Exp. Brain Res. 23, 471--489 (1975) 9 by Springer-Verlag 1975 Perception of Linear Horizontal Self-Motion Induced by Peripheral Vision (Linearvection) Basic Characteristics and Visual-Vestibular Interactions* A. Berthoz, B. Pavard and L. It. Young** Laboratoire de Physiologic du Travail, CNRS-CNAM, Paris (France) t~eceived April 24, 1975 Summary. The basic characteristics of the sensation of linear horizontal motion have been studied. Objective linear motion was induced by means of a moving cart. Visually induced linear motion perception (linearvection) was obtained by projection of moving images at the periphery of the visual field. Image velocity and luminance thresholds for the appearance of linearvection have been measured and are in the range of those for image motion detection (without sensation of self motion) by the visual system. Latencies of onset are around 1 sec and short term adaptation has been shown. The dynamic range of the visual analyser as judged by frequency analysis is lower than for the vestibular analyser. Conflicting situations in which visual cues contradict vestibular and other proprioceptive cues show, in the case of linearvection a dominance of vision which supports the idea of an essential although not independent role of vision in self motion perception. Key words : Visual-vestibular interactions perception -- Linearvection - - Otoliths -- Motion The role of the periphery of the visual field in selLmotion detection has been extensively studied both in man and in animals; the interaction between visual and vestibular cues is now established from psychophysical and electrophysiological data (Brandt et al., 1971 ; Brandt et al., 1972 ; Brandt et al., 1973 ; Dichgans et al., 1972; Dichgans and Brandt, 1973; Diehgans et al., 1973; Gibson, 1954; Maekawa and Simpson, i972; Math, 1875; Young and tIenn, I974; Young and Oman, 1974; Young et al., 1975). An essential result is that vision can contribute to the sensation of self-motion during rotation at a constant velocity whereas the labyrinthine receptors, whose specific stimulus is acceleration, can only signal changes of velocity. * Preliminary reports of this work have been presented in the form of a short communication (Berthoz et al., 1974) and at the European Brain and Behaviour Society Workshop on: "Vestibular function and behaviour" (Pavia, Brain Research, Special issue, p. 17, 1974). ** Present address: Dept. of Aeronautics and Astronautics, MIT Cambridge, mass. (USA). 472 A. Berthoz et al. Most studies thus far have been restricted to circular movements and the c o r r e s p o n d i n g s e n s a t i o n ( c i r c u l a r v e e t i o n ) . L i n e a r m o t i o n s e n s a t i o n , w h i c h will b e h e n c e f o r t h c a l l e d " l i n e a r v e c t i o n " (LV), w a s h o w e v e r i n v e s t i g a t e d b y L e e et al. (1974), w h o , b e c a u s e o f t h e s t r o n g p o s t u r a l effects p r o d u c e d b y s m a l l disp l a c e m e n t s o f a m o v i n g r o o m , p u t f o r w a r d t h e h y p o t h e s i s t h a t v i s i o n is a n " a u t o n o m o u s " a n d also t h e " p r i n c i p a l " k i n a e s t h e t i e senese. T h e a i m o f t h e p r e s e n t w o r k is t o e s t a b l i s h q u a n t i t a t i v e l y t h e c h a r a c t e r i s t i c s of the sensation of linear motion induced by movements of the peripheral visual field i n a h o r i z o n t a l d i r e c t i o n p a r a l l e l t o t h e a n t e r i o r - p o s t e r i o r a x i s o f t h e h u m a n body. A typical instance of such a sensation occurs while sitting in a non-moving train and a neighboring train starts moving. The current experiments were c o n s e q u e n t l y d e s i g n e d t o i m p o s e c o n t r o l l e d m o v e m e n t s o f e i t h e r t h e v i s u a l field or the body of the subject, thus allowing a rigourously defined visual-vestibular interaction. Methods The experimental apparatus is shown in Fig. 1 A. I t allows the projection of a moving 35 m m film loop on a screen which was fixed on a mobile cart. The screen was located in the upper part of the visual field of a seated subject at a distance of 0.5 m from his head (Fig. 1 C). Two lateral mirrors, inclined by 45 ~ with respect to the vertical, gave two virtual images of the screen which were in a vertical plane parallel to the sagittal plane of the head (Fig. 1 B). These images were 0.8 m lateral to each eye of the subject. The projector was equipped with a servo-motor whose angular velocity was measured by a tachometer mounted on the motordrive itself. The calibration of film velocity was made in m/sec (el. Appendix for conversion into angular velocity). Image velocity on the screen (V~) varied from 0--2 m/see with an accuracy of 5O/o. The subject, whose head position was fixed by a chin rest, could view the moving images through a black box equipped with modifiable side windows which limited the visual field between 20 ~ and 70 ~ away from the sagittal plane on each side (Fig. 1 B). On the upper p a r t of the black box a variable opening allowed a view of the screen. According to the experimental conditions the subject was instructed either to fix a black cross which was drawn on the upper screen, or a luminous cross inside the black box and in front of the subject. The cart could roll on wheels guided by rails under the control of two servo-driven torque motors. Cart velocity varied from 0 - - 2 m/see; it was accurate to 5O/o, and was recorded by a tachometer attached to the cart. B o t h the image velocity (Vi) and the cart veloeity (Vc) were programmed b y a H P 5452 computer and could be controlled independently. I n order to eliminate the influence of auditory cues on velocity detection, the subject's ears were covered with earphones delivering a 75 dB broad spectrum noise. Care was t a k e n to limit all vibrations. I-Iowever, tactile and pressue cues, particularly on the chin, were sometimes reported a,s used by the subjects and were undoubetdly used increasingly as the stimuli were repeated. Magnitude Estimation o/Subjective Velocity The sensation of self-motion experienced by the subject was measured by the method of magnitude estimation (Lipetz, 1971). A lever fixed to the cart on the right side of the subject could move forward or backward in the sagittal plane, starting from an upright zero position. Two springs were attached to the beam so t h a t the lever returned to the upright position when no force was exerted. To enable calibration of the velocity magnitude estimation while the subject was looking in the box, two metal reference posts were placed b o t h behind and in front of the zero level position. These posts, which could be sensed by the fingers as the subjects moved the lever forward and backward, were associated with positions corresponding to 50o/o and 100~ of the deflection corresponding to the reference velocity for magnitude estimation. The rotation of the lever was recorded by a potentiometer whose o u t p u t was fed into the computer, together with the image velocity (Vi) and the cart velocity (Vc). Perception of Linear Horizontal Self-Motion i", , B 473 / ,,' '", H .,fl//, ",,, /// 1 i I Vi L V I-~,; T o ....... x A C . Vi / Lx 0 s .Jr cv Fig. 1. Experimental apparatus. (A) The subject is seated on a cart (c). A b a n d of film (b), on which are printed randomly distributed signs (letters, points, crosses, etc.), is moved by a projector. The image of this band is projected on a screen (s) (surface 1.6 m 2, distance to the horizontal plane of the eyes 0.5 m). The subject, whose head is maintained in a normal upright position b y a chin rest, looks through three openings in a box: two lateral windows (w), and an opening above the head. Two virtual images (i) of the screen (s) are given by two mirrors (m) tilted at 45 ~ angles. The subject is required to give continuous magnitude estimations of his subjective sensation of linearveetion (LV) when the film is in motion at the speed Vi, or when the cart is moving at the speed Ve, b y using the h a n d lever (1) on his right side. This lever is also used to control Vi in the active procedure. (B) Schematic view of image position (i) from above, with respect to the subject's eye (angle of vision: 20 ~ to 70~ L V indicates direction of subjective velocity (forward LV for backward Vi). (C) View from the right indicating the position of the screen (s). Same notations as in A and B 1. For each subject, the film velocity was initially set at 1 m/see, which was the mean value of Vi to reach m a x i m u m LV. The subjects were told t h a t the sensation of LV experienced in this condition was to be t a k e n as the m a x i m u m (100%) of the magnitude estimation and should correspond to the m a x i m u m forward (or backward) position of tile hand lever. 2. Vi was subsequently reduced to 0.5 m/see and the subjects were instructed t h a t the LV they experienced was to be indicated as 50% of the maximum Vs. The position of the hand lever was then to be set at the halfway point between zero and maximum. Although we recognized a non-linear relationship between ViVs, the choice of 0.5 m/see for Vi as corresponding to 50% evaluated LV produced a consistent b u t non-linear scale. 3. Vi was t h e n moved b y randomly distributed steps of 0.1 or 0.2 m/see in repeated sequences. The subjects had to indicate Vs with various positions of the h a n d lever. The test was repeated until the subject could successfully r e t u r n to the 100O/o and 500/o positions whenever V1 reached the initially defined values. In addition, the subjects had to indicate the LV associated with the intermediary values of Vi (in a repeatable manner). 0 n l y those subjects (approximately 80o/0) who could perform this task correctly were subjected to further tests. A. Berthoz et al. 474 Movement Direction Detection The same lever was also used for a simple all-or-none direction indication for visualvestibular experiments when the cart was actually moving. A simplified version of this method consisted of a switch which could signal forward-zero-backward motion. Active Modification o / F i l m Velocity The lever was also used in some experiments for control of the image velocity by the subject. By varying the position of the lever the subject could vary the image velocity forward or backward from 0--1 m/set: this procedure was used for active threshold evaluation or for the study of adaptation to a prolonged exposure to a moving visual scene. Luminance Threshold Measurement The stimulus for all LV tests was s p a t t e r n of randomly distributed black letters (Letraset) moving on a white background (the overhead screen). The mean luminance of the overall screen with patterns wa~ 80O/o of the background luminance (the ratio of white to black areas). The contrast between letters and background was always approximately 0.9. Velocity was constant in each trial and varied between trials in discrete steps between 0 and 1 m/set. An active LV estimation was made b y the subjects who could continuously adjust the visual stimulus luminance level b y a potentiometer regulating the light level of the film projector. The subjects redueecl the luminance level until they felt a disappearance of LV. I n order to reduce the effects of light adaptation, measur?s were m~de at least 5 min after the beginning of each exposure. Luminance was measured on the projecting screen by a S E J photometer. I n the ease of low luminance levels, the photometer sensitivity was increased 250 times by reducing the image of the screen optically, thus allowing accurate measurements at a level of 10 .4 Cd/m 2. A i I 2 ,s 3 Vi, m/sec B 00 3 T '.2 '.4 '.6 Fig. 2 A and B '.8 vj i m/sec Perception of Linear Horizontal Self-Motion 475 C Vi LV j 0,02 HZ /-~0,2 D rn (-. 0~. o -D 0 (1) _.C~-90 0.. o,bl d,1 i Hz Fig. 2. Typical threshold, saturation, and latency of the subjective sensation of forward linear velocity (forward LV) induced by a backward motion of images at the periphery of the visual field. (A) Typical variation of the magnitude estimation of LV with increasing image velocity for one subject. 2 threshold; S = upper limit of increasing LV (saturation). Notice the drop of LV when Vi exceeds 2.8 m/see. Vi increased linearly with time (0.01 m/sec2). (B) Latency T of the onset of forward LV for different steps of image velocity (V0. Mean and standard error of mean for 10 subjects. (C) Records of LV induced by sinusoidal variations of image velocity (Vi). First trace is reference sine wave for Vi (constant amplitude 0.8 m/see). From top to bottom: successive samples of magnitude estimation of LV at different freestimated LV amplitude quencies of input Vi. (D) Gain (log10 = v - - - - - -velocity -linage ~ ) and phase of LV for different frequencies of input V~ 476 A. Berthoz et aI. Results General Features: Latency o/Onset and Frequency Analysis o[ Linearvection The basic characteristics of the motion sensation induced by linear motion of the visual scene were first investigated. I t was observed t h a t linearvection (LV) appeared only above a threshold of fihn velocity (Fig. 2). When the subjects were asked to estimate the magnitude of LV with the hand lever, the curves obtained with increasing image velocity (Vi) generally had the shape shown in Fig. 2A. This curve was obtained from a single trial with Vi increasing linearly with time at a slow rate (0.01 m/sec 2 acceleration). Above threshold the sensation of motion clearly increased with Vi. When the upper screen was used, a distinct sensation of linear motion generally involved the entire frame surrounding the moving scene. A preliminary set of experiments was designed in order to determine the pattern of visual images which would induce the strongest LV. I t was found that the classical optokinetie stimulus made of vertical bars was not as effective as a random pattern. Consequently, the pattern of randomly distributed letters whose characteristics have been described in the methods section was finally used. I t was noted t h a t when naive subjects were submitted to the visual stimulations for the first time, a period of up to 10--20 sec could elapse before they actually reported LV. The current latency for the onset of linearvection was measured by applying a step of velocity and asking the subjects to indicate the onset of LV. This procedure is summarized in the insert of Fig. 2 B. The latency T is plotted for forward LV. The delay introduced by the reaction t i m e and manipulation time of the lever has been deducted (approximately 0.4 sec) following a distinct series of experiments in which this time was measured by itself. The value of Vi which gave saturation of LV (Fig. 2A) was measured by asking 10 subjects to indicate the film velocity above which they did not feel any increase in LV. The mean fell in the range Vi ~ 0.8--1 m/sec for most subjects. The value Vi ~ 1 m/sec was then taken as 100% for the calibration of LV. I t was expected from previous observations on circularvection that visual detection of linear movement would be more effective in a rather low frequency range. This was studied by a frequency analysis of reported LV measured by the continuous magnitude estimation procedure. (The subject indicates the intensity of LV forward or backward with the lever). Figure 2C and 2D show respectively a sample of records and a Bode plot of the input image velocity - - output linearvection, gain and phase data for three subjects. I n Fig. 2C it can be observed t h a t the gain shows a rather small drop (less than l0 dB) between 0.01 and 1 Hz. The phase however decreases sharply and at 0.8 ttz, LV is practically in opposite phase with image velocity. Image Luminance and Velocity Thresholds ]or the Appearance o] Linearvection The thresholds for the appeaarnce of LV have been measured as a function of image luminance and velocity. The purpose of this set of experiments was to determine if LV thresholds were close to physiological thresholds for the detection of image motion in the periphery. The influence of luminance was determined by measuring the thresholds of differential luminance levels (alL) (see methods) for which LV could be elicited at Perception of Linear Horizontal Self-Motion 477 AL (cd/rn 2) .1 .01 .001 FUSION 9 0001 .01 t t , t I I . .02 .05 .1 .2 .5 1 V l (m/see) Fig. 3. Differential luminance thresholds for the onset of motion sensation (LV) at different velocities of the visual image (V~). Differential thresholds of luminance (AL) have been determined in the experimental conditions of Fig. 1 (see text). The data points obtained for different magnitudes of V~ (0) indicate mean values and standard deviations. The values obtained by Daniel (1959) with a stationary l ~ angular diameter spot, projected at the periphery of the visual field (50 ~ angle) are plotted for comparison (O) different image velocities (Vi). Figure 3 shows the results for 6 subjects. The values of AL which are necessary to elicit linearvection decrease very rapidly with increasing Vi. They reach a minimmn for values of Vi where there is a fusion between the black moving pattern and the white background. These data have been compared with results obtained by Daniel (1959) 9 In the experimental conditions used by Daniel the subject was shown a white spot at the periphery of the visual field and was asked to detect the presence of the spot. This spot had a variable angular diameter and was presented on a background of different luminances, at various positions at the periphery. L for which the subject could detect the presence of the spot for all these conditions was measured. The results of Daniel for 1 ~ angular diameter spot (which is approximately the size of the letters, dots and signs used in our experiments) presented at a 50 ~ lateral angle, show a similarity with the luminance threshold AL for LV. Even if the experimental conditions of Daniel and ours were somewhat different, this suggests the luminance thresholds for LV are very close to the absolute luminance thresholds for image detection. The thresholds (2) of image velocity which can induce linearveetion were measured, as already discussed under methods, using two procedures, passive and active. The results of the passive procedure are sho~m in Fig. 4 A. The insert shows how 2 was determined for increasing (2 +) and decreasing (2-) variation 478 A. Berthoz et al. Vi X(m/sec) ~'1~)~- 04, A Vi backward p 40 80 sec 02 0 , 02. Vi forward X (m/sec) T B l x='/,(x++x-) 04. l 02, Vi backward eLV 1 nLV2 oVD1 "VD2 Vii///~ 10 X 20 sec 1 o 02. { i { V{ i forward Fig. 4. Thresholds (~) of image velocity (Vi) for the appearance of linear veetion (LV) and for the velocity detection (VD) of image motion. The mean thresholds and standard deviations have been obtained for 10 subjects. The stimulus is presented either peripherally in the side windows alone (circles, LV 1 and VD1) or simultaneously in the side windows and overhead with a fixation point within the overhead screen (squares, LV 2 and VD2). Forward Vi induces backward LV and vice-versa. (A) Passive procedure: The stimulus velocity (Vi) is controlled by the experimenter. Thresholds (~.) for forward or backward LV have been determined for increasing (2+) and decreasing (Z-) image velocities (see insert). In the moving field, the linear vection thresholds (LV2) are as low as the thresholds for detection of stimulus velocity (VD2). (B) Active procedure: The subject himself adjusts the stimulus velocity (Vi) to the thresholds values for development of LV. 2 has been determined asymptotically (see insert). All notations are identical with those in A (0.002 m/see steps) of Vi. The values of 2 are plotted as positive when foward LV was i n d u c e d (i.e. backward Vi) a n d vice-versa. The aim of this e x p e r i m e n t was also to compare LV thresholds with image velocity detection (VD) thresholds. tIenee, for each subject these thresholds were measured successively i n two conditions : 1. with fixation of a p o i n t inside the black box, straight ahaed of the Perception of Linear Horizontal Self-Motion 479 subject, but dissociated from the moving visual field; and 2. with fixation of a cross on the upper screen (see methods). These two conditions gave two sets of results, plotted as LVI, VD I (linearvection and image detection without field centered fixation) and LV2~ VD~ (same but with field centered fixation). These results show that thresholds for LV arc of the order of 0.03 m/sec in Condition l, and at the limit of image motion detection (less than 0.01 m/sec) in Condition 2. It can then be concluded that with our experimental device the thresholds for forward linearvection were of the order of magnitude of image velocity detection by the visual system. The thresholds for backward LV were found to be systematically smaller than those for forward LV. The active procedure (Fig. 4B) gave results which were very similar to those obtained with the passive procedure. Adaptation o/Linearvection I t is k n o w n from field e x p e r i m e n t s w i t h vehicle drivers t h a t exposure to a long d u r a t i o n c o n s t a n t v e l o c i t y leads to u n d e r e s t i m a t i o n of vehicle v e l o c i t y (Denton, 1971; S a l v a t o r e , 1968; S c h m i d t a n d Tiffin, 1969). T h e effects of p r o l o n g e d exposure to l i n e a r v e c t i o n 1 were t e s t e d b y asking t h e subjects to m a i n t a i n t h e i r L V c o n s t a n t t h r o u g h v a r y i n g t h e v e l o c i t y of t h e m o v i n g field with t h e h a n d lever. T h e results are shown in Fig. 5 for one s u b j e c t w i t h different initial velocities of t h e m o v i n g film. The s u b j e c t increased Vi w i t h time, i n d i c a t i n g a slow a d a p t a t i o n t o t h e effect. The m a g n i t u d e of t h e increase is d e p e n d e n t u p o n t h e initial film velocity. I n Condition I (low initial velocity) t h e increase of Vi is only from 0 . 1 5 - - 0 . 2 m/see. I t is m a x i m u m for Condition I I I (0.4--0.85 m/see). T h e maxim u m film v e l o c i t y a t t a i n e d in Condition I V could n o t be g r e a t e r t h a n 1.2 m/see since a b o v e this v a l u e t h e s a t u r a t i o n effect of L V described in Fig. 2 A comes into p l a y a n d t h e s u b j e c t s c a n n o t o b t a i n a n y increase in L V for higher values of Vi. T h e t i m e c o n s t a n t of t h e a d a p t a t i o n of L V as d e m o n s t r a t e d b y this e x p e r i m e n t is a p p r o x i m a t e l y 3 0 - - 5 0 sec. I t is possible t h a t if e x p e r i m e n t s h a d been p r o l o n g e d u p to several hours, o t h e r fluctuations would h a v e been n o t e d b u t long t e r m a d a p t a t i o n has n o t been i n v e s t i g a t e d here. Qualitative Investigation o/Visual-Vestibular Interactions I n o r d e r to d e t e r m i n e q u a l i t a t i v e l y t h e n a t u r e of v i s u a l - v e s t i b u l a r i n t e r a c t i o n in t h e case of linearvection, h o r i z o n t a l linear m o v e m e n t s of t h e visual field a n d of t h e s u b j e c t h i m s e l f (cart motion) were combined. I n this case t h e subjects were i n s t r u c t e d to i n d i c a t e t h e continuous v a r i a t i o n s of c a r t v e l o c i t y ( m a g n i t u d e e s t i m a t i o n s of c a r t velocity) u n d e r t w o conditions: 1. w i t h eyes closed (Figs. 6, 7A) a n d 2. when viewing a visual scene which was m o v i n g a t c o n s t a n t v e l o c i t y 1 The term 'adaptation' seems here to be approximate as it is often used when a physiological response decreases with prolonged stimulation. However, this term is also restrictive in the sense that it tends to be used for describing a peripheral event. In the present case no proof can be obtained, with the procedure used, that this decrease of LV with time is due to a central organisation or to an adaptive property of the retina itself. The alternative term is 'habituation' which is generally attributed to a central process and it is known (Young et at., 1973) that a selective habituation of vestibular nystagmus can be due to visual stimulation. However, habituation is used to designate a decrease in response to repeated (and not prolonged) stimulation. The difference between these two conditions will be dealt with in a subsequent paper. 34 Exp. Brain Res. Vol. 23 A. Berthoz et al. 48O Vi ( m / s e c ) IV saturation LV S III f II J .5 J J o I I 5o lOO T (sec) Fig. 5. Adaptation of linearvection during prolonged exposure to a constant velocity moving scene. In this experiment the subject actively controls the velocity of the visual stimulus (Vi) so as to keep his sensation of self-motion (LV) constant. Reponse curves (I, II, III, IV) are drawn for different initial film velocities. The increase of film velocity (Vi), due to the progressive adaptation of linearvection, is limited in the case of LV by a saturation effect (loss of LV for high film velocities as shown in Fig. 2) A B Eyes closed Eyes open 0 Vc VS 9 2 0 sec I I Fig. 6. Interaction of subjective velocity sensation (Vs) induced by conflicting motion of the visual scene and by the translational velocity (Vc) of the cart. (A) Vestibular detection of linear motion with eyes closed. From top to bottom: cart velocity (Vc), subjective velocity (LV) - - (arbitrary units) and image velocity (Vi). Note that the subject detects the variations of cart velocity well, with a slight delay. (B) Conflict between a constant velocity visual scene and oscillating cart velocity. From top to bottom: cart velocity, subjective velocity (combination of vestibular induced motion sensation and LV) and image velocity (which is constant, equal to 0.5 m/sec moving backward and inducing a forward LV). Note that the subject indicates a subjective velocity which is always forward, even during periods when the cart is moving backward Perception of Linear Horizontal Self-Motion 481 backward inducing forward LV (Vi = - - 1 m/see) (Fig. 6B). Vi was always measured relative to the cart and was purposely chosen at a saturation value for LV in order to induce a maximum forward motion sensation. The cart velocity consisted of pseudo-random slowly varying movements forward and backward as indicated in the upper tracings of Fig. 6. The subjects indicated continuously with the hand lever the magnitude of their subjective velocity (Vs), which here was a combination of visuMly-induced LV and vestibular sensation of motion. Although the velocity of detection was rather good in the eyes closed condition (Fig. 6 A), in the Vi = constant condition, for which the visual cues were frequently in conflict with vestibular and other kinaesthetic cues, the subjects generally reported the sensation of a constant forward LV (as indicated by the second trace on Fig. 6B). Only transient suppressions of LV were occasionally observed associated with rapid changes of cart velocities either forward or backward. These results indicate the very powerful central action of visual information from the periphery of the visual field concerning linear self-motion. They also suggest that whenever the visual motion information is presented, it tends to dominate vestibular cues. However, the experimental conditions of Fig. 6 are such that visually induced motion sensation was saturated and vestibular stimulation could have been close to threshold values; this would lead to an underestimation of vestibular motion detection due to these limitations. For this reason another set of experiments was designed, in order to carefully quantify both visual and vestibular stimulation. The necessity for carefully defining the condition of vestibular stimulation led us to perform, in addition, a simple but quantitative prediction of vestibular detection before presenting conflicting visual and vestibular stimulations. Prediction o/ Vestibular Linear Detection and Experimental Paradigm/or VisualVestibular Interaction Although the purpose of this paper was not to study vestibular detection of linear motion, it was useful to take advantage of some previous experimental work of Young and Meiry (1965) which provided numerical values for linear acceleration thresholds. The curve given by these authors allowed the prediction of the time necessary to detect constant accelerations between 0 and 1 m/see ~. Our experimental conditions were designed so as to apply steps of cart accelerations (ramps of velocities) whose duration was at threshold or above threshold for such a detection ~. The acceleration amplitude was constant (0.4 m/see 2) and the duration varied from 1.6--3 sec. The model, as shown in Fig. 7, predicted the signal provided by the graviceptors during steps of horizontal linear cart acceleration. The corresponding curve, indicated as vestibular response (VR) in the figure, was computed by applying the transfer function shown to the acceleration waveform. 2 Absolute vestibular thresholds for linear acceleration are as low as 0.05 m/see2 in man. In our experimental setup (see methods) the maximum available amplitude of acceleration steps (velocity ramps) was 0.4~ 2 because of limitations in the total displacement for the cart (2 m). This in turn led to a detection time of about 2 sec according to the model of Young and Meiry (see Fig. 6). 34* 482 A. Berthoz et al. Vc % Pc 1.5 ( s + 0 . 0 7 6 ) ~ZR fwd R+ bwd R- s [s+O.19)(s+1.5) I I I I I 15sec F~ o .5 1 m/see 2 Fig. 7. Experimental paradigm for the study of visual and vestibular detection of linear motion, and use of an otolithic model for the prediction of subject's response. This diagram shows a typical sequence of cart motions. Vc is the velocity of the cart. Triangular wave forms have been chosen because they correspond to steps of constant accelerations (0.4 m/sece). Amplitude of velocity is irregular in order to give a pseudo-random movement to the cart. d2x When such a pattern of acceleration (dT2) is sensed by the graviceptors, the model of Young and Meiry (1967) (see block diagram on the right of the third line) allows a calculation of the theoretical output (vestibular response: VR). The model assumed an absolute threshold of approximately 0.006 g for upright subject which placed the 0.4 m/see 2 latency time for subjective vestibular detection of linear movement at 1.6 sec. R + (forward) and R- (backward) are the predicted all-or-none responses of the subject. VR peaks are above the vestibular threshold T h e a c t u a l s u b j e c t i v e m o t i o n d e t e c t i o n was m e a s u r e d b y an all-or-no p r o c e d u r e (see methods) which is different from t h e continuous m a g n i t u d e e s t i m a t i o n described in previous sections of these results. The subjects were o n l y r e q u i r e d to signal t h e d i r e c t i o n of t h e i r linear m o v e m e n t as shown on t h e l a s t line of Fig. 7. This p r o c e d u r e gave t h e p e r c e n t a g e of correct v e l o c i t y d e t e c t i o n f o r w a r d (R +) a n d b a c k w a r d (R-). The e x p e r i m e n t a l p a r a d i g m for v i s u a l - v e s t i b u l a r i n t e r a c t i o n was t h e n d i v i d e d into t h r e e successive conditions as shown in Fig. 8. I n all t h r e e conditions t h e s u b j e c t h a d his eyes open. I. T r i a n g u l a r velocities were a p p l i e d to b o t h t h e cart (Vc) a n d t h e visual scene (Vi) in such a w a y t h a t a t a n y i n s t a n t Vi ~ Vc. This is e q u i v a l e n t to a real life s i t u a t i o n in which t h e s t a t i o n a r y visual scene has a m o v e m e n t of equal a m p l i t u d e a n d opposite d i r e c t i o n to t h e o b j e c t i v e m o v e m e n t of t h e head. I n this c o n d i t i o n direction d e t e c t i o n was e x p e c t e d to be 100% correct. I I . T r i a n g u l a r velocities were a p p l i e d to t h e cart while t h e visual scene was s t a t i o n a r y w i t h r e s p e c t to t h e s u b j e c t (Vi ~ 0). I n this conflicting s i t u a t i o n t h e Perception of Linear Horizontal Self-~otion 212 483 -IF fwd - I Vi T,/V~ A/~ , v v v i fwd R- I I I I I I I I I I I I 5 sec Fig. 8. Experimental paradigm for the analysis of the influence of linearveetion on vestibular detection of linear acceleration. From top to bottom: Vo = velocity of the cart; V1 ~ velocity of visual images; R + and R - have been defined in Fig. 7. The analysis is made for three experimental conditions. (I) Vi = ~ V c The velocity of the visual stimulus (V~) is opposite to the cart velocity. This corresponds to the real-life situation of body movement in a stationary environment. 'fwd' indicates the direction of forward Vc, Vi, or subject's motion detection. (II) Vi = 0 The visual world is fixed relative to the cart (moving with the subject). This corresponds to an observer inside a moving closed vehicle. (III) Vi = constant (0.4 m/see) A constant forward linear vection (fwd) is induced by a visual field moving backward at constant velocity while the cart is moved fore and aft. Responses have been considered correct if they correspond, to the direction of e-~rt velocity at that moment. Subject responses are analyzed as percentages of correct and incorrect detections of forward (R +) or backward (1%-) motion detection s u b j e c t ' s d e t e c t i o n of m o t i o n was p r o v i d e d b y v e s t i b u l a r a n d o t h e r p r o p r i o e e p t i v e cues. As s t a t e d in m e t h o d s , i t was difficult to isolate p u r e v e s t i b u l a r cues. T r a i n e d s u b j e c t s l e a r n e d to use cues o t h e r t h a n v e s t i b u l a r ones. This p a r t l y explains w h y t h e m o t i o n d e t e c t i o n thresholds, for these subjects, were often m u c h lower t h a n t h e t h r e s h o l d s p r e d i c t e d b y t h e v e s t i b u l a r model. I t is also possible t h a t t h e t h r e s h o l d values p r o v i d e d b y this m o d e l are o v e r e s t i m a t e d . H o w e v e r , t h e p r e s e n t e x p e r i m e n t s c a n n o t p r o v i d e an a n s w e r to this point. I I I . A n o t h e r conflicting s i t u a t i o n consisted of t h e a p p l i c a t i o n of t r i a n g u l a r c a r t velocities (as in I a n d I I ) , c o m b i n e d w i t h b a c k w a r d V1 (0.4 m/see) which i n d u c e d a c o n s t a n t f o r w a r d LV. A c c o r d i n g to t h e direction of c a r t velocity, t h e v e s t i b u l a r cue d i r e c t i o n t h e n either s u p p o r t e d (Vc forward) or conflicted w i t h (Vc b a c k w a r d ) t h e visual cue. Results o[ Visual-Vestibular Interaction Experiments The results o b t a i n e d from t h e s u b j e c t whose m o t i o n p e r c e p t i o n was m o s t influenced b y Vi are shown in Fig. 9. A l t h o u g h t h e m o v e m e n t s were delivered in c o n t i n u o u s runs of 16 a c c e l e r a t i o n steps, when t h e d a t a were a n a l y z e d after t h e e x p e r i m e n t , t h e acceleration steps (ramps of velocity) which gave non-conflicting m o v e m e n t s of t h e c a r t in t h e d i r e c t i o n of t h e L V (Fig. 9A) a n d those which gave conflicting m o v e m e n t s in an opposite d i r e c t i o n (Fig. 9B) were s e p a r a t e d a n d p l o t t e d in d i s t i n c t d i a g r a m s . T h e reference conditions I of Fig. 8 (Vi = - - V c ) gave a 100% correct direction d e t e c t i o n for this subject, as p l o t t e d on t h e left side of each d i a g r a m . Condition H is p l o t t e d as t h e o r d i n a t e of Vi = 0 a n d gave 75~o to 80% correct detection. 484 A. Berthoz et al. A 1oo ,' %R+ / I s 50 . . . . . . . . . . . . . . . . . . no conflict 0, ! I (9 ~'l, O V i = c s t lOO ' ' \ %R5o ~ conflict LVTIVc~.-~--~'-"-'-" ~.. o , O .25 , .5 .75 , vi 1 mlsec Fig. 9. P e r t u r b a t i o n of movement direction detection by conflicting visual and vestibular cues in the case of a very LV-sensitive untrained subject. This diagram shows the percentage of correct detections of cart movements either forward (~ or backward (~ for increasing values of constant image velocities (Vj). Each data point corresponds to the mean value for 30 accelerations repeated in a pseudo-random manner either forward or backward as described in Fig. 8. The three conditions shown in Fig. 8 have been used successively, and the responses have been divided in two groups according to whether the linearvection and cart (Vc) velocities were in the same (A) non-conflicting or in opposite (B) conflicting directions. I n both cases the first result plotted on the left vertical scale concerns the condition Vi - - V c (Condition I of Fig. 8) in which all responses are correct (R + -~ 100%) When the visual scene is moving (Condition I I I ) , the detection remains good and even improves for non-conflicting cues. However, the detection decreases very sharply with increasing Vi in the conflicting situation of an opposite movement between the visual scene and the cart. This degradation is m a x i m u m for image velocities much smaller t h a n saturation (0.4--0.6 m/see). I t can be noted t h a t the perturbation decreases when Vi reaches saturation level. This is another indirect evaluation of the range of m a x i m u m LV in these experimental conditions. These rather exceptional effects of visual cues on motion sensation can be found, though to a lesser degree, in most untrained subjects as shown in Fig. 10. In general, the total degradation reached 50% of initial detection. Trained subjects, however, showed increasingly good results. This led us to verify the effect of repetition in a separate set of experiments. The same pattern (sequence 485 Perception of Linear Horizontal Self-Motion A LvTtvc I . it Vi= cst I I I D :~:~ B loo 1 , O/oR- 50. - O~ 0 - . LV~Vc .25 .5 I i .75 i Vi lm/sec I Fig. 10. P e r t u r b a t i o n of movement direction detection b y conflicting visual and vestibular cues for 10 subjects. Same notations as in Fig. 8. The data points are mean and standard deviations for 10 untrained subjects of 64 steps of acceleration) of cart velocities, together with a constant velocity movement of the visual scene (Condition III), was presented 10 consecutive times. The perturbation of motion detection decreases with the repetition of the presentation. For the first presentation the percentage of correct detection is about 40% in the case of conflict between Vi and Vc (Vi = 0.5 m/see). I t raised up to 90~o by the 10th consecutive presentation. Whether this recovery is due to a substitution of tactile cues to vestibular cues, or to a habituation of LV remains to be determined. The possibility of a learning effect restricted to the vestibular detection of the acceleration pattern cannot be excluded, but the use of a pseudorandom distribution of steps of different accelerations and direction, as described above, decreases the probability of this learning occurring. Discussion The above results confirm the very powerful character of the sensation of linear motion induced by moving visual fields symmetric to the body. The thresholds of velocity and luminance for self-motion detection are within the same limits as those for detection of the visual images themselves. An interesting result is the clear difference between the thresholds for backward LV which are smaller than those for forward LV. 486 A. Berthoz et al. I n normal environment, as man progresses forward, through locomotion or passive movement, he usually observes relative linear movement of his peripheral surroundings backwards, and enlargement of approaching objects in front of him. When moving backward (a less common experience) there are no approaching objects to be seen, but only a diminution of the size of objects in the foreground. Thus, one possible explanation for the relative ease of developing backward LV is t h a t the lack of rearward vision imposes no inhibition, unless the lack of approaching objects for forward LV is a significant missing cue. I n addition, the experiments in which conflicting visual and vestibular cues were presented show the great dominance of the visually-induced sensation of motion. These facts are in agreement with the observations of Lee t h a t a very limited translation of a moving room can interfere with the regulation of posture. However, the latencies of onset (1 sec) are in the lower range for LV compared to those reported by Brandt et al. (1973) for circularvection around a vertical axis (1--14 set). Two hypotheses can be proposed to account for this discrepancy. The simplest and most obvious possibility is that experimental procedures were different. For instance, in the case of CV around a vertical axis (rotating drum), the optokinetic stimulus was made of vertical bars which are, according to our own experiments, less efficient in inducing motion sensation than a random pattern. I n the case of a visually-induced tilt around a horizontal axis (rotating disc), the experiments were performed on standing subjects, and proprioception from the legs and feet or the necessity of maintaining an upright posture m a y have provided conflicting cues delaying the onset of CV. I t is indeed certain t h a t the latency to onset of self-motion sensations, for either circularvection or linearvection, cannot be ascribed to purely visual factors, as the moving field is detected with minimal delay. I t was suggested (Young, 1970) t h a t in eases of visual-vestibular conflict, the vestibular signals dominate in the short-term subjective determination of acceleration, but t h a t the visual cues dominate in the long-term sensation of velocity. These views have eontributed to the theory of vestibular constraint on visually-induced tilt (Dichgans et al., 1972; Udo de Haes and SehSne, 1970; Young et al., 1973; Young et al., 1975). As applied to the current case, a step change in field velocity, with the cart stationary, implies a visual acceleration impulse which is above the threshold of the otolith system, but is definitely not confirmed by vestibular signals which continue to indicate constant (zero) velocity. Hence, the short term vestibular dominance delays the onset of LV. (The sensation of LV when a neighboring train pulls out of a railroad station is most effective when the acceleration of the visual field is low - - perhaps below otolith threshold. A step change associated with the sudden passage of a train on the next track does not normally produce LV). With prolonged stimulation, however, LV can develop and dominate, since any constant linear velocity is consistent with the otolith signals at rest. The detailed differences in latencies between LV and CV m a y be related to the convergence of visual signals with otolith afferents in the former case and semicircular canal afferents in the latter, recalling their differing dynamic response. The difference m a y well be equally due to details of the visual field presentation - just as times to onset of CV m a y vary among different experimental setups. Perception of Linear Horizontal Self-Motion 487 Another difference between LV and CV is the very small visual after-effect induced b y prolonged LV compared to the striking after-effect observed for instance after exposure to a rotating disc in the frontal plane. However, it has been noted (unpublished observations) t h a t a visual after-effect can be observed when the subject associates locomotion with LV. This difference between passively induced LV and the sensation associated with locomotion activity is yet to be studied. At this point it should be remembered that, as shown by the involvement of brain stem structures in visual-vestibular convergence, the mechanisms dealt with here concern fundamental elements of sensori-motor activity which are generally set into work during active motion, and t h a t in addition, the processing of a motor program can alter motion sensation. These considerations do imply t h a t some difference will be found between a passively induced motion, such as the one used in the present experiment, and active locomotion. Altogether, the view t h a t peripheral vision would provide an "autonomous" kinaesthetic sense, as proposed by Lee, is not convincing. Visual and vestibular information have been shown to converge on vestibular nuclei (Dichgans et al., 1973 ; H e n n et al., 1974), and it is also known t h a t visual CV can habituate vestibular rotatory nystagmus (Young and Hcnn, 1974). The question would then seem to be to characterize, as we have tried to do above, the complementarity of visual and vestibular cues. Dichgans et al. (]973) proposed t h a t vision "improves the speedometer function of the vestibular system". However, whether LV and vestibular linear motion sensation are actually summed algebraically or whether they interact through a switching hierarchical process remains to be determined. The observed reduction in the dominance over conflicting vestibular cues with repeated trials shows that visual-vestibular interaction is modifiable. One occurrence for this modification is when a repeated association of tactile, proprioceptive and vestibular cues opposes visual cues. A p p e n d i x Relationship between the velocity of visual images projected on the screen and the line-of-sight angle. The angular velocity da of a point M belonging to these images is a function of: a) the velocity of visual images (Vi) b) the distance between the eye and the screen (H) c) the angle of incidence (a). Elementary trigonometric considerations give: da Vi -sin2a dt H For the upper screen H : 0.4 m. Then for an image velocity of I m/set we have: da - = 143 sin2a (deg/sec) Curve I dt For the lateral images: H ---- 0.5 m, then da -115 sin~a (deg/sec) Curve I I dt 488 A. Berthoz st al. 15o.d"T (~ ) 100. .) E 5O- o o((o) o 2'0 ~ Bb Bb Fig. 11 Acknowledgements. This research was supported by a grant from Organisme National de S6curit6 Routibre and by C.N.I%S. and C.N.A.M. (France). L.R. Young was supported by NASA Grant NGR-22-009-701(LR4). References Berthoz, A., Pavard, B., Young, L.R.: R61e de la vision p@riph~rique et interactions visuovestibulaires darts la perception exoeentrique du mouvement lin~aire chez l'homme. C.R. Acad. Sci. (Paris) 278, D, 1605--1608 (1974) Brandt, Th., Wist, E., Dichgans, J. : Optisch induzierte Pseudo Coriolis-Effecte und Circularvektion. Arch. Psychiat. Nervenkr. 214, 365 389 (1971) Brandt, Th., Dichgans, J., Koenig, E.: Perception of self-rotation induced by optokinetic stimuli. Pfiiigers Arch. 332, P~ 98 (1972) Brandt, Th., Diehgans, J., Koenig, E. : Differential effects of central and peripheral vision for egocentric and exocentric motion perception. Exp. Brain I~es. 16, 476--491 (1973) Clark, B. : The oculogravic illusion as a test of otolith function. In: Third Syrup. on the role of the vestibular organs in the exploration of space. ~NASASP 152, 331--339 (1968) Clark, B., Graybiel, A. : Apparent rotation of a fixed target associated with linear acceleration in flight. Amer. J. Ophthal. 32, 549--557 (1949) Daniel, K. : Dcr Einflug dcr ReizmarkengrSge auf die Lichtunterschiedscmpfindlichkeit der Netzhaut bei verschiedenen Adaptationszust~nden. Inaugural dissertation. Tfibingen. (1959). In: Handbook of sensory physiology: visual psychophysics, p. 126--127. (ed Jameson, D., Hurvich, L.M.). Berlin-Heidelberg-New York: Springer 1972 Denton, G.C. : The influence of visual pattern on perceived speed, p. 409--420. Crowthornc, G.B.L.P~.: Road Research Laboratory Report 1971 Dichgans, J., Brandt, Th. : The psychophysics of visually induced perception of self-motion and tilt. In: The Neurosciences III, p. 123--129. MIT Press 1974 Dichgans, J., Held, 1~., Young, L., Brandt, Th. : Moving visual scenes influence the apparent direction of gravity. Science 178, 1217--1219 (1972) Dichgans, J., Schmidt, C.L., Graf, W. : Visual input improves the speedometer function of the vestibular nuclei of the goldfish. Exp. Brain Res. 18, 319--322 (1973) Fischer, 1VLtt., Kornmiiller, A.E. : Optokinctisch ausgelOste Bcwegungswahrnehmungen und optokinetischer Nystagmns. J. Pscyhol. Ncurol. (Lpz.) 41, 383--420 (1930) Gibson, J.: The visual perception of objective motion and subjective movement. Psychol. Rev. 61, 304--314 (1954) Graybiel, A. : Oculogravic Illusion. Arch. Ophthal. 48, 605--615 (1952) Hcnn, V., Young, R.L., Finley, C. : Vestibular units in alert monkeys are also influenced by moving visual fields. Brain Res. 71, 144--149 (1974) Perception of Linear Horizontal Self-Motion 489 Jongkees, L.B.W.: On the otoliths: their function and the way to test them. In: Third Symposium on the role of vestibular organs in space exploration. 1967. NASA SP 152, 307--331 (1968) Lee, D.N., Aronson, E. : Visual proprioceptive control of standing in human infants. Perception and psychophysics 15, 529 532 (1974) Lishman, J . R . , Lee, D.N. : The autonomy of visual kinaesthesis. Perception 2, 287--294 (1973) Lipetz, L.E.: The Creation of physiological and psychological aspects of sensory intensity. In: Handbook of Sensory Physiology, Vol. I, p. 191--225. (ed. Loewenstein, W.). BerlinHeidelberg-New York: Springer 1971 Mach, E. : Grundlinien der Lehre yon den Bewegungsempfindungen, 129 pp. Leipzig: Engelmann 1875 Maekawa, A., Simpson, J.: Climbing fiber activation of Purkinge cells in the flocculus by impulses transferred through the visual pathways. Brain Res. 39, 245--251 (1972) Salvatore, S.: Velocity sensing. Highway Res. Record 282, 79--90 (1968) Schmidt, F., Tiffin, J. : Distortion of drivers' estimates of automobile speed as a function of speed adaptation. J. appl. Psychol. 53, 536--539 (1969) Sch6ne, I.i., Mortag, H.G. : Variation of the subjective vertical on the parallel swing at different body positions. Psychol. Forseh. 32, 124 134 (1968) Sharpe, C. 1~., Tolhurst, D.J. : Orientation and spatial frequency channels in peripheral vision. Vision l~es. 13, 2103--2112 (1973) Udo de Haes, H.A., SchSne, H.: The effectiveness of the statolith organs in human spatial orientation. Acta oto-laryng. (Stockh.) 69, 25--31 (1970) Young, L. 1% : On visual-vestibular interaction. In: 5th Symposium: The role of the vestibular organs in space exploration (Pensacola). NASA SP 314, 205--210 (1970) Young, L.R., Meiry, J . L . : A revised dynamic otolith model. Aerospace Med. 39, 606 (1965) Young, L.R., Dichgans, J., Murphy, R., Brandt, Th.: Interaction of optokinetic and vestibular stimuli in motion perception. Acta oto-laryng. (Stockh.) 76, 24--31 (1973) Young, L. g., Henn, V. S. : Selective habituation of vestibular nystagmus by visual stimulation. Acta oto-laryng. (Stockh.) 77, 159--166 (1974) Young, L.R., Oman, C. : Influence of head position and field on visually induced motion effects in three axes of rotation. In: Proc. of 10th Annual Conference in Manual Control Wright Patterson AFB, (Ohio) (1974) Young, L.R., Oman, C., Dichgans, J.: Influence of head orientation on visually induced pitch and roll sensation. Aviation and Environmental Medecine 46, 264--268 (1975) Dr. A. Berthoz Laboratoire de Physiologic du Travail 41, rue Gay-Lussac F - 75005 Paris France