BF 321 .C6 Copy 2 VISUAL IMAGERY AND ATTENTION AN ANALYTICAL STUDY BY HELEN CLARK A. B. Vassar College, 1913 THESIS Submitted in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY IN PSYCHOLOGY IN THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS 1916 # -3>\ o c*t^ ^ 1921 £ CONTENTS I. Introduction 1 II. Experimental: An Analysis of Visual Imagery. ...... 4 III. Experimental: Imagery, Ocular Movement, and Attention 14 Experiment 1 14 Experiment 2 20 Experiment 3 23 Experiment 4 36 IV. Summary and Conclusions 42 VISUAL IMAGERY AND ATTENTION: AN ANALY- TICAL STUDY 1 By Helen Clark I. Introduction Ever since the appearance of Galton's " Inquiries into Human Faculty," 2 the subject of imagery has received, almost constant attention from psychologists, and, moreover, its culti- vation has been strongly marked by an increasing diversity of new problems. It is not necessary to pass the general literature of the subject in review; but we shall briefly men- tion those phases of it which are most closely related to the present study. First of all, we must note that the current tendency to view mind in its historical aspects has prompted the psychologist to seek for the antecedents of the image; to discover, that is to say, its dependence upon perception. Experimentalists now generally " incline to the view that there is no intrinsic difference between sensation and image." 3 Investigations in the field of suggestion reveal the fact that, under certain circumstances, simple or total images may be mistaken for sensations or perceptions. 4 On the other hand, " a visual perception of distinctly supraliminal value may . pass — even with specially trained observers — for an image." 5 Related studies have dealt with the development and the decay of imaginal complexes in the course of time. The associationists' conception of stable psychical enti- ties has been discarded, and the view commonly held at present is that " the memory is not to be regarded as a store- house of perfectly conserved images, but that the most simple memories are continually exposed to change, and that it is, at times, only by the combination of various memorial re- 1 From the Psychological Laboratory of the University of Illinois. 2 Galton, F., Inquiries into Human Faculty and its Development, 1883. 3 Perky, C. W., An experimental study of imagination, Amer. J. of Psychol, 1910, xxi, 435. Cf. Kiilpe, O., Philos. Stud., xix, 5o8ff., and Washburn, M. F., Mind, n. s., viii, 1899, 32. 4 Cf . Seashore, C. E., Measurements of illusions and hallucinations in normal life, Studies from the Yale Psychological Laboratory, 1893- 97, i-v, 32. 5 Perky, op. cit., p. 435. S 462 CLARK sources that retention is made definite and exact." 6 Experi- mental and theoretical inquiries have also been directed toward the general functions of the image in relation to thought and meaning, 7 discrimination and judgment, 8 and reproduction and recognition. 9 Moreover, there have been several attempts made at classification of total images on a basis of divers char- acteristics, usually functional. Visual images have here re- ceived most consideration. Four important investigations which have recently appeared are those of Perky, 10 L. J. Mar- tin, 11 K. Koffka, 1 ^ and R. M. Ogden. 13 And finally, images have been studied in their relations to motor phenomena, espe- cially to ocular movement. 141 The problems and methods in Ogden's and in Perky's in- vestigations closely resemble those of our own experiments. There are, however, three important differences, (i) Both Ogden and Perky discovered a number of intermediate or " equivocal " images 15 which did not seem to belong either among their " images of imagination " or with their " images of memory." We decided to include in our classification all non-verbal (or primary) images which were reported by our observers. (2) As regards ocular movement, Ogden obtained negative results (p. 381) ; while Perky, by means of another procedure, discovered a striking correspondence between the kind of image and the occurrence of ocular movements (pp. 6 Bentley, M., The memory image and its qualitative fidelity, Amer. J. of Psychol., 1899, xi, 47L Cf. Philippe, J., L'image mentale; evolu- tion et dissolution. Paris, 1903, 76ff, u6f. 7 Cf . Betts, G. H., The distribution and functions of mental imagery. Teachers College, Columbia University, 1909, p. 41. 8 Cf . Whipple, G. M., An analytical study of the memory image and the process of judgment in the discrimination of clangs and tones, Amer. J. of Psychol., 1901, xii, 409ff. 9 Bentley, M., op. cit., 46. 10 Op. cit. 11 Die Projecktionsmethode und die Lokalisation visueller und an- derer Vorstellungsbilder, Zsch. f. Psychol, 1913, lxi, 321. 12 Zur Analyse der Vorstellungen und ihrer Gesetze, Leipzig, 1912. 13 "Experimental criteria for differentiating memory and imagination in projected visual images, Psychol. Rev., 1913, xx, 378. < 14 Perky, op. cit., and Ogden, op. cit. Cf. Meakin, F., Mutual inhibi- tion of memory images, Harvard Psychological Studies, i, 1903, 244; Moore, C. S., Control of the memory image, ibid., 296; Slaughter, J. W., Behavior of mental images, Amer. J. of Psychol., xiii, 1902, 548; Kuhlmann, F., Analysis of the memory consciousness, Psychol. Rev., xiii, 1906, 338f ; Murray, E., Peripheral and central factors in memory images, Amer. J. of Psychol., xvii, 1906, 241 ; Kiilpe, Outlines, 1909, 187; Burtt, H. E., Factors which influence the arousal of the primary visual memory image, Amer. J. of Psychol, xxvii, 1916, 87. 15 Ogden, p. 406; Perky, 436. , ' ■ VISUAL IMAGERY AND ATTENTION 463 442, 451). We decided to observe movements of the eyes by three different methods, one of them being somewhat simi- lar to Perky's. (3) We proposed to discover, if possible, the factor or factors most closely connected with ocular move- ments, not simply the sort of image which was usually involved. 16 Before proceeding to the account of our experiments, we must observe that the term " image " is ambiguous. As it is used by psychologists to-day, it stands for (1) a mental pro- cess, (2) a group or constellation of mental processes, (3) a kind of reference or meaning, and (4) a particular sort or class of objects. As regards the first two meanings, we shall distinguish the "simple image" (a single imaginal process taken at the level of sensation and affection) from the " total image " (the single imaginal constellation or complex) ; and so far as the functional treatment of imagery is concerned, — the treatment which discovers such marks as " particularity," " locality," and " personal reference," — we shall try to make clear in the context when we are dealing with meanings and objects and when we intend to indicate processes. The attempt is worth while because it is still frequently assumed, even by serious writers, that the imaged object is a " mental " object and so ultimately different for the psychologist from the physical objects of perception. II. Experimental: An Analysis of Visual Imagery Problem and method. Our first investigation was an analy- tical study of the visual images aroused by a series of words. The purpose of the experiment was to obtain a descriptive account of the images evoked with respect to twelve different characteristics, to classify the images with reference to two of these peculiarities, and to discover whether there was a resemblance in other details among the images of each group. There were two observers, A and B. 17 was seated in a dark room facing a black wall. The experimenter took her place behind the observer at a table provided with a shaded light for the recording 36 Our observers, named in the order in which they served in the experiments, were : an advanced undergraduate — Miss B. V. Copley (A) ; two graduate students— Mr. C. E. Holley (B), and Miss A. S. Rogers (C) ; three instructors in psychology — Dr. C. A. Ruckmich (D), Dr, J. E. DeCamp (E), and Dr. C. Rahn (F). The author wishes to express her grateful appreciation of their services and to acknowledge her especial indebtedness to Professor Bentley, under whose supervision the investigation was pursued. 17 " o " has been used in the following pages as an abbreviation for "observer." ; . 464 CLARK of O's descriptions. The experimenter transcribed O's account as O dictated. For evoking the images, 114 substantives were used with A; with B, 75 words taken from the same list. 18 The instructions were as follows : " After a signal ' ready,' I shall pronounce a word. You are to note any visual image that may occur. As soon as it has dis- appeared, give a full description of it." It is to be observed that the instructions were quite general. The subject was not required to call up an image, but merely to note any that should occur. Furthermore, the image was not supposed to be that of a particular object, but simply any visual image. The subject was not asked voluntarily to hold the image, which was to take its natural course. The description was not interrupted; but, when O had finished his account, he was questioned as to any points which he had overlooked. The experimenter sought complete information concerning (1) spe- cificity of reference ("particularity"), (2) familiarity, (3) the posi- tion of the imaged object in relation to O, (4) the visual background, (5) the size of the imaged object, (6) its stability, (7) the number and the degree of clearness of its details, (8) color, (9) accompanying affective tone, (10) somatic reference, (11) associative processes, and (.12) temporal reference. Results. The images reported were first divided into three main groups. Those which referred to specified objects and which were apprehended as familiar were called F-images. Memory images (those which not only referred to specified objects and were apprehended as familiar, but which also involved a reference to a particular place and to O's past ex- perience), were considered as a sub-class of the F-images. Images which were specific and unfamiliar were called U- images. These were not necessarily accompanied by a feeling of strangeness ; there was, in most cases, merely absence of familiarity. Finally, images of general reference (signifying a class or a member of a class of objects), which also involved no familiarity, were designated as G-images. 19 With regard to position, the imaged objects were classified as usual when they had the same position in relation to the observer as the perceptual objects commonly occupied; they were called unusual if the position was not characteristic of the perceptual object, or if the imagined object seemed detached from any setting, or if the localization was extremely indefinite. Images were classified as unstable if they fluctuated, faded quickly, or were soon replaced by other images. When describing the 18 Some of the words were : knife, river, dog, church, soap, cloud, grass, king, lyre, Cerberus, druid, tournament, King Arthur, wand, nightingale, baron, Marmion. 19 Both the generality of reference and the definiteness of spatial orientation we found, as did Koffka, to present a large number of degrees and gradations. These differences seem, however, to be unim- portant for our main purpose. VISUAL IMAGERY AND ATTENTION 465 details of images as many or few, the experimenter took into consideration the amount of detail which corresponding objects revealed in perception. Color was described as absent if color was normally present in the perceptual mode of the object. We took associative processes to mean relevant processes not completely assimilated to the total image. Such associative processes might occur at the same time as the principal image, although they were more apt to appear subsequently. Where the corresponding perception had occurred at some time within a year, the image was classed as recent. Somatic reference 20 was present in several forms. In some cases, the observer simply was at the place presented in imaginal form. " The church was in the position in which it would be if I were walking up to it." "Then the image changed, as I went inside." "I was there. I seemed to be younger than I am now." " I was lying on my back and looking at the clouds." Some images stood in definite relation to certain parts of the body. " The gun was parallel to my arm as though my arm were extended." "The bird was back of my arm at the left. I thought of turning my head and looking at it." " The shoes were on my own feet." In some cases, the somatic reference was more explicit. " I imagined myself as there." " I seemed to be coming along the walk." " I knew that I was there." In one case, the subject had a definite visual image of herself. " I saw myself take the key out of the rubber where it is kept." 20 The word " somatic " is used to avoid the ambiguous term " per- sonal." Perky uses, without definition (p. 436), "personal reference" as one of her two cardinal distinctions between " images of memory " and " images of imagination." We mean by " somatic reference " that the observer apprehended the imaged object as related to his own per- ceived or imaged body. The words " personal reference " might have a similar significance; but they might also imply a reference to the ob- server's past experience. Our own analyses seem to indicate that " par- ticularity," Perky's second cardinal distinction, is — even when taken with the first — an unsatisfactory mark of memory. We found many particular and specified objects which could not, save by v the most arbi- trary designation, be called " memories." Similar criticism upon these terms has been passed by Koffka (p. 226) and Ogden (p. 384ff). Perky's determination to study " imagination " appears to have in- duced her to accept, without refined analysis, a " rough and ready cri- terion " of the simpler imaginal complexes. Although she uses this criterion "for preliminary purposes" (p. 451), it nevertheless deter- mines her whole conception of the simpler processes and modes of integration by which, as she assumes, "imaginations" are built up. 466 CLARK All of these functional distinctions are, of course, more or less crude. They should, however, suffice to reveal any strik- ing dissimilarities existing among our classes. Table I shows the distribution of the different kinds of images for each observer. TABLE I Observers F-i mages G-images U-images Totals Primary Primary Secondary Primary A 36 61 43 140 B.. 36 12 7 4 59 Totals 72 73 7 47 199 Most of the visual images were primary (non-verbal), but some of B's images were secondary or symbolic (verbal). The 7 secondary visual images mentioned in the table appeared as independent total images. They are not included in the following discussion of results unless specifically mentioned. Table II shows a more detailed analysis of the images. As regards the temporal relations of the F-images, over half (53%' an d 61%) were referred to objects seen habitually and recently (within a year), while more than a third (36% and 39%) referred to objects perceived on some recent and specific occasion. In 11% of A's F-images there was recognition without definite temporal relations. Under three of the prin- cipal headings of classification there seemed to be little or no correspondence between the kind of imagery and a particular characteristic. These headings were color, stability, and size. 1. Color was present in all but six images. Three of these were F-images, one was a G-image, and one was a U-image. 2. Stability was characteristic of 61% of A's F-images, but only 39% of B's F-images were stable. Of B's G-images 83% were unstable. In the other cases, the percentages are not significant. 3. Normal size was reported in from 70% to 100% of the cases in each group. In the case of A, the G-images char- acterized by abnormal size were three times as numerous as the corresponding F-images, while the abnormal U-images were four times as numerous as the F-images. Of B's images, on the other hand, the F's were the only ones that were ever characterized by abnormal size. These reports, however, may VISUAL IMAGERY AND ATTENTION 467 Si gs Id UlB}J33Un -=* | ; | j M | ■; | : | Dypads — ;u3D3jj co CD CO -■+ cd CO ;| ; [Bn^iqEH — }U903}I cd co LO CM CM H CD £c/> 38 Sig pjOAi sni O 00 CM t*. SUI5J3ET co CM CD CO CD CO CM LO LO 00 t> oo LO ■* co CO CM 8 juasajj CO CD CO co 1—1 CD CO CD LO CO LO CM CD CM CD CM o LO a SuijpBq CO CD CO <* cd co LO CD 00 3 CD CO CD 00 CM o }U3S3Jj CO CM CD CM CM CD t> H - 00 CD -* CM o LO ea as b < !UBSB3ldUQ CD O -f co CM CM 00 CM IBj^naN «* £ CD r-l CO LO o CM CO CO CM 8 rf 01 CO LO ^uBseau CO CM CM r-l CM CM CO CD - LO CM « O .J o U SuiipBq CO CO - CM - CM H LO CM }U3S3JJ CD CO O o CO CO CO cd 8 00 CD CM 8 CM 00 CD CO I" J OS snSEA — ^^I CO 00 CO CD CO 00 CD r-l CM rH CM CD CO CD - LO CM juaj^—AVSjj ■<* 1—1 t> CD CO CM 00 CO CD ^* anSu^— ^ubj^ CO 00 LO ^H H CM - CO CO c- CM O LO JB3J3 AlIB]A[ CD CM CM t> £ CO CD CO CD CM o CD ^ LO CM 1 3[qB}sun "tf co CM CM CD CM CO CM LO o fH CO CO CM CD (M O LO aiqB^s CM CM CD -# CD CO CD CM 00 CM t> CM CM LO CM o LO c73 P3JJBMQ * CO LO 00 £ CD CM JBUUOfyf CO CO CD to co 05 CD LO CM o o O CO B «* o o 1— ( p3}BJ3SSEXg CO 00 O i-( CD CM LO Q I O M u 3U3DS ■ IB3 I3 CM CD £ CO o 0-3 CM ^ LO CM 3U30S 3n3B^ 00 CM CM o CM CD LO O 00 CM H 00 CD «* >* - LO CM szBfj — apBiJS — JO I°3 eg CD N CD 00 CM CD CM LO Sux^DBq -* s CO 00 CM H O CM a CM CD CM 00 CM CM 8 IS IBnsnufi ** a ^ CO CM CD CO 00 CD C^ CD - LO CM IBnsn CM co cd oo LO CO CD CO CD ■* CO CO CD CO 8 CO LO t> : 6 Z £S 6 2 £ 6 Z £§ 6 2 £ 6 Z ^5 d z £ < | n | < | w | < | n s aov wi- * ■ aov WI .i D S aov wi-j "1 468 CLARK be significant when considered in another connection. 21 Table III shows that it was usually the images of large objects that were under-sized and also those of small objects that were over-sized. TABLE III Images of Exaggerated Dwarfed Small objects 10 1 2 Large objects 3 16 The small objects which were exaggerated when imaged were a picture of Queen Elizabeth, a pin, a lemon, a key, a cake of soap, a dinner favor, a tomato, a maple leaf, and two flies. The image of a snake was under-sized. The two medium-sized objects which were exaggerated in imagery were a coat and a bag. Of the large objects, a king, a giraffe, and Ceres were exaggerated, while a tower, a valley, a castle, a horse, a temple, a tabernacle, King Arthur, a crusader, a baron, a serf, Marmion, the Delphic oracle, the Nile, and three forest scenes were under-sized when imaged. With 15 of the 17 images that were dwarfed there was a visual background. With 11 of the 15 that were exaggerated there was no visual background or only a shade or haze, while in 2 more cases the scene was vague. Objects of a certain size are probably perceived with less effort than objects or scenes considerably larger or smaller. Perception of the small objects may involve unpleasant mus- cular strain resulting from the convergence of the eyes. This is especially true when the object is close to the observer. Large objects or scenes necessitate considerable ocular move- ment if all the different details are to be clearly perceived. In some cases, images seem to show a tendency to take that size which, in actual perception, would involve the least ocular strain. Although the three kinds of images did not present marked differences as regards color, stability, and size, certain other peculiarities were more characteristic of one sort of imaged object than of another. We distinguish six cases. 21 J. M. Baldwin and W. J. Shaw (Memory for square size, Psychol. Rev., ii, 1895, 236) found that the size of comparatively small objects was over-estimated. H. K. Wolfe (Some judgments on the size of familiar objects, Amer. J. of Psychol, 1898. ix, 137) discovered that familiar objects (e. g., bank-notes) were under-estimated when drawn on paper. In criticism of this experiment it has been said that " this fact does not necessarily report the visual image of the note,, since reproduction might easily be biased by the introduction of alien senses, i. e., pressure, muscular exertion, etc." (M. Bentley, The memory image and its qualitative fidelity, Amer. J. of Psychol., 1899, xi, 1.) VISUAL IMAGERY AND ATTENTION 469 i. The F-images occupied a "usual" position (89% and 97%) ; the U-images not quite so frequently (84% and 75%). A's G-images were less apt to occupy a usual position (61%), and, in the case of B, the position was even less likely to be usual (33%). 2. The visual background in 83%' and 87% of the F-images and in 67% and 50% of the U-images was a scene. On the other hand, there were scenic backgrounds in only 35% and 8% of the G-images. 3. The details of the images were both many and clear in 72% (A) and 31% (B) of the F's and in 60% (A) and 25% (B) of the U's. The details were few and vague in 29% (A) and 92% (B) of the G-images. 4. Somatic reference occurred with 64% and 61% of the F-images, but with only 11% and 8% of the G-images and with 14% and 50% of the U-images. 5. Associative processes were present with a considerable number of U-images (67% and 50%), with a smaller number of F-images (36% and 36%), and with still fewer G-images (15% and 25%). 6. As regards affective tone, F-images were often accom- panied by pleasantness (72% and 47%), U-images were less frequently associated with pleasantness (63% and 25%), while with G-images consciousness was even less often pleas- ant (44%' of A's images) and might be predominantly neutral in tone (100% of B's images). B reported no cases of un- pleasant affective tone. Of A's 140 images, 32 were connected with unpleasantness. In 17 of these cases, however, the un- pleasantness might be explained by the nature of the object imaged, while in 6 other cases the unpleasant tone is to be referred to associative connections. Only two images involved a definite feeling of strangeness or unfamiliarity. One of these was accompanied by pleasantness, the other by unpleas- antness. These results would seem to show that images re- ferring to agreeable experiences are more apt to be recalled than are those referring to the disagreeable. 22 22 The belief that there is a tendency to forget the disagreeable is held by Colegrove (Memory, 1900, n), Freud (Zur Psychopathologie des Alltagslebens, 1907), and Hollingworth (The oblivescence of the disagreeable, /. of Phil, Psychol, etc., 1910, vii, 709). This view is questioned by E. N. Henderson (Do we forget the disagreeable? ibid., 191 1, viii, 432), who found that one-third of the memories of ten observers were disagreeable. He considers it possible that the average person actually has more pleasant than disagreeable experiences, and that an incident which is remembered with pleasure might really have involved unpleasantness when it occurred. " We forget not so much disagreeable ideas as useless ideas." 470 CLARK A certain combination of characteristics might be regarded as peculiar to the pure type of each sort of image. The pure F-image may be distinguished by its usual position, its scenic background, many clear details, a pleasant affective tone, somatic reference, and associative processes. The pure G- image is in an unusual position and has no background, or merely a shade or haze as background, few and vague details, a neutral affective tone, no somatic reference, and no asso- ciative processes. Like the F-image, the pure U-image has a usual position, a scenic background, many clear details, a pleasant affective tone, and associative processes; but it has no somatic reference. If, however, one attempts to classify the individual images under these detailed headings, it is evi- dent that there are very few examples of the pure types and many deviations from them. Table IV summarizes the variations of the images from the pure types. Only 17 of the whole number are " pure " under our definitions, while 48 vary from the " type " in one, and 68 in two particulars. Transition from one class to an- other was shown in various ways. Let us illustrate. In con- nection with one image there was a feeling of familiarity, but the observer was unable to place it in any spatial or temporal relations. In other cases there was no recognition, though scenes associatively appeared which the observer thought might possibly have been the original settings of the object. In four cases recognition occurred after the image had begun to fade. vSeveral G-images and one U-image had as back- grounds scenes that were recognized. Once the image re- mained the same while one scenic background was replaced by another. In 44 cases the pronunciation of the word was TABLE IV A's Images B's Images F-Im. G-Im. U-Im. F-Im. G-Im. U-Im. 4 1 7 1 4 Variant in 1 characteristic 11 13 10 9 5 2 characteristics 14 16 15 19 3 1 3 characteristics 4 15 10 4 1 4 characteristics 1 13 1 23 2 5 characteristics 1 3 6 characteristics 1 1 VISUAL IMAGERY AND ATTENTION 471 followed by two successive total-images. Of these successions nine were from one G-image to another of the same sort, nine were from an F-image to another F-image, and nine were from a G-image to a representative of the F class. In 13 cases, images from other sense departments were assimilated to the total image. Five of these were temperature-tactual images, 5 were kinaesthetic-tactual, 2 were olfactory and one was auditory. So far as it is permissible to base general conclusions upon our material, these results seem to indicate that, although visual images may be classified according to their genesis or function, there are no definite groups which show peculiar pat- ternings of the ten characteristics, intrinsic and functional, which we have mentioned. Individual differences are consid- erable, and intermediate images are very numerous. F-images, G-images, or U-images may, perhaps, best be considered as designating various points in a developmental or an elaborative series, rather than sharply differentiated classes. III. Experimental: Imagery, Ocular Movement and Attention Experiment 1 Problem and method. The purpose of this experiment was to discover whether there exists any connection between the general character and functions of the visual image and the occurrence and extent of ocular movements. A and B again served as observers. The observer was seated in the dark room with his head fixed in a head-rest. A sheet of blank cardboard was placed at the distance of a meter from his eyes so that a white fixation-point appeared opposite the fovea of the right eye. The experimenter, who was seated at the observer's right, observed the latter's eye through a hand lens. A black box with a very narrow slit in one side was so arranged that the light from an electric bulb inside produced a bright point of light upon the cornea. The posi- tion of the box was shifted until the point of light fell upon the iris of O's eye near the edge of the pupil. When the eye shifted, the margin of the pupil moved noticeably nearer to or farther away from the dot of light, which remained approxi- mately stationary. For the practice series, a sheet of black cardboard was used with a horizontal and a vertical line cross- ing at right angles. Distances of 5 cm were laid off as units along each line. O looked at the fixation-point, then at some point along one of the lines, then again at the fixation-point. After practice, the experimenter was able to tell, with approxi- 472 CLARK mate accuracy, from the appearance of the eye, over how great a distance upon the cardboard the observer's glance had passed in a vertical or a horizontal direction. From the list used in the experiments of Chapter II, we selected the words which had aroused images differing but slightly from the pure types. The selection was made so that the proportion of images in each of the three classes should be approximately the same. Sixty-nine words 23 were pronounced for A, while only the first 35 were used for B. The instructions were those of the preceding experiments with the following additions : " You are to fixate the white dot upon the black screen. When a visual image appears, try to keep it constant, but, if it dis- appears, do not make an effort to recall it. You are to start the stop-watch when the image first appears, and to stop it when the image finally disappears." After the observer had given his account of the image, he was questioned as to any points which he had overlooked. The experimenter sought complete information concerning (i) the visual background of the imaged object, (2) its size and (3) details, (4) somatic reference, and (5) accompanying kinaesthetic processes. Results. Table V gives a general tabulation of the images of observers A and B. Some visual-verbal images occurred, but they are not included in these results. Reports are not included under every heading for every image, since in some cases the image would shift, or O would be uncer- tain as to an introspective item. The percentages are based upon the numbers tabulated. As regards the nature of the different sorts of image, these results are, on the whole, in accord with those of the experi- ments already recorded. F-images and U-images generally had a scenic background, while, with G-images, an imaged visual background was either only a haze or was altogether lacking. In F-images the details were usually numerous and clear. This was less often the case with U-images. G-images were more apt to have few and vague details than were those of either of the other groups. Somatic reference was char- acteristic of F-images, while it did not seem to be present with the other kinds. In this experiment, U-images seemed 23 The list of words was as follows : sword, soap, ogre, leaf, ghost, orange, nymph, broom, pin, oracle, cat, tomato, druid, pipe, catacombs, horse, key, hamlet, convent, church, Marmion, lyre, snake, banana, ice, centaur, aeroplane, mirror, child, knife, serf, dungeon, stone, gun, tournament, boat, cloak, river, w r agon, nail, tiger, helmet, coin, witch, moat, grave, wand, rose, dog, coat, cloud, ring, lion, grass, lemon, sled, hand, bag, chimney, purse, griffon, temple, shoe, Holy Grail, vase, Jupiter, sultan, shrine, Nile. VISUAL IMAGERY AND ATTENTION 473 ill ^uasajj o CM •* ^ - CO CM so CO SujjpBq LO CM to LO co en to co co o td S W H H < (■urji) saSetui -tf 05 o o SUOI}BSU9S J3tl^O - CM o o uiej^s -iEinoo t> CD CO s ">* CM SuijpB'i CM CO CO o LO CM CD oo 3 co IS SuiJpBq t> co I— 1 CO o o LO o o 1U9S9JJ co CO 2 o o ' ■J < H W Q 9n3B^\ M9j[ CS1 LO CM 05 CO LO 00 CM JB9J3— M9J 1* o T* CO - LO 9nSBA — Aue]/^ o CM CO Tt< CM CM JB9J3 — Auep\[ co LO CO CO Cxi 00 ■xt > w N <35 p9ZIS"J9pun i-H cm CO 3 co < p9}BJ983BXg o o - CO Q Z J g < o en O < 9U9DS 00 ^ CO CO B 9ZBH— 9peqs— Jopo o CO CM o Sui^pBq - cm o o CO o A^UQ U99JDS 3pB[£[ o Tf CM LO 00 CM s33euii jo aaquinjsj LO CO 00 00 CO 1 &5 6 ^ 6 £ a (I < CO w u < s 6 & c < 2 ' < H o H 474 CLARK the most likely to be exaggerated or dwarfed. This was less often the case with G-images and it seldom occurred with F-images. Ocular movements and sensations of ocular strain seldom occurred in connection with G-images, but were fre- quently noticed with the other varieties. Table VI shows the number and the extent of ocular movements. Although these TABLE VI Average number of movements per image Average extent of movement (in cm.) Right to Left Left to Right Up or Down Total No. Ex- tent Total No. Ex- tent Total No. Ex- tent F-images 1.84 5.64±2.9 13 4.66 13 7.11 8 3.75 U-images 1.33 5.15±2.4 6 3.33 9 6.44 1 2.5 movements were more frequently present with U-images than with F-images, the average number of movements per image was less and the extent was less in the former case than in the latter. " Extent " refers to the distance, as on the card- board screen, over which the glance of the observer was esti- mated to have passed, not to the amount of movement of the eye-ball. Complete data concerning the duration of the images cannot be given, since 18 of the 33 G-images were immediately followed by F-images. In such cases, the duration of each could not be accurately determined. In 52 instances, how- ever, the time-interval could be ascertained. These results are shown in Table VII. F-images and U-images were, in gen- eral, longer than G-images. The significance of the figures is dubious, however, because of the wide variations in time. TABLE VII Number Average duration M. V. 23 24.0 sec. 13.6 35 10.8 sec. 8.1 14 26.8 sec. 12.0 The fact that F-images and U-images are longer than G-images might account in some measure for the occurrence of ocular move- ments with the two former classes and their absence in other cases. The eyes might tend to move because there was long fixation, regard- less of the nature of the accompanying image. The time-interval VISUAL IMAGERY AND ATTENTION 475 could not, however, be the sole factor involved, for ocular movements occurred with some very short F- and U-images, and did not occur with G-images which were considerably longer. Ocular movements occurred toward the end of 6 images which lasted longer than 15 sec, but in 3 cases other movements had also occurred toward the beginning of the period. The results of this experiment seem to indicate that some relation obtains between ocular movement and different sorts of image. The existence of such a relation may seem strange in view of the results of our earlier experiment, which showed that, when grouped according to characteristics, there are few images of any pure type, each class tending to merge into the others. But the correspondence may be due to the fact that the amount of ocular movement accompanying any visual image is related to its general setting, rather than to the definite patterning of characteristics, which, as we have seen, varies considerably within the groups. It is possible that different characteristic motor attitudes or general conditions of the organism occur with F-images, G-images, and U-images, and that these attitudes involve certain reactions upon the ocular muscles. The fact that kinaesthetic sensations or images were reported in a comparatively small number of cases would not necessarily invalidate such a conclusion. The attention of the observer was directed toward the visual image, and the effects of muscular strain or movement might easily have been overlooked. The experimenter purposely refrained from all unnecessary emphasis upon this aspect of the problem in order that conditions might be as natural as possible. Experiment 2 Problem and apparatus. It occurred to the writer that the results of Experiment 1 would be more significant if con- firmed by investigations involving an objective record of ocular movements. It also seemed advisable to try to find the particular factor or factors which were most closely connected with ocular movement, instead of resting content with the dis- covery of the association of such movement with certain kinds of imagery. These were our chief purposes in the present experiment. For recording ocular movement, we employed the writing tambour of the laryngeal recorder described by Krueger and Wirth 24 which was connected by rubber tubing to the nystagmograph of A. Schackwitz. 25 A record of the 24 Psychol. Stud., 1905-06, i, 103. 26 Zsch. f. Psychol, 1913, lxiii, 442. When the spectacle-frame to which the tambour was attached proved to be unsatisfactory, we re- moved the tambour and fastened it to a support made of a system of adjustable rods and attached to the head-rest 476 CLARK time in seconds was kept by means of an electric marker actu- ated by a metronome, or, in the later experiments, by a time- clock. A second pneumatic system was used to record the course of the image. For this purpose a rubber bulb was so fitted into a cardboard frame that O could produce variations in the height of the recording line by changing the degree of piessure upon the bulb. Strong pressure meant a high degree of clearness in the imaginal processes. Method. C, D, and'E served as observers. The observer was seated with his eyes closed and his head fixed in a head- rest. The first pneumatic system was inflated, and the small tambour of the second system, adjusted to the head-rest, was placed against the right eye-lid. Movements of the eyes were thus automatically recorded beside the clearness-line upon the kymographic paper. The instructions were as follows : ''After a signal ' ready ' a word will be pronounced. If a visual image appears, try to hold it constant, but do not recall it when it disappears. With the right hand, compress the bulb when the image appears. Increase the pressure whenever the image grows clearer and diminish the pressure whenever the image becomes less clear. After the image has disappeared, you will be interrogated as to (i) shifts in attention from object to object, (2) changes in the clearness, the intensity, and the content of the imaginal object, (3) associative pro- cesses, (4) somatic reference, (5) the visual background of the image, (6) specificity of the imaged object and its tem- poral and spatial relations, and (7) familiarity." Sixty-three words were used with each of the three observers. 26 Results. Nothing of importance resulted from this experi- ment in respect to the characteristics of the different sorts of images. The descriptions are in general accord with the more detailed accounts discussed above. The results are sig- nificant, however, in that they show an appreciable corre- spondence between fluctuations in the clearness of the imagery and ocular movements. The records were divided into 5 classes according to the degree of correspondence. We de- scribed a record as showing no correspondence when one line fluctuated and the other did not, or when there were striking differences in the number and the position of fluctuations. Figure 1 shows typical examples of the records in which there was correspondence. Records I, 2, and 3 are samples of high correspondence, record 4 of close correspondence, while in 5 26 Some of the words were as follows : pickle, wizard, maypole. shell, moat, Hercules, shepherd, thimble, rainbow, scythe, commence- ment, dell, note-book, Scotchman, Cinderella, mouse. VISUAL IMAGERY AND ATTENTION 477 Fig. i 478 CLARK the resemblance of the lines is general, and in record 6 the correspondence is limited. Table VIII shows the classification of the results for each observer. TABLE VIII Amount of Correspondence None Some General Close High Totals c 17 8 24 6 10 65 D 17 21 9 5 8 60 E 10 19 7 7 4 47 Totals 44 48 40 18 22 172 128 In 128 of the 172 cases, there appeared, then, a correspond- ence between the records for fluctuations of the clearness of the image and the records for ocular movements. This agree- ment is the more striking in view of the fact that very slight variations in the adjustment of the apparatus or in the time of O's response in varying the pressure upon the recording bulb might have changed the records in such a way that a correspondence which really existed would have been obscured. Experiment 3 Problem and apparatus. In all of the preceding investiga- tions, the images evoked differed widely in their nature and in their origin. Some were very schematic, while others pre- sented many specifying details. A considerable number re- ferred to recent experiences, but some were related to habit- ual perceptions, and still others symbolized classes of objects. In view of these facts, it seemed advisable to investigate the occurrence of ocular movement in connection with images resulting from similar perceptions. We decided to employ a series of objects which were homogeneous except in respect to meaning, and to note whatever differences in ocular move- ment might occur when the objects were imaged. For this purpose, we devised a series of thirty sense and nonsense figures. Each figure or diagram was cut from white paper and pasted on a black card 5 inches square. Half of the pictures represented meaningful objects, while the others were intended to be meaningless. VISUAL IMAGERY AND ATTENTION 479 Ocular movements were detected by a method different from the methods used in our previous investigations. In Experiment i, changes in the position of the eye had been observed by the experimenter; in Experiment 2, a mechan- ical device had been employed. In the following series, the observer himself reported the occurrence of ocular movement. Our plan was to place before the observer a set of points of light arranged in such a way that, when fixation was steady, all of the lights would fall upon the blind-spot. A slight movement of the eye, however, would bring some of the bright points within the field of vision. 27 The apparatus (repre- sented in Figure 2) consisted of two black metal light-boxes designed for the investigation of indirect vision. The smaller light-box supplied the fixation-light and the larger box the lights for determining the limits of the blind-spot. Rotating discs upon the face of the larger box provided for a radial series of light-spots, 5 mm apart, along any one of 360 diameters. Method. C, D, and F served as observers. The observer was seated in a chair with his head fixed in a head-rest and his right eye covered with a black cloth. His left eye was at a distance of i}/ 2 meters from the face of the small box, 27 Perky {op. cit., 437) employed a similar method. Our arrange- ment provides for the detection of considerably smaller movements of the eyes and also for a nice control of the position and intensity of the light-spots. 480 CLARK and the point of light at the intersection of the two lines of holes appeared opposite the fovea. This dot served as a fixa- tion-point; the other openings were covered. Light was ad- mitted through apertures successively nearer to the center of the face of the large light-box until those points were found which cast beams of light just within the area of the blind- spot. As a check upon the accuracy of the adjustment, the slide and the disc on the small box were so moved as to admit light through apertures 5 mm above or below, and at the right or the left of the original fixation-point. When either of these four adjacent spots was fixated, some of the points of light on the large box became visible. An extremely slight change in the position of the eye, then, brought the lights partially within the field of vision. Before each experiment, the head-rest and the boxes were adjusted with reference to O's blind-spot, since a different arrangement was necessary for each observer. When O had taken his place in the dark room, he was given the following instructions : " You will be shown a series of four drawings, the experimenter naming each one as it appears. 28 Shortly after a ' ready ' signal, a drawing will be exposed for 20 seconds. After a further interval, of 20 seconds, another will be exposed, and so on." The drawings were placed over the fixation-point of the small box and were illuminated from above. During this part of the experiment, the apparatus was concealed by a black screen with an opening (2"xi/ 2 ") through which the card was visible. After the perceptual series, the observer was instructed as follows : " You are to fixate the single point of light. The experimenter will pro- nounce the names of the drawings, and you are to call up a visual image of each. Try to hold the image constant, but, if it disappears, make no effort to recall it. Press the key when the image appears and release and press it whenever the lights at the side become visible. Release the key and say ' now ' when the image disappears. Then give complete introspections for the period of the image." A kymographic record was made showing the appearance and disappearance of the lights as indicated by O. The experimenter timed each image with a stop-watch. During the last half of the series, each record was made for a period of 20 seconds, and the observer was required to recall the image if it disappeared during that time. Several other records were made as con- 28 The following names were given to the figures: wij, bottle, ked, butterfly, fak, turtle, ged, star, pam, spool, tis, daisy, rel, key, bam, fish, hof, ring, lat, apple, zum, Jeaf, doj, cup, mej, knife, dak, pipe, kif, scissors. VISUAL IMAGERY AND ATTENTION 481 trols with each observer. Under the controls, O was told to fixate the point of light until the experimenter gave a signal to stop at the end of 20 seconds. Under these conditions, no attempt was made to arouse imagery. After each experi- mental period O's descriptions were transcribed from his dictation. When necessary, the experimenter questioned him with respect to (1) the kinds of conscious processes integrated with the total visual image, (2) associative processes, (3) the clearness of the image, and (4) the accompanying affec- tive tone. O was then asked to sketch the object as perceived and also as imaged. Results. Although the drawings did not serve the purpose for which they were primarily intended, the results were sig- nificant in other respects. Most of the pictures devised for nonsense drawings either proved to be meaningful or were accompanied by meaningful associative processes. Only 7 of the 88 images had no meaning for the observer or signified simply " drawings-shown-a-few-minutes-ago." There was no appreciable correspondence between images of this sort and the presence or absence of ocular movement. The experiments did, however, supplement the results of the investigations which were mentioned in Section II with respect to the size and the accompanying affective tone of the image. We have already cited a possible objection to the observer's repro- duction of the image by a drawing. Since kinaesthetic and tactual sensations were involved in such a depiction, it was possible that the results were not exact representations of the images. But even if allowances are made for constant errors, the total results seem sig- nificant. Since the drawings were 13/2 meters away, we might expect the observer's representations to be slightly smaller than the originals. There was, however, a strong tendency, with two observers, to exaggeration in the image. No drawing was smaller than the figure by more than an inch, while 7 exagger- ated it by 1-6 inches. All observers reported a predominance of pleasantness over unpleasantness in the images (43:19). As a rule, however, the affective tone was connected with the free or impeded fulfillment of instructions, not with the character or the sig- nificance of the total images. Too often writers have assumed without warrant that coincident affective tone and image were inherently related. The correspondence between ocular movement, the sort of image, and the condition of attention appears when we compare all the results obtained from the three observers. The num- 482 CLARK ber of images associated with ocular movements is shown in Table IX, and the duration of these movements with any TABLE IX C. D F I. Co. I. Co. I. Co. Ocular movement 29 4 7 4 28 3 No ocular movement 23 1 1 image, in Table X. (I. stands for the period of the image; Co., for that of the control). Not only did D have fewer images accompanied by ocular movement than did C or F, table x Duration of Ocular Movement Second 2 3 4 5 6 7 8 9 10 11 13 15 20 D I 1 2 1 3 Co 1 3 I 1 5 4 4 2 8 2 2 1 C Co 1 2 1 I 1 3 1 2 7 1 2 4 1 2 4 F Co 1 1 1 but the number of movements with a given image was apt to be less in his case. There was also a noticeable difference in the general character of the images of the three observers. (i) The size of D's images, according to his drawings, varied but slightly from that of the diagrams, while both C's and F's images were exaggerated in approximately 68% of the cases. (2) None of D's images were incomplete, but 11 of C's and 15 of F's lacked some or many details. (3) Four of D's images, 8 of C's and 9 of F's revealed inaccuracies of detail. (4) Three of C's images and 6 of F's were of objects other than the diagrams. Four times, in the case of F, these representations were incomplete. D was always able to arouse an image of the perceived drawing. (5) D's total images never contained imaginal visual elements which were not represented in the corresponding perceptual objects. No pro- cesses from other sense departments were intimately integrated with the images. Such integration occurred with 5 of C's images. Four times the added elements were visual, and once they were kinaesthetic. VISUAL IMAGERY AND ATTENTION 483 In 14 instances, F's total images were elaborations of the perceptual object. The added elements were visual in 3 cases, kinaesthetic in 8, tactual in 2, and organic in one. (6) D always drew the same diagram as representing both the image and its corresponding percept. This was the case with only 11 of Cs images and with none of F's. Individual differences were noticeable, not only in the char- acter of the image, but also in the state of attention. (1) In the case of 22 of D's images, there were a few (from one to three) simple groups of associative processes, usually secondary (verbal). Twelve of Cs images and 19 of F's either lacked such pro- cesses or were accompanied by a great many. Attention was apt to fluctuate when associative processes were so numerous as to act as distractions or when the absence of these processes made the images uninteresting. (2) Further differences existed with respect to the degree of clear- ness and its fluctuations. Over half of D's images were very clear and steady. Twenty of Cs images were very clear, but 19 of them fluctuated. Moderate clearness was characteristic of 15 of F's images, and fluctuations occurred in 24 cases. (3) D's attention was less frequently secondary (or active) than was Cs or ,F's. D reported a conflict between groups of processes, or attention with effort, 6 times, C 11 times, and F in 9 cases. In brief, fixation was usually steady when an image was very clear and did not fluctuate, when it was accompanied by few associative processes, and when attention was without effort. Under other conditions ocular movement tended to occur. The dissimilarities among the images of a single observer are associated with ocular movement in the same way as are variations in all the images of different observers. In the six cases in which D's attention was secondary, fixation was unsteady. During the course of 4 other images the eyes moved, al- though attention was without effort, but in all of these instances there were many associative processes or irrelevant ideas. The one image which was noticeably vaguer than any of the others was also accom- panied by ocular movement. All of Cs images were accompanied by movements of the eyes, but the one image with which there was only a single movement in 20 seconds was very clear and constant, and attention was without effort. One of F's images was accompanied by no ocular movement. Attention was without effort, and the image was clear and constant. With 4 of his other images, there was only one movement in 20 seconds. In all of these cases, attention was without effort. Two images were very clear, and the other 2 were moderately clear, while the images remained constant or fluctuated but slightly. If we turn to the kymographic records of the control-experi- ments, we see that they are in accord with the results already mentioned. Ocular movement frequently occurred during the controls, and, when it did, the general conditions of conscious- ness closely resembled those which were characteristic of the images accompanied by ocular movement. 484 CLARK To sum up, we may say that there is usually no movement of the eyes with images which are fairly faithful representa- tions of the corresponding perceptual objects ; while ocular movement tends to occur with images which differ from the perceptual objects in size, and in completeness and accuracy of detail, and, in some cases, when processes from other sense departments are integrated with the images. Moreover, fixa- tion is apt to be steady when there are few associative pro- cesses (but not so many as to distract attention from the image and not so few as to allow attention to wander), when the image is clear and stable, and when attention is without effort. When these conditions do not exist, ocular movements tend to occur. The results of Experiment 3 are in general accord with those of a preliminary series in which C alone served as observer. The apparatus was a rough wood and cardboard model like the metal boxes in all essential details. A series of words was used to arouse images. O was required to time the images with a stop-watch and to press a record- ing key whenever the points of light became visible. During the course of 15 of the images, a beam of light was thrown upon the cornea of O's eye as in Experiment 1. Unfortunately, the light had to be of such a low intensity that the experimenter's observations of ocular movement are not reliable. The experimental conditions, especially the presence of the beam of light, distracted the observer. For this reason, the value of the results might be questioned if they were not in agreement with those of the other series. Records of 22 images were obtained. Two of these were U-images, 13 were G-images, 7 were F-images. Of the F-images, 6 had somatic reference, 5 had a recognized scenic background, and 3 had fairly many associations. Of the G-images, none had a scenic background, and 11 had few or no associations. Ocular movements were reported by the observer with 5 (71%) of the F-images and with 7 (54%) of the G-images. As regards attention to the image, there was no ocular movement in the one case in which the image was very clear and constant, or with another image which showed a gradual decrease in clearness. Ocular movement occurred 1, 2, or 3 times in 5 of the 10 cases in which the image was vague but exhibited no appreciable shifts in clearness, and in 8 of the 10 cases in which there were marked fluc- tuations in the clearness of the image. There was evidently a degree of correspondence between the frequency of ocular movement and the kind of image, and a rather marked relation of ocular movement to attention. Experiment 4 Problem and method. While Experiment 3 was in progress, it occurred to the writer that, in spite of the apparent accuracy of the apparatus, the observer might sometimes be unaware of the spots of light even when the eyes moved. If O's atten- tion were focused upon the visual image, there was a possi- bility that the light might fall upon a sensitive portion of the VISUAL IMAGERY AND ATTENTION 485 retina without touching off a report. To test this hypothesis, we combined the methods of Exps. 2 and 3. With the left eye, the observer fixated the single point of light as in the preceding investigation. The right eye was closed, and the tambour used in Exp. 2 was adjusted against the lid. In general, the procedure was the same as in Exp. 3, except that, instead of exposing the drawings in a perceptual series and later arousing the corresponding images, a number of words were used as in Experiment 2. C, D, and F again served as observers. As a preliminary control, before each record was made, the observer intentionally moved his eyes. In all of these cases the movement was recorded and the dots of light became visible. Each record was taken for a period of 20 seconds, as in the latter part of Exp. 3. Results. Six records were made with D as observer, 5 with F, and 8 with C. In D's case, no ocular movement was detected by either method. The dots were never visible during the 20 seconds of fixation, and the record made by the pneu- matic system showed no fluctuation. During the course of 4 of F's 5 images, dots became visible, but the other recording line did not waver. Very slight ocular movements evidently sufficed to bring the dots within the field of vision, but were not large enough to affect the tambour. One of C's records indicated that the dots became visible 5 times while there were no corresponding fluctuations in the other line. In the other 7 cases, ocular movement was indicated by both lines. There was a general conformity between the two records, although in some cases the record made by the pneumatic system indi- cated more fluctuations than the other line. There are several possible explanations of these apparent incon- sistencies. (1) C sometimes saw the points of light 8 or 10 times in 20 seconds. It is very likely that she did not react quickly enough to record all the changes. (2) A twitching of the right lid might have changed the one record, although the eyes themselves remained steady, so that no dots appeared. (3) Where the excursion of the eyes was fairly wide, a fluctuation in the pneumatic record might have occurred when the projection of the cornea crossed the bulge of the tambour, and again when the eye resumed its original position, although the dots were visible but once. While one record was being made, the electric current was accidentally reduced, and the points of light became very faint. It was probably due to this fact that the dots were visible less often than in the previous periods. The results of this experiment seem to indicate that the suggested hypothesis is incorrect. If O's eyes had moved while he failed to notice the points of light, the pneumatic record would have fluctuated, while the other was unaffected. But this was not the case. When movement was recorded 486 CLARK by only one line, it was the record made by O that showed fluctuations. The pneumatic system seemed the less accurate of the two. When slight movements of the eyes occurred, the observer noticed the dots, but the tambour was apparently not affected, while larger movements were recorded by both systems. From the results of Exp. 4 we infer that the perception of the dots of light during the experimental period indicated actual movement of the eyes. The results of the short trial preliminary to Exp. 3 revealed the same correspondence be- tween F-images and G-images which was discovered in Exp. 1. Ocular movement usually occurred with the former images, but not with the latter. In Exp. 3, we found that the amount of resemblance among imaged objects and the corresponding perceived objects seemed to be connected with ocular move- ment. As has already been suggested, this correspondence may reasonably be referred to an association of ocular move- ments with certain conditions of attention. Both Exp. 3 and the preliminary trial confirm the correspondence of ocular movements with the degree of clearness and fluctuations of attention which was the most significant discovery in Exp. 2. In Exps. I and 2, the arousal of a particular image was not required, and the experimental conditions did not seem to impose any tax upon the observers' ability to hold the atten- tion steady. In Exp. 3, however, there occurred, as a result of the exacting instructions, numerous cases of secondary attention. There proved to be a positive correspondence be- tween this sort of attention and ocular movements. Were we to attempt an explanation of the correspondence which we have discovered between different sorts of image, attention, and ocular movement, we should turn naturally to the facts of perception. A series of special experiments would be necessary fully to substantiate a transfer of general motor attitude or of ocular movement from perception to visual imagery. Our results suggest, however, that there are sev- eral possible likenesses among perceptions and images in re- spect to movements of the eyes. (1) When an extended perceptual object close at hand has many details, we must, in order to gain an accurate im- pression of it, glance now at one portion, now at another. In a similar way, we have frequently discovered ocular move- ment and shifts of attention where an imaged object is rich in detail or is accompanied by numerous associative processes (as was the case with many of the F-images and with some of the images of drawn figures in Exp. 3). VISUAL IMAGERY AND ATTENTION 487 (2) When a perceptual object has very few details, we are apt to grasp its significance at a glance and then to turn our gaze and our attention to something else. A similar tendency was manifested, in our experiments, in the case of some un- interestingly simple images. (3) In perception, when an object lacks cognitive clear- ness, our eyes are often focused first on one part, then on another, in order to gain a more definite impression. Such waverings of the glance were also characteristic of images. (4) Under normal conditions, when we attend successively to different aspects of a perceptual object, our glance almost invariably follows the direction of our attention. We have already given instances of the agreement between fluctuations of clearness and shifts of fixation in imagery. (5) Finally, when two rival objects claim the perceptual attention, the glance is apt to waver between them. Our results showed that secondary attention to imagery also in- volved ocular movements. All of these resemblances seem to point in the same direction. In the absence of contradictory evidence we may, then, reasonably suppose that the relations existing between the character of the object, attention, and ocular movements in perception are transferred to analogous cases of imagery. Such an hypothesis is in harmony with the current view of the physiological functions underlying per- ception and imagery. When an object formerly perceived is imaged there occurs, presumably, through central excitation, a reinstatement of the appxoximate physiological conditions of perception. The imaged object presents much the same color, shape, size, and details as does the perceptual object. There might easily be a similar reinstatement of muscular innerva- tions. In fact, the recurrence in imagery of the movements of the eyes involved in perception is a more probable phe- nomenon than is the accurate repetition of the visual details of the perceived object, since ocular reactions which are alike (at least in their gross characteristics) occur with a large number of perceptions, while a comparatively small number of objects present a close resemblance in their qualitative aspects. That is to say, there may be more likelihood of the integration of generic muscular reactions in imagery than of the formation of generic visual images. IV. Summary and Conclusions The results upon which we base our conclusions were ob- tained from six trained observers, four of whom served in two or more of the four principal experiments. The ocular 488 CLARK movement occurring in connection with visual images was recorded in the various investigations by three different meth- ods. There was a general agreement in the results whenever two or more experiments involved the same problems. Al- though individual differences were sometimes conspicuous, they were of such a sort as to substantiate our conclusions, not to invalidate them. Agreement between our results and those of other investi- gators is as close as we could reasonably expect in view of the differences which obtain in instruction and method, and, especially, of the differences in the criteria used for classi- fying images. (See Table XI). The "memorial image" of TABLE XI Specific Reference Clark Perky Ogden Koffka F-Images Memory (part, and pers. ref.) Memory (fam.?) U-Images Imagination Individual Individual (degrees of ind'ing and loc'ing) Generic Reference G- Images Imagination General (in part) both Ogden and Perky should probably be considered as a sub-class of our F-images. Perky's images of imagination we should generally have classed — so far as we can tell from her descriptions — as G-images ; Ogden's images of imagina- tion, 29 as U-images. We found that our G-images (Perky's imaginations) 30 and our U-images (Ogden's imaginations) were usually unlike in most of their characteristics. It is not strange, then, that there should be considerable disagreement between Perky's and Ogden's results when these two classes are compared. Koffka's general images would coincide in part with the G-images of our classification, while his indi- vidual images would include both our F and U classes. i. Five of the characteristics of images which we have ex- amined were also considered in much the same way by 29 Ogden appears to have left his observers free to apply their own criteria to their images : he says, " The criteria which seemed to be em- ployed primarily by our observers were those of ( familiarity ' and ' un- f amiliarity '" (p. 384). 30 Perky writes of " generic images " and of " habitual imagina- tions " (pp. 440, 447) in contrast to "particular images." VISUAL IMAGERY AND ATTENTION 489 Perky, Martin, and Ogden. The following summary gives a comparison of the results. (a) There was a general agreement in respect to associa- tive processes, which frequently accompanied the memorial images of Perky, 31 Martin, 32 and Ogden, 33 and were found with our F-images, and also with Ogden's images of imagina- tion and our corresponding U-images. With Perky's 34 and Martin's 35 images of imagination and with our corresponding G-images these accompanying processes occurred less often. (b) Three of the four investigations indicated similarities of clearness and of detail. The memorial images of Ogden 36 and of Martin, 37 and our F-images, were apt to be distinct, not " scrappy "■ and " filmy," as were Perky's. Instead of being " substantial " and " complete," as were Perky's images of imagination, 38 our corresponding G-images often presented few and vague details. Many of Ogden's images of imagina- tion 39 and of our U-images were, detailed and clear. (c) The results of Ogden 40 and of Martin 41 also agree with ours in failing to show any correspondence between the kind of image and the presence or absence of color. Perky gives no statistical summary, but describes the images of memory as " colorless etchings," and those of imagination as " often- times, on the contrary, very highly colored." 42 Her state- ments, however, appear to be contradictory, for she also states that the images of memory " retained their daylight char- acter," while those of imagination took the illumination of their surroundings. 43 This statement might be interpreted as meaning that it was the memories that were colored and the imaginations that had no color of their own. (d) The results of Ogden's investigation and of our own do not agree with Perky's with respect to affective tone. Ogden 44 found no evidence of Perky's memorial " mood of recognition " or. imaginative " mood of surprise." In our in- vestigation, the images (especially in the F class) were fre- quently pleasant, but an explicit feeling of familiarity was not always found with F-images, and a feeling of strange- ness or surprise occurred with only two U-images. (e) As regards stability, there is no correspondence of results. Perky's memorial images were described as " scrappy and fleeting," 45 her images of imagination as " substantial." 46 81 Perky, p. 451. 32 Martin, p. 408, 412. 33 Ogden, p. 405. 34 Perky, p. 450. 35 Martin, p. 408, 412. 36 Ogden, p. 398. 37 Martin, p. 400. 38 Perky, p. 451. 39 Ogden, p. 398. 40 Ogden, p. 396. 41 Martin, p. 405. 42 Perky, p. 447. 43 Perky, p. 446. * 4 Ogden, p. 398. 45 Perky, p. 447. 48 Perky, p. 451. 490 CLARK Martin's images of memory were usually, though not always, more stable than those of imagination; 47 and Ogden found about the same number of stable images of his two kinds. 48 Our investigation revealed no significant correspondence be- tween the kind of image and its stability. Where a com- parison is possible, the total results of Ogden and Martin agree, in general, with those of the writer, while all of these three investigations give evidence which is contradictory to that of Perky. 2. As regards ocular movements, our results agree with Perky's in so far as these movements usually occurred with her images of memory and frequently with our larger class of F-images, but were not, as a rule, reported with her images of " imagination " or with our corresponding G-images. The frequent occurrence of movement with our U-images, how- ever, suggests that no essential relation obtains between images of " memory," as contrasted with other forms or functions of imagery, and these kinaesthetic processes. That is to say, the conditions of ocular movement with imagery seem to be general conditions, — not the conditions which distinguish two imaginal functions, memorial and imaginative. Indeed, it is doubtful whether it is feasible and proper to attempt to oppose these two functions at the level of the total image. Our own analyses, as well as those of Koffka, suggest, instead, a very large number of functional gradations for the simpler imag- inal complexes ; — gradations of specifying, of individualizing, of generalizing, of symbolizing, and of reference to the ob- server, to objects, to times, to places, and to contexts. The alleged functional distinctions based upon " particularity " and " personal reference " are gross, and, moreover, we are not persuaded* by the facts that they are correlated with such differences in process and integration as temporal course, clearness, substantialness, and affective and organic accom- paniments. As for the presence and absence of ocular move- ments under imagery, we are not certain that familiarity, " the mood of memory" (Perky, p. 451), is essentially connected with this special kinaesthesis ; while we have failed to discover by introspection an antithetical or contrasted mood of un- familiarity or " novelty " to mark the absence of such move- ments. These specialized movements we are inclined to ac- count for in another way. They seem to us to be symptomatic of those general central conditions which underlie determina- tion and which fix the state and the configuration of conscious- 47 Martin, p. 407. 48 Ogden, p. 404. VISUAL IMAGERY AND ATTENTION 491 ness, — i. e., fix attention and associative completion. 49 When- ever these general conditions approximate or tend to repro- duce the usual psychophysical accompaniments of visual per- ception, then the ocular movements also tend to occur. Visual perceptions normally flow in trains, passing from feature to feature of the object and likewise from object to object. The kaleidoscopic change of clear processes bears with it a suc- cession of ocular movements. We believe that whenever this diffusive and exploratory kind of visual attention occurs, whether or not the organism is immediately affected by a visual stimulus, there normally appear, save under some special inhibition, the small movements of the eyes which have now been detected in many experiments set for the study of imagery. If our view is correct, then the presence and frequency of ocular movements would have no immediate or intrinsic con- nection either with familiarity or with " personal reference " and " particularity ;" though it might well be true that those images which make most explicit reference to the observer and to his past experiences also most frequently fulfill the general conditions just referred to. 50 It follows that isolated, detached, unset and un-detailed objects, such as appeared in our G-images, would least reinstate the usual conditions of perception ; but it does not follow that these objects furnish the materials of " imagination." We are thus led to connect ocular movements more directly with general central and con- scious conditions than with special functional differences dis- played by the image. In so far as it is permissible to base generalizations upon our experimental results, we feel justified in drawing the fol- lowing conclusions: i. If visual images are divided into three groups upon the basis of the presence or the absence of (i) specific reference and (2) familiarity of the object, certain other functional characteristics are found to be rather closely associated with a particular group or with groups of images. Pure types, however, are comparatively rare, and intermediate forms are relatively numerous. 2. A correspondence discoverable between kind of image 49 Perky found that attention wandered in memory but was nar- rowly focused in imagination (p. 449). 50 As regards audition, olfaction, and the other senses, the facts are not sufficient to warrant general conclusions. Remembering the his- tory of the doctrine of " fluctuations of attention," the author is sus- picious of any general functional difference which does not take into account the bearing of sensory adjustments. 492 CLARK and amount of ocular movement is probably to be referred to certain general conditions of attention. 3. Changes in the clearness of the image correspond, to an appreciable extent, to movements of the eyes. 4. Ocular movement is more likely to occur in secondary attention than in primary or derived primary attention. 5. Characteristic ocular movements, and possibly general motor attitudes, seem to be transferred from visual percep- tions to visual imagery. VITA Helen Clark was born in Cortland, New York, in 1890. She prepared for college in the grades and in the State Normal School at Cortland, and was graduated from Vassar College in 1913, with the degree of A. B. She was graduate scholar at the University of Illinois in 1913-14, and graduate fellow in 1914-16. In 1914 and 1915, she was instructor in psychology at the State Summer School of Rutgers College. She has published (jointly) "A suggested coefficient of affective sensi- tiveness" (Amer. J. of Psychol., 1913, XXIV, 583), and has in press a research in social psychology upon 'The crowd." Deacidified using the Bookkeeper process. Neutralizing agent: Magnesium Oxide Treatment Date: Oct. 2004 PreservationTechnologies A WORLD LEADER IN PAPER PRESERVATION 1 * 1 Thomson Park Drive Cranberry Township, F A LIBRPRY OF CONGRESS 013 370 682 3 $