- Email: [email protected]
Human cerebral potentials evoked by moving dynamic random DOT stereograms
5()
Electroencephalography and Clinical Neurophysiology, 1981, 52: 50-- 56
Elsevier/North-Holland Scientific Publishers, Ltd
HUMAN CEREBRAL POTENTIALS EVOKED BY MOVING DYNAMIC RANDOM DOT STEREOGRAMS M.J. HERPERS, H.B. CABERG and J.M.F. MOL
Department of Clinical Neurophysiology, Rijksuniversiteit Limburg, 6214 PA Maastricht (The Netherlands)
(Accepted for publication: March 24, 1981) Julesz published in 1960 for the first time a m e t h o d to make patterns of random dot stereograms which, when viewed by each eye separately appeared to be completely random, but viewed with a stereoscope formed threedimensional images. His random dot stereograms were static images. Using these static images in search for 'electrophysiological correlate of binocular depth perception in man', Regan and Spekreijse ( 1 9 7 0 ) a n d Regan and Beverly (1973) found cerebral evoked potentials. Later, Julesz (1971), Julesz et al. (1976) and Atkinson and Braddick (1979)used random dot stereograms which were made dynamic by changing frames of dots at a high rate. In 1976 Ross extended the ideas of Julesz (1960) and Julesz (1971) by generating dynamic random dot stereograms on an oscilloscope in which the dots were continuously generated one after the other. Ross (1976) programmed a computer to produce the stereograms on two oscilloscopes, one for the left and one for the right eye, each eye seeing only one oscilloscope. The position of each pair of dots was random. One dot of the pair was displayed on the oscilloscope for the right eye, and the homologous dot on the other scope for the left eye. In that way several thousand pairs of dots were generated per second. By shifting the dots in a defined a r e a a square in the centre of each scope, a few millimetres to the right in the left scope and to the left in the right scope -- this area appeared to be nearer to the observer than the background. Each random dot stereogram pro-
vided no information at all if seen with one eye only. On shifting the points in the opposite direction the 'target area' moved to the far field, also free in space, vivid and sharp edged. By programming the computer in such a way that shifting of the target could be steered, the observer saw an advancing or receding target square in reference to its surround w i t h o u t any loss in its delineation. Investigation of random dot stereograms in general is aimed at understanding the physiological mechanisms of stereopsis. This is of importance to analyse the organization and capacity of the visual system. It is expected that this will contribute to the diagnosis of local brain disturbances. The aim of the current investigation was to describe in further detail the stereoptic potential evoked by dynamic random dot stereograms, that had been found in our laboratory before (Mol and Caberg 1977). Like Lehmann and Julesz {1977) and Julesz et al. (1980), Mol and Caberg (1977) showed that the cerebral evoked potentials thus formed were produced only when both eyes looked simultaneously at the h o m o n y m o u s l y presented dynamic random dot stereogram. They also showed that the maximum of the stereoptic potentials starting at about 150 msec after shifting the target, were found in the centro-parietal region. This study also addresses the question of what change in the latency of the response might occur if the random dots were not presented simultaneously to the two eyes, but with an interocular interstimulus interval, as applied by Ross (1976) in psychological experiments.
0013-4649/81/0000 -0000/$02.50 © Elsevier/North-Holland Scientific Publishers, Ltd,
DYNAMIC RANDOM DOT STEREOGRAM EPs Metl~ods In 1979 the investigations were started with an 8K HP 5450A minicomputer which continuously generated the random d o t stereogram images, consisting of 1000 pairs of dots per second. The pairs of dots were generated sequentially. The t w o dots of a pair were generated with an interval of 50 psec. In 100 psec the light intensity of one d o t diminished to one-third of its initial brightness. The random d o t stereograms were displayed on the c o m p u t e r oscilloscope, the right part of the stereogram on the right half (4 cm X 5 cm) and the corresponding part on the left half of the equally divided scope. The centrally placed target area, a square of 2 cm X 2 cm, covered almost one-fifth of the total surface of the stereogram. The subjectscreen distance was 40 cm, so the stereogram subtended an angle of 5o40 ' by 7°10 ' of arc; and the target area an angle of 2 ° 54' by 2 ° 54' of arc. In all our experiments the basic position of the target was in front of the background. The stimulus consisted of an abrupt forward m o v e m e n t of the target to a new nearer position, in which its angular subtense was approximately twice the basic one. This m o v e m e n t of the target was internally steered by a stepfunction. The subject looked through a 40 cm long double case which completely deprived the left eye of the information intended for the right eye, and vice versa. To prevent vergence reactions of the subject, the fixation point was set at infinity b y using a 2.25 dioptric lens for each eye, built into the double case. Besides the intrinsic basic interocular delay time of 50 #sec between the t w o dots of a homologous pair, investigations were made with rand o m d o t stereograms in which the c o m p u t e r produced an interocular display delay of up to 50 msec (left eye l e a d i n g ) b e t w e e n t w o homologous dots. While generating the 1000 pairs of dots per second the c o m p u t e r calculated at the same time on line the average of 8 channels of the subjects electroencephalogram (EEG). Later
51 (1980) an HP 5451C minicomputer generated the same random d o t stereograms, b u t n o w consisting of 5000 pairs of dots per second, which were recorded on a video recorder (Sony VO-2630) and displayed on a television screen (Trinitron colour receiver CVM2250 E). The equally divided TV screen was covered with t w o polaroid plates at excluding polarization angles, while the subject sitting in front of it wore a pair of prism spectacles with matched polarizing glasses. Again, to prevent vergence reactions the subject's fixation point was set at infinity with t w o prism glasses each of 7.5A diopters (basis temporal). The displayed stereogram was 21.5 cm X 17 cm and the centraUy placed target area approximately 8.5 cm X 8.5 cm. Since the subject sat at a distance of 1 m from the screen, he saw the stereogram at an angle of 12°20 ' b y 9040 , of arc and the target area at an angle of 4 ° 54' b y 4 ° 54' of arc. Basic target position, stimulus form and the possibility of introducing an interocular display delay were the same as in the stereogram consisting of only 1000 pairs of dots per second. Chlorided silver disks were attached to the scalp according to the international 10-20 system. Though different derivations were used depending on the problem investigated, derivations to the ipsilateral ear gave the clearest responses. An average was made of 32 EEG epochs from 100 msec before to 900 msec after the described target shift. The bandpass of the recording system was 0.26--70 c/sec (3 dB down). For all experiments 11 normal healthy subjects, between 20 and 30 years of age, were tested in 34 sessions. The subject was comfortably seated in a semi-darkened room in front of the computer or TV set, looking through the adapting optics that separated the information presented to each eye. He was asked to relax as much as possible and to concentrate on the target and its m o v e m e n t towards him. After a short period of habituation to the target m o v e m e n t the averaging of the EEG potentials was started. (A) In the basic experiment we were con-
52
cerned with the spatio-temporal distribution and latency of the response. Each of the different investigations described below was preceded by this basic experiment, using at least 10 electrodes. (B) To study the influence of the described delay procedure, 5 subjects were tested with 4 different interocular interstimulus intervals (from n o w on called interocular delays). In all delay experiments random d o t stereograms consisting of 1000 pairs of dots were used. In the current investigation only delay with the left eye leading was used. One control recording was made of a subject in which the right eye was leading. (C) The influence on the stereoptic response of the number of dots presented has been investigated by recording responses to random d o t stereograms consisting of 1000 pairs o f dots/sec and 500 pairs of dots/sec. (D) To see what might happen to the response if the subject's attention was distracted from the target shift, 3 subjects concentrated on a marked point in the middle of the stereogram (5000 pairs of dots/sec) in the background plane, disregarding the moving target. (E) The possibility that an accommodation reflex might produce the cerebral evoked potentials had to be ruled out. Therefore records were made before and after accommodation in both eyes was blocked. For this purpose 4 drops of Cyclogyl 1% (cyclopentolate hydrochloride) were used after one drop Novezine 0.4% (oxybuprocaine). The loss of accommodation was proved by reading tests.
Results
(A) Like Mol and Caberg (1977) we found the maximum response in the centro-parietal region. Though we searched for a response earlier than 130--150 msec after the stimulus, no earlier response components could be f o u n d as is shown in Figs. I and 2 {note the latency differences between the parietal and occipital leads in Fig. 2).
M.J. HERPERS ET AL.
I/ r
I
Fig. 1. Cerebral response to a sudden change in depth plane of an imaginary target provided by a pair of random dot stereograms. Lead: C3-A1. In this and subsequent figures, target shift occurs at 100 msec. RC, 0.6 sec HF, 70 Hz (--3 dB). Plots of 32 averages each. Note first negative peak N1 at 200 msec after the stimulus (STIM).
The mean responses to random dot stereograms consisting of 5000 pairs of dots/sec of the 11 subjects examined started 130--150 msec after the target shift, beginning with a negative peak which reached its maximum
/
l.vJ
Fig. 2. Spatial distribution of the stereoptic cerebral response. Note the fairly small occipital response with respect to the centro-parietal response, and the latency differences between the parietal and occipital leads.
DYNAMIC RANDOM DOT STEREOGRAM EPs
53
(NI) at a b o u t 200--250 msec after the shift. A positive dip ( P : ) w a s usually found at 280-300 msec and a second negative peak (N2) at 320--350 msec after the stimulus. Of the various response c o m p o n e n t s the N I was the most constant. The earliest onset of the response was always in the central or parietal region. Fig. 3 shows the s y m m e t r y of the response in the two hemispheres during the first 250 msec after the stimulus. After that an asymmetry appears. However, great variations in amplitude of the various response components were seen between the different subjects tested, ranging from 20 to 80 pV. Individually, each subject showed fluctuations in the largest amplitudes, comparing several records. However, 3 subjects showed a marked constancy of both amplitude and latency of the evoked responses (Fig. 4). (B) Because of its constancy, the latency of the N1 peak maximum was used as a reference point in the experiments with interocular delay. In 9 experiments 5 different subjects were tested, showing marked constancy of
STIM
STIM
N1
C3->P3
C3->C4
N2
P2
P3->O1
C4->P4
I20/uV 2
I~
4
~
MSEC
I~
2
~
40~
MSEC
Fig. 3. Interhemispheric asymmetry of the late response components. Leads as marked in the figure. At 250 msec after the stimulus the left hemisphere is more negative than the right. At 325 msec after the stimulus the right hemisphere is more negative. Note the latency differences of the N1-P2 response components in the C3 ~ P3 and C4 -> P4 leads compared to the P3 -* 01 and P4 -~ 02 leads.
STIM
[]
4~0
80'0 MSEC
Fig. 4. Constancy of stereoptic response, obtained at a 3 week interval (A and B).
latency shifts. An interocular delay (left eye leading) of 15 msec was accompanied by an N1 peak latency shift of 58 msec in one subject. In t w o subjects an interocular delay of 20 msec (left eye leading)caused a mean NI peak latency shift of 75 msec (S.D. + 7 msec). An interocular delay (left eye leading) of 25 msec (3 subjects) and 27 msec (3 subjects) was accompanied by mean N1 peak latency shifts of 93.3 msec (S.D. + 8 msec) and 96.2 msec (S.D. + 11 msec) respectively. One subject was tested once with an interocular delay of 20 msec (right eye leading). An N1 peak latency shift of 79 msec was found, so no further investigations with the right eye leading were done. Apart from the latency shift the amplitude of N1 was smaller in most cases, and smoothing of the P2 and N2 peaks of the response complex occurred. The maximum interocular delay that could be used in our tests was 27 msec. B e y o n d that interval the depth perception became ambiguous, and subjects did n o t perceive the target or its m o v e m e n t as clearly as before. (C) In this investigation a small latency shift of the N~ peak was found, and its amplitude was markedly reduced by reduction of
54 the rate of generation of dots from 1000 to 500 pairs/sec. P2 and N2 peaks were smoothed or even hard to distinguish, as in the records with interocular delay. However, using rand o m d o t stereograms consisting of 5000 pairs of dots per second, no appreciable shortening of the N~ peak latency occurred. (D) Distracting the subjects attention from the target shift by fixation on a black point in the middle of the background plane had no appreciable influence on the response. (E) After accommodation had been completely blocked, the target was seen moving clearly and sharp edged as before. An observation of psychological interest was made by all experimental subjects. They all reported that the square target seemed to become congruently smaller as it moved towards them, and bigger as it moved away behind the background.
Discussion
In general it is not easy to interpret electrical phenomena recorded from the outside of the skull, in terms of the spatio-temporal relations of potential fields in humans (Pockberger et al. 1979). Still we think, based on the results described above, that the evoked potentials do n o t originate in the occipital region. Records as shown in Fig. 3 could indicate an asymmetry of activity between t w o hemispheres concerning stereoptic depth perception, though the significant potential differences only start at 250 msec after the stimulus. This finding is remarkable in relation to the suggestion of 'long-lasting independent data handling in the two hemispheres up to 300 msec after the stimulus' of Lehmann and Julesz (1977, 1978). Using 4 different interocular delays, 4 N~ peak latency shifts, much longer than the imposed delay, were found. Though the relatively constant correlation of interocular delay and latency shift seems intriguing, this correlation remains far from established because of the limited number of interocular delays and subjects. The reasons
M.J. HERPERS ET AL. for the smoothing of the P2 and N2 peaks needs further investigation. Though there must be a minimal amount of dots with disparity before a target can be formed by the brain, this could not be established with certainty in our experiments. Ross (1976) stated 'that for single shapes of a reasonable size the lower limit is about 2000 pairs of points/sec.' In contrast to this finding all our subjects clearly saw the sharp edged target, even though only 1000 pairs of dots/ sec were generated. (This means only 200 dots/sec in the target area.) That the displayed number of dots/sec must reach a 'saturation point' concerning the latencies of the various response peak maxima is obvious because no appreciable shortening of the N~ peak latency evoked by 5000 pairs of dots/sec was established. So, since every random d o t of the used stereograms was generated one after the other, the number of dots and the rate at which they were generated can influence the latency of the response provided that at least a part of the evoked potential stands for the electrical manifestation of 'global stereopsis' (Julesz 1971; Bishop 1975). This is probably the reason why Regan and Spekreijse (1970) found a latency of only 94 msec in their investigations using static random d o t images. In static random dot stereograms all points with retinal disparity are shifted at the same time, and no time for 'collecting' enough dots to delineate the target is needed. These investigations with changes in the number of dots showed a smaller amplitude and a slight increase of the N~ peak latency. This could be explained by the fact that, with fewer points presented the chance that any given point falls in the target area on the display scope was also diminished. So it t o o k longer before enough points with disparity were collected to form the actual target by means of global stereopsis. The maximum interocular delay before depth perception became ambiguous was in our experiments only 27 msec. This finding is n o t in accordance with the results of Ross (1976), who found that the maximum delay before depth
DYNAMIC RANDOM DOT STEREOGRAM EPs perception disappeared could be as much as 50 msec. An explanation for this difference could be the smaller a m o u n t of dots presented in our experiments. With more dots, more reference points are available at a given m o m e n t and the influence of an interocular delay on the a m o u n t of displayed information is n o t so grave, resulting in a longer permitted interocular delay. Consistent with the work of Lehmann and Julesz (1977) and Lehmann et al. (1978) are the results found in distracting the subjects from the target shift. The response does n o t differ essentially from that found in the normally used test situation. We showed that with random d o t stereograms depth perception persists if a c c o m m o d a t i o n is blocked, so it seems very unlikely that the evoked stereoptic potential involves a neural accommodation mechanism. Since the size of the target remained constant before and after the shift, the p h e n o m e n o n of the illusory magnitude changes of the target m a y well be explained as follows. In general, an approaching target, of constant size, is perceived with an increasing visual angle. In random d o t stereograms, an approaching target is perceived with a constant visual angle. This is apparently interpreted by the brain as an approaching target with diminishing size, compatible with daily life experience. The same reasoning goes for the target moving to the far field, seemingly becoming bigger. A fundamental question concerns the neural processes that generate the evoked potentials. Are these potentials only caused by the neural mechanism comparing dots or clusters of dots with retinal disparities, the so-called 'local stereopsis' of Julesz (1971), or are they a response generated by the 'global stereopsis', forming the sharp edged target o u t of the local disparities. A third possibility would be a combination or interference of the t w o mechanisms, because b o t h are actually involved in the perception of the target shift.
55 Summary In 11 normal healthy human subjects an evoked potential was elicited b y moving dynamic random d o t stereograms. The random dots were generated by a minicomputer. An average of each of 8 EEG channels of the subjects tested was made. The maximum of the cerebral evoked potentials thus found was localized in the central and parietal region. No response earlier than 130--150 msec after the stimulus could b e proved. The influence of fixation, the number of dots provided, an interocular interstimulus interval in the presentation of the dots, and lense accommodation movements on the evoked stereoptic potentials was investigated and discussed. An interocular interstimulus interval (left eye leading) in the presentation of the dots caused an increase in latency of the response much longer than the imposed interstimulus interval itself. It was shown that no a c c o m m o d a t i o n was needed to perceive the depth impression, and to evoke the cerebral response with rand o m d o t stereograms. There are indications of an asymmetry between the t w o hemispheres in the handling of depth perception after 250 msec. The potential distribution of the evoked potentials strongly suggests that they are not generated in the occipital region.
R~sum~ Potentiels dvoquds cdrdbraux humains aux mouvements de stdreogrammes formgs de 'random dots' dynamiques Des potentiels c~r~braux ont ~t~ ~voqu~s chez 11 sujets adultes ~ partir de st~r~ogrammes ~ base de configurations de 'random dots' dynamiques (ces random dots ~taient g~n~r~s par un mini-ordinateur). La m o y e n n e de chacune des 8 d~rivations ~lectroenc~phalographiques auxquelles chaque sujet ~tait soumis, a ~t~ calcul~e. Le maximum des potentiels ~voqu~s, d~termin~ de cette fa~on, se situait dans les r~gions cen-
56 trale et pari~tale. Des c o m p o s a n t e s de r~ponse s u c c ~ d a n t au stimulus en m o i n s de 1 3 0 - - 1 5 0 msec, ne s ' o b s e r v a i e n t pas. I1 a ~t~ examin6 l ' i n f l u e n c e que les facteurs suivants e x e r c e n t sur les p o t e n t i e l s ~voqu~s: la f i x a t i o n des sujets ( d ' e x p ~ r i e n c e ) ; la quantit~ de points; le d~lai inter-oculaire dans la pres e n t a t i o n des ' r a n d o m d o t s ' d y n a m i q u e s ; les m o u v e m e n t s d ' a c c o m m o d a t i o n de la lentille. L a latence des r~ponses ~voqu~es par ce d~lai inter-oculaire se m o n t r a i t plus longue que l'intervalle impos~ de celui-ci (les ' r a n d o m d o t s ' d y n a m i q u e s o n t ~t~ pr~sent~s d ' a b o r d l'oeil g a u c h e et apr~s ~ l'oeil droit). I1 a ~t~ d ~ m o n t r ~ que l ' a c c o m m o d a t i o n n ' e s t pas n~cessaire p o u r percevoir en p r o f o n d e u r , ni p o u r ~voquer des potentiels c~r~braux ~ partir de st~r~ogrammes ~ base de ' r a n d o m d o t s ' dynamiques. D e u x c e n t c i n q u a n t e millisecondes apr~s le stimulus s'observait une asym~trie entre les d e u x h~misph~res. La d i s t r i b u t i o n spatio-temporelle des p o t e n t i e l s ~voqu~s, telle que n o u s l ' a v o n s constat~e, n~cessite u n e autre explicat i o n q u e celle de la p r o v e n a n c e occipitale.
References Atkinson, J. and Braddick, O. Assessment of vision in infants. Trans. ophthal. Soc. U.K., 1979, 99: 338-343. Bishop, P.O. Binocular vision. In: R.A. Moses (Ed.), Adler's Physiology of the Eye. Mosby, Saint Louis, Mo., 1975: 558--614. Julesz, B. Binocular depth perception of computer-
M.J. HERPERS ET AL. generated patterns. Bell System Technol. J., 1960~ 39: 1125--1162. Julesz, B. Foundations of Cyclopean Perception. University of Chicago Press, Chicago, Ill., 1971. Julesz, B., Breitmeyer, B.G. and Kropfl, W. Binoculardisparity-dependent upper-lower hemifield anisotropy and left-right hemifield isotropy as revealed by dynamic random dot stereograms. Perception, 1976, 5: 129--141. Julesz, B., Kropfl, W. and Petrig, B. Large evoked potentials to dynamic random dot corretograms and stereograms permit quick determination of stereopsis. Proc. nat. Acad. Sci. (Wash.), 1980, 77: 2348--2351. Lehmann, D. and Julesz, B. Human average evoked potentials elicited by dynamic random dot stereograms. Elelctroenceph. clin. Neurophysiol., 1977, 43 : 469. Lehmann, D. and Julesz, B. Lateralized cortical potentials evoked in humans by dynamic random dot stereograms. Vision Res., 1978, 18: 1265--1271. Lehmann, D., Skrandies, W. and Lindenmaier, C. Sustained cortical potentials evoked in humans by binocularly correlated, uncorrelated and disparate dynamic random dot stimuli. Neurosci. Lett., 1978, 10: 129--134. Mol, J. and Caberg, H. Contingent stereoptic variation as a response to moving random dot stereograms. Electroenceph. clin. Neurophysiol., 1977, 43: 513. Pockberger, H., Petsche, H., Rappelsberger, P., MiillerPaschinger, I.B. und Prohaska, O. Epi- und intrakortikale Aspekte visuell evozierter Potentiale. Z. EEG-EMG, 1979, 10: 184--193. Regan, D. and Beverly, K.I. Electrophysiological evidence for existence of neurones sensitive to direction of depth movement. Nature (Lond.), 1973, 246 : 504--506. Regan, D. and Spekreijse, H. Electrophysiological correlate of binocular depth perception in man. Nature (Lond.), 1970, 225: 92--94. Ross, J. The resources of binocular perception. Scient. Amer., 1976, 569: 80--86.
Electroencephalography and Clinical Neurophysiology, 1981, 52: 50-- 56
Elsevier/North-Holland Scientific Publishers, Ltd
HUMAN CEREBRAL POTENTIALS EVOKED BY MOVING DYNAMIC RANDOM DOT STEREOGRAMS M.J. HERPERS, H.B. CABERG and J.M.F. MOL
Department of Clinical Neurophysiology, Rijksuniversiteit Limburg, 6214 PA Maastricht (The Netherlands)
(Accepted for publication: March 24, 1981) Julesz published in 1960 for the first time a m e t h o d to make patterns of random dot stereograms which, when viewed by each eye separately appeared to be completely random, but viewed with a stereoscope formed threedimensional images. His random dot stereograms were static images. Using these static images in search for 'electrophysiological correlate of binocular depth perception in man', Regan and Spekreijse ( 1 9 7 0 ) a n d Regan and Beverly (1973) found cerebral evoked potentials. Later, Julesz (1971), Julesz et al. (1976) and Atkinson and Braddick (1979)used random dot stereograms which were made dynamic by changing frames of dots at a high rate. In 1976 Ross extended the ideas of Julesz (1960) and Julesz (1971) by generating dynamic random dot stereograms on an oscilloscope in which the dots were continuously generated one after the other. Ross (1976) programmed a computer to produce the stereograms on two oscilloscopes, one for the left and one for the right eye, each eye seeing only one oscilloscope. The position of each pair of dots was random. One dot of the pair was displayed on the oscilloscope for the right eye, and the homologous dot on the other scope for the left eye. In that way several thousand pairs of dots were generated per second. By shifting the dots in a defined a r e a a square in the centre of each scope, a few millimetres to the right in the left scope and to the left in the right scope -- this area appeared to be nearer to the observer than the background. Each random dot stereogram pro-
vided no information at all if seen with one eye only. On shifting the points in the opposite direction the 'target area' moved to the far field, also free in space, vivid and sharp edged. By programming the computer in such a way that shifting of the target could be steered, the observer saw an advancing or receding target square in reference to its surround w i t h o u t any loss in its delineation. Investigation of random dot stereograms in general is aimed at understanding the physiological mechanisms of stereopsis. This is of importance to analyse the organization and capacity of the visual system. It is expected that this will contribute to the diagnosis of local brain disturbances. The aim of the current investigation was to describe in further detail the stereoptic potential evoked by dynamic random dot stereograms, that had been found in our laboratory before (Mol and Caberg 1977). Like Lehmann and Julesz {1977) and Julesz et al. (1980), Mol and Caberg (1977) showed that the cerebral evoked potentials thus formed were produced only when both eyes looked simultaneously at the h o m o n y m o u s l y presented dynamic random dot stereogram. They also showed that the maximum of the stereoptic potentials starting at about 150 msec after shifting the target, were found in the centro-parietal region. This study also addresses the question of what change in the latency of the response might occur if the random dots were not presented simultaneously to the two eyes, but with an interocular interstimulus interval, as applied by Ross (1976) in psychological experiments.
0013-4649/81/0000 -0000/$02.50 © Elsevier/North-Holland Scientific Publishers, Ltd,
DYNAMIC RANDOM DOT STEREOGRAM EPs Metl~ods In 1979 the investigations were started with an 8K HP 5450A minicomputer which continuously generated the random d o t stereogram images, consisting of 1000 pairs of dots per second. The pairs of dots were generated sequentially. The t w o dots of a pair were generated with an interval of 50 psec. In 100 psec the light intensity of one d o t diminished to one-third of its initial brightness. The random d o t stereograms were displayed on the c o m p u t e r oscilloscope, the right part of the stereogram on the right half (4 cm X 5 cm) and the corresponding part on the left half of the equally divided scope. The centrally placed target area, a square of 2 cm X 2 cm, covered almost one-fifth of the total surface of the stereogram. The subjectscreen distance was 40 cm, so the stereogram subtended an angle of 5o40 ' by 7°10 ' of arc; and the target area an angle of 2 ° 54' by 2 ° 54' of arc. In all our experiments the basic position of the target was in front of the background. The stimulus consisted of an abrupt forward m o v e m e n t of the target to a new nearer position, in which its angular subtense was approximately twice the basic one. This m o v e m e n t of the target was internally steered by a stepfunction. The subject looked through a 40 cm long double case which completely deprived the left eye of the information intended for the right eye, and vice versa. To prevent vergence reactions of the subject, the fixation point was set at infinity b y using a 2.25 dioptric lens for each eye, built into the double case. Besides the intrinsic basic interocular delay time of 50 #sec between the t w o dots of a homologous pair, investigations were made with rand o m d o t stereograms in which the c o m p u t e r produced an interocular display delay of up to 50 msec (left eye l e a d i n g ) b e t w e e n t w o homologous dots. While generating the 1000 pairs of dots per second the c o m p u t e r calculated at the same time on line the average of 8 channels of the subjects electroencephalogram (EEG). Later
51 (1980) an HP 5451C minicomputer generated the same random d o t stereograms, b u t n o w consisting of 5000 pairs of dots per second, which were recorded on a video recorder (Sony VO-2630) and displayed on a television screen (Trinitron colour receiver CVM2250 E). The equally divided TV screen was covered with t w o polaroid plates at excluding polarization angles, while the subject sitting in front of it wore a pair of prism spectacles with matched polarizing glasses. Again, to prevent vergence reactions the subject's fixation point was set at infinity with t w o prism glasses each of 7.5A diopters (basis temporal). The displayed stereogram was 21.5 cm X 17 cm and the centraUy placed target area approximately 8.5 cm X 8.5 cm. Since the subject sat at a distance of 1 m from the screen, he saw the stereogram at an angle of 12°20 ' b y 9040 , of arc and the target area at an angle of 4 ° 54' b y 4 ° 54' of arc. Basic target position, stimulus form and the possibility of introducing an interocular display delay were the same as in the stereogram consisting of only 1000 pairs of dots per second. Chlorided silver disks were attached to the scalp according to the international 10-20 system. Though different derivations were used depending on the problem investigated, derivations to the ipsilateral ear gave the clearest responses. An average was made of 32 EEG epochs from 100 msec before to 900 msec after the described target shift. The bandpass of the recording system was 0.26--70 c/sec (3 dB down). For all experiments 11 normal healthy subjects, between 20 and 30 years of age, were tested in 34 sessions. The subject was comfortably seated in a semi-darkened room in front of the computer or TV set, looking through the adapting optics that separated the information presented to each eye. He was asked to relax as much as possible and to concentrate on the target and its m o v e m e n t towards him. After a short period of habituation to the target m o v e m e n t the averaging of the EEG potentials was started. (A) In the basic experiment we were con-
52
cerned with the spatio-temporal distribution and latency of the response. Each of the different investigations described below was preceded by this basic experiment, using at least 10 electrodes. (B) To study the influence of the described delay procedure, 5 subjects were tested with 4 different interocular interstimulus intervals (from n o w on called interocular delays). In all delay experiments random d o t stereograms consisting of 1000 pairs of dots were used. In the current investigation only delay with the left eye leading was used. One control recording was made of a subject in which the right eye was leading. (C) The influence on the stereoptic response of the number of dots presented has been investigated by recording responses to random d o t stereograms consisting of 1000 pairs o f dots/sec and 500 pairs of dots/sec. (D) To see what might happen to the response if the subject's attention was distracted from the target shift, 3 subjects concentrated on a marked point in the middle of the stereogram (5000 pairs of dots/sec) in the background plane, disregarding the moving target. (E) The possibility that an accommodation reflex might produce the cerebral evoked potentials had to be ruled out. Therefore records were made before and after accommodation in both eyes was blocked. For this purpose 4 drops of Cyclogyl 1% (cyclopentolate hydrochloride) were used after one drop Novezine 0.4% (oxybuprocaine). The loss of accommodation was proved by reading tests.
Results
(A) Like Mol and Caberg (1977) we found the maximum response in the centro-parietal region. Though we searched for a response earlier than 130--150 msec after the stimulus, no earlier response components could be f o u n d as is shown in Figs. I and 2 {note the latency differences between the parietal and occipital leads in Fig. 2).
M.J. HERPERS ET AL.
I/ r
I
Fig. 1. Cerebral response to a sudden change in depth plane of an imaginary target provided by a pair of random dot stereograms. Lead: C3-A1. In this and subsequent figures, target shift occurs at 100 msec. RC, 0.6 sec HF, 70 Hz (--3 dB). Plots of 32 averages each. Note first negative peak N1 at 200 msec after the stimulus (STIM).
The mean responses to random dot stereograms consisting of 5000 pairs of dots/sec of the 11 subjects examined started 130--150 msec after the target shift, beginning with a negative peak which reached its maximum
/
l.vJ
Fig. 2. Spatial distribution of the stereoptic cerebral response. Note the fairly small occipital response with respect to the centro-parietal response, and the latency differences between the parietal and occipital leads.
DYNAMIC RANDOM DOT STEREOGRAM EPs
53
(NI) at a b o u t 200--250 msec after the shift. A positive dip ( P : ) w a s usually found at 280-300 msec and a second negative peak (N2) at 320--350 msec after the stimulus. Of the various response c o m p o n e n t s the N I was the most constant. The earliest onset of the response was always in the central or parietal region. Fig. 3 shows the s y m m e t r y of the response in the two hemispheres during the first 250 msec after the stimulus. After that an asymmetry appears. However, great variations in amplitude of the various response components were seen between the different subjects tested, ranging from 20 to 80 pV. Individually, each subject showed fluctuations in the largest amplitudes, comparing several records. However, 3 subjects showed a marked constancy of both amplitude and latency of the evoked responses (Fig. 4). (B) Because of its constancy, the latency of the N1 peak maximum was used as a reference point in the experiments with interocular delay. In 9 experiments 5 different subjects were tested, showing marked constancy of
STIM
STIM
N1
C3->P3
C3->C4
N2
P2
P3->O1
C4->P4
I20/uV 2
I~
4
~
MSEC
I~
2
~
40~
MSEC
Fig. 3. Interhemispheric asymmetry of the late response components. Leads as marked in the figure. At 250 msec after the stimulus the left hemisphere is more negative than the right. At 325 msec after the stimulus the right hemisphere is more negative. Note the latency differences of the N1-P2 response components in the C3 ~ P3 and C4 -> P4 leads compared to the P3 -* 01 and P4 -~ 02 leads.
STIM
[]
4~0
80'0 MSEC
Fig. 4. Constancy of stereoptic response, obtained at a 3 week interval (A and B).
latency shifts. An interocular delay (left eye leading) of 15 msec was accompanied by an N1 peak latency shift of 58 msec in one subject. In t w o subjects an interocular delay of 20 msec (left eye leading)caused a mean NI peak latency shift of 75 msec (S.D. + 7 msec). An interocular delay (left eye leading) of 25 msec (3 subjects) and 27 msec (3 subjects) was accompanied by mean N1 peak latency shifts of 93.3 msec (S.D. + 8 msec) and 96.2 msec (S.D. + 11 msec) respectively. One subject was tested once with an interocular delay of 20 msec (right eye leading). An N1 peak latency shift of 79 msec was found, so no further investigations with the right eye leading were done. Apart from the latency shift the amplitude of N1 was smaller in most cases, and smoothing of the P2 and N2 peaks of the response complex occurred. The maximum interocular delay that could be used in our tests was 27 msec. B e y o n d that interval the depth perception became ambiguous, and subjects did n o t perceive the target or its m o v e m e n t as clearly as before. (C) In this investigation a small latency shift of the N~ peak was found, and its amplitude was markedly reduced by reduction of
54 the rate of generation of dots from 1000 to 500 pairs/sec. P2 and N2 peaks were smoothed or even hard to distinguish, as in the records with interocular delay. However, using rand o m d o t stereograms consisting of 5000 pairs of dots per second, no appreciable shortening of the N~ peak latency occurred. (D) Distracting the subjects attention from the target shift by fixation on a black point in the middle of the background plane had no appreciable influence on the response. (E) After accommodation had been completely blocked, the target was seen moving clearly and sharp edged as before. An observation of psychological interest was made by all experimental subjects. They all reported that the square target seemed to become congruently smaller as it moved towards them, and bigger as it moved away behind the background.
Discussion
In general it is not easy to interpret electrical phenomena recorded from the outside of the skull, in terms of the spatio-temporal relations of potential fields in humans (Pockberger et al. 1979). Still we think, based on the results described above, that the evoked potentials do n o t originate in the occipital region. Records as shown in Fig. 3 could indicate an asymmetry of activity between t w o hemispheres concerning stereoptic depth perception, though the significant potential differences only start at 250 msec after the stimulus. This finding is remarkable in relation to the suggestion of 'long-lasting independent data handling in the two hemispheres up to 300 msec after the stimulus' of Lehmann and Julesz (1977, 1978). Using 4 different interocular delays, 4 N~ peak latency shifts, much longer than the imposed delay, were found. Though the relatively constant correlation of interocular delay and latency shift seems intriguing, this correlation remains far from established because of the limited number of interocular delays and subjects. The reasons
M.J. HERPERS ET AL. for the smoothing of the P2 and N2 peaks needs further investigation. Though there must be a minimal amount of dots with disparity before a target can be formed by the brain, this could not be established with certainty in our experiments. Ross (1976) stated 'that for single shapes of a reasonable size the lower limit is about 2000 pairs of points/sec.' In contrast to this finding all our subjects clearly saw the sharp edged target, even though only 1000 pairs of dots/ sec were generated. (This means only 200 dots/sec in the target area.) That the displayed number of dots/sec must reach a 'saturation point' concerning the latencies of the various response peak maxima is obvious because no appreciable shortening of the N~ peak latency evoked by 5000 pairs of dots/sec was established. So, since every random d o t of the used stereograms was generated one after the other, the number of dots and the rate at which they were generated can influence the latency of the response provided that at least a part of the evoked potential stands for the electrical manifestation of 'global stereopsis' (Julesz 1971; Bishop 1975). This is probably the reason why Regan and Spekreijse (1970) found a latency of only 94 msec in their investigations using static random d o t images. In static random dot stereograms all points with retinal disparity are shifted at the same time, and no time for 'collecting' enough dots to delineate the target is needed. These investigations with changes in the number of dots showed a smaller amplitude and a slight increase of the N~ peak latency. This could be explained by the fact that, with fewer points presented the chance that any given point falls in the target area on the display scope was also diminished. So it t o o k longer before enough points with disparity were collected to form the actual target by means of global stereopsis. The maximum interocular delay before depth perception became ambiguous was in our experiments only 27 msec. This finding is n o t in accordance with the results of Ross (1976), who found that the maximum delay before depth
DYNAMIC RANDOM DOT STEREOGRAM EPs perception disappeared could be as much as 50 msec. An explanation for this difference could be the smaller a m o u n t of dots presented in our experiments. With more dots, more reference points are available at a given m o m e n t and the influence of an interocular delay on the a m o u n t of displayed information is n o t so grave, resulting in a longer permitted interocular delay. Consistent with the work of Lehmann and Julesz (1977) and Lehmann et al. (1978) are the results found in distracting the subjects from the target shift. The response does n o t differ essentially from that found in the normally used test situation. We showed that with random d o t stereograms depth perception persists if a c c o m m o d a t i o n is blocked, so it seems very unlikely that the evoked stereoptic potential involves a neural accommodation mechanism. Since the size of the target remained constant before and after the shift, the p h e n o m e n o n of the illusory magnitude changes of the target m a y well be explained as follows. In general, an approaching target, of constant size, is perceived with an increasing visual angle. In random d o t stereograms, an approaching target is perceived with a constant visual angle. This is apparently interpreted by the brain as an approaching target with diminishing size, compatible with daily life experience. The same reasoning goes for the target moving to the far field, seemingly becoming bigger. A fundamental question concerns the neural processes that generate the evoked potentials. Are these potentials only caused by the neural mechanism comparing dots or clusters of dots with retinal disparities, the so-called 'local stereopsis' of Julesz (1971), or are they a response generated by the 'global stereopsis', forming the sharp edged target o u t of the local disparities. A third possibility would be a combination or interference of the t w o mechanisms, because b o t h are actually involved in the perception of the target shift.
55 Summary In 11 normal healthy human subjects an evoked potential was elicited b y moving dynamic random d o t stereograms. The random dots were generated by a minicomputer. An average of each of 8 EEG channels of the subjects tested was made. The maximum of the cerebral evoked potentials thus found was localized in the central and parietal region. No response earlier than 130--150 msec after the stimulus could b e proved. The influence of fixation, the number of dots provided, an interocular interstimulus interval in the presentation of the dots, and lense accommodation movements on the evoked stereoptic potentials was investigated and discussed. An interocular interstimulus interval (left eye leading) in the presentation of the dots caused an increase in latency of the response much longer than the imposed interstimulus interval itself. It was shown that no a c c o m m o d a t i o n was needed to perceive the depth impression, and to evoke the cerebral response with rand o m d o t stereograms. There are indications of an asymmetry between the t w o hemispheres in the handling of depth perception after 250 msec. The potential distribution of the evoked potentials strongly suggests that they are not generated in the occipital region.
R~sum~ Potentiels dvoquds cdrdbraux humains aux mouvements de stdreogrammes formgs de 'random dots' dynamiques Des potentiels c~r~braux ont ~t~ ~voqu~s chez 11 sujets adultes ~ partir de st~r~ogrammes ~ base de configurations de 'random dots' dynamiques (ces random dots ~taient g~n~r~s par un mini-ordinateur). La m o y e n n e de chacune des 8 d~rivations ~lectroenc~phalographiques auxquelles chaque sujet ~tait soumis, a ~t~ calcul~e. Le maximum des potentiels ~voqu~s, d~termin~ de cette fa~on, se situait dans les r~gions cen-
56 trale et pari~tale. Des c o m p o s a n t e s de r~ponse s u c c ~ d a n t au stimulus en m o i n s de 1 3 0 - - 1 5 0 msec, ne s ' o b s e r v a i e n t pas. I1 a ~t~ examin6 l ' i n f l u e n c e que les facteurs suivants e x e r c e n t sur les p o t e n t i e l s ~voqu~s: la f i x a t i o n des sujets ( d ' e x p ~ r i e n c e ) ; la quantit~ de points; le d~lai inter-oculaire dans la pres e n t a t i o n des ' r a n d o m d o t s ' d y n a m i q u e s ; les m o u v e m e n t s d ' a c c o m m o d a t i o n de la lentille. L a latence des r~ponses ~voqu~es par ce d~lai inter-oculaire se m o n t r a i t plus longue que l'intervalle impos~ de celui-ci (les ' r a n d o m d o t s ' d y n a m i q u e s o n t ~t~ pr~sent~s d ' a b o r d l'oeil g a u c h e et apr~s ~ l'oeil droit). I1 a ~t~ d ~ m o n t r ~ que l ' a c c o m m o d a t i o n n ' e s t pas n~cessaire p o u r percevoir en p r o f o n d e u r , ni p o u r ~voquer des potentiels c~r~braux ~ partir de st~r~ogrammes ~ base de ' r a n d o m d o t s ' dynamiques. D e u x c e n t c i n q u a n t e millisecondes apr~s le stimulus s'observait une asym~trie entre les d e u x h~misph~res. La d i s t r i b u t i o n spatio-temporelle des p o t e n t i e l s ~voqu~s, telle que n o u s l ' a v o n s constat~e, n~cessite u n e autre explicat i o n q u e celle de la p r o v e n a n c e occipitale.
References Atkinson, J. and Braddick, O. Assessment of vision in infants. Trans. ophthal. Soc. U.K., 1979, 99: 338-343. Bishop, P.O. Binocular vision. In: R.A. Moses (Ed.), Adler's Physiology of the Eye. Mosby, Saint Louis, Mo., 1975: 558--614. Julesz, B. Binocular depth perception of computer-
M.J. HERPERS ET AL. generated patterns. Bell System Technol. J., 1960~ 39: 1125--1162. Julesz, B. Foundations of Cyclopean Perception. University of Chicago Press, Chicago, Ill., 1971. Julesz, B., Breitmeyer, B.G. and Kropfl, W. Binoculardisparity-dependent upper-lower hemifield anisotropy and left-right hemifield isotropy as revealed by dynamic random dot stereograms. Perception, 1976, 5: 129--141. Julesz, B., Kropfl, W. and Petrig, B. Large evoked potentials to dynamic random dot corretograms and stereograms permit quick determination of stereopsis. Proc. nat. Acad. Sci. (Wash.), 1980, 77: 2348--2351. Lehmann, D. and Julesz, B. Human average evoked potentials elicited by dynamic random dot stereograms. Elelctroenceph. clin. Neurophysiol., 1977, 43 : 469. Lehmann, D. and Julesz, B. Lateralized cortical potentials evoked in humans by dynamic random dot stereograms. Vision Res., 1978, 18: 1265--1271. Lehmann, D., Skrandies, W. and Lindenmaier, C. Sustained cortical potentials evoked in humans by binocularly correlated, uncorrelated and disparate dynamic random dot stimuli. Neurosci. Lett., 1978, 10: 129--134. Mol, J. and Caberg, H. Contingent stereoptic variation as a response to moving random dot stereograms. Electroenceph. clin. Neurophysiol., 1977, 43: 513. Pockberger, H., Petsche, H., Rappelsberger, P., MiillerPaschinger, I.B. und Prohaska, O. Epi- und intrakortikale Aspekte visuell evozierter Potentiale. Z. EEG-EMG, 1979, 10: 184--193. Regan, D. and Beverly, K.I. Electrophysiological evidence for existence of neurones sensitive to direction of depth movement. Nature (Lond.), 1973, 246 : 504--506. Regan, D. and Spekreijse, H. Electrophysiological correlate of binocular depth perception in man. Nature (Lond.), 1970, 225: 92--94. Ross, J. The resources of binocular perception. Scient. Amer., 1976, 569: 80--86.