\ 5 4 - 
 
 ^ Lc n. 
 
 [Front Mind. Vol. IV. Ho. 13.] 
 
 IV.—THE RELATION OF ATTENTION TO 
 MEMORY. 
 
 By W. G. Smith. 
 
 •I* J 
 
 It is obvious to every one that there are endless degrees of 
 certainty and security in the knowledge which we have ac¬ 
 quired, whether it be practical or scientific. Some things we 
 cannot help recollecting: other things pertinaciously refuse to 
 be recalled, or they come into our consciousness only after great 
 effort. If we try to give an account of the reasons why some 
 given fact or experience has impressed itself on our memory, we 
 find ourselves at once able to name a large number of factors 
 which may have had some share in bringing about the result. 
 Our memory may be distinct and ready because our interest 
 was aroused when the fact first became known to us. We may 
 have given special attention to the fact apart from its emotional 
 character, or we may have had it presented to us in a specially 
 vivid and intense form. It seems plain also that the power of 
 recollecting will depend in part on the general physiological 
 condition of the individual, on the degree of his health or 
 strength, at the time when he first acquires his knowledge 
 as well as at the time when he attempts to reproduce it. 
 
 Some of the factors which condition memory have already 
 been the subject of important researches. One problem among 
 others of which Ebbinghaus 1 has given the solution, has been 
 that of ascertaining what effect the lapse of time since learning 
 series of meaningless syllables has upon the ability subsequently 
 to reproduce such series. A similar though much simpler 
 problem has been studied by Wolfe 2 in his investigation re¬ 
 garding the duration of our memory for musical tones. Recently 
 Muller and Schumann 3 have studied the effects of rhythm 
 and other modifications of the internal relations between the 
 syllables in series like those used by Ebbinghaus. Evidently 
 it is to such investigations as those, carried out according to 
 
 1 Uber das Gedachtnis. 
 
 2 Philosophische Studien , in. p. 354. 
 
 3 Zeitschrift fur Psychologie und Physiologie der Sinnesorgane, vi. p. 80. 
 
 b40903 
 
48 
 
 W. G. SMITH: 
 
 the experimental method, that we must look for definite and 
 exact information regarding the conditions of memory and the 
 actual processes involved in its working. 
 
 One of the elements in memory of whose efficiency we are 
 sure, but whose exact function has not yet been studied, is that 
 of Attention. It is this problem, the relation of Attention to our 
 power of associating and recollecting objects presented to con¬ 
 sciousness, that is dealt with in the following investigation. It 
 was in the course of experimental work dealing with the 
 subject of Mediate Association 1 that its importance and the 
 need of a separate study of the problem became evident. My 
 aim has been to isolate this factor and shew what is the effect 
 when a subject or observer submits himself for the purposes of 
 experiment to conditions which bring about various forms of 
 distraction. I arranged and began the series of experiments 
 which are here described in the Leipzig Institute for Experi¬ 
 mental Psychology in the summer of 1893: they have been 
 carried on since in the Physiological Laboratory, Oxford, in the 
 summer of 1894 2 . To all those who have assisted, whether as 
 reagents or in giving advice and help, I wish to express my 
 heartiest thanks. 
 
 The method employed by Ebbinghaus of repeating aloud 
 series of meaningless syllables till they could be said by heart 
 seemed inapplicable to the present problem. If it is difficult to 
 keep a given state of attention constant for short intervals, the 
 difficulty could only be heightened by the increase in the 
 length of time required for learning when the attention is 
 distracted. And if the object be to isolate the process of 
 attending from the other processes which normally enter into 
 the act of memorising, then we must have some means for 
 analysing the complex elements involved in such a mode of 
 memorising as that adopted by Ebbinghaus. On the other 
 hand the method employed by Munsterberg in his investigation 
 Die Association successiver Vorstellungen 3 , and elsewhere, seemed 
 capable of being used for the purpose of this investigation. 
 The principle of this method consists in presenting the subject 
 matter which the reagent has to commit to memory in such a 
 way that he acquires a certain knowledge of it; by noting the 
 number and kind of errors committed in the attempt to 
 reproduce what has been learned one obtains a measure of 
 
 1 This Journal, in. N.S., p. 289. 
 
 2 A paper describing these experiments was read before the Physio¬ 
 logical Section of the British Association for the Advancement of Science 
 at the Oxford meeting, 1894. 
 
 3 Zeitschr. f. Psych, i. p. 99. Cf. Beitrdge zur experimentellen Psy¬ 
 chologic, Heft iv. p. 121. 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 49 
 
 the strength and accuracy of the associations which have been 
 formed. Obviously, if the conditions under which the subject 
 matter is presented remain constant, the total of errors commit¬ 
 ted ought to remain fairly constant; there seems theoretically 
 no reason why the process of learning should be carried to 
 the point where the memory becomes perfect. The method 
 employed in the following experiments is in principle the same 
 as that initiated by Munsterberg: in detail however there are 
 many variations. 
 
 Letters of the alphabet, presented in such a way that 
 disturbing ideas and associations should not readily be sug¬ 
 gested, formed the subject matter which the reagent had to 
 commit to memory. They furnish a convenient and easily 
 variable material, and though objections, which will afterwards 
 be referred to, may be made to their use, yet practically it was 
 found that they were sufficient for the purpose in hand. After 
 a considerable number of experiments involving a trial of 
 various modifications, the following method was adopted and 
 with slight changes employed in all the experiments whose 
 results appear in the Tables. Twelve letters arranged in three 
 lines, one above the other, were written on a card 1 : the card 
 was placed on a stand so that the subject who sat close by 
 could read the letters with perfect ease. Previous to the actual 
 experiment the part of the room where the reagent sat was in 
 darkness; by a suitable arrangement the person in charge of 
 the experiments could illuminate the card for any desired 
 length of time 2 . The time chosen for exposure was ten seconds; 
 at the close of this interval, or in the later experiments after a 
 pause of approximately two seconds, the subject was required to 
 repeat as much as he could remember of what he had learned 
 while the card was visible: this together with his remarks was 
 then recorded. 
 
 Perhaps the greatest difficulty in the investigation lay in 
 the arranging of conditions which would secure variation of 
 attention without involving any too complicated mental activity 
 on the part of the subject. The forms of distraction used must 
 be simple and must at the same time have a distinct and 
 appreciable effect. In various researches dealing with the 
 
 1 A copy of one of the cards taken at random may be given:— 
 
 b f 1 g 
 s w k r 
 t x m d 
 
 2 In the first experiments the light was furnished by a small incan¬ 
 descent electric lamp : later this was replaced by an ordinary gas-burner 
 the supply of which could be regulated at will. During the actual 
 experiment the reagent was completely screened off from the person 
 conducting the experiments. 
 
 M. 
 
 4 
 
50 
 
 W. G. SMITH: 
 
 relation of attention to reaction-time, the distraction has been 
 brought about by different kinds of sensory impression, such as 
 the playing of music, the beating of a metronome, or successive 
 flashes of light. It is plain however that such a means of 
 producing inattention would be of little use here: the conflict 
 between a series of sense-presentations and the somewhat en¬ 
 grossing activity involved in learning a combination of letters 
 would be too one-sided and unequal. In one group of ex¬ 
 periments a metronome was kept beating while the experiment 
 was going on; the subjects however remarked that they did 
 not notice the sounds at all while learning. It was suggested 
 that by requiring the subject to introduce into the regular 
 beating of a metronome which was set agoing beside him 
 rhythms of varying complexity, one could obtain distinct 
 and simple forms of distraction. As however no definite or 
 constant result was obtained by this method, it was aban¬ 
 doned for the one which was employed in all the experiments 
 which are summarised in the Tables. According to it the 
 activity involved in memorising is interfered with by another 
 form of activity. 
 
 A distinct but not too complicated device for securing what 
 one may call mental distraction consisted in requiring the 
 subject, while learning the letters on the card, to perform a 
 simple sum in addition, viz., that involved in repeating the 
 series 2, 4, 6, 8,... or rarely, when the former series tended to 
 become too mechanically easy, the series 3, 6, 9.... In order 
 that an effective control over the behaviour of the subject 
 might be secured, he was required further to repeat the 
 numbers aloud and to make each step in the addition coincide 
 with the stroke of a metronome beating at the rate of 60—70 
 strokes per minute. In order to compare the results of this 
 form of distraction with those gained when the vocal organ was 
 exercised, but the mental activity implied was relatively small, 
 the subject was required in the experiments of the next group 1 
 to repeat with each stroke of the metronome a simple syllable 
 such as la. The distraction caused by speaking was next 
 compared with that caused by employment of another set of 
 muscles, viz., those employed in tapping the table with the 
 forefinger, each tap coinciding with a beat of the metronome. 
 Lastly experiments were made in order to gain a normal result 
 
 1 In order to facilitate reference to the different sets of experiments, all 
 those involving one special form of distraction are taken together and 
 referred to as a group. The groups are given here in the order in which 
 they appear in the various Tables. Speaking generally, we may say that 
 in group I. the distraction was mental, in group II. vocal, in group in. 
 muscular. 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 51 
 
 and see the value of reading alone without any distraction save 
 that caused by the fact that the metronome continued to beat. 
 The order in which the distractions were introduced was varied, 
 but they were always arranged so that before any one of the 
 four variations could recur, the other three had been tried. 
 The rate of the metronome was kept constant for each set of four 
 experiments, and in each hour the total number of successful ex¬ 
 periments numbered four or a multiple of four. When owing to 
 any cause an experiment did not succeed, further experiments 
 were made until an even number was reached: in this way 
 there was some guarantee that conditions favourable or un¬ 
 favourable to the memory would affect all groups equally. 
 No attempt was made to keep the interval between the ex¬ 
 periments constant; such a constancy seems by no means so 
 necessary with this method as it is when an engrossing and 
 wearying effort has to be kept up for a prolonged interval. 
 In any case the work of recording the observations of the 
 reagent makes such constancy very difficult. No system was 
 used in choosing the cards for the experiments; it could only 
 be rarely and by accident that different individuals had to 
 memorise the same combinations of letters in the same groups. 
 Not more than twice or thrice in all the experiments was the 
 same card used over again with any one individual. 
 
 There are two methods by which one can numerically 
 estimate the value of the answers given by the subjects. The 
 method of summing errors has already been referred to. The 
 other method consists in estimating the value of what has been 
 actually given, instead of paying attention only to what has not 
 been given or has been given wrongly. Each takes expressly 
 into account what the other has neglected: with a sufficiently 
 large number of experiments the two methods ought to give 
 similar results. The first or negative method of estimation 
 does not resemble any procedure employed in practical life: 
 the second or positive method, on the other hand, relies on the 
 principle of positive valuation commonly employed in esti¬ 
 mating the answers given in examinations. 
 
 In dealing with the errors according to the negative method 
 the following classification was adopted. The first class (o) con¬ 
 sists of cases in which a letter is omitted which was written on 
 the card, the second ( i ) of all cases in which a letter is inserted 
 which ought not to be present, whether that letter was not on 
 the card, or though upon the card is wrongly given a second 
 time. In the third class ( d ) are included errors due to dis¬ 
 placement or disorder, whether that is owing to reproduction in 
 partly wrong order, or to complete ignorance of the position. 
 Each error has the same weight attached to it and counts 1; 
 
 4—2 
 
52 
 
 W. G. SMITH : 
 
 the total number of errors gives a basis for estimating the work 
 of the memory in any instance. 
 
 The arrangement of values according to the positive method 
 was more difficult; one is liable indeed to the charge of 
 arbitrariness in affixing any kind of value. The following 
 classification however seems based on a natural and definite 
 principle of division. The letters which were reproduced were 
 arranged in three classes: (1) the letter named is given in an 
 entirely wrong position, or its position is quite unknown: 
 (2) the subject has a certain imperfect knowledge of the 
 position and order; the letter is put in the right line, or in 
 its correct place in a group of letters which is in a wrong 
 position: (3) when everything is right as regards the letter, full 
 value is given. Corresponding to the three degrees of correct¬ 
 ness, three values are assigned; to the letters of the first class 
 is given the value 1, to the third the value 3; the letters of the 
 second class receive the intermediate value 2. A perfectly 
 correct answer then would receive 36 marks; according as the 
 answer is more or less incomplete the total value varies. A 
 more complicated system of valuation, which differentiated more 
 exactly between the different kinds of displacement or disorder 
 included in the second class, was employed in the first estima¬ 
 tion of the results. A special value, for example, was given 
 where the order was simply reversed: then, however, the 
 question arises, what is to be done with a case of reversed 
 order inserted in the wrong line ? All such complications are 
 avoided by the division given above, which is based on a 
 perfectly simple principle. As a matter of fact the results 
 gained by the two systems agree in all essential points. 
 It happened not unfrequently in the earlier experiments 
 that an individual gave several letters in a line without in¬ 
 dicating whether he had any recollection of their position in the 
 line. After some hesitation I decided to regard this lack of 
 positive information as equivalent to uncertainty on the part of 
 the subject as to the actual position of the letters. When a 
 letter was wrongly inserted or repeated, it was simply dis¬ 
 regarded ; in the case of repetition the benefit of the doubt was 
 given, and that letter taken into account which approached 
 most closely to the correct position. It may be said that 
 according to this procedure an essential feature in the answer 
 is unjustifiably neglected. It is plain, however, that if the 
 method of positive valuation is to keep strictly to its principle 
 of recognising what is actually correct in the answer, cases of 
 insertion cannot properly be taken into account. In fact, as 
 will be seen from an inspection of Table III. which gives an 
 analysis of the errors in each group, these cases form a pretty 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 53 
 
 constant, and in most cases relatively unimportant element in 
 the total of errors. It is however easy to subtract from the 
 mean totals in Table IV. the average errors of insertion: the 
 result, it may be remarked, is either to remove or lessen the 
 discrepancies which exist between the two Tables. 
 
 As was before remarked, the reproduction by the subject of 
 what he recollected followed in the earlier experiments imme¬ 
 diately after the illumination of the card ceased. The result of 
 this arrangement often was that the subject, fearful lest all 
 should disappear, tried to give with a rush what was in his 
 consciousness at the moment, or seemed in perfect readiness for 
 being reproduced. As this tended in some measure to defeat 
 the purpose of the experiments by reducing the part played by 
 memory, the mode of reproduction was changed. Accordingly, 
 in the second division of the experiments the subject was 
 required to give what he remembered at a signal which 
 followed two seconds later. This, it is plain, introduces a 
 new and disturbing factor, for the question at once arises; 
 what is being done in the two seconds ? In order to prevent 
 further repeating and memorising during this pause the subjects 
 were directed to continue the activity which had during the 
 time of exposure served as a means of distraction. It may be 
 objected that the original inequality of these activities as 
 regards their ease of execution would interfere here and cause 
 varying loss of memory during the pause: this must be ad¬ 
 mitted, especially since in the last group there was no 
 distraction. But it must be pointed out that this inequality, 
 so far as our knowledge goes, would have its effect not through 
 varying difficulty, but mainly through the varying completeness 
 with which the field of consciousness was occupied by the 
 activity. The amount of work we do in the interval between 
 learning a task and reproducing what we have learned may 
 have an effect on memory, but this factor, so far as I know, has 
 never been directly taken into account and experimentally 
 studied. Had a new and uniform mode of occupying conscious¬ 
 ness been employed in the pause, there would still have been 
 possibilities of error: the reagent must suddenly recollect and 
 start upon something new, an effort whose difficulty would vary 
 according to the difficulty of the preceding effort. 
 
 It has not seemed necessary to give in separate tables the 
 results of the experiments involving the two different methods 
 of reproduction. On calculating the results separately it was 
 found that the difference between the two sets of results was 
 not great or important. There are variations of course, not 
 merely in the absolute totals but also, though seldom, in the 
 relative positions assigned to the groups in the two divisions. 
 
54 
 
 W. G. SMITH : 
 
 But these variations seem to be of the same nature and extent 
 as those which were found when each half of the experiments, 
 furnished by the first three reagents mentioned in Tables II. III. 
 and IV. was calculated separately for each individual. That no 
 constant and appreciable factor is neglected by presenting the 
 results together can be definitely shewn. The following Table 
 gives the mean of the average values assigned to the subjects 
 in the first (A) and in the second (B) division of the experi¬ 
 ments. The upper line in each division gives the results 
 obtained by the negative, the lower those given by the positive 
 method. The numerals in the first horizontal column designate 
 the groups in the way already mentioned; the total number 
 of experiments in each group, except the third, is for the first 
 division 29. The third group was started later and includes 
 19 separate results. In the second division, owing to a slight 
 irregularity in arrangement, the numbers differ slightly for the 
 different groups; in the first there are 33, in the second and 
 third 34, and in the fourth 33 experiments. 
 
 Table I. 
 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 A 
 
 io-o 
 
 11*9 
 
 9-5 
 
 15*3 
 
 8-2 
 
 16-4 
 
 7*55 
 
 21*4 
 
 B 
 
 10T 
 
 11*3 
 
 8-5 
 
 156 
 
 8*0 
 
 176 
 
 7*4 
 
 21*3 
 
 This Table, it ought to be remarked, contains the results of 
 experiments made with the last six subjects whose experiments 
 are given in the succeeding Tables. The experiments made 
 with Wh., A. and B., the first three subjects, all involve a pause 
 of two seconds after memorising; they differ however as regards 
 the way in which this pause was employed. Wh. did nothing 
 during the two seconds, being directed to remain perfectly 
 passive; A. in many cases forgot to carry on the activity which 
 had served as a means of distracting, but did not attempt 
 actively to memorise. B. on the other hand fulfilled the 
 directions with much greater exactness. The resemblance of 
 the results gained under the various conditions gives ground 
 for believing that the relative values assigned to the groups 
 represent constant and definite processes, and that the slight 
 modifications in method do not in any appreciable degree 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 55 
 
 interfere with the main variations which it was the purpose of 
 the experiments to study. In Table II., which is given 
 below, the results are analysed according to the first or 
 negative method. The first vertical column contains the 
 letters which have been used to designate the different 
 subjects. Under N is given the number of experiments made 
 by the subject in each of the four groups: the total number 
 of his separate results is thus got through multiplying this 
 number by 4 1 . The character of the various groups has already 
 been described. Under a is given the average value of each 
 experiment; under mv the mean variation of each actual 
 experiment from the average, this figure being got by adding 
 together all the amounts by which each separate result differs 
 from the average, and then finding their mean. Under r is 
 given the probable error of the average a , calculated according 
 
 / %v 2 
 
 to the formula r = 0’6745 ^, 'tv 2 being the sum of the 
 
 squares of the amounts by which each result differs from the 
 average, and n the number of results concerned. In Table III. 
 is given the average number of errors of each reagent in the 
 three classes (o, i, d,) already described. 
 
 Table II. 
 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 
 N 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 Wh. 
 
 24 
 
 8*8 
 
 2-5 
 
 0*37 
 
 8‘5 
 
 1-75 
 
 0-31 
 
 7-7 
 
 1*8 
 
 0-35 
 
 7-0 
 
 17 
 
 0-28 
 
 A. 
 
 19 
 
 8-3 
 
 1-3 
 
 0-31 
 
 7 3 
 
 1-8 
 
 0*35 
 
 6*7 
 
 1-7 
 
 033 
 
 6*8 
 
 1*5 
 
 0-29 
 
 B. 
 
 22 
 
 10T 
 
 1*5 
 
 0-28 
 
 8-9 
 
 1-2 
 
 0-31 
 
 7-9 
 
 2-0 
 
 0-38 
 
 8-5 
 
 2-2 
 
 0*41 
 
 He. 
 
 15 
 
 93 
 
 1-2 
 
 0-25 
 
 7*7 
 
 2-5 
 
 0*49 
 
 7*4 
 
 2-8 
 
 0-66 
 
 8*7 
 
 1-7 
 
 0*40 
 
 W. 
 
 15 
 
 9-9 
 
 1*4 
 
 0-31 
 
 9T 
 
 2-0 
 
 0-46 
 
 8*2 
 
 1-6 
 
 0-40 
 
 7-4 
 
 1*8 
 
 0-40 
 
 G. 
 
 10 
 
 11-6 
 
 2-1 
 
 0*54 
 
 11-7 
 
 2-6 
 
 0-59 
 
 10-6 
 
 2-5 
 
 0-79 
 
 8-8 
 
 2-0) 
 
 0-53 
 
 M. 
 
 7 
 
 8*6 
 
 0-9 
 
 0-29 
 
 8*4 
 
 1-0 
 
 035 
 
 5-7 
 
 2-5 
 
 0-77 
 
 4-4 
 
 1-6 
 
 0-59 
 
 K. 
 
 8 
 
 11T 
 
 2T 
 
 0*68 
 
 8-75 
 
 0-85 
 
 0-36 
 
 8-75 
 
 1-0 
 
 033 
 
 7-4 
 
 1-6 
 
 0-52 
 
 H. 
 
 7 
 
 9-4 
 
 1*5 
 
 035 
 
 7-8 
 
 1*4 
 
 0-58 
 
 8-6 
 
 1-6 
 
 0-53 
 
 8-0 
 
 1-1 
 
 0*39 
 
 1 The deviations from this rule are so few that they are best given in a 
 note. The following corrections should be applied in Tables II. and IV.:— 
 the number of experiments made by He. W. G. in Group III. should be 
 11, 10, 8, respectively instead of 15, 15, 10, and by G. in Group II. 11 
 instead of 10. 
 
56 
 
 W. G. SMITH : 
 
 Table III. 
 
 
 I 
 
 II 
 
 III 
 
 IV 
 
 
 0 
 
 i 
 
 d 
 
 0 
 
 i 
 
 d 
 
 0 
 
 i 
 
 d 
 
 0 
 
 i 
 
 d 
 
 Wh. 
 
 5-8 
 
 0*9 
 
 2T 
 
 5-2 
 
 1-0 
 
 23 
 
 4-4 
 
 IT 
 
 2-2 
 
 4*2 
 
 0-9 
 
 1*9 
 
 A. 
 
 5*7 
 
 0-9 
 
 1-7 
 
 4-5 
 
 1*2 
 
 1*6 
 
 3-8 
 
 IT 
 
 1*8 
 
 39 
 
 IT 
 
 1-8 
 
 B. 
 
 43 
 
 1-9 
 
 39 
 
 4*0 
 
 1-2 
 
 37 
 
 3-7 
 
 1-4 
 
 2-8 
 
 3 05 
 
 2T5 
 
 33 
 
 He. 
 
 7-7 
 
 0-2 
 
 1*4 
 
 5-9 
 
 0*5 
 
 1*3 
 
 5-2 
 
 IT 
 
 IT 
 
 4-65 
 
 1*65 
 
 2-4 
 
 W. 
 
 6*9 
 
 0-9 
 
 2T 
 
 5-9 
 
 IT 
 
 2T 
 
 5T 
 
 IT 
 
 2-0 
 
 4*0 
 
 0*8 
 
 2-6 
 
 G. 
 
 5-6 
 
 2-9 
 
 3T 
 
 5-0 
 
 32 
 
 3*5 
 
 4-0 
 
 2*7 
 
 39 
 
 3*5 
 
 2T 
 
 32 
 
 M. 
 
 7*6 
 
 0T 
 
 0*9 
 
 6-4 
 
 0*7 
 
 1*3 
 
 33 
 
 1-0 
 
 1*4 
 
 1-7 
 
 1-6 
 
 IT 
 
 K. 
 
 5-9 
 
 1-2 
 
 4*0 
 
 4-0 
 
 0-9 
 
 385 
 
 4-5 
 
 0-9 
 
 335 
 
 2*5 
 
 1*5 
 
 34 
 
 H. 
 
 8*3 
 
 0-55 
 
 0-55 
 
 6-7 
 
 0T 
 
 1-0 
 
 7-6 
 
 0-3 
 
 0-7 
 
 40 
 
 0-4 
 
 36 
 
 In Table IV. are given the figures representing the values 
 of the different groups when the results are analysed according 
 to the positive method. The Table is constructed on the same 
 plan as the second. The amounts assigned according to the 
 three degrees of value assumed by the positive method have 
 not been summed up separately 1 . 
 
 Table IV. 
 
 
 I 
 
 H 
 
 III 
 
 IV 
 
 
 N 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 a 
 
 mv 
 
 r 
 
 Wh. 
 
 24 
 
 15-0 
 
 3*6 
 
 0-62 
 
 16*8 
 
 3*2 
 
 0-55 
 
 19-2 
 
 4-0 
 
 0*72 
 
 20*0 
 
 2-9 
 
 0-49 
 
 A. 
 
 19 
 
 15-8 
 
 34 
 
 0-68 
 
 19-8 
 
 38 
 
 0*76 
 
 21-4 
 
 43 
 
 0-77 
 
 21*2 
 
 32 
 
 0-58 
 
 B. 
 
 22 
 
 15-9 
 
 2-9 
 
 0-53 
 
 18T 
 
 40 
 
 073 
 
 20-4 
 
 3T 
 
 0-57 
 
 27*2 
 
 33 
 
 0*64 
 
 He. 
 
 15 
 
 11-0 
 
 1-6 
 
 0-34 
 
 16-5 
 
 5*9 
 
 1T7 
 
 18T 
 
 4*8 
 
 1*23 
 
 17-9 
 
 2*9 
 
 0-63 
 
 W. 
 
 15 
 
 11-6 
 
 2-5 
 
 0-55 
 
 14*2 
 
 33 
 
 0-75 
 
 17*4 
 
 38 
 
 1-04 
 
 19*9 
 
 39 
 
 0-94 
 
 G. 
 
 10 
 
 13*4 
 
 4-4 
 
 I'll 
 
 14*5 
 
 38 
 
 0-95 
 
 18T 
 
 2-85 
 
 0-82 
 
 20-6 
 
 2-8 
 
 0-88 
 
 M. 
 
 7 
 
 12-0 
 
 2-0 
 
 0-67 
 
 14-7 
 
 23 
 
 0*77 
 
 23 3 
 
 6*4 
 
 1-97 
 
 28-6 
 
 32 
 
 1T4 
 
 K. 
 
 8 
 
 11*5 
 
 34 
 
 1-06 
 
 17T 
 
 2-9 
 
 0-89 
 
 159 
 
 2'1 
 
 0-68 
 
 22-0 
 
 35 
 
 1-05! 
 
 H. 
 
 7 
 
 10-3 
 
 2-2 
 
 0*76 
 
 14-6 
 
 3-4 
 
 1-35 
 
 12-3 
 
 3*2 
 
 0-93 
 
 19-7 
 
 1-8 
 
 0*56 j 
 
 1 It may not be amiss to give the figures representing the average 
 result for all the subjects in each group : they present a sort of typical 
 curve. The numbers in the upper and lower lines give the results of the 
 negative and positive analysis respectively. 
 
 I. 9-7 II. 8*7 III. 7-95 
 12-9 16-3 18-5 
 
 IV. 7-4 
 21-2 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 57 
 
 Before going on to discuss the theoretical value and meaning 
 of the numerical results given in the Tables, I wish to refer 
 shortly to the question how far the methods of experimenting 
 and handling the results are trustworthy. One can judge best 
 of such matters by presenting data such as we have in the 
 foregoing tables. The question, for example, whether com¬ 
 binations of letters of the alphabet will not, when used as 
 subject matter for memorising, give results too uncertain and 
 variable to allow of any conclusion as to the actual processes of 
 association and memory, receives its answer in the foregoing 
 numerical analysis. It is difficult to see how the regularity and 
 constancy of the numerical values can be explained if the 
 method be unreliable. To say that in a single experiment any 
 one letter or combination of letters reproduced may be due to 
 accident is true; to conclude that therefore the same error 
 attaches to prolonged series of experiments is not justified. 
 Nor does the fact that the same letters are employed more or 
 less often during the course of an hour create a difficulty: the 
 combinations of letters employed are thoroughly uninteresting, 
 and have no tendency to remain fixed in memory after a 
 particular experiment has been finished. In any case the 
 possible errors arising from this cause will obviously distribute 
 themselves over all the groups, and disappear as the number of 
 experiments increases. 
 
 This last observation applies equally when we consider the 
 errors due to slight variations in the time of exposure, in the 
 length of the pause, in the rapidity with which the subjects 
 react to the signal for reproduction in the various cases. That 
 different combinations of letters will be learned with varying 
 ease or difficulty is also a feature which necessitates having a 
 fairly large number of observations. Two practical sources of 
 error remain to be mentioned: first, the tendency to form 
 intelligible words out of the letters presented; second, the 
 irregularity with which the distracting activities were carried 
 on during the time of exposure and during the pause. As 
 regards the first error, it has to be noted that with almost no 
 other observers than A. and B. did it make its appearance. 
 Where it does appear however, this tendency is very annoying; 
 when once the idea of forming intelligible words out of the 
 letters occurs to the subject, it is very difficult to suppress it, 
 however much he may wish to do so—in fact the more he 
 wishes the more difficult is it. I have excluded the cases where 
 this error distinctly endangered the result; in the case of the 
 observer A. very many experiments had to be rejected for this 
 reason. The second error illustrates the difficulty of carrying 
 out two conscious activities simultaneously. Whenever the 
 
58 
 
 W. G. SMITH : 
 
 subject broke down during the experiment, for example, when 
 the attempt to do the sum in addition proved a complete 
 failure, then the experiment was stopped. When however the 
 irregularity in keeping time with the metronome and in the 
 other processes was not too great, the observation was accepted. 
 As the state of mind during the pause is not so important, and 
 as it was advisable not to disturb the reagent by insisting too 
 rigidly that the distracting activity should be carried on through 
 the pause, a greater latitude was allowed in this respect. In all 
 cases however it was understood that the pause was not to be 
 used for the purpose of actively memorising. The question 
 may be raised, what happens when the reagent remains passive 
 during this interval ? As far as I can gather, a sort of after 
 image of a line or part of a line seems to be present; it is faint 
 and indefinite in the most of cases, and gradually fades away. 
 
 In order to give opportunity for judging of the extent to 
 which the averages of the total errors and of the positive values 
 given in Tables II. and IV. really represent the actual observa¬ 
 tions, both the mean variation and the probable error have 
 been calculated. The former measures the average extent of 
 deviation from the mean, and in doing so naturally gives great 
 weight to a large deviation. One or two such deviations occur 
 in almost every series, usually one positive deviation being 
 nearly compensated by an opposing negative deviation: they 
 have in every case been included, and no number has been 
 rejected because it was unusually large or small. It is 
 obvious, however, that such deviations are abnormal, and that 
 the mean variation, which makes one bad observation outweigh 
 several good ones, does not give any true representation of the 
 character of the deviations; the larger the error is the less 
 likely is it to occur. For this purpose the probable error is the 
 best guide: in it both the extent of the deviations and the 
 probability of their occurrence are taken into account: the 
 resultant figure r gives the range on either side of the mean 
 within which the observations probably fall. With the help of 
 the mean variation and the probable error we can compare the 
 results here obtained with those given in other researches: it 
 will be granted I think that, comparing the extent of the 
 probable error given in Tables II. and IV. with that reached in 
 the observations of Ebbinghaus, or the mean variation with 
 that which Oehrn 1 finds in experiments similar to those of 
 Ebbinghaus, one finds a sufficient regularity and steadiness in 
 the values assigned by both methods to the various stages of 
 memory. 
 
 1 Experimentelle Studien zur Individualpsychologie. 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 59 
 
 The reliability of the positive method of valuation seems 
 confirmed by the close correspondence which the relative posi¬ 
 tions assigned by it to the various stages of memory shew with 
 those indicated by the other method: with few exceptions what 
 is proved to be a weakened memory by one method is likewise 
 proved by the other. With absolute values we have here of 
 course primarily nothing to do; it is the relative positions 
 assigned to the different groups that is our chief concern. 
 That the classification of values adopted in the working of the 
 second method is merely arbitrary could in no case be reason¬ 
 ably maintained ; that it is in a high degree trustworthy seems 
 proved not merely by the constancy of the relative values which 
 it assigns, but by the general support which it gives to the 
 negative method. The discrepancies between the two methods 
 which do exist can in most cases, as before remarked, be 
 removed or lessened when errors of insertion are taken into 
 account in the second method as well as the first. The other 
 divergencies are only such as might be expected, for it is only 
 roughly that the two methods supplement each other. This is 
 at once apparent when the values assigned to single experi¬ 
 ments are considered; it is only when a larger number of 
 observations are taken that the correspondence becomes defi¬ 
 nitely established. With respect to single experiments it is the 
 second method which gives the truer estimate—truer in the 
 sense that the relative positions indicated by it correspond more 
 closely to the final relations of the groups as established by 
 both methods than do those gained where the sum of errors is 
 considered. 
 
 In beginning experiments care was always taken that the 
 subject should by actual experiments become familiarised with 
 the task he was required to perform. A considerable amount 
 of practice was gone through by several of the subjects before 
 the final experiments began, owing to the number of tentative 
 experiments that were carried through at the beginning of the 
 investigation. That the influence of practice and exercise can 
 not be so great here as in experiments where a certain more or 
 less difficult task has to be done perfectly, and to be persisted in 
 till finished, seems clear; where the learning on the other hand 
 is continued only for a short time with no aim at final correct¬ 
 ness, it is obvious that the subject has far less drill and that 
 there will be less growth in the facility of learning than where 
 series of syllables have to be learned by heart. And at the 
 same time we may expect that the influence of weariness will 
 also shew itself much less where a method such as the present 
 is used. As a matter of fact, while one observer complained of 
 being tired at the end of an hour’s work, another was surprised 
 
60 
 
 W. G. SMITH: 
 
 at such a complaint; the others made no remarks on the 
 subject. Only in regard to one of the groups, the third, was 
 the remark made that the task seemed to become distinctly 
 easier. That in general there was a gradual improvement in 
 the memorising ability of the subjects may be concluded from 
 an inspection of the figures in Table I.; in this Table however 
 the results in the second division shew the effect of exercise com¬ 
 plicated with other factors. In order to gain more exact data I 
 have analysed separately the results of the three observers who 
 have furnished the longest series of experiments, the average 
 values for the first and second halves of the series being 
 calculated for each group. In some cases there is a fall in 
 the value of the memory: in the majority of cases there is a 
 rise, which however is not specially conspicuous in any one 
 group. With reference to the capabilities of the persons who 
 assisted in the investigation, it may be remarked that all save 
 one are graduates or students of philosophy and psychology, 
 while three are teachers of philosophy. 
 
 The general features of the results presented by this in¬ 
 vestigation may now be summed up. The greatest distraction 
 is that caused by the activity involved in summation, while the 
 use of the organ of speech causes a smaller, but still distinct, 
 disturbance. The effort required in tapping the table produces 
 generally a certain disturbance: its effect however is neither so 
 universal nor so distinct. These conclusions may be taken as 
 proved by both methods; the negative method, in addition, 
 shews that the most prominent factor in the weakened memory 
 is the diminution in the number of letters retained. The 
 average number of errors of displacement and of insertion is, 
 on the other hand, smaller and remains for each group practically 
 constant; relatively those errors take a larger place in the 
 results of the first and second groups when we consider them in 
 connection with the decreasing number of letters which are 
 recollected. 
 
 There can be little doubt that the factor to which we must 
 look in interpreting the results of the l#St three groups is that 
 on which various psychologists have insisted, viz., the presence 
 of a motor element in much of our mental activity. And in 
 particular the loss of memory due to the effort of repeating a 
 syllable in an audible voice while one is memorising seems 
 caused by the fact that the reading or learning of any series of 
 letters or words is accompanied normally by a more or less 
 distinct articulation, this articulation being interfered with by 
 repeating the syllable. To what extent our memory or asso¬ 
 ciating processes in any case can act apart from direct or 
 associated motor activity is a further question which cannot be 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 61 
 
 dealt with here. For the interpretation of the present results 
 it seems necessary to assume that this motor activity plays a 
 very large part in the formation of our associations. The full 
 significance of this assumption can be seized only when we 
 remember that what the subject is called upon to do in all 
 cases is simply to read the series of letters, i.e., to let the series 
 of visual objects which are presented to him impress themselves 
 on his memory. Apart from experimental evidence one might 
 suppose that a series of visual objects would associate them¬ 
 selves together as well as any other series of perceptions or 
 sensations. But, as we see, when the reiterated articulation of 
 a syllable obstructs the further innervation of the articulatory 
 mechanism, the memory is much worse than it is where no 
 distraction is present, and where the unskilled observer would 
 detect no process other than that of simple reading. Such a 
 result indeed is only a confirmation of the fact already ascer¬ 
 tained by Pathology of the close connection between sensory 
 and motor activities, and in particular between the under¬ 
 standing of words and the ability to articulate them, or in some 
 other way give to them a motor expression. 
 
 These theoretical considerations are amply confirmed by the 
 observations contributed by the subjects. They insist that a 
 distinct and embarrassing obstruction is felt when they attempt 
 to read over the series of letters, an obstruction which is 
 specially marked at a point where a letter such as z stands, 
 whose pronunciation is unusually difficult. They see the 
 letters and realise them distinctly as visual objects, but are at 
 the same time impressed with a sense of their inability to 
 connect the letters and form them into a series which can be 
 remembered. And they notice how in the intervals of audible 
 articulation the attempt is involuntarily made to insert an 
 inaudible articulation, to employ Bewegungs-antriebe, which 
 shall help to form the letters into a connected whole. 
 
 An additional support is lent to this explanation by a 
 consideration of the data presented by the third group of 
 experiments. The object of this group was to allow of a 
 comparison between the motor activity of the vocal organ and 
 that involved in the use of a set of muscles not directly 
 connected with any mental operation. That an exact com¬ 
 parison is hardly, if at all, possible is plain from the fact that we 
 must in any case employ two quite different groups of muscles 
 habituated to different functions. Apart from theoretical ex¬ 
 actness however, a basis for comparison may be got by observing 
 what effect activity of the vocal mechanism has as compared 
 with that implied in moving the forefinger. Possibly the act of 
 saying ‘ do ’ or * la ’ is more complicated than that of moving 
 
62 
 
 W. G. SMITH : 
 
 * 
 
 the forefinger; yet after all a very considerable degree of 
 facility in executing more complicated vocal movements must 
 be granted to most persons. We may then assume equivalence 
 between the two kinds of movements to this extent at least, 
 that regarded physically the energy expended and the effort of 
 innervation required in the two cases are fairly similar. 
 
 In almost every instance mere muscular movement produces 
 less of a disturbance in memory than does activity of the vocal 
 mechanism. And the reason for this is evident: the articula¬ 
 tory innervations involved in memorising are most conspicuously 
 interfered with by conflicting activity of the vocal mechanism. 
 The observations made by the reagents are only a concrete 
 expression of this statement: what they notice in the ex¬ 
 periments of the third group is a conflict of two motor series 
 such as, for example, is involved in the effort to play on a 
 musical instrument when one hand is required to move in a 
 time different from that of the other. Had the reagents been 
 able to render the two series coincident, and make the rhythm 
 of the finger movement the same as that of the articulatory 
 movement, then the difficulty would have been lessened, and one 
 impulse or innervation would have served to carry out both 
 movements. But owing to the arrangement of the experiments 
 such coincidence would have made the learning too fast or too 
 slow; too fast if a whole line was to be gone through at each 
 stroke of the metronome, too slow if only one letter was to 
 accompany each stroke. In some cases indeed the observers 
 seemed to find a help in the rhythm of the finger movements; 
 when they succeeded in combining the innervations for articula¬ 
 tion and for movement an added emphasis was lent to the 
 letters, which seemed to help in fixing them in the memory. It 
 is evident however that experiments directed to the isolation 
 and direct proof of this fact are needed before anything definite 
 can be said. 
 
 If the series of letters presented to the reagent were 
 associated as visual objects, then it would be still more difficult 
 than in the experiments of the second group to understand why 
 an indifferent muscular activity, such as that of the hand, 
 should interfere with the associations. It may be said that 
 this activity implies a certain degree of inattention to the 
 letters presented, and no doubt this is true. But the dimi¬ 
 nution of attention must have been very small; there is a 
 distraction, but it is almost wholly a distraction of the motor 
 mechanism. And it is important to distinguish here between 
 these two things, predominance of a state of consciousness or 
 attention, and that state or attitude of mind which we term 
 activity. In many of its manifestations, and here in particular, 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 63 
 
 psychical activity, so far as it is a factor different from mere 
 movement of ideas, seems intimately connected with some form 
 of function of the motor system. 
 
 According to the testimony of more than one subject, the 
 disturbance caused by the finger tended to disappear in pro¬ 
 portion as the effort required to make its movement keep time 
 with the beating of the metronome became more automatic. 
 This would mean that the two motor series conflict and compete 
 with each other only so long as they both remain in direct 
 connection with conscious experience: two disparate activities 
 on the other hand can be carried on simultaneously if one at 
 least be automatic, or carried on apart from our ordinary 
 consciousness. Binet concludes, as the result of an experi¬ 
 mental study of such questions 1 , that an automatic movement 
 is able to coexist with a voluntary movement because it is not 
 preceded by a conscious mental representation, and consequently 
 does not involve conflict of ideas. This it is obvious is not an 
 explanation, but simply a restatement of facts: why two con¬ 
 scious series of ideas must conflict is not explained either 
 psychologically or physiologically. The theory, that inability to 
 attend effectively to more than one series of ideas is due to a 
 limitation in the available psychophysical energy, does not seem 
 at all convincing, but what should be put in its place is by no 
 means equally clear. 
 
 It will be observed that while, with regard to the majority 
 of the observers and the average of the results given by all 
 the individuals, the value of memory shews a diminution 
 in the experiments of the third group as compared with the 
 fourth, yet divergencies occur. For this fact various explana¬ 
 tions are available. Apart from accidental variation in the 
 results, which is not a probable explanation, there will evidently 
 be a difference in the effect of these distractions upon memory 
 according as the subject is of a motor, a visual, or an auditory 
 type. I feel inclined to believe that this is partly the expla¬ 
 nation of the irregular results of He.; this individual was 
 particularly emphatic in his statements as to the prominence of 
 visual images in the process of reproduction. How far this 
 explanation can be carried is however doubtful; I tried to 
 ascertain from the subjects what their usual kind of imagery 
 was, and to relate it with the results, but was not able to find 
 any definite connexion: to establish such a relation it would be 
 necessary to ascertain by direct experiment to what extent and 
 in what relations the imagery conformed to the different types. 
 
 1 Revile Philosophique , xxix., “La concurrence des etats psychologi- 
 
 ques.” 
 
64 
 
 W. G. SMITH : 
 
 Another explanation may perhaps be found in the fact that the 
 reagents were left entirely free in their method of learning; 
 they could learn the series slowly, or they could run over the 
 letters quickly and often: especially in the fourth group was 
 there scope given for the varying preferences of the individual. 
 In the other groups the subject, it is plain, had to memorise not 
 so much as he wished, but as best he could. These individual 
 variations may take shape also in another form; according to 
 his momentary disposition he will tend to spend greater or less 
 energy in learning what is presented to him. And apart from 
 this general variability there are the differences which arise 
 from the fact that, as the experiments were actually arranged, a 
 severe distraction alternated with one less severe, or with a 
 state of freedom from distraction. The subject will obviously 
 feel relieved when an easy task is put before him; this 
 however leaves it quite undecided whether he will shew 
 this feeling of relief in memorising much better, the energy 
 expended being the same as before, or whether he will un¬ 
 consciously with the removal of the difficulty lessen the intensity 
 of his efforts and, sure of learning in any case more than before, 
 will content himself with memorising less vigorously. These 
 divergencies, it is plain, furnish the point of departure for 
 further research and exacter analysis. In particular it would 
 be desirable in a more advanced stage of this research that 
 the conditions of experiment should be altered so that while 
 becoming more artificial, they would still in the end be more 
 precise. A greater similarity in the method of learning would 
 in part be secured by the adoption of a method like that of 
 Muller and Schumann, which secures that each object shall be 
 presented to the reagent in an order and for a length of time 
 which can be exactly regulated 1 . 
 
 In order to facilitate the analysis of the various phenomena 
 I have begun by discussing the nature of the processes involved 
 in the experiments of the last three groups. It is however only 
 in the first group that we have facts bearing directly and 
 immediately on the phenomena of attention. In fact the three 
 last groups were arranged in order to analyse and illustrate the 
 factors that are involved in the processes of the first group. 
 That we have in this group a real disturbance of attention 
 might be supposed on theoretical grounds; that such a state of 
 mind actually resulted from the experimental conditions was 
 
 1 The series of meaningless syllables used in their experiments were 
 written on sheets of paper; these were fastened on a cylinder which 
 rotated at a known rate and shewed the syllables in succession to the 
 reagent, who was looking through a slit in a screen set before the 
 cylinder. 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 65 
 
 asserted by the observers. The movements of speaking, said K., 
 one of the most skilled observers, do not become more difficult 
 in the fpuffh group; what the addition does is to diminish the 
 energy available for learning. 
 
 Here again, as in the relation between the second and third 
 sets of experiments, it is not possible to effect an exact 
 comparison between the effort of articulation in the second 
 group with that implied in the repeating the series of numbers 
 audibly. The words uttered in the course of the addition differ 
 with each stroke of the metronome: towards the end of the 
 period of memorising words of more than one syllable are 
 involved. As regards the first point, it is not probable that 
 the effort of repeating different but perfectly easy and well- 
 known words is much greater than that of uttering the same 
 syllable over again. That as the addition progresses we have 
 first words of two syllables and then two or, at most, three words 
 of three syllables, is a more important consideration. The 
 reagents however used constantly a device which minimised 
 the difficulty; emphasising only one out of the two or the 
 three syllables to be spoken, they allowed the voice to die away 
 in pronouncing the rest of the word, which thus formed a sort of 
 indistinct sound or Nachklang. That this diminished intensity 
 of articulation greatly diminished the amount of disturbance 
 otherwise involved seems clear from the cases where the reagent, 
 while attending too exclusively to the work of memorising, 
 allowed the voice to die away and ceased to speak distinctly, 
 thus increasing his power of memorising. The only data of an 
 experimental kind available for the confirmation of this 
 account of the matter are those gained in the course of 
 various tentative experiments. The purely vocal disturbance 
 was in these experiments brought about by pronouncing in 
 
 their order the series of numbers 1, 2, 3,. So far as these 
 
 experiments go, their results correspond very closely to what is 
 found when it is a syllable that is pronounced. 
 
 Simple as the act of addition may seem, the effort required 
 to combine it with the process of memorising was extremely 
 difficult—“painful” was the term applied to it by one of the 
 subjects. For its effect was to make it almost impossible to do 
 more than stare in a helpless way at the different parts of the 
 combination of letters presented. The reagent was conscious of 
 the presence of the visual objects, but it was as if their meaning 
 was lost and he was unable to grasp the letters as anything 
 more than mere signs, as if part of their content was wanting. 
 Accompanying this state of mind was a feeling of helplessness, 
 and a conviction of the uselessness for the purposes of learning 
 and recollection of this merely visual consciousness of what was 
 
 5 
 
 M. 
 
66 
 
 W. G. SMITH : 
 
 presented. According to one observer it was during the earlier 
 and easier stages of addition that memorising was really possible : 
 according to others the opportunity for grasping the connection 
 of letters presented itself in the pauses between the beats of 
 the metronome, or at the close of the time of exposure 1 , when 
 it seemed possible as it were to snatch part of the series 
 and give it a moment’s concentration. In this way short 
 groups of letters, small pictures, could be impressed on the 
 memory, and in this way the full effect of the distraction was 
 not felt; the attention oscillated, so to speak, during the process 
 of memorising. In any case however the total attention together 
 with the ability to reproduce larger groups and a connected 
 series was much lessened. How low the value of the memory 
 would have sunk had the state of distraction involved in 
 each separate step of the addition been maintained uniformly 
 throughout, it is difficult to say. 
 
 Reference has already been made to the fact that as we pass 
 to the second and then to the first group, the number of errors 
 of insertion and of displacement increases relatively to the 
 number of letters actually remembered. As regards errors of 
 insertion this statement seems justified, in spite of considerable 
 variations. But in any case this kind of error is not of great 
 importance in comparison with the errors of disorder: they 
 shew with considerable clearness and persistency an increase 
 from the fourth group to the first. The effect of mental 
 distraction may then be said to consist not only in lessening the 
 total area of consciousness available for the reception of new 
 impressions, but in confusing the recollection and thus lessening 
 the power of forming exact associations, which shall enable the 
 reagent in the first place to locate definitely what has been 
 impressed on his memory, and secondly to exclude ideas which 
 did not form part in the complex first presented to him. 
 
 In his research on the Association of Successive Ideas, which 
 has already been referred to, Miinsterberg comes to the opposite 
 conclusion. In these experiments, which were carried out in 
 order to answer the question whether association through mere 
 succession was possible, letters forming series of varying length 
 were successively presented to the subject. Some of the series 
 were learned with full attention ; others had to be learned while 
 at the same time he was carrying out in an audible voice 
 complicated arithmetical operations. The distraction of the 
 attention lessened the. number of letters wffiich could be 
 
 1 The latter fact was noticed only when the electric light was em¬ 
 ployed. It is difficult to say whether it was illusion, or was due to the 
 circumstance that the cessation of light in the incandescent lamp is not 
 quite instantaneous. 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 67 
 
 retained; according to Miinsterberg it could not be the reason 
 why in series including only four or five letters the right letters 
 almost without exception were reproduced, but at the same time 
 were given in wrong order. The real reason lay, he contends, in 
 the exclusion of any opportunity for association by simultaneous 
 contiguity. “The only possible effect of the greater or less 
 amount of attention devoted to the perception of single letters 
 is to cause them to be impressed more or less firmly on the 
 memory.” 1 
 
 That the main conclusion regarding successive association— 
 viz., that apart from simultaneous association or connections 
 formed by means of the motor mechanism, it has no existence,— 
 is rendered probable may be granted. But it does not follow 
 either from the theory or from the experimental results that 
 the only effect of inattention is to lessen the number of ideas 
 which can be retained; in fact the suggestion that variation in 
 attention can vary the intensity or completeness with which 
 ideas are “impressed” on memory, without at the same time 
 thereby altering the intensity or strength of the associative bonds 
 between them, seems somewhat improbable. And as regards 
 the actual results obtained in Mtinsterberg’s research there is 
 room for the working of all these factors; the greatest share in 
 the disturbance of order may be due to the cause he suggests, 
 but part also may very well be due to the two factors in 
 attention; inattention or distraction may cause fewer ideas to 
 be recollected, and may lessen the strength with which these 
 are held, but it may also diminish the security and firmness of 
 their associative connections. That it does so is rendered 
 highly probable by the results which have already been brought 
 forward in this paper. 
 
 The words Attention and Memory have been used through¬ 
 out these pages as if they corresponded to definitely and 
 generally accepted conceptions. Practically we know quite 
 well what attention is, though the theory of the matter 
 may be hard enough. It will be most profitable here to 
 avoid theories and simply state what view of the function of 
 attention seems most appropriate to the facts which we have 
 been considering. The facts relating to attention are, as we 
 saw, furnished mainly by the experiments of the first group. 
 And the chief fact we found to be that inattention not merely 
 lessened the extent to which ideas could be taken up and 
 retained in consciousness, but in a still greater degree confused 
 and deranged the associative connections in what was actually 
 recollected. This again is simply the objective expression for 
 
 1 Zeitschr. f. Psych., i. p. 105. 
 
 5—2 
 
68 
 
 W. G. SMITH : 
 
 the subjectively felt inability of the reagent to apperceive and 
 relate what was presented to him. Through the process of 
 mental distraction, said one of the observers, Wahrnehmung was 
 turned into Empfindung. 
 
 It is in this that the essential operation of mental distraction 
 seems to lie. In ordinary attentive experience each impression, 
 each idea, as it occurs, is met and transformed—to speak in 
 Herbart’s language—by a mass of apperceiving ideas which 
 render it known, familiar, recognised, and in doing so relate 
 and associate it with similar experience we have had in the 
 past. It is this mass of apperceiving ideas which enables us 
 so easily without conscious or voluntary reflection to follow 
 and understand some new fact in a science with which 
 we are familiar. And in distraction of the kind we are con¬ 
 sidering the really effective element seems to be an inhibition 
 of these associative processes which enable us unconsciously 
 to relate and recognise what is presented to us, the effect 
 of this inhibition being to produce temporarily a state to 
 which one may apply the term Seelenblindheit. How far 
 these processes are in any case represented in consciousness 
 is another question; probably hei;e, as in many other relations, 
 apperception is to be understood as referring simply to the 
 total indefinite state or attitude of mind, “ the feeling of 
 tendency” resulting from the minimal excitation of certain 
 “ideational centres.” The essential fact in attention on the 
 other hand would lie, apart from its accompaniments, muscular 
 adjustment and the like, in the strengthening of an idea or 
 impression by the processes of blending and redintegration, 
 those processes which are operative in all learning and knowing: 
 attention on this view of the matter would be knowledge in its 
 most perfect form, that form being attained not merely by 
 positive support and strengthening, but by an inhibition or 
 repression of all that might compete with and distract energy 
 from the apprehension of what is attended to. There do not 
 seem positive reasons, either theoretical or experimental, to 
 forbid the hypothesis that there are degrees of clearness and 
 distinctness in mental states: it seems however equally im¬ 
 possible to prove that an apparent increase in the clearness 
 of an idea or sensation is an independent special process 1 , and 
 
 1 In a paper “Uber die Schwankungen in der Auffassung minimaler 
 Sinnesreize,” Phil. Stud., vm. p. 365, Eckener asserts that a stimulus 
 can still affect consciousness though quite unperceived and “dark”: his 
 proof consists in the observation that when a minimal sound stimulus has 
 become subjectively inaudible, the reagent is still in many cases able to tell 
 when the source of sound is objectively interrupted. That this proves 
 nothing in regard to degrees of clearness in ideas is evident when we 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 69 
 
 not an illusive appearance due to some change in the associative 
 or emotional complex surrounding it. 
 
 When we speak of memory as being a storehouse for ideas, 
 of psychical states impressing themselves, being retained and 
 last of all reproduced, we are using metaphors which are 
 adequate for all practical purposes, and which are about as 
 true as the theories often brought forward as giving the real 
 truth of the matter. On this subject Physiology seems a 
 better guide than Psychology, for Psychology has been too 
 much burdened by its theory of ideas, and by its conviction 
 that the elementary and fundamental psychical processes must 
 be explained in terms of the developed mind. Physiological 
 memory means simply that a given nervous centre has been so 
 changed—how we do not know—by the occurrence in it of a 
 certain set of processes, that when one of these occurs again 
 the others tend to follow. It does not occur to anyone to say 
 that the old processes are reproduced, and that physiological 
 memory is inconceivable unless we suppose that numberless old 
 innervations fuse with a new one, when that new one occurs, 
 and so enable it to start a series of movements over again. But 
 it does occur to psychologists, and is confidently asserted, that 
 such a fusion of old and new is one of the fundamental 
 associative processes. The physiological account of the matter 
 is simple and natural: old psychical states are no more 
 reproduced than are physical states, but similar states are 
 reproduced; these in their turn tend to recall other states, 
 which again had something like them in past experience. The 
 fusion of old and new which occurs in organic memory means that 
 owing to habit or repetition of an experience the physiological 
 processes have become changed: they are modified owing to 
 what has gone before, so that a fresh reaction is correspondingly 
 modified. There seems no reason why, when we speak of storing 
 up and of reproduction of ideas, we should introduce anything 
 more into the processes than we would require were we to 
 apply like metaphors to purely physiological memory. 
 
 According to the usual statement of the Law of Association 
 by Simultaneous Contiguity, objects or ideas which have once 
 been present together in consciousness tend in consequence to 
 recur in the same connection. It is obvious that in such 
 statements emphasis is laid on the fact that the ideas have 
 been simultaneously elements of a conscious state, and that the 
 process is one concerned solely with sensory or intellectual 
 
 remember that cessation of a stimulus physiologically may have for its 
 first effect an increase of the reaction ; e.g., in experiments demonstrating 
 the electrical response of the frog’s eye to light, the first effect noted when 
 the light is withdrawn is an increase of the response. 
 
70 
 
 W. G. SMITH : 
 
 experience. Observations made during the course of this 
 research give occasion for some remarks on these two points. 
 
 It was pointed out again and again by the reagents that 
 mere perception of a number of letters, mere presence together 
 in consciousness, was useless for the purpose of associating and 
 learning; that unless they were able systematically to go 
 through the series at a moderate speed and with a fair amount 
 of attention, no abiding impression was left on the memory. 
 And the remark suggested by these observations is that those 
 factors which form the real constructive basis of associations 
 are such as are usually recognised merely as helps and supports 
 to a connection which is due to simultaneous apperception. 
 Interest, Attention, Repetition are recognised, though only as 
 modifying factors, as influences which contribute to the efficacy 
 of contiguity. But it is obvious that Interest in itself has no 
 necessary connection with and is not implied in contiguity: 
 Attention again is a factor which as we have seen is of great 
 importance in the processes of associating and recollecting 
 ideas; but that too is not implied in the description commonly 
 given of the nature of association by contiguity. And, so far as 
 the need of Repetition, Exercise, Habit for the formation of 
 firm associative connections is emphasised, it is admitted that 
 mere simultaneity is of little avail. It may well be that the 
 method of entering into one’s breast and observing what 
 happens in the play of ideas whose connections have already 
 been formed, has been the reason why a description which 
 applies to so many of our actual associations is taken as a 
 statement of the law of their formation and growth: that 
 description is here taken for explanation seems in any case to 
 be the fact. Ideas which are associated have very often been 
 contiguous;—on the other hand ideas which have been merely 
 contiguous are in the vast majority of cases not associated. 
 Contiguity is indeed a formal characteristic which associated 
 ideas possess : it seems necessary that ideas should form parts of 
 one conscious state before they can be associated. It is however 
 in these dynamic factors, Interest, Attention, Repetition, that 
 the real causes, the formative influences, are found; it is this 
 which should form the matter of any law intended to state the 
 nature of the processes by which our actual assqciative connec¬ 
 tions are built up. 
 
 The second set of facts points to the need of emphasising, 
 in our account of Association by Contiguity, the part played by 
 the agency of the motor system. That habitual activities, or 
 any acquired physical ability, illustrate the operation of con¬ 
 tiguity has long been recognised. But in order to understand 
 the formation of associative connections in the more purely 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 71 
 
 sensory or intellectual sphere, it seems necessary here also to 
 seek an explanation in the functions of the motor mechanism. 
 How that operates may not be clear; the subject, whose 
 importance has been recognised perhaps most by Bain, Strieker 
 and Miinsterberg, has not been studied with sufficient exactness 
 either psychologically or physiologically. But this factor offers 
 itself without straining as a reason why visual objects presented 
 in succession should refuse to relate themselves as members of a 
 series when the help of articulatory innervations is withdrawn, 
 and why the reagents in these experiments insisted so strongly 
 on the necessity of going over the letters serially and systemati¬ 
 cally, and on the great hindrance caused by disturbance of the 
 vocal mechanism. These facts it would seem cannot be explained 
 except by supposing that, as purely sensory objects, the associa¬ 
 tive tendency of the presentations was much less strong than 
 where the reagent was left free to combine both sensory and 
 motor elements in the process of memorising. And, if this be 
 so, it will only be the psychical expression for the conviction 
 which is gaining ground in Physiology that there are no purely 
 sensory nervous centres, but that what bears this name really 
 subserves a double function. 
 
 Miinsterberg, in the paper from which I have already 
 quoted, states that no special or predominant type of in¬ 
 correct order was detected in the answers which were given 
 in his experiments. Bolton 1 , experimenting on school-children 
 to whom he read over lines of five to eight figures, found 
 that among the errors that of inversion was prominent. 
 Errors of inversion were not infrequent in the experiments 
 already described; still the number was not at all remarkable. 
 Another type of error which occurred fairly often is more inter¬ 
 esting ; it consisted in letters changing their places in the lines 
 so that, for example, the first letter of one line would take the 
 place of the corresponding letter in another, this second letter 
 then occupying the place of the first; in another instance the 
 first letter of the last line is followed by the second of the 
 middle line, by the third of the last line, and then by the 
 fourth of the middle line. Whether this is to be interpreted 
 as a disarrangement of a visual picture it is impossible in the 
 absence of direct experiments to say. The question still waits 
 solution, what sort of error is committed when each element in 
 memory, visual sensations, auditory sensations and so on, is 
 isolated and its capabilities tested. To such a task the method 
 applied in this investigation seems specially fitted. 
 
 1 The Growth of Memory in Schoolchildren. Am. Journ. of Psych ., 
 
 v. p. 363. 
 
72 
 
 W. G. SMITH : 
 
 The number of cases where similar letters were confused was 
 too great to allow of the supposition that the confusion was due 
 to accident: it was however not great enough to allow of a 
 definite account of this tendency. Letters similar in sound, 
 such as m and n, or t and d, were interchanged: similarity in 
 shape seemed to be the reason why h and k , u and v were 
 confused. It may well be that with such letters as t and d, the 
 reason of the interchange lay also in the easy transition from 
 the one position of the vocal organ to the other. What letters 
 are most liable to confusion and what disappear most readily, 
 together with the relation of these facts to the phenomena of 
 aphasia 1 , are matters of interest, but can be studied only by 
 extended and precise experiment. A series of experiments was 
 begun to test what effect audible pronunciation of the letters to 
 be learned had upon the memory: it was however only carried 
 far enough to shew that probably very great individual differ¬ 
 ences exist in regard to the help or hindrance which such 
 express articulation occasions in the effort to memorise. 
 
 It is plain that one great drawback in the results gained 
 first by Ebbinghaus, and next by Muller and Schumann, in their 
 investigations on memory, is that the work of memory is 
 formulated in terms of time only, and that a qualitative 
 analysis of memory is still to seek. It is obvious that 
 results into which visual, auditory, motor and other elements 
 enter as unknown, though probably relatively constant factors, 
 can form only the first step in an account of memory. W T e 
 have still, as was pointed out before, to find out the capabilities 
 of these various factors and to shew further what changes 
 they undergo in the gradual decay of memory. A few observa¬ 
 tions regarding the various factors may be adduced from the 
 foregoing experiments. In the experiments of the last group 
 the memory image seemed to be a somewhat indefinite complex 
 of various factors: as the distracting activity became more 
 difficult, and the aid given by articulatory innervations grew 
 less, the observer had to depend more and more on the visual 
 consciousness. Auditory ideas, as is intelligible, came little 
 into play: they were indeed said to be present, but it is 
 doubtful whether sensations connected with the organ of 
 speech were not mistaken for them, a confusion which seems 
 particularly easy when the two elements are present only in 
 imagination or memory. Motor elements seemed to be present 
 in the process of learning wherever they were not expressly 
 excluded: it was they that lasted longest and enabled the 
 
 1 It might be possible also to establish some relation between such 
 changes and the philological facts embodied in “ Grimm’s Law.” 
 
THE RELATION OF ATTENTION TO MEMORY. 
 
 73 
 
 reagent to recollect when the other sensory images failed. It 
 is not without interest to note the different modes of reproduc¬ 
 tion of these elements. Whereas with the sensory memory 
 the reagent seemed expressly to perceive and recognise each 
 successive idea that appeared in consciousness, the process in 
 motor memory was automatic or mechanical. The reagent 
 pronounced the letters because an impulse to speak presented 
 itself; often the appropriate movements came one after the other 
 without thought or reflection, and without their being recognised 
 as forming the required elements in the complex that was to 
 be recollected 1 . In some cases what was retained in memory 
 was only a vague general idea of the geometrical shape of the 
 letters—their curves or straight lines—which on reflection might 
 become more definite: in other instances there seemed to be a 
 sort of vague idea of the spatial arrangement of the letters 
 without any definite observable images being present. It often 
 happened that, when the reagent believed himself to be in full 
 possession of a line and ready to repeat it perfectly, he was yet 
 unable when the time came to reproduce any part of it: to his 
 surprise the line simply disappeared. Apparently this often 
 happened while he was engaged in repeating the first letters: 
 during this effort the rest of the memory image seemed to be 
 driven out of consciousness. What the conditions of such a 
 belief and its subsequent effects are, could not be exactly 
 determined: according to one observer K. the basis was a 
 Gesammtvorstellung, a sort of all-embracing complex idea in 
 which the parts have an indefinitely felt unity. 
 
 1 Observations of similar phenomena are given by Muller and Schumann 4 
 loc. cit. §§ 23 ff.