Perception

Monocular masking studies show that the visibility of a one-dimensional sinusoidal grating remains unchanged in the presence of masking noise filtered so as to contain spectral components that are at least two octaves away from the spatial frequency of the grating (Stromeyer and Julesz 1972). In the present study, random-dot stereograms were bandpass filtered in the two-dimensional Fourier domain, and masking noise of various spatial frequency bands was added to the filtered stereograms. Masking noise bands containing equally effective noise energy were selected such that their bands were either overlapping with the stereoscopic image spectrum or were two octaves distant. The first case resulted in binocular rivalry; however, in the second case stereoscopic fusion could be maintained in the presence of strong binocular rivalry owing to the masking noise. This finding indicates that spatial-frequency-tuned channels are not restricted to one-dimensional gratings but operate on two-dimensional patterns as well. Furthermore, these frequency channels are utilized in stereopsis and work independently from each other, since some of these channels can be in binocular rivalry while at the same time other channels yield fusion. The main binocular experiments are demonstrated.

Some new subjective contours are described and it is suggested that both these and previously observed disparity-based illusory contours can be explained simply in terms of the magnetic dipole model of stereopsis without any reference to gestalt or cognitive factors.

The amount of depth seen in a random-line stereogram steadily diminishes as corresponding lines in the two fields of view are made increasingly dissimilar in shape. This depth reduction effect is more acute in high- than in low-density stereograms. Both of these results can be explained in terms of the magnetic dipole model of stereopsis.

Two experiments are reported which investigated the effects on stereopsis perception times of including monocularly conspicuous features in random-dot stereograms. It was found that such features facilitated stereopsis in large-disparity but not in small-disparity stereograms, perception times for the latter being relatively short with or without monocular features. Facilitation in the large-disparity stimuli came about both from features which delineated the shape of the whole disparate area and from features which merely happened to lie in the same depth plane as the disparate area, but which did not give any shape cues. It is argued that these various results can be well accounted for by a ‘vergence hypothesis’, which supposes that the long perception times often found with random-dot stereograms are due in part to the absence of stimulus features which can guide the vergence movements necessary for fusing the display.

Many observers of complex random-dot stereograms find that the depth effect takes several seconds, or even minutes, to develop. Julesz (1971) has noted that giving a priori information to such observers about the nature of the ‘hidden’ cyclopean object appears to facilitate their stereopsis. An experiment is reported which investigated this possible facilitation. Naive subjects were shown a complex stereogram following various kinds of preliminary assistance, ranging from simply telling them about the amount of depth they could expect to see to showing them a full-scale model of the cyclopean object. Surprisingly, no benefit from such assistance could be demonstrated. All observers improved their stereopsis perception times with repeated presentations of the stereogram, showing that they could, in principle, benefit from assistance. A follow-up study three weeks later revealed that a substantial part of this improvement was maintained, indicating that the perceptual learning involved can last for a considerable period of time.

The amount of depth seen in a multi-line stereogram composed of horizontal lines steadily decreases as the lines in one field of view are rotated about their midpoints. This effect of orientation difference on stereopsis is more acute the longer the lines in the stereogram. It is suggested that the critical factor underlying the depth reduction is not orientation difference per se, but rather the vertical disparity which an orientation difference introduces into the display between the tips of corresponding lines. This interpretation is supported by the fact that similar vertical disparities caused similar depth reductions regardless of the length of the lines in the stereogram.

Adaptation to a random-dot stereograting with no monocularly visible contours produces a tilt aftereffect in a briefly-viewed test stereograting. The effect is maximal for adapting orientation at ±20–30° from the test orientation. Similarly, the perceived spatial frequency of a stereograting is altered by adaptation to a stereograting of adjacent spatial frequencies.

A stereogram was presented with patterns of opposite contrast–one positive the other negative. One eye received only the positive pattern; the other received both positive and negative patterns superimposed. Subjects reported apparent reversals of perceived depth: crossed (convergent) disparity made the fused stereogram appear further away, whilst uncrossed (divergent) disparity made it appear nearer. It is believed that spatial summation in the visual system blurs the superimposed positive-and-negative contours and shifts their effective positions, leading to reversals in perceived disparity.

The inappropriate constancy scaling theory of visual distortion illusions is tested by optically projecting typical models giving these figures by perspective. Appropriate or inappropriate stereoscopic disparities are then added–with the prediction that when perspective and stereo are geometrically correct the distortion should vanish. This is confirmed with measurements for the Müller-Lyer illusion and by observation of several other classical examples. It is suggested that much previous work has investigated ‘end stop’ conditions, given by angles too extreme to be generated as perspective. Conditions for appropriate scaling, giving zero or small distortions, are found to be critical but readily attainable.

A device has been designed and built for drawing in three dimensions. Drawing may be free-hand; or three-dimensional structures or stereograms (such as stereo x-ray pictures) may be traced in three dimensions. The device produces coordinate data, which may be fed to a plotter to produce permanent stereo pairs, and may be fed to a computer for storage and analysis. Stereoscribe can also be used to plot the ‘subjective space’ of observers, to record perceptual knowledge.