AVA 2003 annual meeting at London page

We have previously shown that a template model for edge perception successfully predicts perceived blur for a wide variety of edge profiles (Georgeson MA, 2001, Journal of Vision, 1(3), 438a). This talk asks whether the model is also able to explain perceived contrast of edges. Our edge-coding model spatially differentiates the luminance profile, half-wave rectifies this first derivative, and then differentiates again to create the 'signature' of the edge. The spatial scale of the signature profile is evaluated by filtering it with a set of Gaussian derivative operators. This process finds the correlation between the signature and the operator kernel at each spatial position. These kernels can thus be considered to be templates, and the position and scale of the best-fitting template indicate the position and blur of the edge. Our finding that perceived blur decreases at low contrasts can be explained by applying a threshold to the first derivative before rectifying. For a Gaussian edge with contrast C and blur standard deviation , the response, R, of the best-fitting kernel is given by: . Hence the contrast of the edge can be estimated from response magnitude and blur: . On this model, the gradient threshold predicts that perceived contrast will decrease with increasing blur. We will show contrast-matching data that support this prediction. We will also show that the model correctly predicts the perceived contrast of more complex edge stimuli formed by adding a linear ramp to a Gaussian edge.

2. Motion opponency and flicker detection

Mark A. Georgeson, Edward J Powell §

Neurosciences Research Institute, Aston University, Birmingham B4 7ET,

§ School of Psychology, University of Birmingham, Birmingham B15 2TT

email: m.a.georgeson@aston.ac.uk

In a classic study Levinson & Sekuler (1975, Journal of Physiology, 250, 347-366) showed that a counterphase flickering grating is detectable (at contrast threshold) when one or other of its two oppositely moving components is detectable. Thus flicker appeared to be detected by independent, motion-sensitive mechanisms. Since then, much evidence has emerged in favour of motion opponency – the idea (rather like colour opponency) that detectors for opposite directions are antagonistically or subtractively coupled. We have previously reported results from a number of motion masking experiments which are well-fitted by a model that combines subtractive motion opponency and divisive contrast gain control. However, opponency and independence appear to be incompatible. We describe an extension of our earlier model that resolves this paradox by having opponent and non-opponent outputs to encode motion and flicker respectively. The extended model has a simple processing architecture and accounts well for new and old psychophysical data on motion and flicker detection, at and above threshold.

3. The extra-retinal motion aftereffect

J H Sumnall, T C A Freeman & R J Snowden (School of Psychology, Cardiff University, P O Box 901, Cardiff CF10 3YG, UK; e-mail: SumnallJH@Cardiff.ac.uk)

The motion aftereffect (MAE) has a long and venerable history in vision science. Most accounts assume MAEs arise from retinal motion signals. However, this need not be the case. For instance, moving stimuli also induce eye movements that create extra-retinal motion signals. Here we explore extra-retinal contributions to the motion aftereffect. Specifically: (1) We examine the type of eye-movement (intentional vs. reflexive) needed to induce an extra-retinal MAE. (2) We determine whether extra-retinal MAEs exhibit storage in the same way as retinal MAEs. Adaptation stimuli were large moving patterns that gave rise to nystagmus eye movements. Following 60s adaptation, a stationary test pattern was displayed in a part of the visual field not previously stimulated and MAE duration was recorded. All observers reported compelling MAEs in the direction opposite to the adapting motion. Our results suggest that: (1) Both intentional and reflexive eye-movements can induce extra-retinal MAEs. (2) Extra-retinal MAEs do not exhibit the property of storage that is associated with retinal MAEs. These findings are consistent with the idea that extra-retinal MAEs arise through a process of nystagmus suppression when observers fixate the stationary test pattern (Chaudhuri, 1991, Vision Research, 31, 1639-1645).

4. Dich-, Mon- and Bi-optic masking revisited: Summation and suppression.

(Tim S. Meese, Neurosciences Research Institute, Aston University, Birmingham B47ET, UK; Fax: +44 121 333 4220; e-mail: t.s.meese@aston.ac.uk)

Legge (1984, Vision Research, 24, 385-394) presented an influential model of binocular summation which can be described as follows. Linear contrast mechanisms for each eye (L & R) are followed by accelerating (approximately squaring) nonlinearities and are linearly summed to produce a binocular contrast signal (L2 + R2). This signal passes through a second accelerating nonlinearity and is divided by the sum of a semisaturation constant (Z) and the binocular contrast signal to give: (L2 + R2)1.2/(Z + L2 + R2). This model is broadly consistent with several empirical observations. (1) Quadratic summation of binocular contrast at detection threshold. (2) Diminished summation for binocular stimulation above threshold. (3) The shape of monocular and binocular ‘dipper functions’ and log-log ‘handle’ slopes of about 0.6. (4) The considerably steeper slope of the contrast masking function produced by dichoptic masking. (5) The absence of a ‘dip’ in the masking function for dichoptic masks. In the light of more recent work in the context of suppressive contrast gain control (e.g. Foley, 1994, Journal of the Optical Society of America A, 11, 1710-1719), I have derived an alternative architecture which incorporates inter-ocular suppression (L2.4/(Z + L2 + R2) + R2.4/(Z + L2 + R2)) and also predicts the five results above. In contrast to Legge’s model, suppression occurs at the monocular stage before binocular summation. Formally, however, the basic models differ only in their numerator. There are several reasons why the new model might be the favoured candidate. First, degradation can remove binocular summation without any loss of dichoptic masking, consistent with the amblyopic visual system. Second, it can accommodate asymmetries in dichoptic masking between amblyopic and fellow eyes that cannot be explained by simple attenuation of one pathway (Harrad & Hess, 1992, Vision Research, 32, 2135-2150). Third, it offers a natural means by which second-order dichoptic masking can be accommodated. Fourth, it offers a route for binocular rivalry.

5. From binocular input correlation to disparity tuning and ocular dominance in cortex

Li Zhaoping, Laboratory of Natural Intelligence, Department of Psychology, University College London. email: z.li@ucl.ac.uk

Stereo images are highly redundant; the left and right frames of typical scenes are very similar. Li and Atick (1994, Network, computations in neural systems 5, 157-174) have measured the binocular correlation of the input signals and hypothesized that the striate cortex is concerned with, among other things, removing binocular correlations in the inputs to achieve more efficient input coding. We can thus show the principle components in the stereo input signals and derive the most efficient coding strategies that achieve binocular decorrelation. It is shown that multiscale coding combined with a binocular decorrelation strategy leads to a rich diversity of cell types: monocular/binocular and disparity selective cells, among which one can identify cells that are tuned-zero-excitatory, near, far, and tuned inhibitory. Consequences for the formation of cortical ocular dominance columns in abnormal developmental conditions such as strabismus and monocular eye closure can also be explained (Li 1995, The neurobiology of computation. pp. 397-402. Ed. J. Bower, Kluwer Academic Publishers). The theory also provided testable predictions on the correlations between the following receptive field properties: ocular dominance, receptive field size, optimal orientation, and disparity selectivities. In particular, it predicted that striate cortical cells tuned to near horizontal orientations are more likely to be binocular than cells tuned to near vertical orientations. This prediction has been subsequently confirmed using data from cats. A psychophysical test of the prediction will also be shown.

6. Extra-retinal cues and motion-in-depth perception

Welchman AE¹, Harris JM² & Brenner E³

1. Max-Planck-Institut für biologische Kybernetik, Spemannstraße 38, 72076 Tübingen, Germany; Tel: +49 7071 601609, Fax: +49 7071 601616, email: Andrew.Welchman@tuebingen.mpg.de

2. School of Biology (Psychology), University of Newcastle-upon-Tyne, NE2 4HH, UK

3. Department of Neuroscience, Erasmus MC, PO Box 1738, 3000 DR Rotterdam, The Netherlands

In natural settings, our eyes tend to track interesting or dangerous objects moving towards us. This complicates the relationship between an object’s motion and the binocular retinal projection of its motion. To perceive the motion correctly, the brain must take account of eye movements. This could be done using retinal cues (the retinal slip of static scene structures) or extra-retinal cues (e.g. copies of motor commands). Previous studies have found that extra-retinal cues provide a very poor cue to eye rotation for objects moving in depth (Erkelens & Collewijn, Vision Research, 1985, p.583-588; Regan et al., Invest Opthal Vis Sci, 1986, p.584-597) thus suggesting that extra-retinal cues are not used to perceive motion-in-depth. Here, we re-evaluate this issue using conditions designed to isolate extra-retinal cues. We measured motion-in-depth discrimination thresholds when the eyes were moving or stationary. To get the eyes moving without the subject seeing any motion-in-depth we presented the target (a small square) in a plane that contained a large structured background and then varied the simulated depth position of the plane over time without changing retinal size. Motion-in-depth is imperceptible under these conditions (due to cue conflict) whilst the large scene promotes accurate pursuit. After bringing the eyes into motion, the background disappeared and the target continued moving; observers then indicated whether the target approached or receded. Observers were able to make fine discriminations of motion direction whether the eyes were initially stationary or moving. This suggests that, in contrast to previous reports, observers can use extra-retinal cues to make accurate motion-in-depth discriminations. Thus, extra-retinal signals may routinely contribute to our perception of objects moving towards us.

7. The utility of binocular vision for the perception of motion in a 3-D world.

Julie M. Harris, Philip J. A. Dean (School of Biology (Psychology), University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, UK; email: j.harris@ncl.ac.uk)

How does the human visual system use binocular motion signals and changing binocular disparity to see motion in depth? We have previously reported that 3-D motion trajectory perception can be highly inaccurate (Harris, 2000, Perception, 29, 83-84). When observers view an object moving towards them in the plane of their eyes, but slightly to the left or right of straight ahead, they perceive the trajectory angle as being much wider than is physically specified (up to 5 times the physical angle). That study used motions which would pass close to, or intercept with, the head (± 7 deg). Observers were asked to move a pointer to the perceived trajectory. The pointer could be moved through ± 90 deg. A possible explanation for the very large inaccuracies found, is that observers were subject to response bias. They might use a large portion of the available response range, simply because it is available. Here we repeated the experiment several times for each observer, but using 3 different stimulus ranges: ± 7 deg, ± 30 deg and ± 45 deg. If observers were subject to a range bias, we would expect a similar range of perceived trajectories for each condition, despite the physically specified trajectories being very different. In fact, we found only slight overlap between the trajectories in the 3 conditions. Observers were consistent in their responses, with wide physical trajectories typically perceived as wider than narrow physical trajectories. However, despite their responses being precise (observers responded similarly each time the same physical angle was presented), all observers were highly inaccurate, showing a large bias towards seeing wider trajectories. We believe that this misperception is real: the human visual system is very poor at using binocular disparity and binocular motion information to perceive 3-D motion trajectories.

8. Saccade planning and object structure

John M Findlay, Dorine Vergilino-Perez ¶

(Department of Psychology, University of Durham, South Road, Durham, DH1 3LE; e-mail j.m.findlay@durham.ac.uk ; ¶ Laboratoire de Psychologie Expérimentale, Université René Descartes, Paris, 71 avenue Edouard Vaillant, 92774 Boulogne Billancourt Cedex, France)

Work on visual attention has demonstrated that object-based segmentation of visual displays can influence the way attention is structured. Here we demonstrate that such influences are important in the planning of eye saccades. Subjects were asked to perform a visual search task in which eye movements were required, comparing search material that either grouped into a single visual object or segmented into several discrete visual objects. We compared saccades made when searching between objects to those made while searching within an object. The former were influenced by both saccade launch site and the object dimensions, indicative that a new planning framework was initiated, while the latter appeared to be pre-planned movements of a fixed magnitude.

ABSTRACTS - POSTERS

1. Evidence for early synthesis of multiple chromatic mechanisms in the human visual pathway

C Vakrou, P V McGraw, D Whitaker, D J McKeefry (Department of Optometry, University of Bradford, Bradford BD7 1DP, UK; e-mail: C.Vakrou@bradford.ac.uk)

Previous studies examining the effects of chromatic adaptation suggest that the cardinal post-receptoral colour mechanisms undergo a cortical transformation, resulting in the synthesis of multiple chromatic mechanisms tuned to different directions in colour space. In the present study we attempt to establish where in the visuo-cortical pathway this re-organisation takes place.

A three-blob Vernier alignment task was used, in which subjects were required to judge the horizontal position of the central Gaussian-windowed chromatic blob relative to identical reference blobs. Prior to this test phase, subjects adapted to an antisymmetric (first derivative of a Gaussian) stimulus, which was spatially coincident with the reference elements of the test stimulus. Following adaptation, a misalignment of the central element was perceived and the magnitude of this offset was established using a method of constant stimuli. All stimuli were isoluminant.

When the adapting and test stimuli contain the same chromatic content, maximum misalignment occurs. However, when the adapting and test stimuli are orthogonal in colour space little or no perceived offset occurs following adaptation. These findings were identical for non-cardinal, as well as cardinal directions, suggesting the existence of multiple chromatic mechanisms. Interestingly, the effects of chromatic adaptation decay rapidly in a manner consistent with retinal after-image formation, and exhibit little inter-ocular transfer, thus placing the locus of synthesis at a point prior to the site of binocular integration. Furthermore, a simple model based on the linear combination of after-image and test stimulus predicts the measured offsets. Taken together, these results suggest that the synthesis of multiple chromatic mechanisms takes place much earlier in the visual pathway than previously thought (Webster & Mollon, 1991, Nature 349 235-238).

Acknowledgment: P V McGraw is supported by the Wellcome Trust.

2. An investigation of visual masking using the Standing Wave of Invisibility illusion

S R Abdelaal, B T Barrett, P V McGraw (Department of Optometry, University of Bradford, Bradford, BD7 1DP; e-mail: s.r.abdelaal@bradford.ac.uk)

In the "standing wave" masking illusion a flickering central bar (target) can be rendered invisible by adjacent counter-phase flickering flanking bars (masks). Previous studies suggest that maximum masking occurs when the temporal offset of the target is approximately 100 ms before the offset of the mask (Macknik SL, Livingstone MS, 1998, Nature Neuroscience 1(2), 144-149). However, a fixed asynchrony is inconsistent with the recently advanced "object substitution" theory of visual masking. (Enns JT, Psychological Bulletin & Review , in press). Our experiment seeks to resolve this conflict by measuring masking effects for a range of mask and target durations.

The stimulus consisted of a target bar flanked by adjacent masking bars, with no spatial or temporal overlap. In any single trial, the target was preceded and followed by the masking stimuli. This sequence of mask-target-mask appeared on one side of fixation, while a reference bar, against which contrast of the target bar was judged, appeared on the opposite side. The reference and target could appear randomly on either side of fixation. The perceived contrast of the target was taken as a measure of masking strength. Additionally, we split the illusion into its forward-masking and backward-masking components.

Forward-masking was found to be independent of the separation between target and mask (range 0-0.6 deg.), and showed a modest increase with increasing mask duration for a fixed target duration. However, backward-masking effects were separation dependent and showed a non-monotonic relationship with mask duration, contrary to the prediction of the object substitution theory. Specifically, masking effects were greatest for intermediate mask durations, and varied between observers.

A fixed temporal asynchrony is not critical in determining maximal masking effects in this illusion. Rather, where masking occurs, it is the relative durations of the target and mask that determine the size of the effect in this illusion. These findings do not support the object substitution theory of visual masking.

Acknowledgment: P V McGraw is supported by the Wellcome Trust.

3. 2-D tilt and 3-D slant illusions in perception and action tasks

Mark F Bradshaw, Robert F van der Willigen, Paul B Hibbard

School of Psychology, University of Surrey, Guildford, GU2 5XH, Guildford, United Kingdom.

email: pbh2@psych.st-andrews.ac.uk

There is a well established dissociation between perception and action based primarily on neuropsychological evidence (Milner and Goodale, 1995, The Visual Brain in Action, OUP). Although equivocal, an important source of evidence from normals is that perceptual illusions may affect the systems differently (see Franz et al, 2000 Psychological Science, 11, 20-25). Here we investigate the relative effects of 2-D tilt and 3-D slant illusions in the two domains using similar "posting" tasks to those employed originally by Milner and Goodale. Subjects were required either to post a card through, or set a paddle to match the orientation of a vertical plane which was presented in two conditions: surrounded by a striped surface tilted between ±90°(2-D tilt contrast), or surrounded by a disparity-defined surface slanted or inclined in depth between ±60° (3-D depth contrast). For 2-D tilt, action and perception were equally affected by the illusion whereas in the 3-D condition they were not. Here, the illusion appeared greater in the posting than in the perceptual task. This difference was greater for slant than inclination. We conclude that although no qualitative differences exist there were quantitative differences between perception and action tasks in the 3-D condition.

4. Perceiving emotion from point-light displays of interpersonal communication behaviours.

Tanya Clarke, David Rose, Mark Bradshaw, David Field, Sarah Hampson (Department of Psychology, University of Surrey, Guildford,GU2 7XH, UK; email: t.clarke@surrey.ac.uk)

Point-light displays of the human form provide sufficient information to recognise useful properties of the actor and of the actions being represented. Although it is now well established that observers can identify properties such as gender (e.g. Pollick, Paterson, Brunderlin and Sandford, 2001, Cognition, 82(2), B51-B61) and affect (Dittrich, Troscianko, Lea and Morgan, 1996, Perception, 25(6), 727-738) from a single actor, it remains to be established how interpersonal interactions are interpreted. A database of short point-light displays made from natural body movements was collected from 5 female and 5 male actors where each male and each female acted together to produce short vignettes that expressed five different emotional states (anger, joy, sadness, romantic love and fear). Presentations of these displays in their natural dyadic form along with presentations where one of the actors was occluded (monads) and presentations where the image of this single point-light actor was reflected to create a mirror-image of the actor (reflection dyads), were shown to participants who were asked to judge the emotional content. Results for recognition of different emotions varied depending on the type of presentation viewed. It was found that for joy, viewing the natural dyads improved recognition performance over the monads, whereas for sadness the opposite was true. Anger and fear were unaffected by viewing condition. Performance for the reflection dyad presentations of love was good and was similar to that of the natural dyads acting this emotion. However, the monad ratings for this emotion were at chance levels. These results suggest that the efficiency of emotion recognition in dyads compared to monads was strongly influenced by the type of emotion portrayed. That performance for joy was enhanced in dyad presentations suggests that some cues to the emotional content of a social interaction are jointly specified by the combined spatio-temporal pattern of the two actors.

5. Liverpool v Chelsea: a pilot study of applied vision and football

Mark A. Georgeson, James Lampard

Neurosciences Research Institute, Aston University, Birmingham B4 7ET,

email: m.a.georgeson@aston.ac.uk

Football is a global business involving huge sums of money, much of which comes from television. Hence the visual presentation of football, and the visual factors that may enhance or hinder perception by viewers and by players, is of more than academic interest. Some combinations of kit colours may promote good 'reading' of the game by enabling rapid visual search and visual segregation of one team from the other, while other combinations seem to make matches confusing and almost unwatchable. A 'good' example might be Liverpool v Chelsea (all-red vs all-blue, on a green background) while a 'bad' example might be Man Utd v Southampton (red top + white shorts vs white top + red shorts). The Laws of the Game make no prescription at all, but in practice referees may check that the shirt colours are different, and the shorts are different, and then suppose that all will be well. We devised an experimental test, using simplified static images with coloured, shaded rectangles to represent the players against a textured green background. In a given image, eight players were shown in random locations without overlap, and the task was to indicate by a key-press whether 3, 4 or 5 of 'your team' (previously instructed) were present. Different combinations of red, blue and white were tested in different blocks of trials. The results strongly confirmed the hypothesis that crossed colours (e.g. red-white vs white-red) were much more difficult than same-colour strips (red-red vs white-white). Accuracy was worse for crossed-colours, and response times around 800 ms in the same-colour conditions rose to about 1500 ms for crossed-colours. This large response-time penalty (700 ms) implies that the wrong choice of strip colour combinations could seriously impede the ability of viewers, and perhaps officials, managers and players, to 'read' the spatial layout of fast-moving football matches.

webdesign by ablen