Hierarchy of spatial interactions in the processing of contrast-defined contours.

Both psychophysical and neurophysiological evidence suggest that there are two visual cortical processing streams, a linear stream that processes first-order stimuli and a nonlinear stream that also processes second-order stimuli. This evidence also suggests that before the extraction of the second-order signal, the nonlinear pathway broadly but not completely pools signals across initial linear filters that encode the orientation of the carrier of the second-order signal. The evidence suggests that such pooling does not occur across carrier spatial frequencies. We show that similar results are obtained with repulsion tilt illusions but not with attraction effects. Attraction effects exhibit complete orientation crossover (while retaining spatial frequency selectivity), perhaps indicating higher-level processing; an experiment on interocular transfer of the effects supported this conclusion.

Asynchronous processing in vision: color leads motion.

It has been demonstrated that subjects do not report changes in color and direction of motion as being co-incidental when they occur synchronously. Instead, for the changes to be reported as being synchronous, changes in direction of motion must precede changes in color. To explain this observation, some researchers have suggested that the neural processing of color and motion is asynchronous. This interpretation has been criticized on the basis that processing time may not correlate directly and invariantly with perceived time of occurrence. Here we examine this possibility by making use of the color-contingent motion aftereffect. By correlating color states disproportionately with two directions of motion, we produced and measured color-contingent motion aftereffects as a function of the range of physical correlations. The aftereffects observed are consistent with the perceptual correlation between color and motion being different from the physical correlation. These findings demonstrate asynchronous processing for different stimulus attributes, with color being processed more quickly than motion. This suggests that the time course of perceptual experience correlates directly with that of neural activity.

The depth and selectivity of suppression in binocular rivalry.

Binocular rivalry occurs when the two eyes are presented with incompatible stimuli and the perceived image alternates between the two stimuli. The aim of this study was to find out whether the periodic perceptual loss of a monocular stimulus during binocular rivalry is mirrored by a comparable loss of contrast sensitivity. We presented brief test stimuli to one eye while its conditioning stimulus was dominant or suppressed. The test stimuli were varied widely across four stimulus domains--namely, the relative stimulation of medium- and long-wavelength-sensitive cones, duration, spatial frequency, and grating orientation. The result in each case was the same. Suppression depended slightly or not at all on the type of test stimulus, and contrast sensitivity during suppression was around 64% of that during dominance. The effect of suppression on sensitivity is therefore very weak, relative to its effect on the perceived image. Furthermore, suppression was largely independent of the similarity between the conditioning and the test stimuli, indicating that our results are better explained by eye suppression than by stimulus suppression. A model is presented to account for the small, monocular sensitivity loss during suppression: It assumes that test detection precedes conditioning stimulus perception in the visual pathway.

Interaction between first- and second-order orientation channels revealed by the tilt illusion: psychophysics and computational modelling.

This paper examines the interaction between first- and second-order contours in the orientation domain. Using the simultaneous tilt illusion (TI), we show that the apparent rotation of a vertical test grating away from that of a surrounding inducing grating (repulsion effect) occurs when both the inducing and test grating are either first- or second-order. Furthermore, a significant repulsion effect is obtained when a first-order inducing grating surrounds a second-order test. If lateral inhibitory interactions between populations of orientation selective neurons provides a plausible explanation for orientation repulsion effects [Blakemore, C. B. Carpenter, R. H. S. & Georgeson, M. A. (1970) Nature, 228, 37-39], it is likely that the cue-invariant mechanisms that encodes the orientation of first- and second-order contours also exhibit inhibitory interactions. A two-channel computational model of orientation encoding is presented where one channel encodes only first-order stimuli while the second channel encodes both first- and second-order contours. In addition to predicting the orientation repulsion effects we observed, the model also provides a functional account of orientation attraction effects in terms of the responses of populations of orientation-tuned neurons.

Orthogonal adaptation improves orientation discrimination.

We investigated the effect of adaptation on orientation discrimination using two experienced observers, then replicated the main effects using a total of 50 naïve subjects. Orientation discrimination around vertical improved after adaptation to either horizontal or vertical gratings, but was impaired by adaptation at 7.5 or 15 degrees from vertical. Improvement was greatest when adapter and test were orthogonal. We show that the results can be understood in terms of a functional model of adaptation in cortical vision.

Orientation processing mechanisms revealed by the plaid tilt illusion.

The tilt after-effect (TAE) and tilt illusion (TI) have revealed a great deal about the nature of orientation coding of 1-dimensional (1D) lines and gratings. Comparatively little research however has addressed the mechanisms responsible for encoding the orientation of 2-dimensional (2D) plaid stimuli. A multi-stage model of edge detection has recently been proposed [Georgeson, M. A. (1998) Image & Vision Computing, 16(6-7), 389-405] to account for the perceived structure of a plaid stimulus that incorporates extraction of the zero-crossings (ZCs) of the plaid. Data is presented showing that the ZCs of a plaid inducing stimulus can interact with vertical grating test stimulus to induce a standard tilt illusion. However, by considering the second-order structure of a plaid rather than ZCs, it was shown that the perceived orientation of the vertical test grating results from the combination of orientation illusions due to the first- and second-order components of an inducing plaid. The data suggest that the mechanisms encoding the orientation of second-order contours are similar to, and interact directly with, those that encode first-order contours.

Dissociable factors affect speed perception and discrimination.

Factors affecting our judgement of the speed of visual motion were investigated. Two types of judgement were made: perceived speed relative to a standard comparison stimulus, and discrimination between the speeds of similar stimuli. The factors affect ng these two judgements were found to be doubly dissociable, suggesting that they may be constrained by processing at different levels of the visual hierarchy. The results are discussed in terms of the 3-D interpretation of visual image motion, and related to possible neural substrates.

Monocular symmetry is neither necessary nor sufficient for the dichoptic perception of bilateral symmetry.

Barlow and Reeves [1979. Vision Research, 19, 783-793] showed that bilateral symmetry detection in dot patterns is about equally efficient whether the displays are viewed monocularly or binocularly. If there is a binocular process which can be stimulated monocularly, this experiment does not indicate whether symmetry detection occurs before or after the site of binocular integration. This is so because the symmetrical patterns would have stimulated both monocular and binocular mechanisms under both viewing conditions. We presented stereoscopic 20-dot patterns, ten dots to each eye, for 150 ms so that 'false fusion' rather than rivalry occurred. Any axis of symmetry in the patterns was oriented at vertical (90 degrees ) or +/-1, 2, 3, or 4 degrees from vertical. The task was to judge whether the axis was tilted left or right of vertical, using the method of constant stimulus differences. Three kinds of pattern were used: SSS patterns were symmetrical in each eye alone and also dichoptically; NNS patterns were random monocularly but dichoptically symmetrical; and SSN patterns were symmetrical monocularly but dichoptically non-symmetrical. Orientation judgements were accurate, and equally so, for SSS and NNS displays but were extremely poor under SSN conditions. A control experiment showed that the poor performance in the SSN condition was not due to the axes of symmetry being eccentric to the fixation point. Thus, monocular symmetry is neither necessary nor sufficient for dichoptic bilateral symmetry perception; and symmetry mechanisms have no access to monocular signals.

Detection of bilateral symmetry in complex biological images.

The recognition of bilateral symmetry in simple dot patterns is reliably influenced by orientation. Performance is best when the axis of symmetry is vertical. We conducted two experiments to determine whether stimulus orientation also affects detection of the low levels of naturally occurring asymmetry in complex biological images. University students judged whether colour images displayed on a computer monitor possessed perfect bilateral symmetry. Stimuli were generated from high-resolution plan-view images of crabs and insects. In experiment 1, the asymmetric stimuli were the original animals, displayed on a standard black background. Symmetrical versions of each natural image were generated by sectioning the shape at the midline, copying and reflecting one side and then fusing the two halves together. To facilitate comparison of results with those obtained in earlier studies, we also presented dot patterns based upon both the slightly asymmetric and perfectly symmetrical natural images. Experiment 2 was designed to assess whether symmetry detection was dependent upon the markings and patterns on the body and appendages of the animals. The natural images were converted to silhouettes and tested against matched dot patterns. In both studies, images were presented in a random sequence with the axis of symmetry vertical, horizontal, oblique left, and oblique right. Performance with the biological images was consistently better than with the dot patterns. Abolishing fine detail did not appreciably reduce this effect. A pronounced vertical advantage was apparent with all stimuli, demonstrating that this phenomenon is robust despite considerable variation in image complexity. The implications of orientation effects for perception of natural structures are discussed.

The differential effects of simultaneous and successive cueing on the detection of bilateral symmetry in dot patterns.

Corballis and Roldan (1975) obtained speeded judgements of whether dot patterns were bilaterally symmetrical about, or translated across, a line. Reaction times (RTs) were ordered V (vertical) > D (diagonal) > H (horizontal) where ">" means faster than. Similar results occurred with blocked axis orientations, suggesting subjects cannot prepare by rotating a mental frame of reference. Blocking trials may have been ineffective because blocking cannot provide incremental benefits over those already provided by axis lines. Four experiments show that the usual axis orientation ordering of V > H > D is markedly attentuated by simultaneous but not successive axis lines. Also, axis cue lines and axis blocking are not equivalent treatments. Instead, unblocked line cues require finite processing time whereas, under blocking, subjects can prepare for the expected orientation. There was no suggestion anywhere of the V > D > H axis ordering that Corballis and Roldan reported. Successive axis line cues may only direct attention to the orientation being cued, but simultaneous line cues may change the stimulus itself, thus providing an additional means of visual processing that facilitates symmetry detection at non-vertical axis orientations.

A functional angle on some after-effects in cortical vision.

The question of how our brains and those of other animals code sensory information is of fundamental importance to neuroscience research. Visual illusions offer valuable insight into the mechanisms of perceptual coding. One such illusion, the tilt after-effect (TAE), has been studied extensively since the 1930s, yet a full explanation of the effect has remained elusive. Here, we put forward an explanation of the TAE in terms of a functional role for adaptation in the visual cortex. The proposed model accounts not only for the phenomenology of the TAE, but also for spatial interactions in perceived tilt and the effects of adaptation on the perception of direction of motion and colour. We discuss the implications of the model for understanding the effects of adaptation and surround stimulation on the response properties of cortical neurons.

Does cortical motion adaptation exhibit functional properties analogous to light adaptation in the retina?

PURPOSE: The retina codes variations in luminance by adapting to and hence discounting, the mean luminance. During adaptation to a moving pattern, perceived speed decreases. Thus we know that the adapted visual system does not simply code the absolute speed of a stimulus. We hypothesize that adaptation to a moving stimulus serves to optimize coding of changes in speed at the expense of maintaining an accurate representation of absolute speed. In this case we would expect discrimination of speeds around the adapted level to be preserved or enhanced by motion adaptation. METHODS AND RESULTS: After adaptation to motion in the same direction as a subsequent test stimulus, seven of eight subjects showed a reduction of perceived speed in the adapted region and seven showed enhanced discrimination. CONCLUSIONS: We conclude that motion adaptation preserves or enhances differential speed sensitivity at the expense of an accurate representation of absolute speed in a manner analogous to retinal light adaptation.

Neural substrates of the tilt illusion.

PURPOSE: It has been suggested that direct and indirect tilt illusions and after-effects have different mechanisms, namely that the direct effects arise in VI and are sensitive to differences in spatial and temporal parameters between test and inducing stimuli, whereas indirect effects arise in extrastriate cortex and are insensitive to such parameters. When Wolfe (Vision Research 1984; 24: 1959-64) reported that large direct tilt after-effects occurred with short test flashes, he postulated that either there are distinct mechanisms which process brief and longer duration stimuli or that there are distinct mechanisms that are not primarily concerned with duration but are differentially responsive to temporal parameters amongst several others. RESULTS: In three experiments we demonstrate that large direct tilt illusions can be induced when parameters other than duration are manipulated, including contrast and spatial frequency, and that such large effects can occur when stimulus parameters are chosen to favour preferentially either the transient (magnocellular-like) system or the sustained (parvocellular-like) system. CONCLUSIONS: These results are thus consistent with Wolfe's second hypothesis. None of these stimulus manipulations had any effect on indirect tilt illusions, consistent with previous findings and hypotheses about the different mechanisms of the direct and indirect effects.

Second order components of moving plaids activate extrastriate cortex: a positron emission tomography study.

A moving plaid is a composite pattern produced by superimposing two sinusoidal gratings which differ in orientation and motion direction. The perceived drift direction of a plaid appears to be determined partly by a binocular mechanism, which follows intersection of constraint rules (Burke and Wenderoth, 1993b), and partly by a monocular mechanism, which tracks the dark and bright intersects of the plaid, the contrast envelopes. The first neurones that respond to plaids as patterns rather than component gratings are found in area V5, also known as MT, which is exclusively binocular. Therefore, the psychophysical evidence suggesting that the contrast envelope tracking mechanism is monocular is surprising but has been obtained consistently. We aimed to localize the contrast envelope tracking mechanism by undertaking a positron emission tomography (PET) activation experiment in which the subjects were presented with alternating plaid components during the control scan and with the moving plaid resulting from the superposition of these components as the activation scan. The results showed differential activation in area V3. Recent results from macaque single cell recordings have also demonstrated increased sensitivity to moving plaid stimuli compared to the plaid component gratings in V3 neurones.

Determinants of fusion of dichoptically presented orthogonal gratings.

Liu, Tyler, and Schor (1992 Vision Research 32 1471-1479) reported the surprising finding that dichoptically presented orthogonal sine-wave gratings do not always produce binocular rivalry. Gratings of high spatial frequency, and especially of low contrast, fuse to produce a stable percept of a dichoptic plaid. Using a somewhat different perceptual task, we replicated those findings and extended them. The probability of a plaid percept is higher for square-wave gratings than for sine-wave gratings, and higher still for rectangular-wave gratings with high duty cycles (with very thin light or dark bars). Experiments were conducted to test whether this duty-cycle effect was due to changes in overall luminance, or in the size of the regions of luminance congruity (which may reduce the probability of rivalry), but no such effects could account for the results. The presence of locally conflicting contour information in the two eyes was shown to be an important determinant of rivalry onset, but, since removing such regions did not eliminate rivalry, other factors also have a role to play. The spatial frequency composition of the gratings is one such factor which is consistent with all of the findings we report.

Adaptation to temporal modulation can enhance differential speed sensitivity.

During adaptation to a moving pattern, perceived speed decreases. Thus we know that the adapted visual system does not simply code the absolute speed of a stimulus. We hypothesised that adaptation to a moving stimulus serves to optimise coding of changes in speed at the expense of maintaining an accurate representation of absolute speed. In this case we would expect discrimination of speeds around the adapted level to be preserved or enhanced by motion adaptation. Speed discrimination thresholds were measured for sinusoidal gratings (1.25 cpd; 12.5 Hz; 40% contrast) with and without prior adaptation to moving, static, and flickering stimuli. After adaptation to motion in the same direction as the test, seven of eight subjects showed a reduction of perceived speed in the adapted region, and seven showed enhanced discrimination. Similar effects were found for adaptation to motion in the opposite direction to the test and to counter-phase flicker, suggesting that adaptation is driven by temporal modulation rather than by motion per se. We conclude that motion adaptation preserves or enhances differential speed sensitivity at the expense of an accurate representation of absolute speed.

Large repulsion, but not attraction, tilt illusions occur when stimulus parameters selectively favour either transient (M-like) or sustained (P-like) mechanisms.

A vertical test grating appears tilted away from a surrounding inducing grating which is 15 degrees from vertical (repulsion effect) but towards an inducer 75 degrees from vertical (attraction effect). This is the tilt illusion (TI) and similar effects occur when inducing and test stimuli are presented successively (tilt after-effect or TAE). When it was reported [Wolfe, J. (1984). Vision Research, 24, 1959-1964] that large repulsion TAEs occurred with short test flashes, Wolfe postulated that either there are distinct mechanisms which process brief and longer duration stimuli; or that there are distinct mechanisms which are not primarily concerned with duration but are differentially responsive to temporal parameters, amongst several others. Other evidence that TI attraction effects are not modulated by test flash duration resulted in an hypothesis that repulsion and attraction effects are mediated by transient and sustained mechanisms, respectively [Wenderoth, P., van der Zwan, R., & Johnstone, S. (1989). Perception, 18, 715-728]. We demonstrate that large repulsion TIs can be induced when parameters other than duration are manipulated, including contrast and spatial frequency but that these parameters fail to modulate attraction TIs. These results are consistent with some previous hypotheses regarding the origin of repulsion and attraction effects and with Wolfe's latter hypothesis but do not support the view that the two effects are processed, respectively, by transient and sustained mechanisms.

Evidence that both area V1 and extrastriate visual cortex contribute to symmetry perception.

Bilateral symmetry is common in nature and most animals seem able to perceive it. Many species use judgements of symmetry in various behaviours, including mate selection [1-3]. Originally, however, symmetry perception may have developed as a tool for generating object-centered, rather than viewer-centered, descriptions of objects, facilitating recognition irrespective of position or orientation [4]. There is evidence that the visual system treats the orientation of axes-of-symmetry in the same way it treats in orientation of luminance-defined contours [5], suggesting that axes-of-symmetry act as 'processing tokens' [6]. We have investigated the characteristics of neural mechanisms giving rise to the perceived orientation of axes-of-symmetry. We induced tilt aftereffects with symmetrical dot patterns, eliciting perceived angle expansion and contraction effects like those usually observed with luminance-defined contours [7,8]. Induction of aftereffects during binocular rivalry resulted in a reduction of the magnitude of these effects, consistent with the aftereffects being mediated in extrastriate visual cortex, probably between visual areas V2 and MT [9]. In a second experiment in which the aftereffects were induced monocularly, their magnitudes were measured in the unadapted eye. Contraction effects transferred completely, suggesting that they are mediated by binocular cells. Expansion effects did not transfer completely, consistent with their having a monocular component. These data suggest that information about the orientation of axes-of-symmetry may be available as early as area V1, but that processing continues in extrastriate cortex.

Effects of pattern orientation and number of symmetry axes on the detection of mirror symmetry in dot and solid patterns.

It has been postulated that as the number of axes of symmetry in a pattern increases, so pattern 'goodness' increases. Recently, a distinction was made between two different theoretical accounts of regularity or 'goodness' in relation to patterns with mirror symmetry: the 'transformational' and the 'holographic' models. It was argued that the former predicts a 'goodness' ordering of four > three > two > one whereas the latter predicts four > two > three > one, where '>' means greater regularity or goodness. In three experiments, we have tested these predictions. In experiment 1, we measured percentage correct and reaction time to dot patterns which had one, two, three, or four axes of symmetry and were flashed for 150 ms. Experiment 2 was identical except that patterns were presented for 2000 ms. In experiment 3, dot patterns were replaced by solid shapes which also had one, two, three, or four axes of symmetry. Although it was found that stimuli with four axes clearly allowed superior performance to that of stimuli with one axis, results obtained with stimuli with two and three axes were almost identical and in between those obtained with one and four axes. The data thus support the suggestion that extra axes add 'goodness' to symmetrical patterns but not in a monotonic fashion.

Comments on Parks's "Prior experience of form and illusory figures: new demonstrations".

Parks presented two example figures to illustrate that (i) familiarity can enhance illusory contour effects by completing an otherwise incomplete form and (ii) such contours are enhanced when their presence can explain an otherwise familiar but incomplete form. While familiarity probably does enhance illusory contours, additional perceptual factors may be involved in Parks's demonstrations.