An investigation of the relationship between and retinal inlage movement in the perception of movementl

ARIENMACK2 NEW SCHOOL FOR SOCIAL RESEARCH

The question investigated was whether correlative phenomenon of no perceived perception of object mation and or not eye movements accompanied by movement despite the occurrence of retinal correlatively if eye movements are not abnormal retinal image movements, displacement to maintain that retinal image matched by retinal image displacements, movements t1uJt are either or both at a movement is either a sufficient or mavement will be perceived. Thus. when different rate or in a different direction necessary condition for tile perception of retimll: image changes are "matched," than the , predictably lead object movement. An viewed in constancy is predicted, and when those to perceived movement. Os reported the dark by an 0 moving his appears changes are not "matched," some change whether or not they saw a visual target to move (Mack & Bachant, J969), but in is predicted. Constancy move when the movement of the target there is no retinal movement. A moving is predicted when retinal image changes are was either dependent on and simultaneous object in an otherwise empty fJeld, accounted for in terms of 0 movement with their eye movements or when the accurately pursued by the O's eyes, appears information. Constancy here signifies the target movement was independent of their to move, but there is no retinal movement. perception of stability in the visual field, eye movements. In the main experiment, A stationary scene viewed by moving eyes i.e., objects in the visual field appear to observations were made when the ratio appears stationary despite retinal image maintain their positions despite between eye and target movement (em/tm) movements (position constancy). The displacements of the retinal images. was 2/5. 1/5, 1/10, 1/20, and O. All these existence of these and other phenomena Failures of constancy refer to those ratios were tested when the direction of suggests that the relationship between occasions in which image displacements of the target movement was in the same (H+), retinal image movement and perceived stationary objects lead to perceived opposite (H-), and at right angle's to movement is more complex. A theory that movement in the visual field. (V+, V-) the movement of the eye~ t:ve attempts to specify the relationship The postulation of some sort of movements, target movements, and reports between image or stimulus displacements comparator mechanism provides a concept of target movement were recorded. Results and perceived movement has been that allows us to account for many of the indicate that a discrepancy between eye variously referred to as a cancellation, facts of movement perception. There is and target movement greater than 20% discounting, compensating, or now considerable evidence supporting the predictably leads to perceived target taking-into-account theory of movement operation of this sort of mechanism during movement, whereas a discrepancy of5% or perception (von Holst, 1954; Mackay, head movements. Information about head less rarely leads to perceived movement. 1967; Rock, 1966; Teuber, 1960). movements appears to mediate the The results are interpreted as support for If we are permitted to overlook the relationship between visual displacement the operation of a compensatory subtle and not so subtle differences and perceived movement (Wallach & mechanism during eye movements. between the various statements of this Kravitz, 1965; Wallach & Frey, 1969; Hay, theory, it becomes possible to set out its 1968; Rock, 1966). In these studies, What is the basis of perceiving the main outlines. All forms of this theory are abnurmal visual displacements, movement of a single object in an in agreement that retinal image changes displacements that are either faster, slower, otherwise unstructured visual field? The alone can neither account for the or in a different direction than the O's own most obvious answer is that it is the occurrence of movement perception nor head movements and do not simply reflect movement of the retinal image that is alone determine whether the movement the 0 movement, predictably led to crucial,3 since with every movement of an perceived resides in the object or in the O. perceived object motion. Until now there object there is a corresponding movement It is a theory that accounts for perceived has been no available evidence that a of the image of that object across the object motion as well as the failure to comparator operates during eye of the O. This answer is the more perceive object motion despite movements. Since the eyes are rarely still attractive since there is now ample displacements or changes in the retinal and since we generally see our visual evidence that nervous systems image. It assumes, in fact, that these are environment as stationary despite the are well endowed with receptor cells correlative events. The theory asserts that constant displacements of the retinal uniquely sensitive to retinal movements. If, the basis ofperceived object motion resides images produced by the fairly constant however, retinal displacement were the in the relationship between the behavior of movements of the eye, it would appear central factor in perceiving object the retinal image and the available sensory that some kind of comparator mechanism movement, it would follow that every time information about the movement of the o. in fact operates during eye movements.4 there was retinal movement greater than It assumes the existence of some kind of Yet in the only completely reported the threshold of the detecting cell central mechanism, a comparator experiment examining the relationship mechanisms, object movement· would be (von Holst, 1954) that takes into account, between eye movements, image perceived, and in its absence, no such matches, compares, or evaluates retinal movements, and the perception of movement would be seen. image shifts with reference to information movement (Wallach & Lewis, 1965) the It is, of course, a well·known fact that about 0 movement. The theory assumes authors conclude that target movements the case is not so simple. There are too that if not accounted for or matched by are not assessed in terms of eye many instances of perceived movement in appropriate 0 movements, retinal image movements. By devising a situation in the absence of retinal movement and of the displacements will invariably lead to the which the 0 saw a projection of his own

Perception & Psychophysics, 1970, Vol. 8 (SA) Copyright 1970, Psychonomic Journals, Inc., Austin, Texas 291 pupil, the authors produced abnormal target motion. Unfortunately the account approximately 2 mm in diam, were displacements of a visual target dependent given is brief and no details are included. recorded by one channel of the pen on eye movements. This technique made it The contradiction between the Wallach recorder. Simultaneous recording of eye possible to produce a situation in which and Lewis results and those briefly and target trace movemertt permitted a the target disk, the projection of the pupil, reported by Yarbus made further continuous check on the relation between could be made to move at either a slower examination of the relationship between eye and target movement. The volume or faster rate than the eye itself. Their eye movement, image movement, and control knob was calibrated from 0 to I. results indicate that abnormal target perceived movement necessary. Another When this knob was set at 0, the trace was dis p lacements accompanying eye reason for such a study is the fact that an not moved at all by eye movements. When movements, do not reliably cause reports eye- m()ve me n t/ im age -m ove men t this knob was set at 1, the trace moved the of perceived target movement. They comparator capable of doing nothing more same amount as the eyes and thus could conclude: "Our results are incompatible than merely discriminating between approximate a stabilized image (Le., when with the view that the apparent rest of instances of eye movements that are either the direction of trace displacement and eye visual objects whose images shift due to accompanied or not accompanied by movement were the same). When the knob eye movements is to be explained by a retinal displacements would make it very was set at some point between 1 and 0, the compensating process which takes the eye difficult to account for any perception of trace moved some fraction of the amount movements into account." movement during eye movements. the eyes moved. Thus, when the knob was From the Wallach and· Lewis results it By a technique different from that used set at 2/5, the target trace moved 2/5 of might be possible to argue that all by either Wallach and Lewis or Yarbus, a the distance that the eyes moved. (All de tection of retinal displacement is situation was created in which eye calibrations were frequently checked.) The suppressed during eye movements. If this movements were accompanied by increase in what is best described as trace were the case, then, since no retinal abnormal image displacements, that is, jitter with increase in gain were mostly displacements would be registered during displacements that did not directly reflect eliminated by filtering. What jitter an eye movement, there would be no the movements of the eye. Os reported remained was accounted for in the reason to expect a loss of position whether or not the target at which they experimental design. constancy during eye movements, and were looking appeared to move. A switch on the control panel permitted there would be no reason to postulate rapid shifts from the horizontal to the some kind of comparator that could match METHOD vertical input channels of the oscilloscopes. retinal displacement to 0 movement The essential requirements for work of This switch had the effect of making the information. That this cannot be the case this kind are: a visual target whose trace move either horizontally or vertically may be seen from the fact that there is movements are strictly related to the in response to lateral eye movements of the evidence that an afterimage appears to movements of the O's eyes, a means of O. A second switch, a plus-minus switch, move even when it is viewed in complete altering the relation between eye and target changed the direction of trace movement darkness and the only 0 movement is 0 movement, and a means of simultaneously accompanying an eye movement. When eye movement (Mack & Bachant, 1969). recording eye movements and the reports this switch was set at + and the This suggests that at least the absence of ofmovement made by the O. horizontal-vertical switch was set at any retinal displacement during an eye Lateral eye movements were recorded horizontal, the trace moved in the same movement constitutes a baSis for by means of . Beckman direction as the eyes. If the eyes moved movement perception. If the absence of bipotential skin electrodes were attached right, the trace moved right. If the +­ displacement is crucial, then there must be to the outer margins of the eyes and a third switch was set at -, the trace moved left a record of the occurrence of electrode was attached to the center of the when the eye moved right. When the displacements in order that these two kinds O's forehead and served as a ground. The horizontal-vertical switch was set at of events be distinguished. It therefore electrodes were wired through a noise filter vertical, then a + setting of the + - switch appears that there must at least be a to one channel of a Sanborn four-channel caused the trace to move up when the eyes mechanism that records retinal pen recorder with sufficient dc moved right. A- setting caused the trace displacements and eye movements and is amplification to represent the signal from to move down when the eyes moved right. capable of determining when these events the eyes accurately. The output of this The various control switches permitted occur simultaneously, which determination channel was wired in tum to an external four directions of target movement, ordinarily results in no perceived object volume control from which it was fed into horizontal with the eyes (H+), horizontal movement. If no evidence for such a either the horizontal or vertical input against the eyes (H-), vertical and up with mechanism is found to operate during eye channel of either one or two Techtronix the eyes to the right (V+), and vertical movements, then it would seem to mean No. 502A oscilloscopes. The volume down with eyes right (V-), as well as that no real (as opposed to illusory) control permitted continuous control over continuous control over the ratio of movement could be perceived during an variations in gain. The effect of this was to eye-to-target movement, em/tm. The eye movement except when the image of afford continuous control over the amount following ratios ofeye-to-target movement, the visual object is stable on the retina (as of oscilloscope trace movement produced em/tm, were repeatedly tested: 0, 1/20, in the case of an accurately pursued target by an eye movement. The oscilloscope 1/10, 1/5, 2/5. Ratios of 1/2 and I were or afterimage). trace served as the visual target. It was the tested only under conditions of random To our knowledge the only available only object visible to the 0 during testing eye movements (to be explained). empirical evidence that supports the view sessions. In order that the 0 should not see Small instant on-offlamps were attached that a comparator mechanism operates the glow of the oscilloscope screen itself, to both sides of one of the oscnJoscope during eye movement appears in Yarbus Kodak Wratten fIlters, No. 47B, were screens. Each light was a distance of (1967). In his book, Yarbus reports an placed over the screen and during testing 6.25 cm from the center of the screen. experiment in which he produced all Os wore goggles containing Kodak Switches on the control panel turned on abnormal image displacements by means of Wratten ffiter No. 35. The movements of either one or the other of these lights. suction-eontact lenses. Os report perceiving the oscilloscope trace, which was Operations of one of these switches

292 Perception & Psychophysics, 1970, Vol. 8 (SA) resulted in a light flash that lasted less than directed eye movements (DEM) and in which the em/tm was zero were 1/20 sec. The lamps were coated and taped nondirected eye movements (NDEM). presented at least three times during each with an opaque material so that the light DEM. The experimental a was testing session. Zero testing trials were flashes were sufficiently dim that no instructed to look straight ahead at the usually presented after a block of trials in were produced even in oscilloscope trace, and bite on the bite bar. which the em/tm was either 2/5 or 1/20. dark-adapted eyes. The light switches were This signaled the onset of a trial. Blocks of trials were ordered so that four wired to an event marker on the pen Immediately following instructions to look consecutive blocks of trials were presented recorder providing a permanent record of straight ahead and a check of the in which the em/tm was constant while the the onset oflight flashes. eye-movement record to establish that the direction of target displacement was varied Each oscilloscope was housed in a o had properly responded, one of the light from one block of trials to the next. shielded and light-tight cell in which there switches was flipped. The a then turned NDEM. This condition was included as was sufficient room for an a to sit. The his eyes to where he had seen the light an attempt to verify the difference center of the oscilloscope screens coincided flash and was required to keep his eyes in between "incidental" and "voluntary" eye more or less with an eye level of the as. this position until he was once again movements reported by Wallach and Lewis Bite bars were fixed 27.5 cm in front of instructed to look straight ahead. During (1965). They found that abnormal image the oscilloscope screens and prevented the interval between turning his eyes to the dis p lacements accompanying eye head movements during testing. Positioned flash and returning his eyes to straight movements sometimes led to movement at an upright oblique angle and to the right ahead, a reported if there had been any reports when eye movements were of the screens were panels consisting of target trace movement. He reported voluntary but rarely when eye movements four buttons placed at the four points of a whether or not the trace moved when he were incidental. In this condition both Os cross. These were used by the a to report moved his eyes. All instructions were gone received exactly the same instructions. target movements. The top button was over carefully before testing occurred, and They were directed to look continuously at used to report upward movement, the all testing was preceded by at least 10 the target trace and to indicate if it moved bottom button for downward movement, practice trials in which all the 0 was by pressing the right button of the button the right button for rightward movement required to do was to move his eyes to the panel for as long as the trace continued to and the left button for leftward movement. flashed light and hold them in that position move. Unlike the DEM reports, only the The reporting buttons were wired through until once again instructed to look straight presence or absence of movement and not the dc amplifier to a channel on the pen ahead. Despite the apparent complexity of its direction was reported. The task of recorder. Depression of one of these the a's task, most as easily succeeded in constantly monitoring a small and dim buttons moved the pen in that channel a doing what was required of them. target light was sufficiently taxing without specific amount, so that the distance the Control as wore no eye electrodes and requiring as to also report direction of pen moved indicated which button was their eye movements were not recorded. target movement. being pressed. The two oscilloscopes were They were asked to look straight ahead and The following em/tm ratios were tested run simultaneously and at all times bite on the bite bar on instruction from the in combination with the four directions of displayed identical trace movement E (the same instructions were heard by trace movement: 1, 1/2, 2/5, 1/5, 1/10, because the input to and the calibration of both as) and to report if the trace moved 1/20, and O. Each combination of em/tm the oscilloscopes was always identical. within the interval between the ratio and direction of target displacement straight-ahead instructions and the was presented for a IS-sec interval at the Procedure instructions to the experimental a to once end of which the em/tm ratio, the The as were run in pairs, one who again look straight ahead. Both as were direction of target displacement, or both observed a target whose behavior depended therefore reporting on the behavior of the were changed. as were allowed to rest on his eye movements (experimental), and trace during the eye movements of the approximately every minute. the other who observed the same target experimental O. Although no record of the Half the Os began testing with DEM, without its being dependent on his eye eye movements of the control as were while the remaining half began with movement (control). Each a was tested made, it is probably safe to assume that NDEM. Half the as were experimental Os twice, once as an experimental and once as control as' reports were made on the basis before they were tested as controls. The a control O. The advantages of testing each of a period in which their eyes moved very remaining half was tested in the reverse a twice and of running as in pairs are that little since they had been instructed to order. it provides control data for each testing look straight ahead. session, and it allows each a to act as his Each combination of displacement ratio Subjects own control, thus accounting for styles of and trace movement direction was Fourteen Ss (seven pairs) were each run reporting. Since we obtained data from presented in blocks of four trials separated twice_ These Ss were Stanford University two as simultaneously, the occurrence of by at least a brief rest period. Thus a block students who were paid for their any electronic jitter with increases in gain of trials consisted of four trials in which participation. was not significant, because we were able the em/tm and the direction of target to analyze the difference between the displacement was constant. Several of these RESULTS reports of the two as. The experimental a blocks of trials were separated by longer DEM wore the eye electrodes and sat in the rest intervals in which any dark adaptation Data from the DEM condition were oscilloscope cell containing the was erased. Each block of four trials analyzed in terms of the percent ofcorrect light-flashing apparatus. The control a sat consisted of two left and two right light responses in each block of four trials. For in the other oscilloscope cell, which was flashes presented in random order. In all example, if a S reported target movement identical to that in which the experimental testing sessions, each combination of target to the left following eye movements to a a sat except that there were no flashing displacement direction and em/tm ratio right light flash, and target movement to lights. There were two parts to each testing was presented for at least one block of the right following eye movements to a left session. An entire testing session lasted trials. Both light flash in three of the four trials about 'Jll h. The two parts are designated orders of em/tm were used. Blocks of trials constituting the block of trials in which the

Perception & Psychophysics, 1970, Vol. 8 (SA) 293 Table I during a as target movement that is Mean Percent Correct Movement Responses: DEM 2/S as great as the eye movement. H- H+ V+ V- Grand X An analysis of means obtained in control conditions reveals the only significant EM/TM E C E CE C E C EC difference to be that between the 2/5 87.5 96.8 91.8 100 94.6 100 94.6 100 92.1 99.2 condition in which em/tm was 1/20 and all 1/5 78.5 91.1 89.6 94.6 87.5 100 87.5 98.2 85.7 95.9 the remaining em/tm conditions (see 1/10 41.1 83.9 39.2 87.5 51.8 92.9 41.1 92.3 43.3 89.1 1/20 13.2 64.6 11.4 64.3 12.5 75.7 7.8 76.9 11.2 70.3 Table 3). Very small target displacements 0 93.2 81.7 appear to be difficult to detect regardless of whether or not the;' are simultaneous Table 2 with an eye movement. Analysis of Variance Outcome (DEM) Statistical comparisons were made in which each experimental S was compared Factors F DF with himself as a control 0 (see Table 3). Groups The following differences were found to be Experimental and Control 177.546 1/416 <.001 significan t: all differences between Direction experimental and control Os in conditions Horizontal and Vertical 46.269 1/416 .05 in which the em/tm equals 1/10 Direction (significant at greater than .01), and all Positive and Negative 1.044 1/416 differences between experimental and EM/TM 215, I/S, 1/10, 1/20 121.147 31416 <.001 controls for em/tm equals 1/20 (significant at greater than .01). Thus, in the O. 2/5, and l/S conditions, there are no significant differences between experimental and Table 3 Mean Scores and Critical Values for Estimating the Significance control Os in their ability to perceive target of Difference Between Means (Scheffe Procedure) movement, although the differences that do appear between these groups are all HV Grand Mean characterized by the fact that the control EMrrM E C EC E C means are greater. When the em/tm was 2/5 89.64 98.39 94.64 100 92.11 99.20 1/10 or 1/20 Os were able to more 1/5 84.11 92.86 87.68 99.11 85.89 95.98 accurately detect target displacement in 1/10 40.18 85.72 46.43 92.86 43.30 89.29 the control conditions than were those Os 1/20 10.18 76.25 12.33 64.46 11.25 70.36 in the experimental conditions who were CV =48.83, P =.01 CV =23.56, P =.01 reporting on target displacements CV =40.49, P =.05 CV =20.72, P =.05 concurrent with eye movements. CV = Critical Value, E = Experimental, C = Control, H = Horizontal, V = Vertical An analysis of the errors made by experimental Os reveals no particular em/tm level was 1/5 and the direction of the immediately adjacent em/tm level. trend. This analysis is potentially most target movement was H-, he received a These comparisons were again made interesting in the case of errors made when score of 75%. Table 1 presents the mean separately for experimental and control Os the em/tm level is zero, since these errors percent correct scores of all experimental (see Table 3). For experimental Os, the might possibly reveal a tendency to see a and control Os for all combinations of mean percent correct responses for the 2/5 stationary target moving in a direction target-movement direction and levels of em/tm level is not significantly different predictably related to the direction of the em/tm. An analysis of variance was from the mean percent correct responses eye movement. Errors were extremely few computed for this data. The results of this for em/tm equals lIS. Movement was in this condition and reveal no trend. Most analysis appear in Table 2. All three main reported as frequently when the target cases in which the percent correct scores factors had a significant effect on the displaced 1/5 the distance the eyes moved were less than 100% were the result of a perception ofmovement. Using the Scheffe as when the target displaced 2/S of that failure to report any movement rather than procedure for all possible a posteriori distance. The following differences are reports of movement in an objectively comparisons (Winer, 1962) the differences significant: The difference between means incorrect direction. between means obtained in each condition in the l/S em/tm condition and means in An analysis was made of reaction time were evaluated. The outcome of this the 1/10 em/tm condition and the (RT) to the light flashes in order to analysis is found in Table 3. Statistical difference between means in the 1/10 and determine if on any trials the light was comparisons between means obtained means in the 1/20 em/tm condition are all visible when the eyes began to move. Any when target movement was horizontal and significant at greater than .01. For all trials in which this had occurred would when it was vertical and when the em/tm em/tm levels less than l/S, the smaller the have to have been ignored since reports of was constant yield no significance (see em/tm ratio the more difficult it is for the movement derived from these trials might Table 3). These comparisons were made o to detect target displacement during an have been based on the availability of separately for experimental and control Os. eye movement. This may be stated another object-relative displacement information} The direction of target displacement does way. The smaller the absolute extent of In no case did RT approximate the not appear to affect the perception of target displacement during an eye duration of the light flash, which was movement significantly under these movement the less likely it is to be 1/20 sec. Typically, RT was slightly less conditions. detected. The difference between means in than 1/2 sec. The only explanation we can Again using the Scheffe procedure, all conditions in which the em/tm is 2/5 and 0 offer for the prolonged duration ofthe RT means obtained at one level of em/tm were are not significant. Ss appear to be as is that the complexity of the O's task compared with the means obtained from accurate in reporting no target movement tended to make his responses to the light

294 Perception & Psychophysics, 1970, Vol. 8 (SA) Table 4 extent, for example, larger than 3 deg, and significant difference between Range (in Degrees) of Target Movement expressed as the percent of the total experimental and control Os. Also in both EM/TM Range number of eye movements of that extent conditions displacement ratios less than or larger occurring in any given IS-sec trial. 1/10 invariably lead to a significant 2/5 6.2-4.2 1/5 2.1-3.1 Both ways ofanalyzing the data were used. decrement on the part of experimental Os 1/10 1.6-1.1 One of the drawbacks of a condition of in the detection of target movement. 1/20 .8- .5 spontaneous eye movements for an In the NDEM condition for both investigation of this kind is that Os tend to experimental and control Os a decrease in Table 5 make very few spontaneous eye em/tm level produced a corresponding Mean Percent Movement Reports: NDEM movements that are larger than I deg under decrease in the amount of movement E C these experimental conditions, Le., when reported or in the number of movement the only visual target is a dimly glowing reports. This can be seen in the OEM 1 69.6 65.2 spot of light which moves only as some results as well. A clear difference, however, 1/2 55.0 59.8 2/5 52.5 56 function of the O's own eye movements between this data and the data from the 1/5 35.0 40.3 and which they are instructed simply to OEM condition is the marked degree of 1/10 31.3 41.1 observe. When the em/tm level is low, the similarity between the control and 1/20 18.8 27.1 am ou n t of target displacement experimental scores. The decrease in the Scores calculated on basis of percent of time accompanying a relatively small eye amount of movement reported by during each 15-sec trial in which movement movement is infinitesimal. Even with eye experimentals and controls with decreases is reported, expressed as a percent of 15 sec. movements as large as 3 deg, the amount of in em/tm level appears almost as striking in target displacement is almost negligible. the control as in the experimental data. We flashes more deliberate than they otherwise Thus there is only a slim chance that there attribute this similarity between might have been. are a sufficient number of occasions in experimental and control scores to the An analysis of the size of the eye which even a control 0 is able to detect difficulty in detecting any target movements made in response to the light target movement. Furthermore, in this movement in this condition. This may be flashes indicates that eye movements varied condition, Os are required to monitor the explained by the fact that the smaller the in size between 10.5 and 15.6 deg, with the behavior of the target light constantly, em/tm ratio the less the actual target average eye movement covering about which is likely to cause fatigue and displacement, and the less the actual target 13 deg of visual angle. An eye movement , consequent inaccuracies. displacement the less likely was the of 13 deg did, in fact, carry the eye from Table 5 depicts the results of the NDEM experimental 0 to make eye movements its focus at the center of the oscilloscope condition. Despite these drawbacks, the that were sufficiently large to displace the screen to either light, a distance of results obtained generally resemble those target a detectable amount. It is therefore 62.5 mm. Variations in the size of an eye of the OEM condition. Comparing each 0 not surprising that both experimental and movement did not appear to affect with himself as a control, the difference control Os show a decrease in ability to movement reports. Reports of movement betw~ the amount, the percent of each detect movement with a decrease in level were no more frequent or accurate when IS sec in which movement was reported, of em/tm that appears to be independent eye movements were on the large end ofthe for trials in which the em/tm ratio was 1/2, of whether or not target displacements eye-movement range than when they were 2/5, 1/5, 1/10, and 1/20 are as follows: occur during the O's eye movements. on the small end of this range. Variations 3.6%, 4.0%, 5%, 10%, and 12%, Despite this similarity between the in the size of the eye movements did, of respectively. In the 1/2, 2/5, and 1/5 patterJ1 of experimental and control scores, course, affect the absolute amount of conditions, the differences in amount of the pattern of significant differences is not target movement that accompanied any movement reported by both control and totally unlike that found in the DEM data. given eye movement. Since these variations experimental Os are trivial, whereas the As in the DEM results, the only significant in the size of eye movements did not seem differences between experimental and difference that appears in the control to affect movement reports, the control scores when the em/tm was either scores is that between the scores obtained concommitant variations in the absolute 1/10 or 1/20 are significant. This same when the em/tm equaled 1/20 and the amount of target movement also seem to pattern appears if a comparison is made scores obtained from all other em/tm have had no affect on these reports. The between the number of movement levels. For experimental Os, the scores ability to perceive target movement seems responses made by experimentals and obtained from trials in which the em/tm therefore to depend on the em/tm ratio controls following eye movements 3 deg or ratio was 2/5 or larger are significantly rather than on the absolute amount of larger at each of the em/tm levels (again different from scores obtained on trials in target movement. Table 4 describes the treating each 0 as his own control). These whi~h the em/tm was less than 2/5. This range of variation in the amount of target differences are as follows for the 2/5, 1/5, difference is not found in the DEM results movement produced by variations in eye and 1/10 em/tm trials: 11%, 11.4%, and in which the first significant break is movements in the various em/tm 17.7%, respectively. The only significant between scores obtained from trials in conditions. difference is between experimental and which this ratio is less than 1/5. NDEM control scores in the 1/10 em/tm results show no significant difference NDEM condition. between scores from the 1/5 and 1/10 The results of this condition are not Here, as in the OEM condition, direction conditions. Thus, with spontaneous eye simply described. Only two quantitative of target displacement did not significantly movements, target movement is no more descriptions of the data are possible; one is affect movement reports. More important, difficult to perceive when that movement the time dUring each IS-sec trial in which in both conditions there is a significant is dependent on the eye movement and 1/5 the 0 reports seeing target motion, difference between experimental and as great as the eye movement than when it expressed as a percent of 15 sec. The other control Os where the em/tm level is equal is 1/10 as great. As in the DEM results, is the number of movement reports to 1/10, and for both conditions this is the there is a significant difference between all following eye movements of a certain Hrst level of em/tm that produces a clear experimental scores obtained when the

Perception & Psychophysics, 1970, Vol. 8 (SA) 295 Table 6 some differences all results indicate that 15.6 deg to the right over the retina, and Mean Percent Correct Reports: OEM when target displacements are 2/5 as great movement will be seen. This seems to (Closer Viewing Distance) or greater than a concurrent eye movement indicate that when the discrepancy V+ V- H+ H- Grand X (in the main experiment, when target between eye and image movement is 20% displacements are 1/5 as great or greater), or greater, object movement will always be 2/5 98.8 96.3 95.0 87.5 94.4 1/5 92.5 86.1 71.9 58.9 77.2 that displacement is predictably perceived, seen. If, however, the eye moves and the 1/10 72.2 55.1 43.3 44.4 53.5 and when the displacement is only 1/20 as target displaces 1/20 of that distance, 1/20 35.7 33.5 7.5 20 24.1 great as the eye movement that movement is generally not seen. Thus, if displacement is generally not perceived. the eye moves 13 deg to the right and the em/tm ratio was 1/20 and when the ratio target moves .65 deg either in the same was 1/10 or greater. DISCUSSION (H+) or the opposite (H-) direction, the That our results are not limited to eye When the eye scans a stationary object, image will displace either 12.35 or movements of a particular absolute size is the image of that object displaces across 13.65 deg right over the retina. Under clear both from the results of the NDEM the retina of the eye in the same direction these circumstances, movement will not condition and from another study which and to the same extent that the eye has generally be seen, indicating that a predated the one here described. This moved. This is the normal condition for discrepancy of 5% between eye and image earlier study resembled the DEM condition perceiving position constancy and is displacement is acceptable and constitutes of the study reported with the following identical with the experimental condition grounds for the perception of a stable, exceptions: The Os sat closer to the in which the em/tm level is zero and there unmoving visual environment. That a 5% oscilloscope screen; the bite bar was placed is no objective target displacement. A look target displacement is equivalent to target 12.5 em from the screen so that an eye at the data obtained in the DEM condition immobility can be seen by comparing the movement of 26 deg was required to move where the em/tm is zero indicates that, in scores in the OEM condition when the the eye from the center of the screen to fact, movement is generally not reported em/tm was 1/20 with those scores obtained one of the lights; and no control condition under these circumstances. Target when the em/tm was zero. To make the was employed. Each 0 was run for two displacements 1/5 as great or greater than two scores comparable, we can subtract the experimental testing sessions but Os were the accompanying eye movement were as scores in the 1/20 condition from 100%, not tested in pairs. In every other way, this reliably perceived as the identical target arriving at a mean score of no movement experiment was identical to the DEM displacements that were independent of reports amounting to 88.2%, which is condition of the experiment reported. Ten eye movements. This result conclusively entirely comparable to the score of 93.2 in Os were tested. The results appear in establishes the fact that target the 0 condition in which there is actually Table 6. With a few exceptions, they displacements concurrent with eye no target displacement. resemble those results already reported. movements are not suppressed and that The fact that target displacements that The following are the exceptions: V 1/5 there is no complete saccadic suppression are 1/10 as large as concurrent eye scores are notably greater than the scores of movement perception during saccadic movements are not perceived as frequently in the H 1/5 conditions, although this eye movements. Results in the DEM as the identical displacements occurring difference is not significant at .05. (The conditions support the conclusion that independently of eye movements, but are critical value for a difference that is direction of movement perceived more frequently seen than displacements significant at .05 is 35.165, Scheffe, 1953. corresponds to the actual direction of that are only 1/20 as large as the eye The difference value obtained is 23.3.) This target displacement. movement, suggests that a discrepancy as trend, which was absent in the data A second conclusion to be drawn is that great as 10% between eye and image previously described but present in all but these results support the concept of a displacement is close to the threshold of the 2/5 condition in this experiment, comparator mechanism that matches what constitutes an acceptable basis for conforms to data reported by Wallach, available target displacement information perceiving stability during eye movements. Frey, and Rodney (1969) for head with available information about 0 eye It might be argued that these results are movement accompanied by abnormal movements. simply to be interpreted as an index of the displacements. Displacements that are at If this hypothetical comparator were to capacity for movement detection during right angles to head movement appear to operate with extreme precision, only a saccadic eye movements. Looked at in this be more easily perceived. It is not clear retinal-image displacement that exactly way the results indicate that the direction why this tendency is restricted to this set matched an eye movement would of target displacement during a saccade is of data. Another exception is that the constitute the condition for perceived not a factor in determining whether or not mean scores in the 2/5 condition are position constancy, that is, no movement that displacement will be perceived, and, significantly greater than those in the 1/5 perceived despite retinal-image movement. further, that there is the expected raising in condition. (Critical value for a difference at Thus, if the eye turned 13 deg to the right, the movement detection threshold during .05 is 16.3. The difference obtained is only a 13-deg rightward displacement over an eye movement. While this account of 17.2.) A difference between a displacement the retina would constitute a the results is legitimate, it in no way ratio of 2/5 and one of 1/5 does appear in position-constancy match. It is apparent invalidates or contradicts an account of the the NDEM data. The means obtained with from the data that eye-movement/image­ result in terms of a comparator mechanism. an em/tm of 1/20, with the exception of movement comparisons are not so precise. To insist on an account of the data strictly the H+ 1/20, are greater than the analogous I n most instances when the target in terms of thresholds and to rule out as scores in the previously reported DEM displacement is 1/5 as large as the eye illegitimate any other kind of account is to results. This may reflect the fact that in the movement, regardless of the direction, the confuse description and explanation. The previously reported results when the displacement is perceived. Thus, if, for concept of a cOIi~parator or compensatory em/tm was 1/20 an average eye movement example, the eye moves 13 deg right and mechanism is explanatory. The results of displaced the target .78 deg, whereas here the target shifts 2.6 deg in the same (H+) these experiments support this kind of the target was displaced 1.3 deg. or the opposite (H-) direction, the image explanation. To speak of threshold is The point to be made is that despite of the target will displace either 11.4 or simply to employ a well understood and

296 Perception & Psychophysics, 1970, Vol. 8 (SA) efficient terminology for describing results. several possibilities suggest themselves. I-deg image displacement in either the To infer from these results to the operation Perhaps the fact that a 5% mismatch same or the opposite direction as the eye of a comparator mechanism is an attempt between eye and visual displacement moved would "match" the eye movement, to account for them. information is an acceptable basis for and no movement should be perceived. The A puzzling aspect of the results is that position constancy during eye movements results of this experiment do not permit they fail to show any consistent effect of is a function of a saccadic inhibition of any evaluation of the relative merits of the direction of target displacement. For visual displacement detection during eye these possible explanations of the apparent the most part, the results indicate that the movements such that a 10·deg imprecision in the "match" between eye em/tm is the central variable. Since under displacement is indistinguishable from a and image movement. normal position-constancy conditions, the 9-deg displacement. These results reconfirm the fact that image of a stationary object displaces in Still another possibility might be that retinal displacement cannot be considered the same direction over the retina as the the tolerance for a mismatch as a basis for a sufficient condition for movement eye moves, it might be reasonable to position constancy may be the result of a perception since in all cases in which target expect that experimental conditions in faulty memory for egocentric direction. If displacements are in the same direction as which the image of the target displaces in the crucial components in the comparison the eye movement (H+), the higher the the same direction as the eye might lead between eye and visual displacement em/tm level the less the retinal image less frequently to perceived movement, all information were eye·movement displacement, yet the lower the em/tm other things being equal, than those cases information on the one hand and level the less likely is target movement to in which the image displaces in some other information about the visual direction of be perceived. In conclusion, these results direction. The only results that even an object before and after an eye clearly contradict those reported by suggest that this may be the case are from movement on the other hand, rather than Wallach and Lewis (1965) and extend the early DEM study in which the 0 sat information about retinal movement, then those of Mack and Bachant (1969). They rather close to the oscilloscope screen. As the tolerance for a mismatch might be the constitute evidence that "abnormal" reported, the mean percent correct scores result of an inherent inability to remember retinal movements concurrent with eye at each level of em/tm are here somewhat visual directions accurately. If an object movements do lead to perceived higher for the two V conditions than for were straight ahead at the onset of an eye movement. the two H conditions. This is exactly what movement and moved I deg during an eye one might expect, since in both H movement of 10 deg, perhaps the reason conditions, regardless of the level of that target movement is not perceived is em/tm, the image of the target displaced in the result of an inability to remember the same direction as the eye moved, precisely where the object was at the onset REFERENCES whereas in the V conditions the target was of the eye movement. If this were the case, BRINDLEY, G. S., & MERTON, P. A. The always displacing at right angles to the eye a I-deg displacement might be absence of position sense in the . so that the image was always displacing at indistinguishable from a O-deg Journal of Physiology, 1960, 153,127-130. an angle to the direction the eye moved. displacement and would account for the HAY, J. Visual adaptation to an altered No difference could be expected between fact that it is not perceived. In considering correlation between eye movements and head movements. Science, 1968, 160,429-430. the H conditions, and generally none this explanation it is necessary to assume MACK, A., & BACHANT, J. Perceived movement appears. Further testing is required in not only that memory for position of the afterimage during eye movements. which image displacement is directly becomes more inaccurate when the Perception & Psychophysics, 1969. 6, opposite to the direction of eye movement position to be remembered precedes an eye 319-384. MacKAY, D. M. Ways of looking at perception. before any conclusions are drawn about movement but, more important, that In w. Wathen-Dunn (Ed.), Models for the the relevance of the direction of image retinal movement is still registered in the preception of speech and visual form. displacement for the perception of target match between information about eye Cambridge: M.I.T. Press, 1961. movement. movement and change of visual direction, ROCK. I. The nature of perceptual adaptatzon. Another question left unanswered by since otherwise we would have to conclude New York: Basic Books, 1966. SHAFFER, 0., & WALLACH, H. this work is whether it is the percent of that no real movement is perceived during Exten t-of-motion thresholds under discrepancy between eye and image an eye movement. If retinal movement subject-relative and object-relative conditions. movement or the absolute discrepancy information did not playa role, only the Perception & Psychophysics. 1966. I, between these two inputs that determines inference of movement based on a change 441-451. if object movement will be seen. It is TEUBER. H. L. Perception. In J. Field (Ed.). in visual direction would be possible. From Handbook of physiology-neurophysiology 11/. possible that whenever there is a our own work, it would be extremely hard Baltimore: Williams & Wilkins, 1960. discrepancy between eye and image to argue that real movement is not seen Pp. 1595-1668. movement that is larger than 10%, target during a saccade, since Os reported no von HOLST. E. Relations between the central movement will generally be seen-or it is difference in the appearance of target nervous system and the peripheral organs. possible that whenever the discrepancy movement when they were tested under British Journal of Animal Behaviour, 1954, 2. 89-94. between eye and image movement is experimental and control conditions. WALLACH, H., & FREY, K. Adaptation in the greater than, for example, 1.5 deg, Finally, it might be argued that the constancy of visual direction measured by a movement will be seen. The fact that the apparent imprecision in the match between one-trial method. Perception & Psychophysics, results of the various conditions show the eye and target movement is a function of 1910,5,249-252. same pattern suggests that it is the percent inaccurate or imprecise eye-movement WALLACH. H., FREY. K., & RODNEY, G. Adaptation to field displacement during head of discrepancy which is crucial, but more information. If the eyes were to tum movement unrelated to the constancy ofvisual work is required. 10 deg to look at something off to the side, direction. Perception & Psychophysics, 1969, One could ask why the postulated and the information registered in the 5.253-256. comparator operation during eye nervous system is that the eye has moved WALLACH, H.. & KRAVITZ, J. The measurement of the constancy of visual movements is not more precise. This either more or less than 10 deg, perhaps direction and of its adaptation. Psychonomic question is unresolved by this study, but somewhere between 9 and 10 deg. then a Science, 1965. 2, 211-218.

Perception & Psychophysics, 1970, Vol. 8 (SA) 297 WALLACH. H•• & LEWIS, C. The effect of everyone who helped make this research possible. in the visual Held with respect to another is a ab,nonnal displacement of the retinal image I am particularly grateful to Dr. Leo Ganz for his powerful indicator of movement. during .eye movements. Perception & generous help and to Robert Brown who 4. We assume that the information about eye Psychophysics. 1966. 1, 25-29. designed and built some of the apparatus and movements that is matched against retinal WINER. B. 1. Statistical principles in assisted in the testing of Os. I would also like to displacement information is efferent rather than experimental design. New York: McGraw-Hili, express my gratitude to Hans Wallach. Irvin afferent information (Brindley & Merton, 1960). 1962. Rock, and Leon Festinger for their careful 5. Data reported by Schaffer and Wallach YARBUS. A. L Eye movementsand vision. New reading of an earlier version of this manuscript. (I966j indicates that Os are capable ofreporting York: Plenum Press. 1967. 2. Address: New School for Social Research. far smaller subject-relative target displacements. New York, New York 10003. Perhaps the low in tensity of the target in our NOTES 3. All discussion of movement perception is work accounts for this difference in results. 1. This research was done during an NIMH restricted to object movement. which is subject postdoctoral fenowship at Stanford University. I relative since. with movement that is object would like to express my appreciation to relative, the relative displacement of one object (Accepted for publication January 28. /970.)

298 Perception & Psychophysics, 1970, Vol. 8 (SA)