Latest News

We have just published a pre-print in bioRxiv on the work from my graduate student Raffles Zhu.

In human vision, the retinal input is transformed into internal representations through a series of stages. In earlier stages, the signals from a particular visual field locus are passed in parallel from one visual processing area to the next. The connections at each stage may therefore introduce “error”, where incorrect or convergent projections result in a loss of spatial precision. Psychophysical and physiological studies have implicated spatial scrambling of this sort as a cause of the visual deficits in amblyopia. Several methods to measure scrambling (both in amblyopia and in healthy vision) have been developed in recent decades. In this work, we introduce a new approach. We consider two stages of visual processing where scrambling may occur: either at the input to or the output from the simple cell stage in V1. We refer to these as “subcortical” and “cortical” scrambling respectively. We investigated the impact of these two types of scrambling on a letter identification task. A physiologically-inspired decomposition and resynthesis algorithm was used to generate letter stimuli that simulate scrambling at each of these two stages. To establish a performance benchmark, we trained separate Convolutional Neural Networks (CNNs) to perform the task with each scrambling type. Comparing CNN performance against that of eight humans with normal healthy vision, we found humans exhibited greater resilience to subcortical scrambling compared to cortical scrambling. We further investigated performance by comparing confusion matrices. Compared to a simple template matching model, we found the human strategy to be more consistent with our CNNs. We conclude: i) the human resilience for subcortical scrambling suggests this may be the stage at which a greater degree of scrambling is introduced in the visual hierarchy, and ii) humans employ flexible strategies for identifying scrambled stimuli, more sophisticated than a simple template match to the expected target.

We have just published a study in Frontiers in Neuroscience looking at how the relationship between different measures of binocular function changes with aging.

Introduction: Changes in vision that occur in normal healthy aging can be seen in fundamental measures of monocular vision. However, the nature of the changes in binocular vision with age remain unclear.

Methods: A total of 28 older (53–66 years) and 28 younger adults (20–31 years) were enrolled in this study. We performed a battery of tests to assess differences in monocular contrast thresholds and various binocular visual functions including dichoptic masking weight and strength, the binocular balance point for fused stimuli, and stereoacuity in the aging and control groups.

Results: Aging significantly increased monocular contrast thresholds (p < 0.001). Although this suggests that aging reduces the effective “input gain” to vision, we also found a significantly elevated contribution of those weaker signals to interocular suppression (p < 0.001). Consequently, there was no significant net difference in the strength of interocular suppression (p = 0.065). We did not find a significant difference of absolute balance point between the two groups (p = 0.090). Lastly, the mean stereoacuity was worse in the older group compared to the younger group (p = 0.002).

Discussion: Our findings confirm previous results showing differences in contrast sensitivity and stereoacuity with aging. Furthermore, we find a change in interocular suppression that is a possible consequence of the change in contrast sensitivity. It is suggestive of a cortical system that maintains a homeostatic balance in interocular suppression across the lifespan.

We are happy to announce Savannah Dunberry is joining the lab as an undergraduate project student in McGill’s PSYC-396 course. Her project concerns mapping sensitivity to both achromatic luminance and isoluminant colour contrast across the visual field.

We have just published a pre-print in bioRxiv in collaboration with colleagues from Wenzhou Medical University.

Within the population of humans with otherwise normal vision, there exists some proportion whose ability to perceive depth from binocular disparity is poor or absent. The prevalence of this “stereoanomaly” has been investigated in previous studies, some finding the proportion to be as small as 2%, others finding it to be as great as 30%. In this study, we set out to investigate the possible reason for the wide range of results found in these studies. We used a digital stereoacuity measurement tool that could measure performance in tasks requiring either the detection of disparity or the discrimination of the sign of disparity. The stimulus design was otherwise similar between the two tasks. In a cohort of 228 participants, we found that 98% were able to consistently perform the detection task. In contrast, only 69% consistently performed the discrimination task. The 31% of participants who had difficulty with the discrimination task could further be divided into 17% who were consistently unable to perform the task (seeming to behave at chance), and 14% who showed some ability to perform the task. We propose that the greater prevalence of stereo-anomaly is revealed when tasks require the judgement of the direction of disparity.

We have just published a study in Vision Research using a noise-masking approach to investigate stereopsis in amblyopia.

People with amblyopia demonstrate a reduced ability to judge depth using stereopsis. Our understanding of this deficit is limited, as standard clinical stereo tests may not be suited to give a quantitative account of the residual stereo ability in amblyopia. In this study we used a stereo test designed specifically for that purpose. Participants identified the location of a disparity-defined odd-one-out target within a random-dot display. We tested 29 amblyopic (3 strabismic, 17 anisometropic, 9 mixed) participants and 17 control participants. We obtained stereoacuity thresholds from 59% of our amblyopic participants. There was a factor of two difference between the median stereoacuity of our amblyopic (103 arcsec) and control (56 arcsec) groups. We used the equivalent noise method to evaluate the role of equivalent internal noise and processing efficiency in amblyopic stereopsis. Using the linear amplifier model (LAM), we determined the threshold difference was due to a greater equivalent internal noise in the amblyopic group (238 vs 135 arcsec), with no significant difference in processing efficiency. A multiple linear regression determined 56% of the stereoacuity variance within the amblyopic group was predicted by the two LAM parameters, with equivalent internal noise predicting 46% alone. Analysis of control group data aligned with our previous work, finding that trade-offs between equivalent internal noise and efficiency play a greater role. Our results allow a better understanding of what is limiting amblyopic performance in our task. We find this to be a reduced quality of disparity signals in the input to the task-specific processing.

We are happy to have Linda Wang joining us as an undergraduate project student in McGill’s COGS-444 program. Her project concerns contour integration, specifically the ability of the visual system to integrate contours formed from isoluminant variations in colour.

I was honoured to give a talk recently in the McGill University Quantitative Life Sciences seminar series. A recording is embedded below.

In visual psychophysics, we aim to characterise and explain human behaviour when performing a task involving a visual stimulus. One of the simplest abilities we can investigate is the detection of a low contrast target. For example, an experiment might involve asking a participant whether they can see a small pattern shown on a display. In general, when the contrast is very low the target will be too faint to be seen. By measuring detection performance as a function of contrast, one can find the contrast required for reliable detection of that target. It is generally the case that larger targets can be detected at lower contrast. The nature of the relationship between size and detectability has been investigated through studies of spatial summation. One central question has been whether the “summation” that occurs reflects an additive pooling of local responses to parts of the target. This would benefit performance by increasing the signal-to-noise ratio at the decision stage. Other “probability summation” models attribute the performance benefit to the participant having more opportunities to detect a larger target’s individual parts. In this seminar, I will discuss various simple models of summation and their implications for visual processing.

We have just published a study in Ophthalmic & Physiological Optics investigating how image distortions from aniseikonia impact stereoacuity.

Purpose: To simulate both lens-induced and screen-induced aniseikonia, and to assess its influence on stereopsis. Additionally, to determine if screen-based size differences could neutralise the effects of lens-induced aniseikonia.

Method: A four-circle (4-C) paradigm was developed, where one circle appears in front or behind the others because of crossed or uncrossed disparity. This stereotest was used for three investigations: (1) Comparison with the McGill modified random dot stereogram (RDS), with anisometropia introduced with +2 D spheres and cylinders, and with aniseikonia introduced with 6% overall and 6% meridional (×180, ×90) magnifiers before the right eye; (2) Comparison of lens-induced and screen-induced 6% overall and meridional magnifications and (3) Determining if lens and screen effects neutralised, by opposing 6% lens-induced magnification to the right eye with screen-inducements of either 6% left eye magnification or 6% right eye minification. A pilot study of the effect of masking versus not masking the surround was also conducted.

Results: The 4-C test gave higher stereo-thresholds than the RDS test by 0.5 ± 0.2 log units across both anisometropic and aniseikonic conditions. However, variations in power, meridian and magnification affected the two tests similarly. The pilot study indicated that surround masking improved neutralisation of screen and lens effects. With masking, lens-induced and screen-induced magnifications increased stereo-thresholds similarly. With lens and screen effects opposed, for most participants stereo-thresholds returned to baseline for overall and ×180 magnifications, but not for ×90 magnification. Only three of seven participants showed good compensation for ×90 magnification.

Conclusions: Effects of lens-induced aniseikonia on stereopsis cannot always be successfully simulated with a screen-based method. The ability to neutralise refractive aniseikonia using a computer-based method, which is the basis of digital clinical measurement, was reasonably successful for overall and ×180 meridional aniseikonia, but not very successful for ×90 aniseikonia.

We have just published a study in PLoS ONE introducing a psychophysical method for assessing the quality of processed visual images. This publication also presents the Cartesian-Separable log-Gabor in a quite useful manner.

Image processing algorithms are used to improve digital image representations in either their appearance or storage efficiency. The merit of these algorithms depends, in part, on visual perception by human observers. However, in practice, most are assessed numerically, and the perceptual metrics that do exist are criterion sensitive with several shortcomings. Here we propose an objective performance-based perceptual measure of image quality and demonstrate this by comparing the efficacy of a denoising algorithm for a variety of filters. For baseline, we measured detection thresholds for a white noise signal added to one of a pair of natural images in a two-alternative forced-choice (2AFC) paradigm where each image was selected randomly from a set of n = 308 on each trial. In a series of experimental conditions, the stimulus image pairs were passed through various configurations of a denoising algorithm. The differences in noise detection thresholds with and without denoising are objective perceptual measures of the ability of the algorithm to render noise invisible. This was a factor of two (6dB) in our experiment and consistent across a range of filter bandwidths and types. We also found that thresholds in all conditions converged on a common value of PSNR, offering support for this metric. We discuss how the 2AFC approach might be used for other algorithms including compression, deblurring and edge-detection. Finally, we provide a derivation for our Cartesian-separable log-Gabor filters, with polar parameters. For the biological vision community this has some advantages over the more typical (i) polar-separable variety and (ii) Cartesian-separable variety with Cartesian parameters.

We are pleased to announced that the lab’s basic science research on visual perception will be supported by a recently-awarded NSERC Discovery Grant “Changes in visual processing strategy: effects of context, task, and training” from 2022-2027.