Percy E. The reliability of standard scores in adding ability. Ele. Sch. Teacher. 1912, 13:91-105. (222). Peirce, C. S., and Jastrow, Joseph. On small differences of sensa- tion. Memoirs, National Academy of Science, Washington, 1884,373-83- (185). Pentschew, Christo. Untersuchungen zur Oekonomie und Technik des Lernens. Archiv f. d. ges. Psychol., 1903, i :4i7-526. (167). Peters, Wilhelm. Aufmerksamkeit und Reizschwelle ; Versuch zur Messung der Aufmerksamkeitskonzentration. Archiv f. d. ges. Psychol., 1906, 8 :385-432. (114). BIBLIOGRAPHY 251 Philippe, Jean. Un recensement d'images mentales. Rev. Philos., 1897,44:508-524. (167). Sur les transformations de nos images mentales. Rev. Philos., 1897, 43 :48i-493- ( i77) • Pillsbury, W. B. Attention. Lon., 1908. (71, 108, 109, 113, 138). A study in apperception. Am. Jr. Psychol., 1896-7, 8:315 ff. (203). Pohlmann, A. Experimentelle Beitrage zur Lehre vom Gedachtnis. Berlin, 1906. (142, 145). Pritchett, Henry Smith. Fifth annual report of the president and of the treasurer of the Carnegie Foundation for the advancement of teaching. 1910. (3). Radossawljewitsch, Paul R. Das Behalten und Vergessen bei Kind- ern und Erwachsenen nach experimentellen Untersuchungen. Leipzig, 1907. Reviewed in Am. Jr. Psychol., 19:274. (29). Das Fortschreiten des Vergessens mit der Zeit. Univ. Zurich. [Gottingen. n.d.] (19). Raif, Oskar. Ueber Fingerfertigkeit beim Clavierspiel. Zeits. f. Psychol., 1900, 24:352-355. (4). Rail, E. E. Some experimental evidence on the transfer of training in memory. Psychol. Bull., 1912, 9:88 fif. (29). Rein, W. Outlines of pedagogics, (i). Ritter, C. Ermiidungsmessungen. Zeits. f. Psychol., 1900, 24:401- 444. (113). Ruediger, W. C. Indirect improvement of mental function through ideals. Ed. Rev., 1908, 36:364 fif. (4). Schukowsky. Experimentelle Studien iiber das Gedachtniss von passiven Bewegungen. Zeits. d. Psychiatric, 1899 ; Ref. Zeits. f. Psychol., 23:233. (197). Schumann, F. Die Erkennung von Buchstaben und Worten bei momentaren Beleuchtung. Bericht u. d. I. Kongress f. Exp. Psychol., 1904, 34-40. (160, 203). Psychologic des Lesens. Bericht u. d. II. Kong. f. Exp. Psychol., 1906, 153-183. (160). Schuster, Edgar. The promise of youth and the performance of manhood. — A statistical enquiry . . . Whether success in the Exam, for the B. A. degree at Oxford is followed by success in professional life. Galton Eugenics Lab. Memoirs. Univ. of London, 1907, III. (3). Scripture, E. W. Cross-education. Pop. Sci. Mo., 1900, 56:589-596. (4). Recent investigations at the Yale Laboratory. Cross education. Psychol. Rev., 1899, 6:265 E. (4). Ueber den Associativen Verlauf der Vorstellungen. Philos. Stud., 1892, 7:50-146. (185). Smith, Theodate ; and Brown, Emily M. On the education of muscular control and power. Studies from the Yale Psychol. Lab., 1894, 2:115-118. (4). 2Sa JOHN EDGAR COOVER Seashore, C. E. Hearing ability and discriminative ability for pitch. Univ. Iowa Studies in Psychol., 1899, 2:55-64. (135). A method of measuring mental work. The psycher- gograph. Univ. Iowa. Studies in Psychol., 1902, 3:1-17. (52-126). Secor, W. B. Visual reading : A study in mental imagery. Am. Jr. Psychol., 1899-1900, 11:225-236. (37, 69, 158). Segal, Jacob. Ueber den Reproductionstypus und das Reproduzieren von Vorstellungen. Archiv f. d. ges. Psychol., 1908, 12:124-235. (69, 80, 107, 150, 151, 154, 168, 231). Sharp, Stella Emily. Individual psychology : A study in psycholog- ical method. Am. Jr. Psychol., 1899, 10:329-391. (69, 113, 146, ^5^^- . . . , Sherrington, Charles S. Integrative action of the nervous system. N. Y., 1906. (iii, 242). Sidis, Boris. The psychology of suggestion. N. Y., 1898. (185). Slaughter, J. W. A preliminary study of mental images. Am. Jr. Psychol., 1902, 13:526-549. (167, 190). Sleight, W. G. Memory and formal training. Br. Jr. Psychol., 191 1, 4:386-457. (18, 24, 28, 64, 67, 70, 142, 154, 169, 181, 225). Smith, W. G. Antagonistic reactions. Mind, 1903, 12:47-58. (109). Relation of attention to memory. Mind, 1895, N.S. 4:47-73. (80, 150, 151, 154, 155. 295)- Spearman, C. 'General Intelligence' objectively determined and measured. Am. Jr. Psychol., 1904, 15 :20i-293. (134). Starch, Daniel. Transfer of training in arithmetical operations. Am. Jr. Psychol., 191 1, 2:306-310. (4). Starke, Paul. Die Messung von Schallstarken. Phil. Stud., 1886, 3:264-304. (197). Stiles, Percy G. The "All or None" principle and its implications* Am. Phys. Ed. Rev., 1910, 15:1-5. (iii, 242). Stratton, George Malcolm. Experimental psychology and its bear- ing upon culture. N. Y., 1903, (79). Stroh, Marie ; Shaw, A. Margaret, and Washburn, M. F. A study in guessing. Am. Jr. Psychol., 1908, 19:243-245. (185). Stumpf, Carl. Tonpsychologie. 1883. (70). Swift, E. J. Beginning a language. A contribution to the psychol- ogy of learning. Studies in Pnilos. and Psychol. (Garmon vol.), 1906, 297-313. (177)- Learning to telegraph. Psychol. Bull., 1910, 7:149-153. (177,242). Relearning a skillful act : An experimental study in neuro- muscular memory. Psychol. Bull, 1910, 7:17-19. (65). Studies in the psychology and physiology of learning. Am. Jr. Psychol, 1903, 14:201-251. (241). BIBLIOGRAPHY 253 Sybel, A von. Ueber das Zusammenwirken verschiedener Sinnes- gebiete bei Gedachtnisleistungen. Zeits. f. Psychol., 1909. 53 :257-367- {70, 167, 174, 177, 222, 223). Talbot, Ellen Bliss. An attempt to train the visual memory. Am. Jr. Psychol., 1897, 8:414-417. (30, 176). Tanner, Amy. The community of ideas of men and women. Psychol. Rev., 1896, 3 .-548-550. (158). Thomas, Fr. Ein weiteres Beispeil von Assoziation durch eine Ge- ruchempfindung als unbewustes Mittelglied. Zeits. f. Psychol., 1896, 12:60-61. (185). Thompson, Helen Bradford. Psychological norms in men and vi^omen. Univ. Chicago Contrib. to Philos., 1903, 4:No.i (108, 118, 135, 158, 164). Thorndike, Edward L. Educational psychology, ist ed., 1903. (6, 13). Educational psychology. Vol. 2, 1913. The psychology of learning. (45, 57, 188). Effect of practice in case of a purely intellectual function. Am. Jr. Psychol., 1908, 19:374-384. (222). Memory for paired associates. Psychol. Rev., 1908, 15:122-138. (149). Practice in the case of addition. Am. Jr. Psychol., 1910, 2i:483ff. (222). and Woodworth, R. S. The influence of improvement in one mental function upon the efficiency of other functions. Psychol. Rev., 1901, 8:247-261, 384-395, 553-564- (7. 9. 34, 65, 113, 169, 225). Titchener, Edward Bradford. Experimental psychology. Vol. I, Part II, (150) ; vol. II, Part I, (74) ; Part II, (74, 112). Lectures on the elementary psychology of feeling and at- tention. N. Y., Macm., 1908. (71, 108, 134). Simple reactions. Mind, 1895, N.S., 4:74-81. (108). A Text-book of psychology, N. Y., Macm., 191 1. (299). Toulouse, fi. Enquete medico-psychologique sur les rapports de la superiorite intellectuelle avec la nevropathie. (Emile Zola) 1896. (69, 113, 142, 145, 150). Tschisch, W. von. Experimentelle Studien iiber das Gedachtniss von Gehorswahrnehmung. Zeits. d. Psychiatric, 1896. (Quoted in Zeits. f. Psychol, 23:233). (197). Ueber das Gedachtniss fiir Sinneswahrnehmungen. Drit- ter Internationaler Congress fiir Psychol, in Miinchen, 1896, 95 ff. (197,198). Urbantschitsch, V. Ueber den Einfluss einer Sinneserregung auf die iibrigen Sinnesempfindungen. Archiv f. d. ges. Physiol., 1888,42:154-182. (5,6). Vogt, Ragnar. Ueber Ablenkbarkeit und Gewohnungsfahigkeit. Psychol. Arbeiten, 1899-1901, 3:62-201. (5, 113). 254 JOHN EDGAR COOVER Volkmann, A. W. Ueber den Einfluss der Uebung auf das Erkennen raumlichen Distanzen. Ber.d.Kgl.-Sachs.ges.d.Wiss., 1858, io:38ff. (6). Wallin, J. F. Wallace. The doctrine of formal discipline. Jr. Ed. Psychol., 1910, I :i68-i7i. (30). Has the drill become obsolescent? Jr. Ed. Psychol., 1910, 1:200 if. (242). Spelling efficiency in relation to age, grade, and sex, and the question of transfer. Baltimore: Warwick and York, 1911. (4). Warner, Charles Dudley. (Editorial in Harpers Mag., March 1895, on Miss Aiken's work). (3). Wells, F. L. A neglected measure of fatigue. Am. Jr. Psychol., 1908, 19:345-358. (164, 165). Normal performance in the tapping test. Am. Jr. Psychol., 1908, 19:437-483. (82, 164). Practice effect in free association. Am. Jr. Psychol., 1911, 22:1-13. (222). The relation of practice to individual differences. Am. Jr. Psychol., 1912, 23:75-88. (113, 116, 222, 224). Wessely, R. Zur Frage des Auswendiglernens. Neue Jahrb. f. Pad., 1905, 8:297-309, 373-386. (18). Whipple, Guy Montrose. An analytical study of the memory image and the process of judgment in the discrimination of clangs and tones. Am. Jr. Psychol., 1901, 12:409-457, 1902, 13:219-268. (135, 138, 167, 177, 190, 194). Effect of practice upon range of visual attention and of visual apprehension. Jr. Ed. Psychol., 1910, 1:249-262. (160). Manual of mental and physical tests. Baltimore, 1910. (75, 78, 79,. 113, 135, 142, 146, 148, 150, 157, 158, 164). Reaction-time as a test of mental ability. Am. Jr. Psychol., 1904, 15:489-498. (hi). Whitehead, Louis Grant. A study of visual and aural processes. Psychol. Rev., 1896, 3:258-269. (167). Whitley, Mary Theodora. An empirical study of certain tests for individual differences. Archives of Psychol., 1911, 3:No.i9. (69, 113 ,222,224). Winch, W. H. Accuracy in school children. Does improvement in numerical accuracy transfer? Jr. Ed. Psychol., 1910, i :557-589. The faculty doctrine, correlation, and educational theory. Jr. Philos. Psychol, and Sci. Meth., 191 1, 8:337-348, 372-384. (168, 223). Further work on numerical accuracy in school children. Does improvement in numerical accuracy transfer? Jr. Ed. Psychol, 191 1, 2 :262 ff. (4). BIBLIOGRAPHY 255 Immediate memory in school children. I. Visual. Br. Jr. Psychol., 1904, 1:127-134. (80,150,295). Immediate memory in school children. II. Auditory. Br. Jr. Psychol., 2 :52-57. (142). Some relations between substance memory and productive imagination in school children. Br. Jr. Psychol., 191 1, 4:95-125. Transfer of improvement in memory in school children. Br. Jr. Psychol., 1908, 2:284-293. (4) ; 1910, 3:386ff. (4). Wissler, Clark. Correlation of mental and physical tests. Psychol. Rev. Mon., 1901, No. 16, 3:No.6. (107, 108, 113, 135, 145, 158). Wolfe, H. K. Untersuchungen iiber das Tongedachtniss. Philos. Stud., 1886, 3:534-571- (134, 197, 198). Woodworth, R. S. Accuracy of voluntary movement. Psychol. Rev. Mon., 1899, No. 13, 3:1-114. (4). Non-sensory components of sense perception. Jr. Philos. Psychol, and Sci. Meth., 1907, 4:169-176. (190). and Wells, F. L. Association tests. Psychol. Rev. Mon., 1911, No. 57, i3:No.5. (116). Wreshner, Arthur. Die reproduktion und Assoziation von Vorstell- ungen. Zeits. f. Psychol, Erg.-Bd., 3, 1907. (70, 167). Wundt, Wilhelm. Grundriss der Psychologie. lOte auf., 191 1. (160). Grundziige der physiologischen Psychologie. 5te auf., 1902. (43, 74, 77, 80, 103, no, 160, 189). Principles of physiological psychology. Vol. I (tr. by Titchener; § 3 of the Introduction, from the 1893 ed., 18 ff), (2). . . ; Vorlesungen iiber die Menschen- und Tierseele. Leipzig 1911, 5te auf. (112). APPENDIX APPENDIX A TABLE I. Improvement as Shown by Tests on Other Functions Before and After ii Days' Training on e-s (Text p. 35) Reagent Gs Reagent Cr Speed in Sec. Accuracy Speed in Sec. Accuracy Before ^^^^.' No. of Words Before Train- ing After Train- ing No. of Words Training 1 ram- ing Marked B. A. Omitted B. A. Marked B. A. Omitted B. A. e— s 199.5 85.6 35 45 12 2 133-3 91.8 46 47 5 i— t s-p c— a e— r 390 111.5 198 223 108.6 124.8 134 56 9 22 53 40 15 19 53 3 I 9 9 6 7 136.9 89.4 1 16.6 166.5 120.2 66.2 85.6 135-6 37 15 25 52 42 10 27 58 4 2 8 8 I Totals 922.5 462.4 140 127 13 22 509.4 407.6 129 137 '4 11 a— n l-o, e— r 248 178 214 177.8 133-5 138 37 a! 39 6 40 9 1 7 10 4 6 172.8 135-8 212.8 142.2 88.6 134 31 16 62 35 9 42 15 7 14 I 4 Totals 640 449-3 113 85 17 20 521.4 364.8 109 86 22 19 Nouns e — s 160 175 I37.8 99.2 79 38 77 49 2 2 196.7 137-6 184.6 93 66 42 80 46 4 2 11 3 Functions on SameMt '1 922.5 Diff."640 462.4 449-3 140 "3 127 85 13 17 22 20 509.4 521.4 407.6 364.8 129 109 137 86 14 22 11 19 Total 1562.5 911.7 253 212 30 42 1030.8 772.4 238 223 36 30 26o APPENDIX A TABLE II Improvement Shown by Percentages of Tests after Training to Tests before Training (Text p. 35) Reagent Gs. Reagent Cr , « ^ , « ^ Speed Accuracy Speed Accuracy % of After-Test % Words Mkd % of After Test % Words Mkd A3 1.28 .69 I. II i-t" s-p c-a e-r .28 .85 .60 .86 i.ii .76 1.03 .88 .74 .81 •93 1.03 .93 M5 a-n' 1-0 e-r •71 75 .64 .99 .64 '97 .82 .65 •63 1.06 .90 1.02 Nouns e-s .86 •57 1.00 1.0S .94 .68 % Functions on Same Mat'l Diflf. Mat'l .50 .70 .93 .93 .80 .70 1.03 .98 Of Total .58 •93 .76 I. or ^ The Trained Function. ' Employed on Same Material as Training Mat'l, . . . Printed Columns. * Employed on Different Material, . . . Manuscript. iReagent Gs. TABLE III Quantitative Results (Text, p. 39) Total Dev. What % of Avg. Dev. in Whole Weight Lifted Grams Tests in Training Series (40-120 grams) (Avg. 80) Objects inside of field (49-113 grams) (Avg. 67.5) Objects outside of field (146-1870) (Avg. 552.7 g.) Before After 12.1% 8.2% 29.7% 22.1% 52.8% 29.8% Before After 9.7 7-2 20.1 14.9 291.9 165.6 %of Gain 25.8% 28.4% 43-3% oJ I I \ I \ \ 1 I ! e-s i-t s-p c-a e-r a-n l-o e-r Nouns e-s PLATE I. Reagent Gs, Efficiencies (in speed) in other I\mctions fcefora and after Training on the e-s Function. The filled portion shows ctoount of improvement. (From Table I.) 1^ rt"J^ e-3 i-t a-p c-a e-r a-n l-o e-r Nouns e-s PLATE HI* Reagent Gs. Gain shorn in per cent. Checked portion shows amount of improvement transferred. y , . e-e i-t 8-p c-a e-r ' a-n l-o e-r Nouns e-e FL&TB IT. Reagent Cr. (See Text, p. 35) i-t s-p c-a e-r a-n 1-0 e-r Nouns . e-s PLATE V. Gs, reagent. , Accuracy in percent of initial capacity. Checked portion shows amount of change. (From Table II.) 100 % \/// 7777, ZZZZ. ^ y^^^/A '///t 50l_ ZZZZc e-8 i-t s-p c-a e-r a-n l-o e-r Noune e»8 PLATE VI. Reagent Cr. 8 9 10 11 PLATE VIT. Practice- Curve from tests taken at beginning and end of each day's training of the e-s Function. (See Text, pp. 35, 38) APPENDIX A 263 Reagent Cr. TABLE IV (Text, p. 39) Total Dev. What % of Avg. Dev. in Whole Weight Hefted Grams Tests in Training Series (40-120 grams) (Avg. 80 g.) Before After 11.2% 09.0% Objects inside of field 26.1% 18.5% (49-113 grams) (Avg. 67.5 g.) Objects outside of field i5-3% 52-7% (146-1870 g.) (Avg. 552.7 g.) Before After 8.7 7.2 %of Gain 20% 17.6 12.5 29.2% 84.8 291.6 -243.9% TABLE V (Text, p. 39) Reagent Gs. Suggestion Blocks Wt. of Blocks Bi ;fore Training After Train ing ( > r ^ ISt 2d Total 99th looth Total 40 + 15 15 +5 +5 10 45 +15 15 + ro 10 50 —10 — 10 20 — 10 +5 15 55 +15 +ro 25 — 5 +5 ID 60 +15 15 5 — 10 15 65 —IS +15 30 +5 5 70 —5 5 —5 5 75 +5 5 80 +5 + 10 15 — 10 ID 85 — 20 — 20 40 —15 — 10 25 90 +15 — 5 20 +5 +15 20 95 +15 — 20 35 +20 20 100 +35 —15 50 + 10 10 105 —5 5 — 10 10 no +20 — 5 25 +10 — 20 30 115 — 5 —15 20 120 +5 — 10 15 — 10 —15 25 Totals 185 145 330 75 170 245 Average 165 122.5 25.8% Gain 264 APPENDIX A TABLE VI (Text, p. 39) Reagent Cr. Wt. of Be :fore Training After Training Blocks A ^ f \ ( N ISt 2d Total 99th looth Total 40 45 —5 5 —5 —5 10 50 — 5 5 +15 15 55 +5 +10 15 +5 —5 10 60 — 10 —15 25 +10 10 65 +15 + 10 25 + 10 +10 20 70 +5 —15 20 +5 5 75 +20 —25 45 — 10 +5 15 80 —30 30 —20 —5 25 85 —10 10 —15 +20 35 90 +20 +5 25 — ID +15 25 95 +20 20 + 15 —5 20 100 +10 10 — 10 — 10 20 105 + 10 +5 15 —5 —5 10 no — 20 — 10 30 —5 +5 JO 115 +5 5 —5 —5 10 120 — 20 —5 25 —5 5 Totals 17s 175 130 305 115 130 245 Average 2 152.5 122.5 20% Gain- TABLE VII (Text, p. 39) /-•_ Deviations on Objects Gs. Be fore Training Aftei r Training Inside Field Wt. 1st 2d Total 1st 2d Total I. Big Wt. 81 —31 -6 2,7 —26 —16 42 2. Bar soap 113 —13 +2 15 + 13 —28 61 3. Eraser 67 — 22 —17 39 — 2 —7 9 4. Pocket Bk. 81 — I — 21 22 — I — I 2 5. "Outlook" 105 —60 —60 120 —30 —50 80 6. Keys 52 — 2 —7 9 —7 +3 10 7. $3 silver 80 +10 +20 30 —5 5 8. Small Bottle (>i —18 —13 31 —18 —8 26 9. Elec. bulb 49 -H6 +46 52 +21 +26 47 10. Elec. switch 84 +21 +26 47 +6 +11 17 Totals 675 184 218 402 149 150 299 Average 201 149 28.4 Gain APPENDIX A 266 TABLE Vin (Text, p. 39) Outside Field II. Big Bottle 317 —97 —127 224 — 127 +8 135 12. "Outwest" 280 —105 — 100 205 —80 —30 no 13. Mucilage 330 —143 —133 276 +67 —33 100 14. Miinsterberg 714 —479 —364 843 —114 —214 328 15. Mem. App. Wt. 941 — 641 —441 1082 +159 +359 518 16. Ps Molding 146 —16 —16 32 —16 —6 22 17. Hammer 310 -165 — 120 285 —135 +65 200 18. Tin Box 462 —257 — 162 419 —62 +38 100 19. Key Mem. 154 —24 —9 33 +6 —4 ID 20. Psy. Rev. 1870 —1370 — 1070 2440 —920 —870 1790 Totals 5527 3297 2542 5839 1686 1627 3313 Average 2919.5 43 1656.5 .3% Gain TABLE IX (Text, p. 39) Deviations on Objects Cr. Wt. Be fore Training After Training Inside Field ist 2d Total 1st 2d Total I. Big Wt. 81 — I -6 7 — II — II 22 2. Bar Soap 113 —13 -38 51 —3 +2 5 3. Eraser 67 —17 — 22 39 —7 —7 14 4. Pocket Bk. 81 + 19 +29 48 — ^21 +29 50 5. "Outlook" 105 —5 —5 10 —15 —5 20 6. Keys 52 — 2 —7 9 +13 +8 21 7. $3 Silver 80 -|-20 +40 60 + 15 +20 35 8. Small Bot. 63 —7 +27 34 —18 —18 36 9. Elec. Bulb 49 —4 +41 45 + 1 —4 5 10. Elec. Switch 84 —24 +26 50 +26 +16 42 Totals 675 112 241 353 130 120 250 Average 176.5 25 125 1.2% Gain TABLE X ( Text, p. 39) Outside Field II. Big Bottle 2,-^7 —117 —117 234 —77 —77 154 12. "Out West" 280 —5 +120 125 +70 +20 90 13, Mucilege 333 +67 +17 84 +667 +567 1234 14. Munsterberg 714 —214 —114 328 +286 +86 372 15. Mem. App. Wt. 941 +59 +59 118 +59 +59 118 16. Pc Molding 146 +104 + 104 208 — 21 —16 37 17. Hammer 310 — 10 +90 100 +40 +90 130 18. Tin Box 462 -87 —62 149 +238 —162 400 19. Key Mem. 154 +4 +96 100 —14 —24 38 20. Psy. Rev. 1870 +130 — 120 250 +1130 +2130 3260 Totals 5527 797 • 899 1696 2602 3231 5833 Average 848 2916.5 243.9% Loss Grama 100_- 75. 50 _ 25_ ^__^ Weight of Objoct . .. Average deviation After Training —p. Average deviation Before Training Object 9 10 PLATE Vin. Reagent Ga, Weights of Objects Inside the Field, and average deviations of Judgnents. (From Table VII.) 1200, llOO. 1000, 90QJ 800, 700. 60Q_| 500. 40Q_4 300 200_ 100. — r~U rM— — 1 ^---4. ■.„-[- j Lii-t"~T — i 'ObJ9Ct 16 19 \'4 Pl,AT!: I7v. Rsa^aat U 15 Obiacts Outsicio '^f jiald, (See Text, p. 40) Oram* 100—, 75. 50. 25. Weight of Objects Average Deviation Before Training Average Deviation After Training T 10 PLATE X. Reagent Cr, Weights of Objects inside the Field, and average deviations of judgments. (From Table IX.) 120Q_ UOO- 100Q_ 90Q_ 80Q_ 70Q_ 6oa_ 500_ 40a_ 300_ 20CL_ 10Q_ U Objfco'c 1* 19 12 17 11 13 18 14 15 PLATE XI. Reagent Cr. Objects Outside of Field. (See Text, p. 40) 20 APPENDIX A ^ Degrees of White i6o° 16$" 170° 180° 185° 190° Totals TABLE XI Regular Reagents (Text, p. 44) Brightness-Discrimination. Before Training (Judgments on 90 variables) 12 10 10 5 I! 12 % R 66.7 % U 17.8 Aw. Na. Ya. 5 W U %R R W U %R R W U %R .80 .67 .67 •33 ■73 .80 II 10 4 3 8 10 3 5 9 10 6 5 ■73 .67 .27 .20 •54 •67 10 7 4 7 II 60 14 16 46 6 38 42.2 47 21 22 52.2 24.4 •54 .67 •47 .27 •47 .73 (21 Days' Training on Sound during an interval of 64 days) TABLE XII (Text, p. 44) Brightness-Discrimination. After Training (Judgments on 90 variables) Aw. Na. Ya. f •R W U %R r R w u %R f R W U %R ifio" IS 13 10 10 1. 00 14 I •93 12 3 8 6 8 .80 165° 170" 17^" I 2 2 I 3 3 .87 •67 .67 8 7 4 4 4 7 3 4 4 •54 •47 •27 5 4 I 2 5 6 •33 •27 .07 180" 185" 190° 5 II 6 2 4 2 .33 •73 7 12 I 2 7 I 47 .80 12 13 2 I 2 .80 .87 Totals 64 13 13 52 19 19 47 15 28 %R %U 71.1 144 57.8 2I.I 52.2 3I.I TffO APPENDIX A TABLE XIII Regular Reagent Cr. (Text, p. 44) Brightness-Discrimination Before Training Degrees (Judgments on 76 Variables) of White R W U % R no* 8 2 .80 120° 6 I 2 .67 130" 5 3 2 .50 135" 6 4 .60 140" 145" 2 2 6 .20 150" 4 5 •44 160" 6 3 .67 170" 7 2 .78 Totals 44 6 26 %R 57.5 %U 33-7 After Training (Judgments on 80 Variables) R W U %R 10 o o i.op 10 O O I.OO 8 I I .80 4 2 4 40 7 2 I .70 9 I o .90 10 O O I.OO 10 O O I.OO 68 16 85.0 7-5 TABLE XIV (Text, p. 43) Sound-Discrimination. Beginning Training (Judgments on 60 Variables) Intensity 23-57 25.06 26.48 27.83 29.12 30.35 32.68 Aw. Na. Ya. Cr. %R %U R W U%RR W U%RR W \3 %R R W U%R 5 2 .30 S 3 .20 5 4 -lo 1 3 .60 2 3 50 2 3 50 2 .50 4 .30 5 .50 6 .10 5 .30 I .50 1 .30 4 .10 2 .50 5 .40 3 60 I .70 3 -40 .50 .60 3 -40 3 40 2 .40 Totals 22 20 18 36.7 30.0 22 15 23 36.7 38.3 26 18 16 43.3 26.7 27 17 16 450 26.7 Intensity TABLE XV (Text. p. 43) . Sound-Discrimination Ending Training (Judgments on 60 Variables) Aw. Na. Ya. Cr. R W U %R R W 1 U %R R W U %R R W U %R 23-57 6 4 .60 5 2 3 ■50 5 4 I •50 6 4 .60 25.06 3 5 2 .30 4 2 4 .40 3 5 2 •30 6 4 .60 26.48 2 3 5 20 4 3 3 .40 3 2 5 .30 6 2 2 .60 27.83 29.12 8 2 .80 5 2 3 .50 5 5 .50 b 2 2 .60 30.35 6 2 2 .60 2 5 3 .20 4 2 4 .40 4 4 2 .40 32.68 6 4 .60 5 5 •50 4 4 2 .40 6 2 2 .60 Totals 31 20 9 25 19 16 24 16 20 34 18 8 %R 51.7 41.7 40.0 56.7 %U 15.0 26 .7 33.3 U-3 TABLE XVI Control Reagents. (Text, p. 44) Brightness-Discrimination. Before Interval (Judgments on 60 Variables) Rl. An. Wr. Degrees R W U %R 6 66.4 4 81.6 7 64.8 2.5 73.8 10 68.1 2 74-5 2 63.8 0.6 74-7 8 65.3 I 75.8 3 63.5 I 73.1 8 80.9 6 62.4 70.5 2.5 82.6 77.1 76.3 7 5 3 B. After Training CI. Al. Cr. Bs. Sec. Errors Sec. Errors Sec. Errors Sec. Errors 643 67.9 9 65.2 3 69.6 ID 64 I 64.2 6.5 643 5 68.3 8 61 0.5 62.6 5-5 69.2 6 72.8 8 63.5 1.5 65.6 II 68.2 5 69.4 8 62.9 1-5 64-3 10 66.2, 5 72.7 5 63 2.5 66.3 6 65.2 4.5 71.6 8 61.2 I 63.1 12 60.8 2 68.8 3-5 60.9 1-5 61.8 II 61.2 5-5 69.2 2 62.3 I 62.1 15 61.5 7 71.8 5 61 I 66.5 II 6i 3 65.7 0.5 61.6 1-5 61.5 8.5 64.2 6 69.3 7 60.9 62.3 10.5 62 5 69 2 62.6 2 63.2 17 62.2 4 66.4 8 57-9 60.5 11.5 58.8 65.5 3 61.9 1-5 60.8 13-5 62.5 5 59-2 3 60.6 1.5 58.9 10 61 7 69.S II 60.5 I 60.9 15-5 63.5 8 70.7 10 59.2 0.5 61.6 13 61.9 9-5 276 APPENDIX A TABLE XXII (Text, pp. 53, 55, 56) Reactions on Typewriter (Avg. per lOO per Day) A. Before Training CI. Al. Cr. Bs. \ f — "^ \ t ^ / Sec. Errors Sec. Errors Sec. Errors Sec. Errors 71 2 94 73 3 99.5 5 76 0.9 93 4 76.0 3 87.3 3 73-5 I.I 80.0 3 71.2 3-5 84.0 4.5 67.4 I 73.1 6.3 69.1 3 80.1 5.3 69 I.O 79.3 4.5 69.1 5 77-9 4.3 63.3 0.8 72.7 7 67 B. After Training 4 63.3 I 65.2 8 66.4 4.5 70.7 8 61.3 I 62.9 II 62.1 4-5 69 3.3 60.4 I 61.0 13 61.7 5.4 66.3 6 TABLE XXIII (Text, pp. 53, 56, 57) Reactions on Typewriter (per 100) (Control Reagents) Before Interval Mn. Ge. Gs. / ^ 1 \ / \ 9I.I 2 147-3 7 1 07. 1 95.7 3.6 125.7 4 79.1 9I.I 6.5 154-2 3.5 79-9 1.5 93.7 6 157.9 3.5 84-8 2.5 86.1 5 131.8 2.5 85.8 1.5 86.7 8 133.6 2 854 I 76.7 3.5 106.2 3 ys.s 3 1 16.5 2 76& 8 119 2 70.9 8 127.2 0.5 7Z 7 111.7 71.7 14 118.1 I 78.1 5 82.2 I 739 6 90.8 1-5 76.8 8 95.5 2 73,7 6 97.4 I 74 5 106.6 I 67.4 8 100.6 99.7 After I 2 Interval 73 4 91.4 1.5 86 73.4 3 84.2 I 76.4 2 704 7 82.2 83.7 68.4 2 104.9 3 77.1 0.5 68.7 6 91.7 3 82.5 3 70 5-5 88.7 77.4 I 70.7 3 83.4 0.5 69.6 3 81.7 66.7 5 91.5 2 64.6 II 95.1 3.5 63.3 6.5 82.5 I 62.3 8 82.9 I Day 3/3 I 3/5 2 3/8 3/10 3/12 5 3/15 6 APPENDIX A 277 TABLE XXIII (Con.) Second Group, Control Reagents Before Interval Bd. Bh. Bs.2 Cf. [02 7 / ^ 100 3 100 7 128.4 I 93-4 14+ 84.2 6+ 87.6 5 112 I 92.6 13+ 744 8+ 82.8 I 118 I «7.4 7+ 75 5+ 75.8 4+ 100.8 86.6 5+ 78 5 72 2+ 102.6 86 5+ 76.8 7 70S, 3+ 96 80 7 76.4 7 7Z 2 93-4 I 84.2 5+ 79 72.8 3+ 92.4 I 81 7 72.4 I 73 3+ 88 2+ 774 4 80 4+ 68.4 2 80.8 73-2 3+ 76.2 2 66.4 3 87.2 79 3+ 76.2 3+ 68.6 4 87 I 76 4 734 4 68.6 88.6 I 71.2 3 76.6 4 62.6 I 81.8 67.1 3 74.2 2 65.6 I 78.2 72.6 4 72.4 6 67 5 88.4 I 71.6 5+ 72.4 II 62.8 3 80.8 2+ 72.8 4 75 5+ 61.2 3 80.4 68.8 3 74 6 62.4 4 86.2 I 69.4 2 71.8 7+ 64.8 3 80.8 I 66 67.4 8 62.6 I 77.8 68.6 3 66.4 8 63 6 76.8 67 7 64.8 13+ 67.6 4+ 75 66.6 2+ 654 7 62.6 3 74.8 67.6 3+ 66.2 10 60.4 74 65.6 I 70 7 61.6 4+ 73.6 I 63.2 1+ 66.6 6 694 I 68 66.2 3+ 63 7 60 2 70.4 I 64.6 5 62.2 5 62.4 3 74 65.4 4 64.6 II 60.4 6 714 2 65.4 2 60.6 5 62.2 4 75.2 0+ .(4/f^r Interval 64.6 2+ 70.4 2 63 2 75.2 2 62.6 4 62.6 6 58.6 3 65.2 I 65 2+ 62.4 5 60.6 4 694 63.8 2 58.ot 4 60.4 6 71.6 1+ 66 4 61.4 5+* 59.6t 2 65.2 I 65 6 6i.4t 8+* S7.6t 4+ 62.2 I 62.8 4 64.4 4 58 7 67.6 2 634 4 60 4 584 4 64.4 2 60 5 61.2 7 56.6 6 68 61 5+ 61.6 4+ 54 4 67 61 3-t- 59-6 6 54 3 70.4 I 59-2 7+ 57.8 5 55 5 66.2 58.6 7 59-6 3 54.8 4+* 65.8 2 56.8 3 56.6 6+ 57.6 3 57 61 7+ 59 8+ 57.6 5 64,2 60 8 60 5 58.2 3 65.0 2+ 58.6 7 61.6 7 59.2 4 68.6 * Over 10 reactions one place late,— a distinct process of reaction which greatly decreases the time. t Memory of the beginning of the series. 27? IPPENDIX A TABLE XXIV (Text, p. 56) Reactions on Typewriter (Avg. per 100 per Day) (Control Reagents) Before Interval Mm. Ge. Gs. 1 ^ ( \ 90.7 74-1 74 5 7-5 6.3 141 .8 4 1 16.5 1.3 96.1 1.5 After Interval 70.7 66.2 4.5 7 90.S 1.3 86.2 1.3 80.5 TABLE XXIV (Con.) Second Group Control Reagents Before Interval Bd. Bh. Bs.2 Cf. 102 7 100 3 100 ^ r 7 128.4 I 87.7 8.5 77-5 6.3 77.0 2.8 103.8 0.5 78.5 4.3 76.2 2.3 69.6 2.5 87.3 0.8 70.7 3.7 74.1 5.7 63.6 2.8 82.6 0.7 67.5 2.8 67.0 7.2. 63.5 2.8 76.5 0.2 6s.i 2.7 64.5 6.8 62.7 z-:i 72.1 0.7 After Interval 64.5 3.2 62.7 4.8 60.0 3-5 68.8 I.O 61.2 A-7 60.8 S.o 56.0 4.8 67.3 0.8 59.0 6.4 59-4 5.8 57.5 3.8 64.1 0.8 g ps i w w t^ >. >. E^ t> — ! o o o o o o o o 00 t- *o y u> o> I Pt ■*> M « O c o s I > o o o o o o o o c> o a* 00 fr- «o to o o> 00 t- «o 282 APPENDIX A TABLE XXV (Text, pp. S3, 54, 56) Card-Sorting Reaction (per loo) CI. Al. Cr. Bs. 1 1 \ (- \ <■ \ Sec. Errors Sec. Errors Sec. Errors Sec. ] Errors 113 136.6 2 130.2 1JJ.8 I00.6 134 5 122.4 ^ 127.6 I I1I.2 119.2 108.6 3 1 1 7.8 4 I034 1 16.4 2 112 3 129.4 2 96.8 123 I 112.2 2 124.2 9Z-2 I 109 2 103.8 2 111.8 3 99.8 I J05.6 3 96.2 ^/5 101.8 1034 2 97-2 I 107-5 3 92 105.4 2 109.5 4 101.6 4 100.6 6 104.5 2 89.3 89.4 97.6 3 98.4 2 101.4 I 88.6 91.6 I 102.6 7 87.4 5 88 91.4 I 99 2 100 2 90.4 96.2 8 97 I 99-2 2 91.6 92.2 3 95-4 2 100.8 2 87.9 94-8 4 88.2 96 3 93 91.8 6 99-2 I 90.6 91.8 3 92.4 3 98.4 2 904 91.6 3 90.6 4 95.5 I go I 88 2 88.8 2 100.5 4 91.8 89.8 5 93-4 3 96.5 6 93.4 88 5 96 2 97-5 3 92.2 90.2 3 96.2 5 109 7 85.4 91.4 2 89.8 4 103.S 6 94.5 87.2 4 95.5 5 97 5 90.4 91.4 4 92.4 2 95-5 2 «4.5 91.8 4 94.8 5 103 3 81.5 96 4 95-2 2 102 5 81.S 98 9 92.4 I 9^-5 I 85 2 100.4 7 89.2 2 89 89 96.6 4 91-5 I 105 8 8S 91.8 3 88 3 95-5 I 89.2 I 96 4 92 5 90.5 3 86,4 94.8 3 91-5 3 89.5 2 ^5-5 88.4 6 Sp.^ I 89.5 2 82 90.4 4 86.2 6 88 I 79.5 89.8 7 97 4 92 5 79 I 94.2 5 97-5 4 90.5 I 81 93 2 88 83.S 96 5 87 I 82 93.5 6 81 88.5 91-5 95.5 93-5 94 92.5 96 92 92 91 6 2 14 2 7 4 4 5 8 4 3 APPENDIX A TABLE XXVI (Text, p. 53) Card-Sorting Reaction (Avg. per loo per Day) 283 a. Al. Cr. Bs. Sec, Errors Sec. Errors Sec. Errors Sec. Errors 106.8 135.3 3.5 126.3 I 127.2 1.8 107.3 1 17.8 I 110.3 3 118 1.5 97.9 0.5 116 i.S 102.4 1-3' 109. 1 1.5 90.6 102.9 3 102 4-3 97 2.5 89.1 94.2 Z-i 94-9 1.3 98.6 2 90.8 93.S 3-5 90.9 3.8 98.2 3-5 90.2 0.5 90.5 2 93.9 3-5 106.2 6.5 90.7 88.9 5 94.6 3.5 99.4 3.8 92.6 90 3-3 90.8 1.5 95-5 2.3 83.1 0.5 96.5 6 90.8 3 89.4 2 87 94.2 3.5 87.7 3.5 89.4 1.8 87.3 0.5 95.4 3.5 95-9 3.8 81.5 0.3 90.7 5.5 92.8 6.2 8r.9 931 91-5 5.3 3-5 TABLE XXVII (Text, pp. 57, 58) Comparison of Gains Between Training Periods Typewriter-Reaction Training Reagents (4th and 5th Days, in Practice.) Control Reagents (4th and 5th Days after Interval.) Avg. of 2nd and 3rd Days Avg. of 4th and 5th Days a. 70.5 66.1 Al. Cr. 76.6 70.0 76.0 67.9 Bs. 86.0 79 Mn. Ge. 74 105.5 68.4 88.4 Gain % Gain Avg. last two Days before Card Practice First two after 4-4 6 66.1 62.3 0.6 2.1 0.8 3 76.0 67.9 64.0 64.3 7.0 8 79 69.9 5.6 17.1 7 16 Gain % Gain 3.8 6 12.0 3.6 16 5 9.1 12 f I \ I / \ I ( \ . ) y I / ) ( / i) S t r ) <7 & / a m o \ / \ / 2 OO o ooo o o ®cv» O OO'^C* O CO 9 ill o o o o o o <0 CM «H o a> 00 o o o o r« O Ok OO iW rm o o o o r-l O 0» 00 r-f iH II I I I I I § o o CO ca o o o o p-« o a» 00 APPENDIX A TABLE XXVIII (See Text, footnote pp. 58-9) Time per pack of 50 cards, sorted by suit into four compartments, showing the influence of practice in the typewriter-reaction. Total Total Diff. Before Interval Regular Reagents Control Reagents Bd. 55 57-5 I 55 50 Bh, 55.4 51 52.6 48 I I I Bs.2 52.4 54-6 SO 52.6 I Cf. 62.4 I 70 67 67 Cn. 57.4 52.4 53.6 I 48 I Sn. 48.8 50.6 49 50.6 — » I I 217.5 I 207.0 3 209.6 I 266.4 I 21 1.4 2 199.0 2 After Interval 56.6 I 54-2 I 52 52 4 SI SO 49.8 46.8 I 2 3 3 52.8 5 53.4 5 52.2 3 50.6 3 61.4 56.4 57 57 60 I 53.8 I 52.6 s S1.4 49.8 S0.6 52.4 214.8 6 197.6 9 209.0 16 231.8 221.4 9 204.2 — <2.7 — 1.2 —9.4 —4.5 -0.6 —0.3 —34.6 —13.0 +10 +4.7 +5.2 +2.6 APPENDIX B (Presenting data relevant to the Experiment on Attention, pp. 76-17^-) e Q e Q Fig. 2. (See p. 76) Arrangement of com- partments in cabinet for card-sorting. Fig. I Symbol on cards Cardsorting Instructions to reagents 1. This is a test in "reaction with discrimination and choice." On each of the packs of fifty cards the time and errors will be recorded. Speed, therefore, should be aimed at, yet the sorting should be accurate. Time will be saved if you get a dependable mental scheme of the compartments rather than directly matching the cards, which, although necessary at first, makes sorting wait upon the eye. — 2. In the final introspections of the day note : (a) Any special hindrances or helps to the sorting, (b) Your mental scheme, if any, (c) Any development or change in your scheme, (d) What demanded attention most? Any tendency to name or pronounce ? (e) Whether the sorting is fatiguing, (i) Any bodily strain, (2) Any mental strain, (f) Whether sorting is agreeable or otherwise, (g) Any change in these respects from previous introspections. (Ap- plicable only after the first day.) 3. In introspections noted between packs mention briefly a few of the more important points about the process of sorting that occur to you. 4. Look over the cards on the cabinet so you will be able to distinguish the symbols readily. 5. The procedure will be : (a) Arrange first pack conveniently in hand, and take position com- fortably before the cabinet. (b) Throw on the table the blank, at announcement of "Go!" (c) Sort cards, aiming for speed and accurate work. As last card leaves hand announce "Now !" (d) Note brief introspections. (e) Like procedure with remaining three packs, taking them in order. (f) After the last pack of the day, note your introspections in fuller form. Fig. 3 (See p. 76) APPENDIX B 289 aatanetntaeneaatanetntaeneeaeneatntenatta tanetnaene Fig 4. Sample series of stimuli used in the typewriter-reaction (see pp. 52, 76). Poets Philosophers Statesmen Scientists Musicians Homer Plato Pitt Faraday Beethoven Virgil Socrates Gladstone Darwin Wagner Shakespeare Aristotle Washington Huxley Mozart Milton Kant Jefferson Watt Hayden Browning Locke Lincoln Tyndall Bach Tennyson Hume Webster Agassiz Mendelssohn Goethe Hegel Roosevelt Galileo Handel Whittier Pythagoras Napoleon Helmholz Verdi Poe Spinoza Bismark Newton Paderweski Longfellow Descartes Burke Ohm Paganini Fig. 5. The names of men in the classes used for Controlled Reaction. (See p. 77) Series I Series 2 Series 3 Series 4 Series 5 Homer Darwin Socrates Aristotle Browning Pitt Virgil Shakespeare Hayden Lincoln Beethoven Wagner Huxley Kant Bach Faraday Gladstone Milton Watt Tyndall Plato Mozart Washington Jefferson Locke Series 6 Series 7 Series 8 Series Q Series 10 Hume Roosevelt Napoleon Verdi Descartes Agassiz Galileo Goethe Bismark Ohm Mendelssohn Hegel Helmholz Paderweski Paganini Webster Handel Whittier Poe Burke Tennyson Pythagoras Spinoza Newton Longfellow Fig. 6. The series of names as presented. (See p. 77) (a) Intervals (b)S )eries No. 1 \ Scale Intensity*^ I, 2 3, 4 5 ,6 7, 8 9, ID I. 20.16° .5 I 2 3 4 5 2. 23.57 -7 7 8 9 I 2 3. 25.06 .8 S 6 7 8 9 4. 26.48 .9 2 3 4 5 6 5. 27.83 i.o — No rm 9 I 2 3 4 6. 29.12 I.I 4 5 6 7 8 7. 30.35 1-2 8 9 I 2 3 8. 32.68 1.4 3 4 5 6 7 9. 34.85 1.6 6 7 8 9 I * From Fechner : Psychophysik. S. 181. Fig. 7. Intervals of intensity, and series as presented in Sound Discrimination. In the Series, No. 2 is the reverse of No. i. (See pp. 77, 78, 188.) 290 APPENDIX B Fig. 8. Greater, Less, Like, Doubtful, > < III ? Symbols used in recording (See JU( dgments 78) in Sound Discrimination. I 2 Series 3 4 5 3 4 1 3 4 I 2 4 3 2 2 3 4 2 I 4 3 I 4 3 2 4 I 3 2 4 2 3 I 2 2 I 3 2 4 2 3 I 4 2 I 4 2 3 2 I 3 4 2 I Intensities* 1 = 10° = 0.12 2 = 25° = o.Bo 3 = 40° = 2.10 4 = 60° = 5.00 *From Fechner's Psycho' physik, I :i8i). Fig. 9. Series of sounds in the test on Memory of Sounds. (See p. 78) Series of letters and figures 1 / \ f \ < \ 1 \ (a) (b) (a) (b) (a) (b) (a) (b) (a) (b> K 4 C 5 P 7 G 6 L 8 S 7 V 8 H I L 9 G 2 B 5 J 6 Z 8 C 7 M 9 M 2 B 3 K 5 H 4 P 6 F 6 S 7 B 9 P 8 S I P I H 2 S 4 F 3 B 5 H 3 W 4 P 6 M 5 K 7 C 8 T 9 C 2 B I C 3 L 2 D 3 G 5 S 4 H 6 G 9 K I N 3 K 2 P 4 Fig. 10. The series of (a) consonants used in the test on Memory of Con- sonants, and the series of (b) Digits used in the test on Memory of Numerals; also (c) the pairs presented in the test on Memory of asso- ciated Pairs. (See p. 79.) APPENDIX B 291 (Y o A Fig. II. The series of symbols used in the test on Memory of Visual Signs. (See p. 79.) M — G C RN Fig, 12. Reproduction of size and style of letter and spacing used in tests on Learning 12-Conson- ant-Rectangles. (See pp. 80, 82.) W- S- Q- N- S- B- W- - R L M ■ -N W- R T W R ■ -S - -D -N -T -N -L -R -R -N -M Fig. 13. The letters used in the test on Word-CompJetion. (P. 80.) Reagents Practice Curves. BtteneiTS Threshold of Visual Attention. Bcpoeure of 12-letter« rectangles l/lO See. Curves show average number of Points per day} three points is the value of a correctly placed reproduced letter. The curves do not indicate that the training of 19 days had brought the capacity to its nwatlnial efficiency. Says r 5 6 1 8 9 10 U 12 13 Fig. 14. (See text, pp. 83, 93) "R — 15 16 17 »- 2^2 APPENDIX B 15 10 Beagent* SI. Practice Currea, Learning 12-letter-rectanglee. Bipoavre 10 See. Curve* •how average maaber of points per experiment per day^ three points is the value of a reprodueed letter correctly placed. Maximal efficiency had not yet been attained. Days 1 2 3 4 S 6 7 8 910UU13 14 15 16 1fflS Fig. is. (See text, p. 94) Saster Vacation Ssys "9 10 n 12 13 14 15 16 17 IB Fig. 16 Analysis of Rt.'s Practice Curve in Learning 12-Ietter-rectangles, accord- APPENDIX B 293 ing to complexity of process, showing change in complexity during practice. The curves show the number of letters reproduced from rectangles according to the imagery from which they were recorded; e.g., on the first day 80 letters were recorded from rectangles upon which the Kinaesthetic (verbal) imagery alone was used, 40 letters from rectangles upon which Kin. and Visual imagery were coordinated, 17 letters from rectangles for which the three- fold coordination of Kin. + Associations + Vis. imagery was used. (See text, pp. 95, 176.) At the beginning most of the work was done in Kin. A. imagery which as a single-fold method disappeared on the 6th day. The two-fold method was dominant on the 4th, 5th, and 6th days. From the 7th day the three- fold and four-fold coordinations predominated. The course of practice is toward greater complexity of the process. Easter vacation caused a lapse to the status of learning of the 6th day, after which the same development in complexity of coordination takes place as that on the 7th and 8th days. ut- ter* ISO 140 130 120 110 100 90. 80. 70 Reagent Rt« Kinaeethetle _ . . Visual _ Aeaoelatlona "i 3 4 5 6 5 5 § 10 11 i5 13 14 iS Ti i? io Fig. 17 Analysis of Rt.'s Practice Curve in Learning 12-letter-rectangles, according to the number of letters reproduced from each kind of imagery during the day, disregarding the complexity of method in coordinations,^ e.g., on the first day 125 letters were reproduced from Kinaesthetic (Verbal) imagery, 18 from Visual imagery, and 8 from Associations. (See text, p. 174.) Associations for Rt. were usually visual images of words, as "Chemically Pure" for holding the letters C P. It is evident that the principal rise in the practice curve is caused by the growing frequency of Visual and Association letters, i.e., by the growth in the process of the auxiliary forms of imagery. The letters reproduced on the I2th day are similar to those reproduced on the 3d day in that they are alphabetic symbols; in the mind of the reagent, however, they differ greatly in kind of letter — they are more dominantly visual and visually word-bound symbols, less rote-letters in the muscles of the tongue. 294 APPENDIX B ReagaBt Ul — First Third. — Sseond *rhira. — Third Third. Fig. i8 Curves show distribution of Reaction to Sound in the first, second and third thirds of Ly.'s training (iioo reactions) ; only the reactions with the Morse key at lOOngram tension being chosen (270 reactions). Reactions from- 91 to 100 sigma are plotted on 100. (See text, p. lOi). Practice-effect involved change in processes. APPENDIX C Scoring the 12-letter-rectangle If a single score is to measure the power of reproduction, it must give value to a reproduced letter when it is misplaced in the lecord ; and it would seem that this value should be less than the value given to a correctly placed letter, and should vary with the degree of spatial reproduction as indicated by the amount of misplacement. Any system of values, nevertheless, is arbi- trary, since it will favor some methods of learning and reproduction more than others. Let the following scheme represent the relative positions on a 12-letter- rectangle : j c b a a = Correct position, and may be in any space on the card. b = Adjacent space on the line, k g f d c = Second remove on the line. d = Adjacent space in the column. e = Second remove in the column. 1 i h e f = Space adjacent at corner. Etc. If the reproduced letter belongs in position a, in the above scheme, (and that may be in any space on the card), we may compare several ways of scoring noticed in the text (p. 80) : Points Text (i)' Winch= Smith* Text (2)* 3 = a a a a 2 = b, c, j b, d b, c b, d, f I = d, e c, e d, e, f, g, h, i, j, k, 1 c, e, g, h, i, j, k, 1 o = f, g, h, i, k, 1 f, g, h, i, j, k, 1 A rote method of learning by the line would be favored by Text (i) and Smith, for its misplacements would be more frequently made within the line ; by the column, by Winch, for a similar reason. A visual reproduction would be favored by Text (2), since / is a proximate location; and vague repro- duction by Smith and Text (2), since they oflfer no zero value for repro- duced letters. In partial justification of the zero value it may be urged that over half of the consonants of the alphabet are on the card, and that mere guessing would raise the score of reproduction in over half of the guesses, especially when some of the consonants are known and the chance for correct guesses is consequently increased. If the letters are really reproduced, however, they should contribute to the score. Perhaps the assignment of values should not be made until all the mis- placements of a given reagent are tabulated and the various kinds of mis- placement are compared with degrees of reproduction as revealed by an analysis of his mental processes; then the values could be assigned with reference to the degree of reproduction, which might be found to decrease with the frequency of the kinds of misplacement. ^P. 80. =• Winch: Br. Jr. Psychol. 1:129; also Whipple's Manual (ist ed.) p. 369. 'Smith: Mind. N. S. 4:52. The writer's interpretation of the rather indefinite text is given here. *P. 187. 296 APPENDIX C Such a tabulation was made of the results from 20 experiments by each of ten reagents in Tests 13 and 17, and it was found that although there was considerable variation between the different reagents in the number of letters misplaced in 20 experiments (12-44, with an average of 15% of all reproduced letters in Test 13; 7-38, with an average of 29% of all reproduced letters in Test 17), within each test large individual variation in the kind of misplacing involved a few kinds only (in Test 13, misplacings c and d; in Test 17, b, d, and /). If frequency as found in the aggregate results would not call for a division of these particular kinds of misplacing into different classes for evaluation, a system of evaluations upon the basis of aggregate results might be available for general application. This condition is fulfilled, as may be seen from the following tables, but it necessitates a separate system of evaluations for each of the two tests. All of the misplacings of all of the reagents in each test were tabulated and calculated with the following aggregate results (reduced to the average number of letters so placed in one experiment) : r lesi U ■> ' - ■ 1 c; >L 1/ \ J c b a J c b a .01 •215 .42 6.97 .015 .065 .285 2.245 k g f d k g f d .015 .085 .11 .26 .020 .080 .160 .200 1 i h e 1 i h e .01 .025 •03 •03s .000 •045 .015 .045 But these amounts show the actual distribution of misplacings ; if we assume that a misplacement which occurs more frequently should be penalized less in the score measuring reproduction than one which occurs less fre- quently (introspections are not full enough to indicate clearly v/hether it should), our system of arbitrary values could not be taken from the distribu- tions shown above for the reason that the chance of making some misplacings is much greater than that of making others ; e.g., in chance guessing the misplacement / would occur 24 times to / 4 times. We should need this distribution as modified by the relative chances of the misplacings. The Ratio of Recorded to Chance Misplacings would seem to meet this need. If position a is moved systematically over the card, occupying in turn every one of the 12 spaces, there would aggregate 132 possible misplacings distributed over the lettered relative positions as follows : 24, f ; 18, b ; 16, d, g; 12, c, h ; 8, e, i, k; 6, j ; 4, I. According to these chance values of the positions the 1.27 misplacings (per experiment) of Test 13, and the 0.93 misplacings per experiment of Test 17, could be distributed ; position d, in Test 13 would then have the value of .153 instead of .26, the ratio of its actual occurrence to chance occurrence being 1.70. These ratios calculated for all the relative positions are presented in the following table: APPENDIX C 297 Ratio of Recorded to Chance Misplacing. j C ^6 b a J 0.17 1.87 2.44 •35 k S f d k .20 .56 .48 1.70 •35 1 i h e 1 .27 •33 .26 .46 0.00 les c b ■77 2.24 g f •71 •95 i h .80 .18 d 1.77 e .80 But it might be objected that we do not still have a measure of the relative frequency of the respective misplacings, since owing to the reagent's direction of his attention to one portion of the card rather than to another, the letters which in reproduction were misplaced may not have been distributed evenly over the 12 positions. Calculation sustains the objection so far as it suspects irregular distribution over the positions : The range in Test 13 is 13 (in the 1st position) to ss (in the 7th position) ; in Test 17, 7 (positions 9 and 11) to 41 (position 4). When the actual chances for the respective misplac- ings are calculated, however, the results showing relative frequency of mis- placings are substantially the same as in the tables above : Ratio of Recorded to "Actual" chance misplacing. t Test 13 ^ / Test 17 > jcbaj cb a .20 1.87 2.33 .30 .77 2.39 k g •22 .53 f d k g f d •44 1.67 ■33 .69 •96 1.71 h e 1 i h e ■27 •49 .0 .86 .21 .86 1 i •32 .35 And if instead of pooling the results of 20 experiments by each of a dozen reagents we use the 450 experiments (like Test 17) by a single reagent (Mn.), in which the habits of attention may be expected to be more uniform and consequently more disturbing to our calculation on the basis of "regular" chance, the range in irregular distribution being from 4 (position 10) to 47 (position 4), and the number of reproduced letters misplaced being 242 (5%), we get the following remarkable approximation to the ratios with "regular" chance distribution : Ratio of Recorded to /—"Regular" Chance Misplacing--. ^'Actual" Chance Misplacing— ^ jcbaj c b a .28 .41 3.22 .26 .41 3.32 k g f d k g f d .55 .62 .92 1.18 •52 .61 .96 1.19 1 i h e 1 i h e .27 .34 .27 .41 .26 .34 .28 41 298 APPENDIX C There is little doubt therefore that these tables furnish us with fairly reliable relative frequencies of the respective misplacings in these two tests. Upon the assumptions that the more frequent misplacing should be penalized less in the score, that three points shall be the value of a correctly placed letter, that in the interest of facility fractional values shall be avoided, and that great injustice is not likely to result in striking an approximate balance between over-evaluation and under-evaluation of respective misplacings, it would seem that the following values might be recommended : Test 13. (Exposure 10 sec.) Test 17. (Exposure i/io sec.) 3=a 3=a 2 = b, c, d, 2 = b, f , d, I = all others. i = all others. The arbitrariness of forcing this qualitative difference in reproduction into a quantitative difference, has already been remarked. The g or i misplacings are only by courtesy to the strenuous statistician to be regarded as possessing one-half the reproductive value of misplacings b or d, and one-third the reproductive value of a correctly reproduced letter. Disregarded also is the partiality to visual memory: letters retained in vivid visual imagery keep their relative positions ; letters reproduced from equally vivid auditory or kinaesthetic imagery may be free from either temporal or spatial order. The values listed under Test 17 are those employed in our Text (2) method of scoring. It will be remembered that exposure of the card was about a tenth of a second; it is for this reason that /^ is a proximate position. Upon re-scoring Tests 13, and 17, the average scores were found to range in points from 15.5 to 27.3, and from 4.1 to 9.1, respectively, and to increase only 0.36, and 0.61 points, respectively. Our Test (i) method was unsatisfactory, therefore, principally because it is an arbitrary method, and, like many mental test methods of measurement, confuses process and product. Its difference from other methods of scoring, equally arbitrary, has no bear- ing on our treatment of results, which, the reader has noted, is principally qualitative. APPENDIX D Influence of Subliminal Dififerences upon Judgment in Stimulus Comparison (Discrimination) (See Text, p. 185) In experiments on Sensible discrimination in which the Method of Constant Changes, Right and Wrong Cases, is employed, it is commonly observed that, if the steps between stimulus-differences are small enough, iR Cases (correct judgments on "greater" or "less") fall off regularly with the decrease in stimulus-difference even when these steps fall below the difference which is conventionally designated as the "Least Noticeable Difference." (50% R Cases. vid. Ktilpe : Outlines of Psychology, p. 69; or Titchener : A Text-book of Psychology, p. 213). This influence of "subliminal" differences upon judg- ment in stimulus comparison may be illustrated from our own data, if we aggregate the judgments of all reagents upon each of the intervals (stimulus- differences) in the respective tests and practices in Sensible Discrimination. If we call the interval between the variable of least intensity and the norm the first, and the interval between the variable of greatest intensity and the norm the last (7th or 9th), we get the following table of R Cases, in per cent: Interval abed I 77 92 85 78 2 67 65 57 70 3 49 50 31 59 4 39 31 23 38 5 = Norm 6 59 49 41 41 7 79 55 53 • 62 8 76 74 65 9 84 84 75 (180 judgments on each interval, 14 (160 judgments on a ^ Discrimination of Brightness reagents) 1904-5. (Text, pp. 42ff.) b = Discrimination of Sound with Sound Pendulum each interval, 16 reagents) 1910-1911. (Text, pp. I34ff.) c = Discrimination of Sound with Sound Pendulum (120 judgments on each interval, 12 reagents) 1911-1912. (Text, pp. 2o6ff). d = Discrimination of Sound with Fall Phonometer (96 judgments on each interval ( 2 reagents) 1911-1912. (Text, i88ff). The facts of the table are more apparent in the curves drawn from it : Intervals 1 23 45 678 « lOOf. Nora The "Least Noticeable Difference" is 50% R Cases. INDEX (For the Name Register consult the Bibliography). Adding Test, 5. Anomalies in experimental results, 65 ff ; causes of, 68. Anomalous practice-effects, in tests on Learning 12-letter-rectangles, 153-4; on Extensive Threshold of Visual Attention, 163, 173; on Recognition or Choice of i of 2 Letters, 201 ; on Reproduc- tion and Recognition of Letters on 12-consonant-rectangles, 205; on Memory for Visual Symbols, Immediate Memory, 215. Association, with areas of rectangles, 7, 8 ; with weights, 8. Associations, representative vs. auxiliary, in memory experiments 24; a 'common factor,' 26; occurrence of, in training on Exten- sive Threshold of Visual Attention, 85, 92 ; in training on Learn- ing 12-letter-rectangles, 95 ff, 98ff ; in tests on Memory of Con- sonants, 143; Digits, 146; Visual signs, 148; paired associates, 149; Learning 12-letter-rectangles, 151 ff; learning with dis- traction, 155-6. Attention, a 'common factor,' 19, 23 ; Experiment upon, 70 ; measure- ment of, 70 f; tests of, 72, 106 ff; training of, 73, 82 ff; our tests inadequate for measurement of, 183. Blickensderfer typewriter, used in training on typewriter-reaction, 51, in tests, 76, yy. Card-sorting, reported from the literature, 10, 11, 12, 14; in ex- perimentation: {Training) : 50-1, changes in processes in, 62; (Tests) : procedure in, 75-6; cabinet for, 75; test results,_ii8 ff ; variability in processes in, 119 ff; change in processes during test in, 122-3. Cancellation test, reported from the literature, 9; marking out a's and o's, procedure, 75; test results, ii4ff; variability in proces- ses in, ii4ff; see Marking out words, 34. Changes in processes, 68; see also. Tests; as, Reaction to Sound, Cancellation, etc., see also. Practice. Chinese Legation, training and efficiency of, 3. Cicero, 23. College Grades and professional success, 3. Color perception, 6, 7. 'Common Factor,' The, as found by Ebert and Meumann, 19; by Fracker, 22 ; by Sleight, 28 ; in Marking out words, 36 ff ; analy- sis of, 224 ff; formerly thought simple, now complex, 224; quest for, through analysis of processes, 226 ff, and of the effect of practice upon these processes, 227 ff ; formal factors found, 230; is a formal factor in 'Spread of Training,' is material of experience in 'Transference,' 230; relative promi- 302 INDEX nence of these two kinds, in the literature, 230-1 ; both selected from experience, 231 ; provisional classification of possible, 232; summary, 239. Complete learning, in memory experiments, 17, 212. Complexity of process, in discrimination of sound, 47-8 ; in estimat- ing weights, 41 ; see also Tests, Training, and Analysis of Common Factor. Consciousness, Eight-level clearness in content of, in Extensive Threshold of Visual Attention, 88 fif. Control experiments, in Sensible Discrimination, 44 ; in Reaction with Discrimination and Choice, 52 ; in Experiment on Attention, 73 ; on Reproduction, 186. Controlled Reaction, (Test) : Apparatus and procedure, 77; test re- sults, 131 ; variation in processes, 131 ; change in processes dur- ing test, 133. Critique of Test-Training-Test type of experiment, 219; necessity for analysis of processes, 219; illustration by hypothetical ex- periment, 220 flf ; merely quantitative treatment of results worth- less, 221 ; consequent meaninglessness of Probable Error as in- dex of reliability, 221 ; tests should include sufficient number of experiments, 223 ; should be limited in number, 223 ; can be made equivalent by use of Miiller's 'cyclical changes,' 223 ; training should be more rigorous, 223 ; control reagents should be equal to the regular reagents in number and ability, 223-4 ; mathemat- ical treatment of results may be simple, 224 ; summary, 239. Cross-education, 3 f. Danish language discriminates between the general and special efifects of practice, 33. Discrimination of sound intensities, (Test) : In Experiment on At- tention, Apparatus and Procedure, 77-8; test results, 134; analysis of processes, 135; variability in processes, 136 f; (Training) : In experiment in Sensible discrimination, 42fT; In Experiment on Reproduction, 188 ; analysis of processes, 190 ff ; 'Free judgments,' 191; 'classification,' 192-3, 195; 'personifica- tion,' 193 ; Change in processes in practice, 193 ; individual differ- ences, 190-196; quantitative results, 196 ff; no improvement in capacity, 197-8; (Test) : In Experiment on Reproduction: Pro- cedure, 188; results, 206; analysis of processes, 206: individual differences, 207-210; Interference, 209-210; retrogression in ca- pacity due to indirect methods of dealing with stimuli, 210; range in initial capacity, 211. Discrimination of touch, 6; of colors, 7; of pitch, 7. Distraction, Habituation to, 5, 6. 'Double Fatigue Order' for eliminating practice-effect in tests, 223. Drill, coming into vogue again, 242. Eminence and training, 3. INDEX 303 Estimating Weights, Experiment on, 39; effect of training in, 41 ff ; variability in and complexity of processes, 41. Extensive Threshold of Visual Attention, (Test) : apparatus and procedure, 81; test results, 160; variability in processes, 161; with distraction 163; (Training): Results, 82 ff; analysis of processes, 83 ff, 91 ff, 93 f ; development of 8-level clearness of content, 88-89; factors of improvement, 94; variable influences upon the score, 87-88, 91 ; scoring the 12-letter- rectangles, 295. 'Faculty Psychology,' alleged basis of Formal Discipline, i, 2. Formal Discipline. Definition, i ; assailed by eminent educators, i ; evidence to support, 3 ff ; training and eminence, 3 ; college grades and professional success, 3 ; Rhodes Scholars vary with rigor of training, 3; relationship between school capacities, 3; the data of 'cross-education,' 3-4 ; the data from the literature of experimental psychology, 5 ff. Formal training, 241. General Effect of Special Practice, 178 ff, 218-9, see also 'Transfer- ence,' and 'Spread of Training.' Habituation to distraction, 5. Herbartians, 2. Herbartian Psychology, 2. Imagery : Visual forms, 22 ; varying value of, 23 ; vicarious func- tioning of, 22, 23-4, 28, 48 ff, 136 ff, 139 ff, 148, 191 ff, 211 ff, 217; coordination of fourfold content, in Learning 12-letter- rectangles, 96 ; 'maturing' vague, 83, 85 ff, 89 ff. Individual differences, causes of large, 69. Individual variation, in training on Tachistoscopic work, 91 ff, 94 ff ; in test results, 82 ff, 167 ff, 199 ff ; see also Variability. Inducted effect of state of alertness from experimenter to reagent, 83-84. ^Interference,' 10, 23; in the less simple processes, 11; drops away rapidly with time, 11 ; in relation to practice-effect, 14; caused by change of form of response, 2y ; provision for testing for, in Discrimination of sound, 185, 209, 210 ; compared to negative special effect of special practice, 219. Introspection, function of, 181, 220, 224, 226, 240; limitation of, 185, 203, 226; on the 'intervals' or 'moments' of an experiment, 187, 199 ff. Laboratory experimentation, contribution of, and limitations of, 241. Learning 12-Consonant-Rectangles: (Test): Apparatus and pro- cedure, 80; results, 150; analysis oit processes, 151 ; variation and changes in processes, 151, 152, i74ff; method of scoring, 295; (Training): 94 ff, 98; Changes in processes during training, 94 ff, 98 ; factors in improvement, 97 f , loi ; influences upon score in, 97, loi. 304 INDEX Learning process, related to general effect of special practice, 240; see also Practice-Curve. Letter-squares, scoring of, 80, 82, Appendix C. 295. Marbe color mixer, 43. Marking Out a's and o's {Test), procedure, 75; results, 113, 116; variation in processes, 114, 116. Marking out Words, Experiment in, 34 ff ; change in processes in, 37; variability of processes in, 37-9; anomalous processes in, 38, 68 ; 'common factor' in improvement and transference, 36 ff ; see also Word-Marking. 'Maturing' vague imagery, 83, 85 ff, 89 ff, 174, 204. Mean Variation, as a measure of attention, 108. Memory, Immediate, 16. Memory for colors and odors, 28; for non-sense-syllables, 14, 15; for poetry, 15, 20; for intensities of sound, grays, tones, numbers etc., 20 ; for school material, 24 ff ; for poetry and non-sense- syllables, 29; for visual symbols, {Test) : procedure, 188; analysis of processes, variation in processes, 21 1-4; vicarious functioning of imagery, 211 ff; range in initial efficiency, 214, 215; advantage of trained reagents, 216, 217. Memory of intensities of sound {Test) : Apparatus and procedure, 78; results, 138; analysis and variability of processes, 139 ff; change in processes during test, 141; of Consonants {Test): Apparatus and procedure, 79; results, 142 ff ; analysis of, varia- bility and change in, processes, 142 ff; of Numerals {Test): apparatus and procedure, 79; results, 145 ff ; analysis of, varia- bility and change in, processes, 147; of Visual Signs {Test): Apparatus and procedure, 79; results, 148; analysis of, varia- ation in processes, 148; of paired associates {Test) : Apparatus and procedure, 79; results, 149; analysis of and variation in proceses, 149. Memory-training, 23. Memory-training with mnemonics, 103. Mental processes, relations between, 2 ; variability in, 70 ff ; see Tests and Practice-Curve. 'MentalTests,' contribution to popular notion of the nature of, 108. Method of Constant Changes, Right and Wrong Cases, 43, 78, 190 ff, 206. Miiller's 'cyclical changes' for making successive tests equivalent in difficulty, 223. Per Cent, form of statement, for showing change in efficiency, 67-9, 224. Plateaus of growth demand effort, 242. Practice, changes processes, 227 ff ; rate of improvement as related to initial efficiency, 124, 222; changes of processes during, 222; "Good form" in, 240 foot-note. Practice- Curve, 173; Learning 12-letter- rectangles involves changes INDEX 305 in complexity of processes, 95, 175-6, 292-3, and in kind of pro- duct, 174, 293 ; changes in simple reaction to sound, loi f. 294. Practice-effect, permanence of, 64 foot-note, 64, 67; general and spe- cific, 32-3, 64; negative, in learning 12-letter rectangles, 154; involves changes in processes, 174; see 'Transference,' and 'Spread of Training.' Professional success and college grades, 3. 'Psychergograph,' The, 126; improvised from typewriter, 51. Reaction to classes of men, as, Poets, Philosophers, etc., upon auditory stimulus of names of individual men, as Plato, Browning, etc., See Controlled Reaction, yy, 131. Reaction to sound {Training) : In experiment on Reproduction, loi if; change in processes in, loi flf; (Test) : Procedure, 74; results, 108. Reaction with discrimination and choice, Experiment in, 50 ff ; errors are correlated with longer time for series, 54 ; causes of errors, 55 ; see Typewriter-reaction, and Card-Sorting. Reaction-time, 9. "Recognition or choice of one of two letters," (Test) : Procedure, 186; results, 199; analysis of and variation in processes, 199; quantitative analysis, 200; choices distributed according to cer- tainty, 201 ; in "uncertain" choices, time is a function of R cases, 202 ; influence of preceding cards, 202-3 ; 'maturing' of 'fringe' imagery, 204. Reproduction, (Tests) : (i) Recognition or Choice of One of Two Letters, 186, 199; (2) Reproduction and Recognition of Letters, 187, 205; (3) Sound Discrimination, 206; (4) Memory for visual symbols, 201; Experiment on, 184 fif; Procedure in tests, 186; results, 199 fif; results illustrate worthlessness of purely quantitative treatment, 205 ; anomalous results, 206 ; variation in processes, 206; results of training on Sound Discrimination, 188 flf. Retina, transference of fixation motives from fovea to periphery, 30. Reversible perspective, 30. Rhodes Scholars, 3. Rhythm, a 'common element,' 26, 27. Scholarship varies with rigor of training, 3. School capacities, 3. Score, influences upon in Extensive Threshold of Visual Attention, 87-88, 91 ; in Learning 12-letter-rectangles, 97, loi ; influence of method on, 152; made unreliable by including large practice- eiTect, 221 f. Scoring the 12-letter-rectangle, 295. Sensible Discrimination of Sounds and Brightness, Experiment in, 42 f{; transference shown 46; analysis of process, 47 f. Sensitivity, limen for light lowered by simultaneous sensation of sound, 5, 6. 3o6 INDEX 'Special' practice, always somewhat general, 13, 32-3, 172, 180-1 . 218 f. 'Specific training' said to be specific in its effects, 28. 'Spread of Training' as defined by Fracker, in which the common factor is a conscious or subconscious connection, 225 ; as used in this text involves spread of modes of mental processes, which contrasts with 'Transference' of content of experience, 230; experimental test for, 184; a case of negative, in memory for visual symbols, 216, and from simple reaction to typewriter- reaction, 181 ; positive, from 'literal' prose to visual symbols, 215, 216. Statistical treatment of results: Per Cent form of statement, 67, 68, 69, 224; see Critique of the Test-Training-Test type of experi- ment, 219 ff. Subliminal Differences influence judgment in Sensible Discrimina- tion, 185, 299. Tachistoscopic work; see Extensive threshold of visual attention, 81, 83 ff, 160 ff; Learning 12-letter- rectangles, 80 f, 94 ff. 150 ff; Memory tests, 79, 142 ff ; Reproduction tests, 186, 199 ff. Tapping, {Test): Procedure, 81 f; results, 164, variability in pro- cesses, 165. Tests : 'Cyclical changes' for making successive tests equivalent in difficulty, 223; 'Double fatigue order' for eliminating practice- effect, 223; see Marking Out Words, Sensible Discrimination, Reaction with Discrimination and Choice, Reaction to Sound, Cancellation, Cardsorting, Typewriter-reaction, Controlled Re- action, Memory, Learning 12-letter-rectangles, Word-comple- tion, Trains of ideas, Extensive Threshold of Visual Attention, Tapping, Reproduction. Trains of ideas, (Test) : Procedure, 81; results, 158; variability in processes, 159. 'Transference,' causes of, 19, 22, 28, 63-64, 65 ; as defined by Fracker, 225 ; as used in text involves the concrete material of experience, and contrasts with 'Spread of Training,' 184, 225, 230; anoma- lous cases of, 65 f , 68 f ; causes of anomalous cases of, 68 ff ; dis- agreement of data and conclusions, 67 f ; case of negative trans- ference from maps to card-sorting, 181 ; see also General Effect of Special Practice. Typewriter, used to prepare material for Marking Out Words, 35 ; used in Reaction with Discrimination and Choice, 51 ff; used in , typewriter-reaction, 76 ; used in Controlled Reaction, yy ; experi- ment in learning to use the, 30. Typewriter-reaction, reported from the literature, 11 f; (Training) : Procedure and results, 50^1 ; change in processes during train- ing in, 63; (Test) : Apparatus and procedure, 76; results, 126; analysis of, variability and changes in, processes during test, I29ff. INDEX 307 Variability of mental processes : Extent of radical change in method during tests, 167 ff; extent of variation in initial capacity of reagents, 169; causes of, 171 ff ; see test results and training re- sults of the various tests ; as. Reaction to Sound, Cancellation, etc. Voluntary control, 4, 30; in typewriting, 30 ff; in drawing, 30; in reversible perspective, 30. Visual imagery, a 'common factor,' 23, 26 ; increased in use but not in power by training, 30 ; see Imagery. Word-Completion, {Test): Apparatus and procedure, 80; results, analysis of processes, variation in processes, 157. Word-making, reported from the literature, 9; see Marking Out Words. BIOGRAPHY John Edgar Coover was born on a prairie farm near Remington, Indiana, March i6, 1872. In infancy he moved with his parents to their early home in Cumberland County, Pennsylvania, where he started to a village school in the winter of 1877-1878. In the Spring of 1878 his family moved to a farm in central Kansas, near Dorrance, where he attended a country school for about four years. Another removal to a still newer neighborhood, in the same county, resulted in poor school advantages for about six years. In the Autumn of 1890 he went to school at McPherson College (sectarian, German Baptist or Brethren) where he finished his study of the common branches and learned short-hand and typewriting. In the Spring of the next year he entered an office on the Union Pacific Railway where during a little over a year's station work he learned telegraphy. In the Spring of 1892 he left home for Colorado (Denver) where he revived an early purpose to fit himself for teaching school. After considerable effort during a time while he worked as journeyman printer, and publisher of a country newspaper, he was enabled to enter the State Normal School of Colorado (Greeley) in 1893- 1894, from which he was graduated in 1898. After serving as principal of schools in Como, Colorado, for one year, he entered Stanford Uni- versity (California) (Autumn of 1899), where in January, 1904, he received the A.B. degree in Philosophy ; and in May, 1905, the M.A. degree in Psychology to which department he presented a thesis on "Formal Discipline from the standpoint of experimental psychology" (from which a report was published, in collaboration with Professor Frank Angell, in the American Jr. Psychol., 1907, 18:328-340, under the title "General Practice effect of special exercise)." After occupying the principalship of the Tuolumne County High School (Sonora, California) for two years, and of the Dixon Union High School (Dixon, California) for three years, he returned to Stanford University, became assistant in the Department of Psy- chology, and pursued research for two years more, leading to the Ph.D. degree. Secondary and higher education had been pursued for 12 years, during which time expenses were met by earnings from short-hand and typewriting, printing, an assistant librarianship, telegraph oper- ating, and teaching. The present research occupied four school-years ; during three and one-half it was almost wholly experimental. .0 021 339 489 7