Vision Res. Vol. 27. No. 4, pp. 529-536, 1987 Printed in Great Britain. All tights r~servcd MOTION A. MAW, Copyright 0 AFTEREFFECTS ASSOCIATED EYE MOVEMENTS J. GOODWIN, H. THORDARSEN, D. BENJAMIN, 0042-6989/87 $3.00 + 0.00 1987 Pergamon Journals Ltd zyxwvutsrqpo WITH PURSUIT D. PALUMBO and J. HILL Russel Sage Foundation, 112 East 64th Street, New York, NY 10021, U.S.A. (Received 8 Ocfober 1985; in reoised form 9 September 1986) Abstract-Contrary to an earlier report [Anstis and Gregory, Q. JI exp. Psycho/. 17, 173-174 (1965)), we find that the sustained retinal motion caused by tracking a moving target over a statibnary grating does not result in a motion aftereffect (MAE) which is equivalent to that resulting from comparable retinal motion caused by actual motion of a grating. The MAE associated with tracking generally occurs in elements falling on areas not previously exposed to retinal motion. It is in the same direction as the previous retinal motion in the display and is apparently an induced MAE caused by a weak, below threshold MAE in the elements stimulating areas that were previously exposed to retinal motion. Based on an analysis of eye movement records, we do not believe that the weakness of the tracking MAE is primarily a function of the poor quality of the tracking eye movements. Other possible reasons for the weakness of the MAE are suggested. INTRODUCTION Prolonged exposure to the sustained undirectional motion of contours causes a motion aftereffect (MAE). A subsequently viewed pattern appears to move in the opposite direction. In a widely cited study, Anstis and Gregory (1965) assert that this effect is a function of the retinal, rather than the perceived motion of contours. This assertion is based on two parallel findings. When observers tracked a moving point over a stationary set of vertical bars, a condition which causes retinal but not perceived motion, a MAE was subsequently perceived. Conversely, if the observers tracked the moving vertical bars, a condition which eliminates their retinal but not their perceived motion, no MAE was perceived. According to Anstis and Gregory the MAE produced by tracking over a stationary pattern is, in fact, indistuinguishable from that caused by fixating while observing an actually moving pattern. Support for the Anstis and Gregory assertion is provided by a report of findings by Tolhurst and Hart (1972). They found that there were no differences between the adaptation produced by an actually moving grating and a grating whose motion was produced by tracking a moving point across it when it was stationary. In their study the measure of adaptation was the elevation of the contrast threshold for the moving pattern. 529 Contradictory evidence has also been reported, however. Morgan et al. (1976) found that when observers tracked moving stripes across a superimposed stationary or oppositely moving set of stripes, and then looked at a motionless version of this display, a MAE was perceived which was in the same direction as the previous retinal motion of the nontracked stripes. In other words, the MAE was opposite in direction to that reported by Anstis and Gregory. Morgan and his collaborators attribute their effect to induced motion (Duncker, 1929). To summarize their analysis: when observers track a moving stimulus, it is stable on the retina while all other visible contours displace. A normal MAE develops in the area of the retina stimulated by the displacing edges, which then induces an opposite motion in the area of the retina previously exposed to retinally stable stimuli. In one condition, observers tracked moving stripes visible to only one eye, while the other eye viewed only the surrounding stationary background. Since the resulting MAE was equally strong in each eye, they concluded that the effect did not depend on sight of the moving bars but rather on stimulation by the displacing background which induced a MAE in the previously tracked pattern. Morgan et al. attributed their failure to replicate Anstis and Gregory to the nature of their test stimulus. 530 A zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIH &‘tAC K 1’1 al Anstis and Gregory had used a photograph of plexer were used to generate the dispia>h -\ sandpaper as their test pattern whereas Morgan dense raster produced by the function generet zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA al. used the adapting pattern itself. This ators formed a horizontal band of iuminancc explanation seems unlikely, however. particuwhich crossed the CRT screen. This band of Iarly since Tolhurst and Hart also used the luminance was m~lltiplexed on to 3 vertical adapting pattern as the test pattern and found positions on the screen and converted to a results which were consistent with those re- square wave grating by a square wave signal ported by Antis and Gregory. from a third function generator. The output of Because of the importance of the Anstis and the third function generator was synchronized Gregory finding for our understanding of with the horizontal raster forming signal and MAEs and motion perception more generahy, it applied to the oscilloscope”s 2 axis input. .A seemed important to re-examine the comparafourth function generator was used to display bility of MAEs following an observation interthe fixation point. The 3 square wave gratings vaf in which the adapting motion was produced (with contrast levels approaching I) and the either by the actual motion of a grating pattern fixation point could be independently swept across the screen by a ramp wave form proor only retinal motion produced by tracking. The experiment designed to do this had two duced by a fourth function generator. Since the comparable conditions. In one condition the effective length of the gratings was several times observer tracked a set of moving bars and a the width of the screen, the gratings remained continuous across the 127cm width of the superimposed tracking target as they moved screen as they were swept across it. Finally. a between two rows of flanking, stationary bars. logic signal tripped by the ramp generator Thus the tracked bars were essentially stationretrace blanked the gratings during the retrace ary on the retina while the flanking bars moved. interval. The resulting dislays were stable and In the other condition which is considered a free of visible flicker with the individual stimucontrol condition. the observer fixated a stationary point centered on the stationary row of lus elements being refreshed at rates in excess of inner bars while the upper and lower rows of 250 Hz. The visual display was viewed from a distance flanking bars moved. In both cases, assuming accurate tracking, the image of outer bars dis- of 34Scm. It consisted of three rows of tight grey, vertical bars vertically separated by 1.06“ placed and therefore if there is a MAE, they and a centered fixation point. The background should subsequently appear to move. The inner was black and all testing was in the dark. so bars in both cases were stationary on the retina only the pattern itself was visible. The alterand therefore any subsequent perceived motion nating light and dark bars in the outer flanking in these bars must be induced. Were tracking rows each subtended a horizontal extent of perfect, fixation and tracking would yield equiv2.12”, and a vertical extent of 5.3’. The bars in alent amounts of both absolute and relative the inner center row had the same horizontal motion. This may be important since there dimension as the flanking bars but were 3.18” is evidence that MAEs are negligible or vertically. The bars in each row formed a square nonexistent if there is no relative motion (Wohlgemuth. 1911; Day and Strelow, 1971). wave grating with a spatial frequency of 0.236c/deg. When they moved, they covered a distance of 2 1.18”. EX PERI M EN T I During the adaptation period in the tracking Mefhod condition, the fixation point was initially 1.27” to the left of the right edge of the virtual viewing Subjects. Two separate groups of eight window and moved to the left covering a visual observers were paid for their participation. One extent of 19”. It moved in tandem with the group was tested in the tracking condition while middle row of bars and thus maintained its the other served in the zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA control condition. A total position in the center of a bar as it moved across of 16 subjects were tested. the screen. When the fixation point reached the Apparatus and stimuli. The visual display left edge of the screen, the entire display vanpictured in Fig. I was presented on Tektronix ished for 7OOmsec providing the observer with 51 IO oxitioscope with a fast phosphor CRT. time to saccade back to the right. The display The phosphor (PI 5) decays to 10% of its initial reappeared with the tracking target in its initial luminance in less than 3 psec. Wavetck function position again moving leftward with the bars. generators and a Tekronix Type 4701 multi- Pursuit eye movements 531 Fig. 1. Adaptation and test display. MAE was reported in the first trial up to three The velocity of the tracking target and inner additional trials were run. Testing was termibars was 4.4”/sec. The upper and lower flanking nated either on the trial in which the observer rows of bars were stationary. reorted a MAE or after the fourth trial. In the fixation control condition, the inner Observer’s rask. On actual display motion bars and centered fixation point remained trials, subjects were asked to carefully maintain stationary during the adaptation period. The fixation on the stationary point. On tracking upper and lower flanking rows of bars moved trials they were asked to fixate the center point rightward at 4.4”/sec. Every 4.32 set the display and track it carefully as it moved to the left. blanked (the fixation point remained visible) for When it reached the edge of the display window 700 msec simulating the tracking conditions. In where it disappeared, they were instructed to both conditions the stationary adapting pattern close their eyes, saccade back to the right edge with centered fixation point was the test stimuof the display aperture where the point relus. Eye movements were monitored by an SRI appeared, and begin tracking once again. The eye tracker (Crane and Steele, 1978). zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFE instruction to shut the eyes during the saccade Procedure. In the tracking condition caliwas, of course, meant to eliminate the remote bration of the eye tracker and a practice trackpossibility of visual stimulation by oppositely ing trial preceded actual testing. In the tracking moving contours that might interfere with the practice trial the observers tracked the moving adaptation. fixation point to the left and saccaded back to Subjects were told that immediately following the right just as they would during the actual the 90 set period in which they either tracked or adaptation period, but no bars were visible. fixated while the pattern moved, the display During the actual adaptation period the observers tracked the moving point superimposed on would briefly disappear and reappear with a centered fixation point. They were instructed to the moving middle row of bars. The flanking fixate this point when it became visible and bars were stationary. A trial lasted 90 set during which time the bars and fixation point made I8 report any motion in the center and/or flanking sweeps across the field. Following the 18th bars. Prior to testing they were told that sometimes it was possible to have a sense of motion sweep, the display blanked and reappeared with without seeing objects actually change their the fixation point centered and nothing moving. In the control condition the observer mainpositions and that this might be the character of tained fixation on the centered point throughout the motion they perceived. If any motion was the adaptation and test period. In both conperceived, they were asked to verbally report its ditions the observer reported any apparent direction and its duration. The experimenter motion in the middle and/or flanking rows of kept track of whether or not a MAE was bars and its duration and direction. Observers reported, the trial on which it occurred, its viewed binocularly with their heads held in direction and duration. The principle evidence position by a dental impression bite plate. If no of a MAE is therefore the number of times it 53’ zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA A. MACL tv al. was reported followmg the first. second. third or were a function of the quality of these eye fourth trial in the center and flanking bars. It is movements. The description of the eye movement data simple frequency data. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA from the tracking condition is based on averaged results from four typical sweeps of the eye Results across the field from the trial in which the If Anstis and Gregory’s results are to be observer reported a MAE. (Data from the one confirmed. we should find equivalent MAEs observer who failed to report any MAE was in the 2 conditions, since, assuming adequate taken from the fourth trial.) If we include data tracking, the retinal adapting motions are from this subject in our calculations. then an largely equivalent. In both conditions the outer, average of 86.6% ( 16.7 ‘) of the distance covered flanking bars displaced rightward which means by the eye in any single 19 sweep can be that they should appear to displace to the left attributed to smooth pursuit eye movements. (If during the test. Since in both conditions the we exclude the subject who failed to report any center bars were retinally stable, they should not MAE, this mean is increased to 900/b.) The appear to move. In the fixation condition in average velocity of pursuit movement for the 7 which the outer bars moved and the observers subjects who reported MAEs was 4 ‘set which fixated a stationary center point, 5 of the 8 was somewhat less than the target velocity of subjects reported a MAE to the left in the outer 4.4’jsec. The mean number of saccades on a single sweep was 6.67 and the mean extent of bars by the fourth adaptation trial. Seven of these saccades was 41.6’. For the seven subjects these 8 subjects also reported a MAE to the who reported a MAE. smooth motion acright in the center bars by the fourth trial, which we take to be an induced MAE generated by the counted for between 59 and 76 set (Z = 69 set) primary MAE in the outer bars. In sharp conof the 77.4 set of the adaptation period in which trast to these results, only one of the 8 subjects the pattern was visible. (A trial actually lasted in the tracking condition ever reported any for 9Osec but the pattern was blanked for MAE in the outer bars and the one subject who 12.6 set of that period.) For the one subject who did, did so on the third trial. However, as in the failed to report a MAE, smooth pursuit acfixation condition, 7 of the 8 subjects reported counted for an average of 53 of the 77.4 sec. an induced MAE to the right in the center bars The analysis of the eye records indicates that by the fourth adaptation trial. There was no on any single adaptation trial. tracking was clear difference in the duration of the MAE in likely to provide an average of about IO set the 2 conditions. In the control condition its (13%) less of smooth motion than actual patmean duration was 10.8 set (SD 6.5) while in tern motion observed while fixating. However. the tracking condition it was 15.9 set (SD 12.1). there were 4 subjects in the tracking condition Thus the difference between the results from the who reported the induced MAE on the first trial 2 conditions resides in the frequency of reports and for these subjects smooth pursuit accounted for an average of 95% of the trial. (The mean of the primary MAE in the outer bars. In the control condition 63% of the subjects reported tracking velocity for these subjects was 4.2, +e.c.) Nevertheless, none of these subjects reported a it, while in the tracking condition, only one subject (13%) reported it. primary MAE in the outer bars. While in conEy e movements. Although there were no trast, 3 subjects in the control condition redifferences between the results of the two condiported the primary MAE on the first trial. Certainly in the case of these subjects the tions with respect to the frequency of reports of difference in the perceived MAE does not seem the induced MAE in the center bars, there was to be largely a function of differences in exa sharp difference with respect to reports of the posure to smooth motion. Furthermore, if we primary MAE in the outer, flanking bars. What assume that the adaptation of the mechanism might account for this difference? One obvious underlying MAEs is cumulative, increasing with source of the difference might be the quality of exposure to the adapting stimulus, then certhe smooth eye motions in the tracking conditainly by the third or fourth tracking trial, the tion, If tracking were poor, periods of smooth amount of exposure to smooth pattern motion image motion would have been less frequent. If was at least as great, and probably greater, than the duration of the exposure to smooth image that provided by the first 2 fixation trials. Yet, motion is important in the production of by the second fixation trial. 50% of the subjects MAEs. then perhaps the differences in results Pursuit eye movements zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPON 533 in the control condition had reported the pri- induce an opposite MAE in a previously stamary MAE, while there was only one report of tionary surrounded stimulus rather than elicit a a primary MAE by the fourth trial in the primary MAE in the stimulus that previously tracking condition. For these reasons we think displaced. Evidence that this is so would lend it unlikely that much of the difference between credibility to the argument that the reason why the 2 conditions can be attributed to the quality tracking produces an induced MAE is because the primary MAE produced by image motion of the tracking. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA associated with tracking is weak. Of course, if Discussion this is true, we are still left with the question, why this should be. These results which fail to confirm Anstis and Gregory ( 1969) raise related questions. The first EXPERIME~ 2 question is why does the retinal motion caused by tracking eye movements rather than actual M ethod stimulus motion fail to elicit reports of a priSubjects. Fourteen observers were paid for mary MAE in the elements of the pattern which their participation. displaced over the retina during the adaptation App~rat~ and st~uIi. IIe display was identiperiod? In asking this question it is necessary to remember that the adapting retinal motion pro- cal to that used in the control condition of the duced by tracking was as effective as that pro- first experiment. Procedure. In order to try to produce a weak duced by actual stimulus motion in eliciting reports of a secondary or induced MAEs in the MAE, the period of adaptation was severely elements of the pattern which had been retinally abbreviated. During adaptation the observer stable during the adap~~on period. This neces- fixated the central s~tiona~ point which was sarily would seem to imply that a primary MAE superimposed on the center row of stationary was in fact produced in the flanking elements in bars. As in Experiment 1, the 2 outer sets of both conditions but, with one exception, was flanking bars moved rightward but now only for 4 cycles. Each cycle was again 4.32 set followed perceived and reported only in the condition involving actual stimulus motion. This must by a 700 mse-cblank period. Adaptation time in a singie trial was 17.28sec. Following the adap mean that for some reason the MAE produced in the tracking condition was weaker than that tation period, the observer reported whether all in the control condition and therefore (except or any part of the display which was now for the one subject who reports it in the tracking stationary, appeared to move and if so the direction and duration of the motion. If an condition) is below threshold. We know from Duncker (1929) that when observer failed to report any MAE after the first motion is below threshold and occurs in a trial, a second trial was run. stimulus which surrounds another stationary Results stimulus, it is only the stationary surrounded Of the 14 subjects tested, only 2 reported a stimulus which appears to move and does so in the opposite direction. Since the flanking bars primary MAE in the outer bars which had which had displaced retinally during adap~tion moved during a~ptation. These 2 subjects resurround the bars which had been stationary ported a MAE to the left, opposite the direction during adaptation, an undetected, below of the adapting motion. Twelve of the 14 subthreshold, primary MAE in the flanking bars jects, however, reported an induced MAE in the could induce an opposite motion in the center center bars to the right. Nine of the subjects bars. Moreover, if Rock et at. (1980) are correct reported this effect following the first adaptation in arguing that induced motion is motion sub- trial and the ~maining 3 reported the induced tracted from the motion of the inducing stimu- MAE following the second adaptation trial. The lus, then on this account as well, a weaker mean duration of the MAE was 5.3 set (SD primary MAE which induced motion in a sur- 2.8 set). rounded stimulus would be less likely to be perceived. The next experiment examines the proposition that a weak MAE, even if produced by actually moving bars which are observed while fixating a stationary stimulus will generally only Discussion As predicted, brief exposure to actual. pattern motion leads to a general failure to report a primary MAE in the elements of the pattern which had displaced during adaptation, but 534 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFE e r uf A. M ACK does cause the perception of an induced MAE which was turned on immediately al’ter :hc in the pattern elements which had been stationadaptation period. ary during adaptation and which are surProcedure. Since eye movements were again rounded by the previously moving eiements. recorded in this experiment, the eye cahbration These results appear to lend direct support to procedure was undertaken prior to actual testthe view that a weak MAE, itself below the ing. Once this was completed, the subject was detection threshold, will induce an opposite instructed to fixate and track the central point motion in a surrounded stimulus. They thereas it displaced to the left over the stationar? fore lend indirect support to the hypothesis that vertical bars. As in the earlier experiment when the induced MAE associated with tracking is the the tracking target reached the edge of the consequence of the fact that tracking causes a display, it and the bar elements blanked for 700msec and reappeared with the tracking tarweak MAE unlike that caused by an equivalent get again 1.27’ from the right edge of the display amount of retinal motion produced by actual and moved leftward at 4.4’/sec. A 45 set adappattern motion. tation period was used. Following the tracking Finding an induced MAE following tracking interval, the subject was instructed to turn is consistent with the report of Morgan et al. immediately to the illuminated photograph of but, because it failed to confirm the Anstis and Gregory results, seemed to demand further ex- the sandpaper and to report any appearance of motion, its direction and duration. ploration. As noted in the introduction, Morgan ez al. attributed their failure to confirm Anstis Results and Gregory to the difference in test pattern. None of the subjects reported seeing any Anstis and Gregory tested for a MAE with a motion in the test stimulus. In other words we photograph of sandpaper while we, like Morgan found no evidence of any MAE under these et al., used the adaptation pattern itself as the test stimulus. Although this did not seem to us conditions. The eye movement records verified the reason for the difference in outcome, we that tracking was reasonably accurate and nevertheless thought it reasonable to attempt to therefore could not account for this complete failure to elicit a MAE. replicate Anstis and Gregory under conditions which more closely resembled the ones they Discussion used. Thus in Experiment 3 observers tracked This failure once again to confirm the Anstis across a pattern of stationary bars for 45 see, the and Gregory findings led us to make one final adaptation interval used by Anstis and Gregory attempt to do so. However, instead of testing and then looked at a photograph of sandpaper, with the photograph of sandpaper which we which served as the test pattern. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA found to be a poor stimulus for eliciting a MAE even if the retinal motion was produced by EXPERIMENT 3 actual bar motion, we chose to test for a MAE Method using the adaptation pattern itself as we had done in the earlier experiments. Since moniSubjects. Five subjects were tested and paid toring eye movements required our adaptation for their participation. and testing to occur in the dark, whereas Anstis Apparatus and stimuli. The apparatus and stimuli used for the adaptation display were a and Gregory adapted and tested their subjects in a lit environment in which the display aperture version of that used in the earlier experiments. A single set of vertical bars were displayed on and other things as well were visible, we added the oscilloscope. These were actually the center a luminous frame to the oscilloscope screen. We set of bars from the previous displays. The did this despite the fact that we did not believe photograph which served as the test stimulus that the absence of a visible surround would was of coarse sandpaper (No. 36) which was make a difference when the subjects tracked. magnified 2$ times. The photograph was placed Relative motion is known to be important for MAEs and therefore the presence of a visible on a wall at right angles to the oscilloscope screen so that the subject had to turn 90” in stationary surround is important when a set of order to see it. It measured 23.5 by 20.9cm and moving elements serve as the adapting stimulus. subtended a visual angle of 10.4” vertically and With tracking, however, the adapting elements 11.7” horizontally at the viewing distance of are physically stationary and caused to displace retinally by pursuit eye movements which, of 114.3 cm. The photograph was lit by a lamp Pursuit eye movements course, also cause all other visible, stationary stimuli to displace in the same way. Thus, a visible stationary surround fails to provide any relative motion stimulation. In fact, when tracking a moving target over a stationary grating, the only relative motion is that between tracking target and everything else that is visible. From the description of the Anstis and Gregory report it would seem that relative motion in their tracking condition was also restricted to that between tracking point and whatever else was visible. Thus in this respect too, the experiment more closely duplicates the conditions present in the Anstis and Gregory testing situation. EXPERIMENT 4 Method Subjects. Ten observers were paid for their participation. Apparatus and stimuli. ne display was again presented on the oscilloscope and was identical to that of the first experiment. A luminous frame 10.5” by 11.7” surrounded the display. Procedure. During the adaptation period which was 45 set the subject tracked the moving fixation point leftward over the static array of bars. All 3 sets of bars were visible during both adaptation and test. When the adaptation period ended, the pattern froze with the fixation point at its center. The subject then reported whether any part of the display appeared to move and its direction and duration. Jesuits Only one of the 10 subjects reported a MAE in the bars. This subject reported motion as leftward which is opposite the adapting motion. In contrast, 7 of the 10 subjects reported an induced MAE to the right in the fixation point. The mean duration of this effect was 7.11 set (SD 6.07 set). Two subjects failed to report any MAE. Again our results fail to support those of Anstis and Gregory.* To make perfectly certain that the stimuli we were using would produce a *We continue lo be extremely puzzled by our failure to replicate the Anstis and Gregory tracking results. Perhaps a lit ~~ronrneRt is the critical difference. However, we have run a series of related MAE experiments which did not involve monitoring of eye movements and were carried out in dim ambient illumination and still failed to obtain their tracking results. 535 typical MAE when the adaptation occurred under more normal conditions, that is where the adapting motion was caused by grating motion, we ran one final control experiment in which the inner set of bars displaced between the set of outer flanking bars which were stationa~. EXPERIMENT 5 Method Subjects. Ten observers were tested. Apparatus, st~~~~ Ed zyxwvutsrqponmlkjihgfedcbaZY proce~re. Tote apparatus and stimuli were identical to those used in the first experiment. In this experiment, however, the inner bars displaced to the right and the outer bars were stationary. Adaptation lasted 45 set during which time the observer fixated the centered fixation stimlulus. All ten subjects reported a MAE to the left in the center bars. Its mean duration was 6.05 see (SD 2.3 NC). These results, like those in the fixation-control condition of the first experiment, thus demonstrate that the stimuli were appropriate for producing a standard MAE when adapting motion was a function of actual pattern motion. Discussion Contrary to prior reports, our results indicate that the retinal motion produced by actual pattern motion and that produced by tracking over an identical stationary display are not equally effective in causing a MAE. Tracking rarely causes the perception of a primary MAE in the elements which previously displaced over the retina, although it frequently causes the perception of an induced MAE in stimulus elements falling on areas of the retina not peviously exposed to motion. This seems to be because the MAE associated with retinal motion produced by tracking is weaker than that produced by an equivalent interval of actual pattern motion. In fact, the aftereffect associated with tracking appears to be sufficiently weak so that its presence is generally apparent only indirectly, through the induced motion of stimuli falling in areas not previously exposed to moving edges. There may be several reasons why this is so. The first and least interesting reason concerns the availability of relative motion information. In Experiment 4 where the adapting motion was 536 A. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJ MACK cl al. produced by tracking, the only relative motion was between the retinally stable tracking target and all other visible elements of the display (the physically stationary bars and the luminous viewing frame). This only caused an induced MAE in the fixation point that must have been induced by an unreported, weak MAE in the stimulus elements which had displaced during tracking. In Experiment 5 in which the inner bars displaced between flanking stationary bars every subject reported a primary MAE in the center bars. Clearly some of the difference between these two sets of results depend on the difference in the availability of relative motion. In Experiment 1, however, where we did not attempt to closely duplicate the testing conditions used by Anstis and Gregory, and every effort was made to equate the amounts of relative as well as absolute retinal motion in the two conditions, the MAE% elicited in the 2 conditions were still not comparable. The MAE produced by tracking was much weaker than that produced by actual pattern motion. Tracking produced only a single report of a primary MAE although it did elicit reports of induced MAEs. In contrast actual pattern motion was far more likely to elcit reports of primary MAEs as well as induced effects. Since, for the reason given earlier, we do not think that these differences can be largely or primarily attributed to the quality of the tracking eye movements, some other explanation must be sought. Although we have no independent evidence, we would like to suggest that at least some of the difference between tracking and fixation may be due to the damping of the motion signal by the compensation process which operates during tracking (Mack and Herman, 1978) and which accounts for the phenomenon of position constancy. If this is correct, it would mean that the compensation process which matches the retina1 motion signal against the eye movement signal, nulling it when a match is found, has an impact on the mechanism underlying MAEs. We suggest that there also may be another possible reason why tracking may cause a weaker MAE. It is, at least possible. that some of the diffrence between tracking and fixation might be due to differences in perceived motion. During tracking the retinal motion of the physically stationary elements is associated with little or no perception of motion. This is an instance of position constancy. On the other hand. the retinal motion caused by actual element motion in the fixation condition is accompanied by the perception of pattern motion. It is therefore possible that this difference contributes to the difference in outcomes. However, even if neither of these factors are responsible for. or even contribute to the difference between the MAEs caused by tracking and actual pattern motion, these experiments demonstrate that the retinal motion caused by tracking generates a distinctly weaker MAE and is therefore not comparable to that caused by actual stimulus motion. Acknowledgemenrs-This research was supported by an NSF research grant BNS 8310811. We thank Bob Fendrich and Dan Reisberg for various kinds of essential help. REFFRENCES Anatia S. hf. and Gregory R. L. (1965) The aftereRect of seen motion: the role of retinal stimulation and of eye moveaunt. Q. JI cxp. Psychol. 17, 173-114. Crane H. D. and Steele C. H. (1978) Accurate three dimensional eye tracker. Appl. Opt. 17, 691-704. Day R. H. and Strelow E. (1971) Reduction or disap pearance of visual aftereffect of movement in the abeence of a patterned surround. Narure, Lend. 238. 55.56. Morgan M., Ward R. and B~sr~ll E. (1976) The aftereffect of tracking eye movements. Perceprion 5, 307-317. black A. and Herman E. (1978) The loss of position constancy during pursuit eye movements. VLGon J&s. 18 55-62. Rock I., Auster M., Schiffman M. and Wheeler D. (1980) Induced movement based on subtraction of motion from the inducing object. /. exp. Psyckol. 6, 391403. Tolhurst D. J. and Hart G. (1972) A paychophyaical investigation of the dfccts of controlled eye movements on the movement detectors of the human visual system. Vision Res. 12, 1441-1446. Woh@muth A. (1911) On the aftereffect of seen motion. Br. J. P.rychol., Monogr.. Suppl. I.