I have recently studied the perception of brightness, shape and motion; and also muscle control during jogging.
Stratton (1897) wore upside-down prisms for 1--3 weeks to see if he could adapt to them. I wore special video goggles that inverted not position but brightness: I lived in a negative world in which blacks were white and vice versa. At first I could not recognise faces at all. I soon learned to recognise emotional expressions (surprise, anger &c), but could never recognise faces, e.g. photos of celebrities. Shadows were white; fuzzy shadows looked like glowing reflections, bright highlights looked like black bugs, and sharp shadows looked like white paper cutouts. My perception of shape from shading did adapt. My postdoc Alan Ho also wore the goggles for a week, and when he took the goggles off his own hands surprisingly seemed not to be his, but to belong to somebody else, a symptom well known in patients with parietal brain lesions!
Alan Ho and I have discovered that Simultaneous Contrast, in which a white surround makes a gray spot look darker, is greatly enhanced if the spot (not the surround) flickers between black and white (Fig. 1). Color contrast is likewise enhanced by chromatic flicker of the test spot. Negative afterimages of luminance [and color] were also enhanced when projected on a test field that flickered between black and white [or between complementary colors.] So flicker also enhances both simultaneous and successive contrast, for both luminance and color. We examined the combination rules for pairs of luminances which were presented either successively as flicker or else dichoptically (one to each eye, and fused binocularly). The brightness averaging functions for spatial increments (light spots) on dark surrounds were quasi-linear for binocular fusion but quadratic for flicker. For spatial decrements (dark spots) on white surrounds, the brightness averaging functions were strongly nonlinear winner-take-all for both binocular fusion and flicker. We also found temporal analogues of Fechner's (1860) paradox and Levelt's (1965) dichoptic contour effect. We conclude that the visual rules for combining luminances, whether in flicker or binocular fusion, favour disproportionately the spot with the higher contrast.
In White's (1979) illusion of brightness, the background is a square-wave grating of black and white bars. Grey segments that replace parts of the black bars look much brighter than grey segments that replace parts of the white bars. White's illusion has been attributed to simultaneous contrast caused by anisotropic lateral inhibition, which is supposedly stronger along the bars than across them. But we ruled out any anisotropic processes by demonstrating illusions similar to White's effect in checkerboard (Fig. 2) and isotropic random-dot patterns. The illusion has also been attributed to assimilation, the opposite of simultaneous contrast. But by using colored patterns we demonstrated that the hue shifts are in the direction of contrast, not assimilation. By decomposing White's stimuli into two component pictures we showed that the illusion is caused by isotropic simultaneous contrast.
With P. Cavanagh, I. Watanabe and I. Shrira, I discovered some brightness illusions from diamond and hourglass shapes (not illustrated). These may throw light on early (low-level) mechanisms of brightness perception.
When a flickering disk was steadily viewed in peripheral vision (20°--40° eccentricity), we noticed that after about 10 s it appeared to fade out and disappear (Spillman 1987). We measured this phenomenon in various conditions, causing our observers to increase the flicker amplitude gradually to keep the flicker just visible, and we hope to develop this procedure into an early diagnostic test for the visual disorder of glaucoma -- a leading cause of blindness in the USA.
Why did El Greco paint people as tall and thin? Was there something wrong with his eyes? I turned students into "artificial El Grecos" by fitting them with a special lens that stretched the visual scene horizontally. When asked to draw a freehand square, they drew a tall, thin El Greco rectangle. However, when asked to copy a square, they copied it perfectly. Would El Greco have painted perfect portraits from life, and distorted people drawn from memory? If El Greco were astigmatic, he would have seen a distorted world for years, not minutes, so I persuaded a volunteer to wear the El Greco lens for two days and make frequent drawings. Result: Her copied squares were always perfect copies. Her freehand squares were at first tall, thin rectangles, but they gradually became squarer as she adapted to the lens, and after two days her freehand squares were perfectly square despite the lens. Conclusion: Even if El Greco were astigmatic, this would not have affected his paintings. His distortions arose from an artistic choice, not from a visual defect.
I have discovered two new, related illusions of motion. 1) In the 'chopstick illusion' a vertical and a horizontal line overlapped to form a cross. Each line moved along a separate counterclockwise circular path in antiphase, without changing orientation. Result: The intersection of the lines was wrongly perceived as rotating counterclockwise , although physically it moved clockwise (Fig. 3). Conclusion: Intersections are not parsed as objects, so their motion path cannot be extracted, but instead the motion of the terminators (tips) propagates along the lines and is blindly assigned to the intersection. 2) In the 'sliding rings illusion', two rings overlapped in a figure-8 and rotated about the centre of the figure-8. With Dana Ballard at Rochester U, I measured eye movements made to these rings. Result: If perceived as a rigid rotating figure-8, the eyes could readily track the intersections of the rings. But if the rings were perceived as sliding over each other, the eyes were completely unable to track the moving intersections, although they could track each moving ring (Fig. 4). Conclusion: Pursuit eye movements are under top-down control and depend upon perceptual parsing of objects.
If a black square and a white square exchange places on a grey surround, does the black square or the white square appear to jump? It turns out (Anstis & Mather 1985) that on a dark background it is the white square, and on a light background the black square, that appears to jump. So motion is perceptually assigned to the higher-contrast square. David Smith and I now find that the same applies to ambiguous stereo or stereo depth. The higher contrast signal of vernier offset or of depth carries the day. The indifference luminance at which the motion, vernier offset or depth became ambiguous or vanished, lay at the mean of the black and white bars. We found it was at the arithmetic mean (B+W)/2, not the geometric mean à(B*W). This showed that the visual system processed luminance linearly, without any log transform. This is puzzling because many other visual processes involve a compressive nonlinearity, often logarithmic. We cannot yet explain this new discrepancy.
George Mather (Dept of Psychology, Sussex U, UK) and I extended these results for jumping bars to second-order apparent motion of textured squares. Two squares of random-dot texture, like sandpaper, exchange places. They resemble fine and coarse sandpaper. On a background of very fine sandpaper it is the coarse square, and on a very coarse background it is the fine square, that carries the apparent motion. This elucidates perceptual mechanisms of second-order motion.
Dr. Herwig Baier (UCSD Biology Dept) and I have devised some novel tests for zebrafish vision, based on the innate optomotor response which causes the zebrafish to swim along in pursuit of any large moving surface. Our computer movies measure the fishes' optomotor responses to special patterns of colour and brightness. For example, we superimposed red stripes moving to the left on green stripes moving to the right. For a human observer, the pattern seems to move to the left if the red stripes are lighter than the green, but to the right if the red stripes are darker than the green. At red-green equiluminance, no motion is seen. With this technique we have confirmed that zebrafish are much more green-sensitive and less red-sensitive than humans. Another stimulus, which we shall not describe in detail, appeared to move to the left [right] if the fish process luminance information in a linear [logarithmic] fashion. Results favoured linear, not logarithmic processing. We have collected data on about 100 fishes so far, and our research is still in progress. We shall hook other visual illusions to motion in special moving displays, such that the fish will follow them if and only if they are subject to the illusion.
I have discovered some new motor aftereffects from jogging. After jogging on a treadmill for 30 s, observers were asked to dismount and jog in place on the ground. Result: they inadvertently jogged forward through 1-2 m. We examined the neural site of this aftereffect by showing that after hopping on the treadmill on the left foot, observers still showed the aftereffect when they hopped on the ground on their left foot, but not when they hopped on their right (unadapted) foot. This lack of inter-limb transfer shows that the aftereffect has nothing to do with optic flow nor with the balance organs, but is located in the neural pathways that control each leg separately. We are now investigating corresponding aftereffects from compensatory rotary motion on a motorized turntable, with Ian Howard and Jim Zacher at York U, Toronto. The turntable (like a rotary treadmill) rotates clockwise but the observer keeps facing the same way (say, north) by making counterclockwise twisting efforts with his feet. During the test period the observer attempts to jog on the spot with eyes closed, but inadvertently rotates on the spot, often through more than 360°. The research is still in progress.
sanstis@ucsd.edu
Back to Stuart
Anstis Home Page
Psychology Dept