illusion and 6.4 mm wide in the Visual capture in visual illusions* Ponzo. The induction lines were displayed on a surface inclined 15 deg from the horizontal. The front edge of JAMES T. WALKERt the display surface was 83.8 cm above University of Missouri-St. Louis, St. Louis, Missouri 63121 the floor. A thin sheet of Plexiglas covered the induction lines; thus, the Visual capture is the resolution of visual-tactual conflicts in favor of vision. In induction lines could be seen but not each previous demonstration of this phenomenon, a distorted optic array, or felt. The test blocks, various lengths of some mechanical analog for a distorted optic array, has produced a conflict black-lacquered Plexiglas about between erroneous visual information and presumably veridical tactual 12.5 mm in cross section, were information. The present experiments demonstrate visual capture in the presented, a pair at a time, 30 em Muller-Lyer and Ponzo illusions in the absence of any distortions of the optic apart on the clear Plexiglas covering array. the induction lines. Two visual va ria b les, vertical pieces of The resolution of visual-tactual Every prior demonstration of visual black-lacquered Plexiglas about conflicts in favor of the of vision capture (see, for example, Rock, 1966; 12.5 mm in cross section, could be has been called visual capture (Hay, Kinney & Luria, 1970; Pick & Pick, adjusted to indicate how long the test Pick, & Ikeda, 1965). In the earliest 1970) has utilized a distorted optic blocks looked. Two tactual variables demonstration of visual capture array (Gibson, 1966) or some were located out of sight, directly (Gibson, 1933), when the 0 viewed a mechanical analog for a distorted optic below the test blocks, on a horizontal straight rod through array. It is possible, however, to surface; each was a modified pair of curvature-producing prisms while arrange a conflict between vision and calipers which 0 adjusted to indicate stroking the rod with the fingers, the touch in the absence of any distortion how long the test blocks felt. On each rod looked curved and felt curved as of the optic array by slightly tactual variable, two vertical metal well. Tastevin (1937, pp. 67-68) modifying some of the classical visual tabs were grasped between the thumb showed an 0 a plaster replica of one of illusions. Each illusion shown in Fig. 1 and middle finger. The thumb tab, the his fingers while the O's real finger was consists of two parts: (1) the tab nearest 0, was immovable, while hidden from view some distance away. induction lines, the radiating lines in the finger tab could be adjusted. The conflict between the visual the Ponzo illusion and the short position of the plaster replica, which converging lines in the Milller-Lyer; Procedure the 0 took to be his own finger, and and (2) the test lines, the vertical lines The experimental design was a the veridical position of the real finger in both illusions. The essential feature three-way factorial, with sex, illusion was resolved in favor of the sense of of these visual illusions is the O's (Mulfer-Lyer and Ponzo), and vision: the felt position of the O's own overestimation of the length of one modality (vision and touch) finger coincided with the seen position test line relative to the other, in spite constituting the three factors. of the replica. Hay, Pick, and Ikeda of the fact that both test lines subtend Repeated measures were made on the (1965) showed that the felt position the same visual angle. Thus, these third factor. Each 0 was shown only of the hand is strongly influenced by illusions occur in the absence of any one illusion, which was presented in the position seen through distortions of the optic array. If the two orientations. The Ponzo illusion displacement-producing prisms; these test lines were three-dimensional workers considered the earlier blocks that the 0 could feel as well as demonstration of Tastevin ", . . a kind see, and if the induction lines were A. MULLER -LYER of mechanical analogue to the prism simply two-dimensional lines that situation .•. " In another could not be felt, then if the 0 looked demonstration, Tastevin (1937, p.60) at the test blocks and induction lines showed an 0 plaster replicas of his while grasping the blocks with the left index and middle fingers, holding and right hands, the visual illusion between them a small box 1 em long, would conflict with the presumably while the 0 's own index and middle veridical tactual information. fingers, hidden from view, held between them a larger box, 5 em long. EXPERIMENT I The tactually perceived size of the The purpose of this experiment was larger box held by the O's own fingers to investigate the occurrence of visual came to coincide with the visually capture in the Muller-Lyer and Ponzo perceived size of the smaller box held illusions. Adjustment methods were B. PONZO by the plaster fingers. This result is used to measure the magnitude of analogous in important respects to the these illusions and the amount of later findings of Rock and Victor visual capture produced by each. (1964), who showed that an object viewed through a minifying optical Observers system felt, as well as looked, smaller Thirty-two introductory than it really was. students (16 men and 16 women) at the University of Missouri-St. Louis participated in this experiment as a *This project was supported by a course requirement. University of Missouri Summer Research Fellowship. A version of this paper was presented to the Midwestern Psychological Materials and Apparatus Association in Detroit, Michigan, May 7, Figure 2 illustrates the apparatus. Fig. 1. Illusion magnitude as a 1971. The induction lines were made of function of adjustment method in t Address: Department of Psychology, black chart tape attached to sheets of University of Missouri-8t. Louis, 8001 (A) and instruction condition in (B). Natural Bridge Road. St. Louis. Missouri white poster board; the lines were Tactual measures in both experiments 63121. 12.8 mm wide in the Muller-Lyer indicate visual capture. & , 1971, Vol. 10 (2) Copyright 1971, Psychonomic Journals, lnc., Austin, Texas 71 significantly greater than zero (t =4.46, df =15, P < .001; t =6.53, df = 15, P < .0\)1). Thus, both illusions produce visual capture. As VISUAL VARIABLES shown by the analysis of variance, the "\---- magnitude of the visual capture effect is significantly less than the visual illusion magnitude in both the Miiller-Lyer and Ponzo illusions (7.76% and 8.24%, respectively).

EXPERIMENT II The purpose of this experiment was 1&«$SLOCKS to measure visual capture in the Ponzo illusion, using a different psychophysical method, the method of limits, and a different experimental • design. Observers Twenty-four students (12 men and 12 women) enrolled in undergraduate psychology courses at the University of Missouri-St. Louis participated in TACTUAL VARIABLES this experiment as a course requirement. No 0 had previously Fig. 2. Apparatqs. taken part in Experiment I. was presented with the radiating lines left and right test blocks (lengths in Materials and Apparatus (1) converging toward the left and millimeters) were presented in an The Ponzo illusion induction lines (2) converging toward the right. The individually determined random order: used in Experiment I were also used Miiller-Lyer illusion was presented 90-100, 100-110, 100-100, 110-100, here. In Experiment II the induction analogously. A set of five visual and 100-90. Since the. sum of the lines always converged to the right, as adjustments and a set of five tactual lengths of test blocks presented on the shown in Fig. 1B. The same basic adjustments were made with the left is equal to the sum of the lengths illusion in one orientation; then the on the right, a difference in the orientation was changed, and the appropriate direction between an O's '0 A. EXPERIMENT I VISUAL !'ZZJ procedure was repeated. All four adjustment of the left and right visual ADJUSTMENT: possible combinations of orientation variable is evidence of a visual illusion. TACTUAL _ orders and visual and tactual An analogous difference between the adjustment orders were used. left and right tactual variables is Os were instructed to adjust the two evidence of visual capture. visual variables until each looked the For each of the two orientations of same length as the corresponding test each illusion, the difference between block looked; these adjustments were mean left and right visual adjustments made by turning knobs on the front of was expressed as a percentage of the the apparatus. Both visual variables smaller mean. Each O's visual were in sight while visual adjustments adjustment score is the mean of the were made; after each adjustment, percentages, above, for the two they were lowered out of sight. All illusion orientations. A tactual Ponzo Muller-Lver adjustments ascended from an initial adjustment score was similarly derived Group Group '2,------~ length of zero. for each O. A positive visual score In making tactual adjustments, Os indicates a visual illusion in the B. EXPERIMENT IT grasped the test blocks simultaneously expected direction, and a positive '0 with the thumbs and middle fingers of tactual score indicates the occurrence both hands and then moved their of visual capture. hands to the tactual variables, which they were instructed to adjust until Results each felt the same length as the The results are shown in Fig. 3A. A corresponding test block felt. In order repeated-measures analysis of variance to make satisfactory adjustments, Os (Winer, 1962) shows a statistically were allowed to move their hands back significant effect of. visual vs tactual and forth between the test blocks and adjustments (F = 10.43, df = 1/28, the tactual variables as often as they P < .01). No other effect approaches needed. The finger tabs on the tactual significance. Since there was no variables were adjusted by pushing significant difference between the sexes and no interaction with the sex with the index fingers and pulling with Ponzo Control the middle fingers. All adjustments factor, the results presented below Group Group were made from an initial separation were pooled for men and women. The Fig. 3. Open vertical rectangles of 25 mm between the tabs. mean tactual adjustment scores for the indicate testblocks. Rectangles are of For each set of visual or tactual Mtiller-Lyer and Ponzo illusions equal length. Solid black lines are adjustments, the following five pairs of '(4.07% and 5.54%, respectively) are induction lines.

72 Perception & Psychophysics, 1971, Vol. 10 (2) apparatus was used as that in Ponzo illusion occurred (F = 31.02, overestimation of the length of the Experiment I. The left test block, the d f = 1/20, p < .001); such a right test block (t = 4.65, df = 11, standard, was 100 mm long. The right comparison under tactual instructions p < .001). Since there were no test block, the comparison, was (9.67% vs 2.50%) shows that the independent measures in the present selected from a graded series of blocks Ponzo illusion produced visual capture study of perceived size by vision alone differing by 2-mm steps. Since the (F = 33.50, df = 1/20, p < .001), and a and by touch alone, these results tactual variables were not used in this similar comparison under objective provide weak evidence-no more than experiment, they were removed from instructions (9.43% vs 2.50%) shows a suggestion-of the occurrence of a the apparatus, making it possible to that objective judgments were also natural, nonexperimental visual lower the display surface to a point influenced by the illusion (F =31.35, capture phenomenon. 1 where its front edge was 65.8 cm df = 1/20, P < .001). The absence of Visual capture was demonstrated in above the floor. any statistically significant effects of the Mtiller-Lyer and Ponzo illusions in instruction conditions shows that Experiment I, where a psychophysical Procedure visual, tactual, and objective method of adjustment was used. In Six men and six women were run judgments are not differently Experiment II, which used a different with the Ponzo induction lines present influenced by the visual illusion. Of experimental design and a different (Ponzo group); the remaining six men the numerous possible interaction psychophysical method, the method and six women were run with a blank effects, only one was statistically of limits, the Ponzo illusion was again sheet of white poster board in place of significant--the triple interaction shown to produce visual capture. The the induction lines (con trol group). as between temporal order of judgment, present experiments show that visual grasped the standard and comparison sex, and illusion vs control (F = 5.16, capture can be produced by visual blocks simultaneously between the df = 2/24, p < .05); this interaction illusions, as well as the visual thumbs and middle fingers of the left implies that the pattern of illusion distortions that have previously been and right hands. While grasping and magnitude as a function of judgment shown to produce this phenomenon. looking at the blocks, all as compared order differs slightly across male and the length of the right test block with female illusion and control groups. the standard block under three REFERENCES instruction conditions: (1) visual, DISCUSSION BROWN. K. T. Factors affecting differences (2) tactual, and (3) objective. as made The visual magnitudes of the in apparent size between opposite halves a practice judgment under each Miiller-Lyer and Ponzo illusions in the of a visual meridian. Journal of the condition before the start of the Optical Society of America, 1953. 43. present experiments are somewhat 464-472. experiment proper. No information as lower than the magnitudes typically GIBSON. J. J. Adaptation. after~ffect. and to accuracy was given at any time. reported (see, for example, Mountjoy, contrast in the perception of curved lines. Instructions emphasized that three 1958; Leibowitz et al, 1969). In most Journal of , 1933, 16, 1-3!. kinds of judgments were to be made. earlier studies. the illusions were GIBSON. J. J. The considered as Under the first condition, above, a was presented essentially perpendicular to percep tual systems. Boston: instructed to indicate " ... whether the a's line of sight. In the present Houghton-Mifflin, 1966. P. 310. HA Y. J. C.• PICK, H. L.• JR., & IKEDA. K. the right block looks longer or shorter experiments, the illusions had to be Visual capture produced by prism than the left, not whether it feels viewed obliquely because of the design spectacles. Psychonornic Science. 1965, longer or shorter, or whether you of the apparatus. The oblique angle of 2, 215-216. think it really is longer or shorter, but regard may have depressed the HOCHBERG. J. E., & BITTERMAN, M. E. Figural after-effects as a function of the whether the right looks longer or magnitude of the illusions. retinal size of the inspection-figure. shorter than the left. " Analogous An apparent discrepancy between American Journal of Psychology. 1951. instructions were given under the the results of Experiment II and some 64,99-102. KINNEY. J. A. S., & LURIA, S. M. other instruction conditions. earlier studies of left-right visual and Conflicting visual and tactual-kinesthetic Judgments of equality were permitted tactual comparisons must be pointed stimulation. Perception & Psychophysics. but not encouraged. Sequences of out. Earlier studies have shown that 1970.8. 189·192- instruction conditions were the size of an object held in the right LEIBOWITZ. H •• BRISLIN. R., PERLMUTTER, L.. 8< HENNESSY, R. determined for all Os by the rows of a (or dominant) hand is significantly Ponzo perspective illusion as a single 3 by:~ Latin square. Each a underesti mated by touch alone manifestation of space perception. made two ascending and two (McPherson & Renfrew, 1953; Science, 1969. 166, 1174·1176. descending judgments in ADDA order McPHERSON. A., 8< RENFREW. S. Wertheimer, 1954), while, by vision Asymmetry of perception of size between under each instruction condition. The alone, the size of an object in the right the right and left hands in normal point of subjective equality, the visual field may be overestimated s u b i e c t s, Quarterly Journal of middle of the C's interval of (Brown, 1953) or slightly, but not Experimental Psychology. 1953. 5. 66-74. uncertainty, was averaged for each significantly, underestimated MOUNTJOY. P. T. Effects of exposure time judgment condition and subtracted (Hochberg & Bitterman, 1951). Thus, and intertrial interval upon decrement to from the length of the standard block a conflict between vision and touch the Miiller-Lyer illusion. Journal of (100 mm); the difference was might occur naturally, that is, in the Experimental Psychology. 1958. 56. 97-102. expressed as a percentage of the absence of any experimentally PICK. A. D.• 8< PICK, H. L.• JR. Resolution standard block. produced illusion or distortion. Ifsuch of sensory conflict in perceptual a conflict were resolved in favor of development. Ceskoslovenska Psvchologie, 1970, 14,52·60. Results vision, then the control as in ROCK, I. The nature of perceptual The results are shown in Fig. 3B. A Experiment II might be expected to adaptation. New York: Basic Books, repeated-measures Latin-square overestimate tactually the length of 1966. Pp. 145-175. 222·249. analysis of variance (Winer, 1962) the right test block, as was the case, ROCK. I.. & VICTOR. J. Vision and touch: An experimentally created conflict shows a statistically significant main rather than underestimate its size, as between the two senses. Science, 1964. effect of illusion vs control (F = 34.93, might be predicted from the earlier 143, 594-596. df = 1/12, P < .001). A comparison tactual studies. The control group TASTEVlN. J. En partant de l'experience d'Aristote. L'}o:ncephale, 1937. 32. 57-84. between means for illusion and control means under visual and tactual 140-158. grou ps under visual instructions instructions (3.14% and 2.50%, WERTHEIMER, M. Constant errors in the (10.04% vs 3.14%) shows that the respectively) indicate significant measurement of figural after~ffects.

Perception & Psychophysics, 1971, Vol. 10 (2) 73 American Journal of Psychology, 1954, NOTE Teghtsoonian, M. Two varieties of perceived 67,543-546. 1. Note added in proof: After this paper length. Perception &: Psychophysics, 1970, was accepted for publication, a Very clear 8,389-392. WINER, B. J. Statistical principles in demonstration of a related natural visual experimental design. New York: capture phenomenon came to the writer's (Accepted for publication December 26, McGraw-Hill, 1962. attention. See Teghtsoonian, R., &: 1970).

74 Perception & Psychophysics, 1971, Vol. 10 (2)