Abstract
Keynote Lectures
Measurement and Correction of Binocular Sensorimotor Disorders
Department of Psychology, Northeastern University, Boston, Massachusetts, United States
Binocular vision disorders affect around 5% of the population and may involve sensory and/or motor deficits, including amblyopia and strabismus. Existing approaches to the assessment and remediation of binocular vision disorders are often insensitive, impractical, and ignore the synergy between sensory and motor aspects of binocular function. I will describe a battery of methods that rapidly provide a comprehensive assessment of sensory and motor deficits and a novel oculomotor feedback-based technique to correct ocular alignment deviations. These approaches may identify why some patients are unresponsive to treatment and whether to continue or modify an intervention.
Where to Look? Eye Tracking in “Real World” Situations
Royal Holloway, University of London, London
As we take in our visual world, we are constantly moving our eyes around, sampling patches of high-resolution information over time, from which our visual experience is formed. One of the fundamental questions of visual perception is what leads the eyes to sample a given patch at a given time. This choice is based on a combination of the reduced quality information available from peripheral vision and the representation of the contents of the scene that has been built up over time. Many drivers of this strategy have been described, often considering a static scene displayed on a screen. In this case, stimulus onset and offset are controlled by the experimenter, whereas in real-world situations, the visual array available to sample over time depends on where we direct our heads and move our bodies. In addition, in real-world situations, eye movements take on the role of stabilizing the image as well as scanning the scene for relevant information. Mobile eye tracking allows us to explore this dual role of eye movements, but it is difficult to control the visual input. On the other hand, virtual reality allows full control and increasingly easy access to eye tracking. As these virtual environments become more widely used in a variety of applications, it is imperative as vision scientists that we explore their visual properties.
Talks
Attentional Selection Counteracts Shifts in Ocular Dominance Induced by Short-Term Monocular Deprivation
School of Psychology, University of Nottingham, Nottingham, England
Depriving one eye of visual input for several hours increases the relative perceptual contribution of this eye to binocular tasks (e.g., Lunghi et al., 2013). We have recently shown that a lack of visual input is not necessary to induce changes in ocular dominance, suggesting that higher level attentional mechanisms may play a role (Wang et al., 2019). Here, we examined (a) whether selectively attending to an eye’s stimulus enhances its dominance during binocular rivalry and (b) how attentional selection interacts with the effect of monocular deprivation on eye dominance. In five subjects with normal binocular vision, sensory eye dominance was assessed using a binocular rivalry task (horizontal and vertical grating, 1.8 cpd, 20% contrast), before and after 30 minutes of monocular deprivation using an opaque patch or a translucent lens. Before each block, participants were instructed to attend to either eye’s stimulus (tagged with a named colour) or given no attentional instructions. The dominant orientation perceived was reported on each trial and the contrast of the stimuli presented to each eye was varied to generate psychometric functions of eye dominance. We found that both opaque and translucent patching shifted eye dominance in favour of the deprived eye, but this shift can be counteracted if the nondeprived eye’s stimulus was selectively attended. These data reveal both top-down and bottom-up contributions to the modulation of sensory eye dominance.
References
Investigating the Relationship Between Eye Dominance and Excitatory/Inhibitory Balance in the Human Visual Cortex Using Combined fMRI-MRS at 7 Tesla
1Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, England
2Wellcome Centre for Integrative Neuroimaging, University of Oxford, Oxford, England
3Ecole Normale Superieur, Paris, France
4School of Health Sciences, Purdue University, West Lafayette, Indiana, United States
5Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, England
6Wellcome Centre for Integrative Neuroimaging, University of Oxford, Oxford, England
The binocular visual system has to combine images from both eyes into a singular whole to serve perception and action. Yet subtle biases in the eye that is preferred for perception and action exist, and this “eye dominance” can be used to study the neural computations behind binocular vision. We ask to which degree subtle variations in eye dominance are related to the balance of excitatory and inhibitory neurotransmitters, glutamate and gamma-Aminobutyric acid (GABA), in the human visual cortex (“Excitation/Inhibition (E/I) balance”). We combined psychophysics and brain scanning during visual stimulation at 7-Tesla field strength. Thirteen observers took part in a 1-hour psychophysical experiment to quantify eye dominance and a separate 1.5-hour combined functional magnetic resonance imaging (fMRI)-magnetic resonance spectroscopy (MRS) scan to simultaneously measure neurochemistry and hemodynamics in the primary visual cortex (V1). During MRI scanning, participants’ dominant or the nondominant eye was stimulated using alternating blocks (64 seconds) of a blank black screen followed by a high-contrast flashing checkerboard. Neurochemical and Blood Oxygen Level Dependent (BOLD)-fMRI signal changes were measured from a 2 × 2 × 2 cm3 voxel in the primary visual cortex using 7-Tesla combined fMRI-MRS. We found that the strength of eye dominance was predicted by the difference in V1 E/I balance between eyes. Stronger eye dominance correlated with a decrease in E/I balance due to an increase in GABAergic inhibition during dominant relative to nondominant eye viewing (r = −.647, p = .023). No such relationship was found when relating eye dominance to the BOLD-fMRI signal. In summary, we show that small biases in eye dominance in the normal binocular visual system are related to the balance of E/I in the primary visual cortex, and that this relationship can only be revealed during active processing of visual input.
Neural Modelling of Antisaccade Performance of Healthy Controls and Huntington’s Disease Patients
1School of Computer Science, University of Lincoln, Lincoln, England
2School of Optometry and Vision Sciences, Cardiff University, Cardiff, Wales
3Division of Psychological Medicine and Clinical Neurosciences, School of Medicine, Cardiff University Cardiff, Wales
In the antisaccade paradigm, subjects are instructed to fixate a central stimulus and then perform an eye movement directly away from a peripheral visual stimulus. The paradigm requires the parallel programming of two decision processes: the suppression of an erroneous prosaccade towards the peripheral stimulus and the initiation of a volitional antisaccade to the mirror opposite position. Although healthy controls (CTL) typically make few errors, patients suffering from Huntington’s disease (HD) make more errors and display increased and more variable latencies of error prosaccades and antisaccades. Deficits in the antisaccade performance of these patients are generally interpreted as an impaired top-down inhibitory signal that fails to suppress the erroneous responses. Neural network models with mutual inhibition implementing nonlinear accumulation of information prior to decision-making (eye movement) are presented. Two decision signals representing the volitional antisaccade from the Frontal Eye Fields and the reactive prosaccade from the Lateral Intraparietal Area are integrated in a competitive neural network of the superior colliculus. The model accurately reproduces the error rates and latency distributions of error prosaccades, antisaccades, and corrected antisaccades of CTL and HD cohorts of subjects. The model predicts that antisaccade performance of HD patients is due to a noisier rate of accumulating information, although they are less confident about their decisions than the controls. Finally, competition between the two decision processes in the superior colliculus, and not a third top-down inhibitory signal that suppresses the erroneous response, accounts for the antisaccade performance in healthy and HD subjects.
Does Vergence Explain the Taylor Illusion?
Centre for Applied Vision Research, City, University of London, London
If observers view an afterimage of their hand in complete darkness, then move their hand forwards and backwards in space, they report seeing their hand growing or shrinking. Studies of this illusion (the “Taylor Illusion”: Taylor, 1941) have concluded that vergence is either wholly (Taylor, 1941; Mon-Williams et al., 1997) or largely (Sperandio et al., 2013) responsible for this phenomenon. These studies provide key evidence for the suggestion that vergence is one of the primary cues for size constancy and therefore visual scale. To test whether vergence does in fact affect our visual perception of scale, we asked subjects to view a 3° target for 5 seconds whilst we increased their vergence from 50 cm to 25 cm. During each trial, we also increased or decreased the angular size of the stimulus and tested whether the changing vergence had any effect in biasing their judgements of whether the target was getting larger or smaller? We found a negligible effect in the wrong direction which, when analysed, most likely suggests that vergence makes no contribution to size constancy. These results have important implications for the Taylor illusion (where vergence scaling can no longer be the explanation), but also (when combined with our earlier work questioning vergence as an absolute distance cue) deeper questions about how we see absolute distance and scale.
References
Summation Within and Across Shapes in Central and Peripheral Vision
University of Plymouth, Plymouth, UK
Neurons in early visual cortex are tuned to particular features, for example, orientation and spatial frequency. However, stimuli that are behaviourally relevant are more complex (e.g., objects, faces). The visual system must contain mechanisms that combine these localised outputs. This leads to the question of how these outputs are combined? Previous studies have reported efficient signal integration across the circumference of sinusoidally modulated shapes and radial frequency (RF) patterns. The decrease in thresholds with increasing number of modulations being greater than predicted by probability summation (PS) across independent detectors was taken as evidence for efficient global pooling of contour information. However, more recent evidence suggest that such global pooling might not be necessary, and that models based on the detection of local curvature might be sufficient. We measured the strength of summation within and across RF4 shapes in central and peripheral vision (0˚–4˚). Summation modelled under Signal Detection Theory (SDT) depends on the amount of uncertainty; hence, we systematically tested the role of spatial uncertainty. Results show that RF detection thresholds increase with increasing eccentricity, but the strength of summation is independent of eccentricity and similar within and across shapes. Modelling of summation under the assumptions of SDT shows that the strength of summation is consistent with PS and is similar whether summation occurs within a single shape or across shapes, independent of eccentricity and uncertainty. This implies that the visual system does not treat single closed shapes any different from various shapes distributed across the visual field.
Visual Stress and Persistent Postural Perceptual Dizziness: Connecting Two Literatures
1School of Psychology, Cardiff University, Cardiff, Wales
2School of Psychology and Neuroscience, University of St Andrews, Scotland, UK
3Department of Audiovestibular Medicine, University Hospital of Wales, Cardiff, Wales
Images that deviate from natural scene statistics in terms of spatial frequency and orientation content can produce visual stress (also known as visual discomfort), especially for migraine sufferers. These images appear to overactivate visual cortex. Here, we connect the literature on visual discomfort with a common chronic condition presenting in neuro-otology clinics, known as persistent postural perceptual dizziness (PPPD). Sufferers experience dizziness when walking through highly cluttered environments or when watching moving stimuli. This is thought to arise from maladaptive interaction between vestibular and visual signals for balance. Here, we find that patients with PPPD, and individuals in the general population with more PPPD symptoms, report heightened visual discomfort to stationary images that deviate from natural spectra. These findings were not explained by comorbid migraine. Indeed, PPPD symptoms showed a significantly stronger relationship with visual discomfort than did migraine. We conclude that atypical visual processing—perhaps due to a visual cortex more prone to overactivation—may predispose individuals to PPPD, possibly explaining why some vestibular patients develop PPPD and some do not.
Age-Related Changes in Spatio-Chromatic Contrast Sensitivity at Mesopic and Photopic Light Levels
1Department of Psychological Sciences, University of Liverpool, Liverpool, UK
2Department of Computer Science and Technology, University of Cambridge, Cambridge, England
3School of Psychology, University of Aberdeen, Aberdeen, UK
The purpose of our study was to investigate the spatio-chromatic sensitivity of colour-normal observers as a function of age. We were particularly interested how the adapting light level affected the contrast sensitivity and whether there was a differential age-related change in sensitivity. Contrast sensitivity was measured under varied mean background luminance levels ranging from low mesopic (0.02 cd/m2) to high photopic light levels (7,000 cd/m2) for observers ranging from 21 to 74 years of age. Stimuli of different spatial frequencies and sizes, varying in three directions in the opponent colour space were displayed on an HDR display with luminance levels up to 15,000 cd/m2. Within each session, observers were fully adapted to the fixed background luminance. Our main findings are as follows: (a) for all age groups, both the achromatic and chromatic contrast sensitivity increase as a function of light level. (b) The sensitivity of the younger age-group (<40 y.o.a.) is higher than that for the older age-group roughly by 0.3 log units. To a first approximation, this difference is constant across colour directions and light levels. Only for the achromatic contrast sensitivity, the old age-group shows a relatively larger decline in sensitivity for medium to high spatial frequencies at photopic light levels. The decline in sensitivity for the older age-group is remarkably uniform across light levels and colour directions. This is somewhat surprising since the age-related yellowing of the lens reduces the contrast along the “lime-violet” direction (S/(L + M)) more than along the red-green and achromatic directions. Future models will include the effect of the ageing lens as well as pupil miosis on the contrast sensitivity functions.
Perceived Speed at Low Luminance: Lights Out for the Bayesian Observer?
School of Psychology, Cardiff University, Cardiff, Wales
Motion perception is a major testing arena for Bayesian models of sensory processing. One reason is that the prior distribution is fairly easy to justify—the world is largely stationary, so imprecise speed signals probably indicate movement that is slow. The “slow-motion” prior is thought to underlie a number of perceptual biases, such as the effects of visual contrast, eye movement, and auditory noise on perceived speed. However, one bias that seems to defeat the Bayesian framework is the increase in perceived speed at low luminance, because, as is claimed (e.g., Hammett et al., 2007), sensory precision should decline at low luminance. Here, we test that claim using moving gratings and standard two-alternative forced choice methodology. Contrary to expectations, we found speed discrimination actually improves at low luminance, in keeping with the Bayesian account. This was true for both light- and dark-adapted observers, who also showed similar increases in perceived speed at low luminance. However, further experiments showed perceived contrast depended substantially on speed at low luminance. With contrast cues removed, the discrimination differences disappeared, while the effect of luminance on perceived speed remained the same. Lights out for the Bayesian observer? A preliminary study, in which dynamic luminance noise is injected into the moving gratings, suggests Bayesian like behaviour once more. This points to a possible hybrid model, in which early differences in luminance-dependent speed transduction can be augmented by later differences Bayesian estimation, should there be significant changes in underlying noise.
Reference
Examining the “Oblique Effect” in the Tactile Domain With Mid-Air Haptic Stimuli in a Two-Interval-Forced-Choice Task
Cardiff University Brain Research Imaging Centre, Cardiff, Wales
Within the visual domain, it has been well-established that perception is better (e.g., increased motion and orientation discrimination) in horizontal and vertical planes compared to oblique planes. This phenomenon has been referred to as the “oblique effect.” This effect may also exist in the tactile domain—for example, in tactile motion discrimination on the hand. However, evidence is weak as, until recently, such studies have been methodologically limited, having to use stimulations like brush strokes or reference objects—thus adding a lot of uncontrolled variance. In this study, we used state-of-the-art mid-air tactile stimulation (UltraHaptics, Bristol, UK), to examine (a) if people can discriminate different mid-air tactile motion stimuli and (b) if the oblique effect can be found in the tactile domain. Thirty-two participants were tested on a two-interval-forced-choice task in a “horizontal,” “vertical,” and “oblique” condition. Accuracy was statistically above chance in each of the three conditions—though individual differences were high. Furthermore, we found evidence for an effect of condition, both for objective performance—showing that discrimination accuracy was lowest in the oblique condition—and for subjective confidence ratings—showing that participants were least confident about the oblique condition. Overall, our results thus agree with the existence of the oblique effect in the tactile domain and show the potential of using classic visual psychophysics tasks in tactile perception with the use of mid-air sensations.
V3a and V4 Show Sensitivity to Colour When Luminance-Channels Are Saturated
1Department of Psychology, University of York, York, UK
2York Neuroimaging Centre, University of York, York, UK
3York Biomedical Research Institute, University of York, York, UK
Ventral visual areas like hV4 are thought to play an important role in colour perception, while dorsal areas like V3a appear to be more aligned with achromatic motion processing. Here, we explored how these areas respond to chromatic and achromatic contrast modulations when achromatic responses are attenuated. We identified areas hV4 and V3A using standard retinotopic mapping in six subjects. In the scanner, we displayed a constantly present background comprised an array (100 × 100) of dynamically modulated 0.2° checks (50%L + M+S contrast, subtending 20° total) updating every 50 milliseconds. Probe gratings were additional modulations of this background that added chromatic (L-M contrast at 2.7% or S-cone contrast at 10.5%) or achromatic (L + M + S contrast at 15%) contrast in a square wave pattern (1.25 cpd or 2.5 cpd) within the central 10°. These probe gratings were phase-reversed at a rate of 1 Hz. hV4 was more responsive to colour than V3A, and hV4, but not V3A, showed a reduced response for higher spatial frequency, particularly for the S-cone stimuli. We show that “pure” colour responses in hV4, but not V3A, are robust and depend on spatial frequency in a way that maps onto human behaviour.
Posters
The Potential Off-Axis Performance and Accuracy of the PowerRef 3 for Measurements in Nystagmus
School of Optometry and Vision Sciences, Cardiff University, Cardiff, Wales
Accommodation is vital to produce a clear, focussed image for close work. In nystagmus, a continuous conjugate oscillation of the eyes, studies on accommodation are lacking. Ong et al. (1993) demonstrated normal static accommodation, apart from increased depth of focus, in nystagmats. However, in determining the dynamics of accommodation in people with nystagmus, it is unknown how accurate refractive findings will be off-axis. We therefore sought to discover the effect of off-axis positions on measured refractive error during convergence. In six presbyopes without nystagmus (age: 43–70), we used the PowerRef 3 (PlusoptiX R09), which provides continual recording of refractive error, to quantify the effect of imposing different values of defocus (+4, +3, +2, +1, −2 and −4 D) off-axis and centrally. At 1 m, measurements were obtained when participants fixated the PowerRef 3 and at four gaze positions spaced c.2° apart, on either side of the instrument. For positions to the left, readings were obtained for the right eye and vice versa. The lens-induced refractive changes did not differ from the expected change both at the centre and off-gaze positions (except for a −2 D lens at 4°; p = .037). However, these values were all significantly correlated. For gaze position up to 8°, refraction measured was not significantly different from that at the central position (F value range: 0.048–0.778; p value range: 0.624–1.000). Therefore, off- axis refraction measured using the PowerRef 3 remains reliable up to 8°, implying that we will be able to record the stability of refraction during nystagmus eye movement.
References
High-Phi and Three-Dimensional Global Motion Processing
School of Psychology, Cardiff University, Cardiff, Wales
Wexler et al. (2013) discovered a visually remarkable effect they called “high-phi.” When a rotating motion pattern (an inducer) is followed by a few frames of noise, a very fast rotational jump of the stimulus is perceived. We were interested in nature of the high-phi effect, in particular, its relationship to 3D global motion (optic flow) processing. We first established the high-phi effect works with random dots (the original demonstration using textured surfaces). Then, we tested whether the effect occurs with radial motion as well as circular motion (yes). Next, we conducted a series of experiments in which we removed a portion (wedge) of the stimuli and used a probe to test for a high-phi effect in the unstimulated region (cf., the phantom motion after effect). We tested (a) whether the effect occurs in the unstimulated wedge (yes); (b) whether, like the original high-phi effect, the perceived jump direction and magnitude changes as a function of the inducer duration (yes but); (c) whether the effect stores, that is, still occurs if there is a delay between the inducer and the probe presentation (yes, over the period we have tested); (d) whether the effect follows optic flow geometry, whether the magnitude of the effect increases with distance from the focus of expansion (yes but). We are currently exploring whether we can generate a stereo motion-in-depth version of the effect, and whether the magnitude of the effect decreases with the size of the stimuli (cf., surround suppression).
References
Assessing the Status of Visual Cortex in Macular Disease
1Department of Psychology, University of York, York, UK
2York Teaching Hospital, York, UK
3York Neuroimaging Centre, York, UK
Macular degeneration (MD) refers to a collection of disorders causing a progressive loss of central vision. The current focus of ophthalmological research concerns physiological changes within the eye, aiming to treat eye disease and prevent further loss of vision. However, far fewer studies assess the consequences of ocular disease on visual cortex. Cross-sectional magnetic resonance imaging (MRI) studies have revealed structural changes in visual cortex in MD, but the rate of change is unknown. To explore the rate of change in cortical thickness within the occipital pole (OP) and calcarine sulcus (CS), we acquired structural MRI data on 11 patients with different forms of MD and 18 age-matched controls over multiple time points in a ∼24 month period. Data were analysed using a linear mixed-effects model. Preliminary data show a significant reduction in grey matter in patients in the OP, and an accelerated rate of decline compared to controls. Whilst patients did have a significantly thinner CS, the rate of change did not differ between groups. Understanding the time course of changes may prove important for visual restoration; the success of interventions aiming to restore functionally useful vision will be limited if visual cortex is no longer viable.
Revisiting the Positive Impact of Visual Exposure to Nature: A Case of Aesthetic Preference?
School of Psychological Science, University of Bristol, Bristol, UK
Exposure to nature, even only as photographic images, impacts positively on physical and mental health. In line with this claim, we recently reported that walking towards images of urban as compared to nature scenes was more effortful, as indicated by a decrease in gait velocity and step length (Joyce & Leonards, 2017). Yet, what causes these gait changes: differences in image statistics, semantic associations related to different image types, or scene aesthetics? Here, we present two studies in which participants walked towards images of natural and urban scenes matched beforehand for their aesthetic properties by an independent observer sample (n = 300). Gait parameters were recorded with three-dimensional-motion capture and analysed with multilevel modelling. In the first study (n = 22), aesthetics but not image category was predictive of gait changes (step length). In the second study (n = 22), in which participants additionally rated each image for visual discomfort while they walked, visual discomfort and its interaction with aesthetics, but not image category, predicted gait changes (step length, velocity). These data suggest that rather than environment type per se higher perceptual constructs such as visual aesthetics are at the core of the often described nature benefits.
References
Perceptual Inference With Structured Representations
1Department of Engineering Mathematics, University of Bristol, Bristol, UK
2School of Computer Science and Informatics, Cardiff University, Cardiff, Wales
3School of Psychology, Cardiff University, Cardiff, Wales
4Department of Neuroimaging, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London
We provide an account of perceptual inference framed as an inverse problem whereby the brain must infer the causes of visual stimuli. We assume that (a) percepts are driven by internal models of the external world, (b) that the brain generates sensory predictions under these models to confirm hypotheses about the state of the world, and (c) That these models may be evidenced by observation and induction. We explore these assumptions with a novel form of generative model and a simple perceptual categorization task. The “forward” component of this model generates visual predictions from a neural network tailored to parsimoniously represent distinct factors of variation in visual stimuli. We draw a parallel here to neurons encoding beliefs (and expectations) via top-down signals in the brain and their reliance upon statistical regularities within the real world. We explore how the objective function affects the ability of the model to make inferences on the causal structure of the world and how this may occur in the brain. The “inverse” or “perceptual” components of our model are formulated as differential equations that subsume the neural network. Perceptual inference is simulated by inverting these models with dynamic expectation maximization (Friston et al., 2008), a model inversion scheme that has been proposed as a theory of neurobiological activity that incorporates elements of planning regions (e.g., prefrontal cortex and basal ganglia) interacting with sensory neocortex.
Funding: This work was funded by the EPSRC grant EP/R030855/1.
References
The Short-Term Effects of Optokinetic Stimulation on Markers of Visual Dependence: A Theory of Visual Vertigo Manifestation
1School of Psychology, Cardiff University, Cardiff, Wales
2Department of Audiovestibular Medicine, University Hospital of Wales, Cardiff, Wales
Visual vertigo is a heterogenous syndrome reported in patients with a history of vestibular compromise (Bronstein, 1995). This is thought develop due to an adaptive reweighting of sensory cues for postural control that up-weights vision when vestibular firing is erroneous. In some, this is thought to become maladaptive after recovery of the vestibular system resulting in a maintained dependence on vision (Staab et al., 2017). Rehabilitation focuses on desensitisation: exposure to visual motion to reduce reliance on vision as a balance cue. Following this logic, exposure to visual flow ought to produce measurable change in the visual contribution to balance or sense of vertical in any viewer. An exploratory experiment exposed participants to either no visual motion (n = 23), recordings of symptom provoking situations (n = 23), playing a racing video game (n = 23), or passively viewing a recording of the game (n = 23) for 15 minutes. We found no changes in the setting of the visual vertical (rod and frame task), but there was a suggestion of nonsignificant decreases in visually induced postural sway. A second experiment was conducted to test this possibility, increasing the duration of the exposure and the number of measurements. Preliminary findings and their clinical relevance will be discussed.
References
Experimental Aesthetics in Virtual Reality: Mondrians, Unicorns, and Squared Circles
Department of Psychology, Royal Holloway, University of London, London
Experimental aesthetics as a research field mainly investigates the experience of art empirically and has strong links to visual psychophysics. The recent developments in virtual reality (VR) allow for the simulation of a wide range of experiences. The last few years have shown a drastic rise of entertainment-related use cases such as games and media consumption, especially using consumer-grade headsets. However, the research potential of VR in vision science is still somewhat overlooked. VR also opens up methodological novelties in empirical aesthetics research, especially by utilizing eye, head, and position tracking. For example, we previously demonstrated that gallery visitors showed similar viewing patterns for Mondrian’s room design in real and virtual environments as indexed by gaze data from mobile and VR eye-trackers. In addition, we also showed that a virtual art-gallery consisting of “physical” and “digital” artworks was an ecologically valid environment for an experiment on aesthetic judgments. Currently, we are exploring other potential directions to generate immersive experiments, such as using digital avatars in gallery spaces as a social component of the aesthetic experience. Arguably, the ability to generate and parametrically modify immersive, virtual objects and environments seems to be a valid methodological direction forward allowing close to real life experiences that remain fully under experimental control.
Salience of Spatiochromatic Patterns
University of Aberdeen, Aberdeen, UK
We investigated the dependence of perceived contrast on cone-opponent stimulus content and its spatial distribution in a larger sample of observers. Participants matched a comparison patch to a white standard of fixed contrast. The first experiment determined the point of iso-salience for gratings, Gabors and Gaussians along cardinal directions in cone-opponent colour space for two-alternative forced choice (2AFC) and adjustment tasks to determine the overall effect of different spatial patterns and tasks on perceived salience. No difference was found between adjustment and 2AFC tasks. No asymmetries were found for gratings and Gabors defined by either reddish-greenish or bluish-yellowish mechanism. However, Gaussians were found to require less contrast to achieve equal perceived salience with a standard compared to gratings and Gabors irrespective of colour. Also, bluish Gaussians were found to have higher salience than yellowish Gaussians at equal levels of contrast. We conducted a second experiment to determine how perceived contrast of bluish and yellowish stimuli was affected by spatial distribution of contrast within uniform and Gaussian-like patches. The S-(L + M) asymmetry decreased as the patch’s spatial distribution of contrast changed from being more smoothly distributed to a uniform patch. In conclusion, perceived contrast (and therefore salience) of stimulus patches varies depending on both cone-opponent content and its spatial distribution within experimental stimuli. This may be due to low-level influences caused by basic physiological differences between single-opponent and double-opponent neurons.
Using a Multiprimary High Dynamic Range Display to Investigate Melanopsin-Mediated Visual Function
1Department of Experimental Psychology, University of Oxford, Oxford, England
2Department of Computer Science and Technology, University of Cambridge, Cambridge, England
In addition to being intrinsically photosensitive themselves, the melanopsin containing intrinsically photosensitive Retinal Ganglion Cells also receive input from rod and cone pathways, making it difficult to assess the contributions of melanopsin to vision in humans in vivo (Dacey et al., 2005). Display systems with at least four controllable primaries enable melanopsin function to be studied using melanopsin-isolating stimuli produced via the method of silent substitution (Spitschan & Woelders, 2018). We present a novel multiprimary high dynamic range (MPHDR) display that is suitable for such experiments and discuss how to use it to study the roles of melanopsin in perception and pupil control. Our display is the first system, to our knowledge, that allows the delivery of spatially controllable, high dynamic range, multiple primary stimuli. The MPHDR display can reach a maximum luminance of 3,200 cd/m2, has a colour gamut 46% wider than sRGB displays, and uses a five independent parameter system to control six effective primaries. Using the MPHDR display we can produce melanopsin stimuli with an available melanopsin contrast of 23% at 2,000 cd/m2 up to a maximum contrast of 117% at 75 cd/m2. We discuss using our display to further investigate the influence of melanopsin on pupil control. The display allows us to probe the influence of melanopsin on pupil control over a range of luminance levels from 75 cd/m2 to 2,000 cd/m2, something that has been previously inaccessible.
References
Changes in Retinal Structure Associated With Schizotypal Traits in Healthy Individuals
1Ulverscroft Eye Unit, University of Leicester, Leicester, UK
2Department of Health Sciences, University of Leicester, Leicester, UK
Diagnosis and disease monitoring for schizophrenia is based on symptoms alone as there are no clinically useful biomarkers. Changes in the retina have been documented using optical coherence tomography (OCT) including changes in the macular region. Confounders such as psychotropic medication and comorbid diseases may underlie these differences. Schizotypal traits evident in a healthy population provide an alternative approach to delineate between disease aetiology and confounders. OCT images were acquired from 34 healthy adults, of whom 14 were first-degree relatives of individuals with schizophrenia. Images were manually segmented to derive foveal and parafoveal measurements. In addition, all participants were assessed for schizotypal traits using the Oxford-Liverpool Inventory of Feelings and Experiences (O-LIFE) score. No significant differences were found between relatives and controls for OCT parameters. Participants with high unusual experience O-LIFE scores had thinner temporal retinal nerve fibre layer (adjusted p = .012, p = .006) and thinner foveal photoreceptor complex (adjusted p = .149, p = .049) measurements. Nasal parafoveal retinal thickness was increased in higher impulsive nonconformity scores (adjusted p = .069, p = .023). Sublayer analysis revealed significant increase in cone outer segment tips and the retinal pigment epithelium layer in individuals with higher scores (p = .034). Subtle retinal differences exist in individuals with high schizotypal traits, especially in the temporal retinal nerve fibre layer which contain the papillomacular bundles. Overall, this is consistent with published findings in individuals with schizophrenia. These data indicate that foveal changes may be a marker of disease aetiology in schizophrenia rather than being caused by psychotropic medications or comorbid diseases.
Sensitivity to Global Form in the Presence of Noise Is Stimuli Dependent
1School of Health Professions, University of Plymouth, Plymouth, UK
2School of Life Sciences, Glasgow Caledonian University, Scotland, UK
Glass patterns composed of varying elements or line segments have been employed previously to investigate global form processing. However, whether or not these different stimuli target similar local or global mechanisms is not known. A total of five visually normal participants discriminated the orientation of glass patterns composed of different component elements (two elements, dipole; three elements, tripole; four elements, tetrapole) and line segments (Line 2 and Line 4 equivalent to dipole and tetrapole) in the presence of various noise levels. At no noise, the thresholds for Line 4 (
The Importance of Contrast for Camouflage
Department of Experimental Psychology, University of Bristol, Bristol, UK
Avoiding detection is important in regards to both predating and survival. Camouflage is an often-used evolutionary strategy and can be achieved in multiple ways, for example, via background matching or by drawing attention away from one’s outline. Disruptive colouration is of the second type and is achieved by high contrast markings. Previous research has explored the comparative effectiveness of these two strategies with mixed results. Consider Stevens et al. (2008), Fraser et al. (2007), and Cuthill et al. (2005). These studies used different levels of contrasts for their disruptively camouflaged targets, which we hypothesised being the reason for the mixed results. We explored the importance of contrast for the effectiveness of disruptive camouflage on human participants. Our results showed that for targets with disruptive colouration search-time decreased as a function of increased contrast. This shows the importance of contrast level for disruptive camouflage and may explain the different results of previous studies. A further novel result was on visual search patterns. As the level of contrast increased, participants more frequently identified the target without fixating on it. When trials without target fixations were removed, there was no difference in the percentage of fixations between the different levels of contrast. This shows the importance of the peripheral vision in the detection of salient targets, a visual search mechanism previously not given due attention.
References
Predicting Time to Contact Across the Visual Image
1School of Computer Science and Informatics, Cardiff University, Cardiff, Wales
2School of Psychology, Cardiff University, Cardiff, Wales; 3Centre for Neuroimaging Science, Institute of Psychiatry, Psychology and Neuroscience, King's College London, London
Estimation of Time to Contact (TTC) for a single approaching object is relatively straightforward. TTC can be estimated on the basis of changes in distance, changes in optical size, and changes in binocular disparity. Estimation of TTC across the whole visual field is more complicated: objects or features move within the visual field and need to be tracked to estimate TTC. We outline a potential method for estimating TTC across the whole visual field. We start with a depth map for time t1, a depth map for time t2, plus independent information about how the viewpoint has moved during the interval delta t, the time that has passed between t1 and t2. Given the map at t1 and information about viewpoint movement we then calculate a “warp field,” effectively a prediction of the optic flow that would arise between t1 and t2. Using the warp field, we match up points (in our implementation, pixels) between the two depth maps. For each point at time t2, we calculate the change in distance, delta d, that is, dt2−dt1. TTC for each point in the image at t 2 as then given as dt2/delta d × 1/delta t. The result is a TTC image. We show the results of simulations, identify limitations of such an approach (and consequently the signature of use of such an approach by the human brain), and describe alternative implementations.
Funding: This work was supported by EPSRC grant EP/R030855/1 (to D. M., S. K. R., and R. J. M.).
The Geometry of High-Level Colour Space
Department of Psychological Sciences, University of Bristol, Bristol, UK
The optimal representation of a signal is determined by the task and the dominant noise. In colour, most work has concentrated on low-level tasks, where the predominate noise is photoreceptor based (Vorobyev & Osorio, 1998). For high-level representations, what matters is not the colour, but what that colour informs the user about the world, and the dominant noise is due to failure of colour memory (Baddeley & Attewell, 2009). Here we estimated the properties of this high-level representation by testing colour memory across hue and saturation in CIE1931 colour space. We identify “basins of attraction” in this space, corresponding to Berlin & Kay’s basic colour terms (Berlin & Kay, 1969). We propose that these biases are due to a nonuniform prior over colours. We compare the predictions of two such priors: an empirical prior, based on the observed distribution of colours in the world, and an optimal prior due to distortions to the geometry induced by the task. To identify the form of this prior, we are training a deep network to identify objects purely based on colour. We are then using the Jensen–Shannon divergence to measure how changes in colour were reflected by changes in object classification. Comparisons between the empirical data and models of these nonuniform priors over colour will be presented.
References
Visual Merits of Individually Tailored Contact Lenses in Presence of Light Scattering
1ISSP University of Latvia, Kengaraga, Riga
2Faculty of Physics, Mathematics and Optometry, University of Latvia, Kengaraga, Riga
3Rizzieri Optometristi, Padova, Italia
We measured and calculated: (a) human eye aberrations without and with contact lenses (CLs) eliminating higher order aberrations; (b) visual resolution and contrast sensitivity in conditions when additional scattering is induced by special filters; and calculated point spread function (PSF) and other retinal visual merits. Aberration errors of patient eyes were measured by Keratron®Onda, or studying scattering effects caused by PRO-MIST 1–5 filters in modified model eye by Hartman–Shack sensor of MultiSpot-2500. Wide angle PSF caused by scattered light was detected using photocamera Olympus photodetector with pixel pitch of 4 µm and 14-bite resolution at 405, 532, and 670 nm directly on detector or projected with lens. Applying of adapted custom-made CL for patients without any symptoms of eye cataract statically justified improvement of psychophysical perception characteristics comparing to conditions of use of conventional goggles. Sets of Zernikes measured in patients and corresponding appearance of PSF still overstep limits of visus Vs = 20/20 region. Presence of light scattering (experimental modeling using thin planeparallel PRO-MIST filters in front of eyes was similar to authors: Ginis et al., 2014) does not reveal strong or remarkable shifts of Hartman-Shack graph maxima centers when measurements performed at 860 nm. However, scattering (induced by PRO-MIST filters strongly dependent in visible) builds a pedestal stretching up to ±5°. This light intensity pedestal reduces PSF peak-to-valley metrics that leads to decrease of vision contrast sensitivity and acuity. Impact of CL rotational and lateral shifts due to blinking is discussed. Studies done with support of PostDoc.No.1.1.1.2/16/I/001.
References
Sensitivity of Gamma Oscillations to the Anisotropy of Visual Stimuli Suggests a Link to Divisive Normalisation
Cardiff University Brain Research Imaging Centre, Cardiff, Wales
A strong oscillatory response at gamma frequencies (30–70 Hz) can be induced in visual cortex by luminance-defined gratings but is typically much weaker (or absent) to natural stimuli. We hypothesised that gamma oscillations might be linked to inhibitory processes involved in the divisive normalisation of neuronal responses and that the strength of the gamma response might therefore reflect the nonlinear pooling of stimulus energy across orientation channels. We tested this by presenting stimuli with varying levels of anisotropy to eight human participants and measuring their gamma response using Magnetoencephalography (MEG). Visual stimuli were created by filtering white noise images with frequency-domain filters that had a Gaussian spread of amplitude across orientation. We used six different filters with orientation bandwidths of 3.75°, 7.5°, 15°, 30°, 45°, and ∞°, respectively (the latter corresponding to uniform distribution of energy across all orientations). Participants passively viewed 80 trials of each of these six conditions. We found that the amplitude of gamma oscillations in visual cortex increased (and their frequency decreased) with increasing stimulus anisotropy. In line with our normalisation hypothesis, the tuning of the response was well fit by a mathematical model in which gamma amplitude reflects nonlinear summation of contrast energy across orientations. Consistent with the idea that more anisotropic stimuli induced greater inhibition of the cortical response, we also found that the evoked and high-frequency gamma responses were both suppressed with increasing stimulus anisotropy.
The Relationship Between Cybersickness, Vection, and Postural Instability Induced by Optimized Fraser Wilcox Illusions in Virtual Reality
1School of Psychology, University of Lincoln, Lincoln, England
2School of Computer Science, University of Lincoln, Lincoln, England
3School of Media, Culture and Society, University of the West of Scotland, Paisley, UK
Cybersickness is often experienced when viewing virtual environments through head-mounted displays. Motion-in-depth compared to lateral motion is believed to result in increased visual discomfort. This discomfort is related to the accommodation–vergence conflict and the conflict between the visual and vestibular systems (e.g., Li et al., 2014). Dominance of visual information can lead to postural instability, the experience of vection (illusory self-motion), and consequently cybersickness. In four experiments (E1–E4), we aimed to understand the relationship between these concepts by presenting rotating and expanding optimized Fraser–Wilcox illusions (E1–E2: artworks by Kitaoka, E3–E4: controlled stimuli) (Kitaoka, 2006) in VR, allowing us to compare motion-in-depth and lateral motion while minimizing accommodation–vergence conflict. Results showed that rotating motion illusions are generally perceived as stronger than expanding motion illusions (E1), which complicates interpreting the effect of motion direction on experienced discomfort. In order to match both illusions for illusion strength, the spatial frequencies were varied (E3 and E4). Contradictory to expectations, discomfort seems to be related to strength rather than direction of the illusion, with stronger illusions causing more discomfort (E4). Vection magnitude is affected by illusion strength and experienced discomfort, particularly the amount of dizziness observers experience while viewing the illusions (E2). Observers that experienced more dizziness also showed more head movements in the anterior–posterior direction. Both postural stability and vection are affected by the amount of dizziness an observer experiences while viewing an illusion, indicating that dizziness is the component of cybersickness that mainly relates to postural instability and vection.
References
Visual Vertigo Is a Spectrum in the General Population
1School of Psychology, Cardiff University, Cardiff, Wales
2Department of Audiovestibular Medicine, University Hospital of Wales, Cardiff, UK
Visual vertigo—or visually induced dizziness—is part of a common chronic condition called persistent postural perceptual dizziness (PPPD). It is generally thought to be a maladaptive long-term consequence of visual compensatory changes elicited by acute vestibular insults. Here, we find high levels of PPPD symptoms in healthy populations, suggesting instead that PPPD is a spectrum the pre-exists in the population and may predispose some vestibular patients to more severe and longer lasting symptoms, rather than arising as a consequence of vestibular insult. We collected responses to two common clinical questionnaires, Visual Vertigo Analogue Scale and Social Communication Questionnaire (VVAS and SCQ) in three healthy cohorts: community public health research volunteers (N = 2011 for VVAS, N = 1616 for SCQ), paid online participants (N = 209 for VVAS, N = 139 for SCQ), and students (N = 202 for VVAS only). We compared scores to a cohort of patients diagnosed with PPPD (N = 25) and found that around 10%, 8%, and 20%, respectively, of the three healthy cohorts scored above the 25th percentile patient score on one PPPD measure (VVAS) and 50% and 54% scored above the 25th percentile patient score on the other measure (SCQ). Scores correlated negatively with age, counter to expectation if PPPD arises following unknown insults that have rising cumulative probability with age. Scores were associated with migraine in two populations, but leaving the majority of variance unexplained, suggesting that migraine is not the major factor underlying the spectrum of PPPD symptoms in the general population.
The Relationships Between Multisensory Processing, Dizziness, and Anxiety
1School of Psychology, Cardiff University, Cardiff, Wales
2Department of Audiovestibular Medicine, University Hospital of Wales, Cardiff, UK
Persistent postural perceptual dizziness (PPPD) is a common chronic condition presenting in neuro-otology clinics. Its cause is generally assumed to lie mainly in the interactions between the three sensory systems underlying balance: vestibular, visual, and somatosensory. Here, we find that PPPD symptoms are also associated with self-reported oversensitivity in other sensory domains: touch, hearing, smell, and taste. We use structural equation modelling to unpack this relationship and its additional associations with anxiety and migraine—which are both long known but underspecified correlates of dizziness. We find that anxiety partly, but not wholly, mediates the association between multisensory sensitivity and PPPD symptoms. Most of the path coefficients and mediation effects in our model were unchanged between participants with and without migraine. Our findings support the idea that PPPD is a complex, neurological condition that includes both psychogenic and perceptual factors and thus may suggest that some brains are predisposed to generalised cross-modal sensory-overload, and this in turn may makes them more prone to severe PPPD should a vestibular insult occur.
Predicting Out Head Movements
1School of Computer Science and Informatics, Cardiff University, Cardiff, Wales
2School of Psychology, Cardiff University, Cardiff, Wales
3Centre for Neuroimaging Science, Institute of Psychiatry, Psychology and Neuroscience, King's College London, London
When an observer walks the head bounces (moves vertically), sways (moves laterally), bobs (changes speed in the forward/backward axis), and rolls (rotates about the forward/backward axis). These movements are typically ignored in computational and theoretical models of perception (and action). The result is that existing models are either incomplete or invalid. Head movements complicate the estimation of time-to-contact, and so on. How might the brain deal with bounce, sway, bob, and roll? One solution is to “predict it out,” to anticipate head movement and then compensate for it. We investigated the viability of predicting head movements. We collected head movement data from natural human walking and used an Long Short Term Memory (LSTM)-based time-series prediction network to learn how the head moves. The network attempted to predict movement during the upcoming 0.5 seconds based on the preceding 2 seconds of movement. To provide an estimate of confidence on the predicted head movement, we built a second network to predict the prediction error of the first. We demonstrate the implementation of the system, show how well it works (which provides a potential upper bound on the performance of the brain), and describe its limitations and strengths.
Funding: This work was supported by EPSRC grant EP/R030855/1 (to D. M., S. K. R., and R. J. M.).
Perceptual Stability Through Prediction: Computational Simulations
1School of Psychology, Cardiff University, Cardiff, Wales
2School of Computer Science and Informatics, Cardiff University, Cardiff, Wales
3Centre for Neuroimaging Science, Institute of Psychiatry, Psychology and Neuroscience, King's College London, London
How does the brain know that changes in the retinal image resulted from translation of the eye, rather than changes within the world? And how during translations of the eye, does the brain maintain a stable and persisting representation of the world on which attentional and other processes can operate? If the brain could predict how self-movement will change the retinal image, this would provide a solution to both problems. We have been exploring, via computer simulation, an image prediction mechanism that could be implemented in the human brain. In our model, the current image is encoded using an array of dynamic memory units that hold their contents for ∼100 milliseconds. Each memory unit has a receptive field, and the units are topographically organised. Using a depth map, plus information about how the eye is moving through space, a prediction is generated of the upcoming optic flow field. For each memory unit, we extract the predicted optic flow vector at the location mapped by its receptive field and shift the receptive field to the location indicated by the flow vector. Once the receptive fields have been remapped, the array of memory units embodies a prediction of the future retinal image. Each memory unit has a mechanism that allows it to compare the stored contents to the current contents of the receptive field. We show how well the predictive mechanism works and explain how it could be implemented in the cortical motion processing pathway.
Funding: This work was supported by EPSRC grant EP/R030855/1 (to D. M., S. K. R., and R. J. M.).
Distance Estimation From Egocentric Point of View: A Comparison Between Natural and Linear Perspective
Fovolab, School of Art and Design, Cardiff Metropolitan University, Cardiff, UK, Cardiff, UK
Humans naturally see the world from an egocentric viewpoint, but important questions remain about how best to represent this point of view. Linear perspective (LP) is a pictorial technique developed by artists during the Italian Renaissance and is still the standard method for rendering three-dimensional space geometry on two-dimensional images. However, our previous research suggests that images created using linear perspective do not accurately represent visual space, particularly under normal viewing conditions and for wide fields of view (FOV), where it can appear to distort object size and shape (Burleigh et al., 2018). We hypothesised that people would prefer and be better at estimating distances in natural rendering (NR), a novel projective geometry developed by the authors based on artistic depictions of the natural structure of human vision. To test this hypothesis, we conducted two experiments comparing LP and NR images. The results showed that participants (a) preferred NR images compared to the LP version, (b) choose wider FOV settings if playing in NR (M = 142°) compared to LP (M = 115°), and (c) were 4.9 times more likely to estimate a target object’s correct distance and 5.12 times more likely to give an accurate distance estimation if looking at NR compared to the LP. We conclude that NR approximates the perceived structure of the egocentric viewpoint more closely than LP, providing a more immersive and natural experience of virtual environments, and we will discuss the implications of these findings for our understanding of visual space.
References
Learning and Adaptation: No Habituation of the Dynamic Motion Aftereffect Measured With Grating Stimuli
School of Psychology, University of Nottingham, Nottingham, UK
Sensory Adaptation and Perceptual Learning are well-studied examples of short-term and long-term visual plasticity, yet relatively little is known about how these mechanisms interact. Recent work found that repeated sessions of motion adaptation leads to a direction-selective attenuation of the motion aftereffect (MAE). Across a series of days, Dong et al. (2016) and Dong & Bao (2019) adapted participants to moving random dot fields and measured the dot motion coherence required to null the resulting MAE. They found a progressive reduction in nulling percentage over time, but no concomitant change in coherence detection thresholds. Here, we tested whether this finding generalises to a comparable design using periodic grating stimuli. Participants adapted to leftward or rightward drifting Gabor patches (1 c/deg, 5 Hz, 30-second initial adaptation, 5-second top-up) then judged the direction of a compound Gabor test pattern comprising variable contrast components drifting in opposing directions. After 9 days of training, we observed no systematic change in the contrast-ratio required to null the motion aftereffect. There was, however, a statistically significant reduction in thresholds for discriminating changes in the contrast ratio of the test pattern. This dissociation between learning-induced changes in perceptual sensitivity, but not bias, is completely the opposite of previous work (Dong et al., 2016; Dong & Bao, 2019), demonstrating that habituation of visual adaptation is not a general characteristic of motion mechanisms.
References
Foveal Surround Suppression of Contrast Perception Is Unchanged in Older Adults
Department of Psychology, University of York, York, UK
The perception of a central stimulus can be inhibited by the presence of a surrounding stimulus: an effect known as surround suppression. This effect can be observed in luminance contrast perception and motion perception and is strongest when the spatial properties of the centre and surround are matched. Spatial suppression of motion discrimination has been shown to weaken with age (Betts et al., 2005) while contrast suppression has been observed to strengthen with age under foveal presentation (Karas & McKendrick, 2009). While these results are based in different metrics, they both measure a form of spatial suppression—meaning they are at odds. What could be the explanation? It is possible that the latter work is measuring overlay masking, a separate form of precortical orientation invariant contrast inhibition (Petrov et al., 2005). We sought to replicate the results of Karas and McKendrick (2009) using stimuli designed to isolate surround suppression from overlay masking. In a sample of younger (<30 years) and older (>60 years) adults, we measured the strength and spatial tuning of foveal surround suppression. Results support the presence of spatially tuned suppression but do not demonstrate an effect of age-group. We propose that the results of existing work may be confounded by overlay masking or a form of temporal suppression.
References
How Long Does It Take for Mobile Eye-Tracker Calibration to Drift?
1School of Psychology, University of Leeds, UK
2Cognitive Science, Department of Digital Humanities, University of Helsinki, Finland
Mobile eye trackers have a number of advantages over traditional chin rest or remote eye trackers (e.g., light, portable, and allowing natural head movements); however, there are some issues. For example, fixing a mobile eye tracker to the head can be difficult, and there may well be shifts in position as the participant moves, which can introduce measurement error over time (calibration drift; Kassner et al., 2014). Manufacturers of mobile eye trackers will often provide accuracy estimates in their documentation which come from ideal calibration circumstances: for example, Pupil Labs state that their core head-mounted eye tracker is accurate to 0.6° (https://pupil-labs.com/products/core/tech-specs). This study aimed to investigate how this accuracy compares to that achieved in our laboratory. Participants completed 4 × 10-minute simulated driving tasks while having their eye movements tracked with a Pupil Labs (version 1.2.7) head-mounted mobile eye tracker. The eye tracker was calibrated at the beginning and end of each block across a 5 × 7 target array projected onto a large field of view screen. Our findings suggest that after 2 minutes of testing, median error >2.5° and after 60 minutes most median error >5°. These error estimates far exceed those stated by the manufacturers and indicate that mobile eye trackers need to be calibrated regularly (i.e., every 1–2 minutes) for gaze location estimates to approach the accuracy stated by the manufacturers.
References
Effect of Cortical Magnification of Target and Flanker Size on Radial Anisotropy in Visual Crowding
1School of Optometry and Vision Sciences, Cardiff University, Cardiff, Wales
2Cardiff University Brain Research Imaging Centre, Cardiff University, Cardiff, Wales
3School of Psychology, Cardiff University, Cardiff, Wales
4UCL Experimental Psychology
Crowding is the deleterious effect of clutter on object recognition and a cardinal limit in peripheral perception. Radial anisotropy is a characteristic feature whereby flanking elements located between fixation and the target interfere with recognition at greater distances than more distal flankers, if otherwise equal in characteristics such as size. Pelli (2008) inferred mathematically that a fixed critical spacing in the cortex is responsible for unimpaired object recognition, which would produce radial anisotropy when translated to visual space due to differences in cortical magnification either side of the target. We propose that maintaining constant flanker size in cortical space may also partly account for radial anisotropy. Crowding zones were determined in four young, healthy participants with Landolt Cs; once with target and flankers of the same size, and secondly with radial flankers scaled according to cortical magnification determined from retinotopic maps of healthy observers in a separate study. Participants reported the orientation of a target at 10
References
Effects of Microsaccades on Global Processing of Shape
School of Psychology, University of Nottingham, Nottingham, UK
Human sensitivity to subtle changes in visual shape is often attributed to the existence of global pooling mechanisms that integrate local form information across space. While global pooling is typically demonstrated under conditions of steady fixation, other work suggests that prolonged fixation leads to a collapse of global structure. Here, we ask whether small ocular movements during periods of fixation affect global processing of radial frequency patterns—closed contours created by sinusoidally modulating the radius of a circle. Observers discriminated the shapes of two peripheral patterns while fixational eye movements were recorded binocularly at 500 Hz. Global processing was assessed from the rapid fall in discrimination thresholds as cycles of modulation increased, whereas changes in mean threshold was used as an index of overall performance. Microsaccades were detected using a velocity-based algorithm, allowing trials to be sorted according to the relative timing of stimulus and microsaccade onset. Results indicate that global processing was unaffected by the occurrence of microsaccades. However, task performance varied depending on the temporal occurrence of these small eye-movements relative to stimulus onset. Performance decreased when microsaccades occurred close to stimulus onset and increased when they occurred after stimulus onset. These findings suggest microsaccades have dissociable effects on global shape discrimination.
Investigating the Role of Extra-Retinal Signals in Saccadic Perception
Cardiff University Brain Research Imaging Centre, Cardiff, Wales
Compensating for the effect of eye-movements on the retinal image is essential to stable visual perception. Knowing whether an object in the environment has moved can involve both retinal information and information about the eye’s position/movements. A number of tasks have been developed that claim to measure the contribution of information about the eye’s position/movement to accurate judgements of object movement. These tasks are not only used to gain insight into motion processing but have also been used to investigate the underlying causes of psychosis, often relating their findings to deficits in the use of outflow, and an impairment in the ability to distinguish sensory changes due to own action from sensory changes due to events in the world (Bansal et al., 2018). However, it is not clear the tasks that have been used actually measure the same thing, nor is it clear what type of eye-position information they are actually measuring. They could be measuring a participant’s use of inflow, outflow, or a combination of the two (Weir et al., 2000). We designed an experiment to explore these issues, comparing performance on four different tasks—an antisaccade task, a saccadic shift task, a smooth-visual pursuit task, and a saccadic compression task—and using differences across individuals to look for inter-task correlations. The pattern of correlations in task results provides insight into whether the tasks are measuring the same thing and what they are measuring.
References
Seeing More Than the Face: Whole-Person Perception of Emotion
1Cardiff University Brain Research Imaging Centre, Cardiff, Wales
2Max Plank Institute for Biological Cybernetics, Tübingen, Germany
3FOM University of Applied Sciences, Augsburg, Germany
Previous research on the perception of facial expression has highlighted the importance of context. For instance, affective body posture influences perception of facial expression, such that observers are more likely to perceive a disgusted face as “angry” when presented in the context of an angry body (Teufel et al., 2019). This integration of face and body emotion cues is highly variable across individuals, offering an opportunity to study the mechanisms underlying integrated whole-person perception. Using psychophysical methods, we indexed the precision with which observers could represent isolated facial expression and body posture cues as well as the influence of each cue on the integrated whole-person percept of emotion. Our results indicate that the perceptual integration that leads to whole-person representation is determined by the precision of the individual cues. These results provide the basis for developing a mechanistic model of how facial expression and body posture cues are combined to create integrated whole-person percepts of emotion and have important implications for our understanding of real-world individual differences in social perception.
References
Detection of Self-Generated Global Motion Is Suppressed by Orthogonal Rotation
School of Psychology, University of Nottingham, Nottingham, UK
Integrating visual and motor cues is key to interacting with the world around us, for instance, when performing visually guided movements. Incongruencies between motor and visual signals may affect sensitivity to self-generated visual motion; however, evidence on the direction of this effect is inconsistent. Predictions derived from motor behaviour could facilitate the detection of visual motion congruent with that behaviour (Christensen et al., 2011). Alternatively, attenuation of self-generated sensory signals could increase sensitivity to unpredictable motion (Lally et al., 2011). Here, we characterise the profile of sensitivity to self-generated visual motion following rotations of that motion. Participants used a touch-screen to control the motion of dots within a random dot kinematogram. Motion inputs were rotated along eight possible angles: 0°, ±45°, ±90°, ±135°, or 180°. In additional control conditions, we passively played back dot motion to participants without their direct motor control. The ratio of signal-to-noise dots was adjusted to estimate motion coherence thresholds for each condition. With no rotation, coherence thresholds were comparable to those in the passive control conditions. By contrast, thresholds were elevated by rotations towards the orthogonal axes. However, thresholds decreased back towards baseline following a full 180° rotation, suggesting tuning to the axis of motion in addition to the direction. These results suggest a reduced sensitivity to self-generated visual motion that is incongruent with motor signals in terms of the direction and axis of motion.
References
“Spying” on Visual Areas Using James Bond and Functional Magnetic Resonance Imaging
School of Psychology, Cardiff University, Cardiff, Wales
The identification of specific visually responsive brain regions (e.g., V1, middle temporal visual area (MT), etc.) has long relied on anatomical landmarks or functional localisers. However, these approaches are respectively often inaccurate and time-consuming. The work of Bartels and Zeki (2004) suggests an alternative. Observers could watch a movie clip (rich broadband, natural stimuli) for 15 to 20 minutes in a magnetic resonance scanner and then independent component analysis (ICA) could be performed on the BOLD data. Different visual areas respond to different features in the movie. Voxels within one functionally specific region would have similar patterns of activation, and voxels in a different functionally specific region would have dissimilar patterns of activation. Therefore, ICA should segment the visually responsive parts of the brain into visual areas. If the procedure works then visual areas can be identified without the need of multiple localisers, anatomical landmarks or stable fixation, in a fraction of the time than would be needed if using the more traditional approaches. We tested this idea. We collected functional magnetic resonance imaging data on a group of participants who watched 15+ minutes of a James Bond movie, performed ICA and correlated the components with a probabilistic atlas of visual areas. These correlations revealed matches between the ICA components and the atlas visual areas. Next, we aim to validate these matches by extracting the time courses of the matched components and correlating them across brains.
References
The Retina–Cortex Relationship in Glaucoma and Healthy Adults
1School of Optometry and Vision Sciences, Cardiff Universit
2Cardiff University Brain Research Imaging Centre, Cardiff University, Cardiff, Wales
3School of Psychology, Cardiff University, Cardiff, Wales
Studies of visual function in glaucoma typically measure sensitivity at multiple locations in the visual field (perimetry) to determine the functional integrity of the retinal ganglion cells (RGCs). However, it is often overlooked that perimetric sensitivity represents the functional integrity of the entire pathway under examination rather than just RGCs. Here, we investigate the relationship between objectives measures of retinal and cortical function in patients with glaucoma and healthy controls. Twenty-one glaucoma patients (median interquartile range [IQR] age: 67.5 [59.8, 72.2] years; Mean Deviation (MD): −4.4 [−11.7, −1.3]); and 19 controls (median [IQR] age: 69.7 [61.9, 73.6] years; MD: +0.08 [−0.4, +0.5]) underwent monocular pattern electroretinography (PERG) and functional magnetic resonance imaging (fMRI). PERG was recorded during an onset-offset grating task (aperture: 5.5–18.0° × 5.5–18.0°) with five different spatial frequencies. The luminance component was subtracted, yielding a spatial tuning curve from which the peak amplitude was extracted. The cortical response was the peak surface-based fMRI response to radial checkerboard stimulation (19.5 × 11.0°). This took place in a SIEMENS 7 T MRI (Repetition Time = 2 seconds; Echo Time (TE) = 25 milliseconds; resolution = 1.2 mm3; 33 slices) and was restricted to the anatomically defined V1. PERG amplitude was statistically significantly lower in glaucoma patients than in controls (p < .01), as was fMRI amplitude (p = .045). The relationship between PERG (µV) and fMRI amplitude (z score) was moderate in healthy controls (slope = 0.16; r2 = .14; p = .055), but the slope was markedly shallower in glaucoma patients (slope = 0.03; r2 = .003; p = .41). Measures of cortical function are altered in glaucoma, in addition to retinal function. However, the influence of the RGC mosaic on upstream cortical responses may lessen in glaucoma.