Journal of Vision

PurposeFoveal vision in individuals with albinism is impaired not only by reduced visual acuity but also by elevated crowding - the disruption of object recognition in clutter. Because albinism is characterised by both retinal underdevelopment and nystagmus (uncontrolled eye movements), it is unclear whether crowding is elevated primarily from image motion due to eye movements or an additional sensory deficit. To disentangle these factors, we examined the spatial and featural selectivity of foveal crowding in albinism, comparing performance with controls and prior data from individuals with idiopathic infantile nystagmus syndrome (IINS), where nystagmus occurs without retinal underdevelopment. MethodsAdults with albinism (n=8) and age-matched controls (n=8; 19-49 years) identified the orientation of foveal Landolt-C targets. In Experiment 1, targets were presented alone or flanked horizontally or vertically to assess spatial selectivity. In Experiment 2, flankers were of the same or opposite contrast polarity to assess featural selectivity. Stimulus size was adaptively scaled using QUEST to estimate gap-size thresholds. ResultsCrowding was substantially elevated in albinism, relative to both controls and IINS. Experiment 1 revealed stronger crowding for horizontally than vertically positioned flankers in albinism, mirroring the predominant direction of nystagmic eye movements. In Experiment 2, opposite-polarity flankers did not reduce crowding, indicating an absence of selectivity for target-flanker similarity. ConclusionsFoveal crowding in albinism is markedly elevated, with a nystagmus-related spatial anisotropy and a lack of featural selectivity. These characteristics suggest that these elevations reflect both retinal image motion and a substantial sensory deficit arising from abnormal visual development.

Each individual person looks at natural scenes in their own unique way, resulting in a distinct perceptual experience of the world. However, little is known about why such differences in gaze emerge. Here, we test the hypothesis that idiosyncrasies in gaze behavior are predicted by inter-subject variations in internal models--expectations about how scenes typically look. In two experiments, we first characterized participants personal internal models by asking them to draw typical bathroom and kitchen scenes. Individual differences in these drawings were quantified using an objective deep learning pipeline and, in turn, related to individual differences in gaze behavior. In Experiment 1, where participants freely viewed a set of kitchen and bathroom photographs, inter-subject similarities in internal models did not predict inter-subject similarities in gaze. In Experiment 2, we encouraged strategic exploration through gaze-contingent viewing and a memory task. Here, inter-subject similarities in internal models predicted similarities in fixation frequency and the sequence in which different object categories were inspected. These findings suggest that the influence of internal models on visual exploration is stronger under increased sensory uncertainty and when expectation-guided sampling of the environment is encouraged. Together, our results provide new insights into how individual expectations shape gaze behavior and help explain why people differ in how they explore the visual world.

Purpose: To quantify whether fundus tracking in microperimetry improves psychometric parameter estimation (in vivo demonstration of improved stimulus-delivery precision), and to derive a psychometrically grounded criterion intensity for suprathreshold (defect-mapping) microperimetry. Methods: Twenty-five healthy volunteers underwent MAIA2-microperimetry at five loci: three outside and two inside the blind spot. Frequency-of-seeing (FoS) functions were measured in four blocks (2 tracking on; 2 tracking off). FoS-data were fit using cumulative-Gaussian psychometric functions estimating sensitivity parameters. Mixed-effect models assessed tracking effects, and posterior simulations defined the optimal criterion intensity for separating 'seeing' from 'non-seeing' loci. Results: Tracking had little effect on threshold estimates at loci outside the blind spot, but lowered threshold estimates within the blind spot (posterior median difference PMD [95% CrI] of -1.46 dB [-2.30, -0.62] at locus 4, and -1.02 dB [-1.94, -0.08] at locus 5). Tracking was associated with steeper psychometric slope parameters at loci 1-3 (PMD of -0.14 dB [-0.29, 0.01], -0.27 dB [-0.43, -0.12], and -0.22 dB [-0.40, -0.04]). Without tracking, false-positive responses were more frequent when fixation shifts displaced stimuli toward the 'seeing' retina. Simulation-based analysis identified 13 dB as nominally optimal criterion for suprathreshold microperimetry (Youden index: 0.76 [0.74, 0.79], comparable to 10 dB (0.74 [0.72, 0.76]). Conclusions: Even in healthy volunteers with stable fixation, fundus tracking measurably reduced sensitivity estimates at 'non-seeing' loci and sharpened FoS curves in the 'seeing' retina. A criterion intensity of 10 to 13 dB is a defensible choice for separating 'seeing' and 'non-seeing' retina in suprathreshold (defect-mapping) perimetry paradigms.

Color is a prominent feature of visual experience, yet humans can recognize objects easily and accurately from grayscale images. We examined whether color becomes more useful when spatial information is degraded due to blurring. Participants viewed naturalistic scenes in color or grayscale, and reported whether a named target object was present across a range of blur levels that simulated optical defocus from 0-8 diopters. With unblurred images, performance did not differ between color and grayscale conditions, but as blur increased, recognition accuracy declined. Color provided a modest but reliable advantage at higher levels of blur, suggesting that color becomes increasingly useful when optical quality is degraded. We hypothesize that the evolutionary shift towards trichromacy may have been partially driven by the need to compensate for optical degradation due to aging and/or accumulated light exposure.

Saccadic eye movements sweep the visual scene across the retina, yet the resulting motion is rarely perceived. Visual factors alone, such as the presence of static pre- and post-saccadic images, can attenuate motion perception, suggesting a masking of the motion signal during early visual processing. Here, we isolated the visual component of this reduction in motion perception using simulated saccades presented to fixating observers. Across two experiments, we manipulated motion amplitude (6-18 dva), duration, and velocity profile and measured perceived amplitude and velocity at varying masking durations. Visual masking strongly reduced perceived motion amplitude and velocity, with short halftimes ([~]15 ms) that were largely invariant across saccade amplitudes. Critically, motion following a naturalistic saccadic velocity profile was perceived as smaller and slower than constant-velocity motion matched in amplitude and duration, even without explicit masking. This additional reduction increased with both amplitude and duration. These results show that visual mechanisms alone can account for substantial motion reduction across a large range of amplitudes and demonstrate a partially separable contribution of the saccadic velocity profile, suggesting that the temporal structure of retinal motion itself supports perceptual continuity across eye movements.

Binocular vision depends on the integration of matching visual features across the two eyes, while conflicting interocular signals can engage active inhibitory processes in the visual system. To investigate the temporal dynamics of these putative inhibitory processes, we examined how transitions between different binocular correlation states influence perceptual detectability and response speed. Using dynamic random-dot correlograms - free of monocular cues and allowing precise interocular manipulation - we presented brief target intervals embedded in longer background sequences. Stimuli varied in binocular correlation: correlated (C) patterns contained identical luminance profiles in both eyes, anticorrelated (A) patterns had inverted luminance dots, and uncorrelated (U) patterns had independent dot arrangements. Across three experiments, we measured (1) the presentation duration threshold required to detect a change in correlation, (2) simple reaction times (RTs) to the same transitions at suprathreshold levels, and (3) psychometric functions across durations for selected transitions. In Experiment 1, A[->]C transitions yielded significantly higher duration thresholds than C[->]A, indicating a suppressive influence associated with prior anticorrelation. In contrast, Experiment 2 showed that A[->]C transitions produced the shortest RTs, while C[->]U transitions were slowest, suggesting a rebound-like facilitation following prior suppression. Experiment 3 confirmed these temporal and contrast dependences, with opposite changes in contrast threshold and reaction times between transitions toward and away from the correlated fusional states. This divergence between perceptual onset and reaction time is consistent with a two-phase account in which binocular anticorrelation is associated with an initial suppressive phase followed by rebound-like facilitation that accelerates responses once the target becomes detectable. These findings are consistent with current models of binocular rivalry and fusion, and provide a temporally resolved behavioral perspective on how inhibitory control in sensory systems may dynamically influence subsequent responsiveness under conditions of perceptual ambiguity.

How does motion contribute to robust object perception when appearance cues are unreliable? In natural scenes, camouflage, clutter, and occlusion can obscure object boundaries in static images, yet humans often resolve these ambiguities once objects move. Here we ask whether modern artificial vision systems capture this motion-dependent computation and whether their internal representations align with those used by biological vision. Using videos from the MOCA (Moving Camouflaged Animals) dataset, we introduce behavioral benchmarks that quantify the accuracy of object position and size estimation in scenes containing either static or moving objects. We first compare human observers with a diverse set of artificial neural networks. For static stimuli, image-based and video-based models achieve similar accuracy in predicting object position and size. However, humans show systematic improvements when objects are presented in motion. Image-based models do not exhibit this motion-dependent improvement, whereas several video-based architectures reproduce this behavioral pattern by integrating information across time. To examine the representational basis of these differences, we record neural population responses from the macaque inferior temporal (IT) cortex during presentation of the same stimuli. Models that more closely match IT representations also better reproduce human motion-dependent behavior. These results show that static accuracy alone is insufficient to evaluate models of visual perception and that alignment with primate visual representations provides a useful guide for developing models that capture dynamic computations in vision.

Some images are consistently remembered better than others, suggesting that memorability reflects intrinsic image properties. We tested whether within-category distinctiveness underlies this effect. Across three experiments (N = 477), participants categorized indoor scenes previously rated for subjective typicality and then completed recognition memory tests. Typical scenes were categorized faster and more accurately, but were remembered worse and showed a more liberal response bias than atypical scenes. These opposing effects were robust across categories. To link subjective typicality to visual representations, we quantified image distinctiveness using a convolutional neural network (CNN). Across layers, CNN-derived distinctiveness closely tracked human typicality judgments and predicted both categorization speed and memorability, with strongest effects in higher, semantic layers. Critically, the memory advantage for atypical scenes persisted even when most images were atypical, ruling out rarity within the experimental context. Together, the results show that intrinsic scene memorability reflects an images position within a category-specific representational space.

Reverse correlation is a widely-used and well-established method for probing latent perceptual representations in which subjects render subjective preference responses to ambiguous stimuli. Stimuli are purposefully designed to have no direct relationship with the target representation (e.g., they are randomly-generated), a property which makes each individual stimulus minimally informative toward reconstructing the target, and often difficult to interpret for subjects. As a result, a large number of stimulus-response pairs must be gathered from a given subject in order for reconstructions to be of sufficient quality, making the task fatiguing. Recent work has demonstrated that the number of trials needed can be substantially reduced using a compressive sensing framework that incorporates the assumption that the target representation can be sparsely represented in some basis into the reconstruction process. Here, we introduce an alternative method that incorporates the sparsity assumption directly into stimulus generation, which holds promise not only for improving efficiency, but also for improving the interpretability of stimuli from subjects perspective. We develop this new method as a mathematical variation of the compressive sensing approach, before conducting one simulation study and two human subjects experiments to assess the benefits of this method to reconstruction quality, sample size efficiency, and subjective interpretability. Results show that sparse stimulus generation improves all three of these areas relative to conventional reverse correlation approaches, and also relative to compressive sensing in most conditions.

Selective attention enables the prioritization of task-relevant information while managing distractors, and steady-state visual evoked potentials (SSVEPs) are widely used to track this process by tagging different visual objects at distinct flicker frequencies. However, whether the choice of tagging frequency itself influences other neural and cognitive measures remains unclear. Here, 27 participants performed detection and 1-back working memory tasks while a central target and peripheral distractors flickered at either 8.6 Hz or 12 Hz. The working memory task produced slower responses, more errors, and greater perceived difficulty than detection. Tagging frequency strongly shaped neural responses, with 8.6 Hz eliciting higher SSVEP signal-to-noise ratios than 12 Hz regardless of stimulus location. Nevertheless, stronger SSVEP responses for centrally attended stimuli were associated with fewer working memory errors and larger early visual ERP responses, while SSVEPs for attended and distractor stimuli were negatively correlated. In addition, the working memory task produced a larger P1-N1 peak-to-peak difference, and tagging frequency altered the timing and amplitude of early ERP effects. Together, these findings show that tagging frequency is not a neutral methodological parameter, but one that shapes both neural indices of attention and their relationship to cognitive performance.

Receptive fields provide a concise description of the stimulus selectivity of visual neurons. But this stimulus selectivity is neither static nor linear, and these nonlinear effects are not well captured by standard linear or pseudo-linear receptive field models. At the same time, receptive field models incorporating nonlinear effects are largely empirical, and are not easily interpreted in terms of underlying cellular and synaptic mechanisms. Here we show that two nonlinear mechanisms in the primate outer retina shape neural responses and that these contribute significantly to responses to natural stimuli and to the retinal output signals. Incorporating these outer retinal nonlinearities into models for visual function will improve our ability to identify the mechanistic origin of specific features of downstream visual responses.

Reaction time is a measure of the speed of our response to stimuli in the environment. Even for a well-trained task, a subjects reaction time varies. One source of this variability is internal state fluctuations (such as changes in arousal). There are few studies that systematically quantify the extent to which reaction time varies across different timescales and link this to measures of systemic physiology associated with arousal. In much of the literature, it is assumed but not demonstrated that behavioral and systemic measurements associated with arousal will be consistently linked because both estimate a common underlying arousal process. In this work, we examined this assumption by simultaneously measuring reaction time, heart rate, and pupil diameter in rhesus macaque monkeys performing several visual tasks over hours and across hundreds of sessions. We found a portion of the variability in reaction time could be linked to systemic physiological signatures of arousal on fast timescales from second to second and slower timescales from minute to minute. This link between reaction time and systemic physiology was also present for different biomarkers of arousal (heart rate and pupil). However, the strength of this relationship varied depending on the arousal biomarker. Our findings support the conclusion that there are multiple arousal mechanisms that act simultaneously to influence behavior and multiple timescales at which they operate.

1Markerless pose estimation has emerged as a powerful technique for animal behavior quantification, capable of high resolution tracking of body parts. Many neuroscience labs rely on tools like DeepLabCut and SLEAP, which provide accessible interfaces but restrict users to a narrow set of models and configurations. In this work, we adopt MMPose an open source, general-purpose computer vision library to build a workflow for training and evaluating multiple state-of-the-art models on animal video datasets. We benchmark these models in two scenarios: (1) a complex maze assay with occlusions and varied backgrounds, and (2) a simpler open field arena with a high-contrast background. Our results show that a bottomup model (DEKR) delivers the highest accuracy in the complex task, whereas lighter-weight models (e.g., SLEAP) offer superior speed highlighting a clear trade-off between accuracy and throughput. We also evaluate a recently published foundation model (TopViewMouse-5K) trained on a large top-view mouse dataset to test its generalization. It performs poorly on our tasks at zero-shot, and even when we combine its data with our training set, we observe no consistent benefit. These findings emphasize the importance of context-specific model selection and the need for more diverse training data to create generalizable pose models. By leveraging a general-purpose vision library, researchers can flexibly choose models that best suit their experimental needs. This work illustrates how adopting advanced computer vision frameworks can accelerate behavioral neuroscience and genetics research, paving the way for more scalable, reproducible, and sensitive analysis of animal behavior.

Inscapes is a low demand abstract animation used as an alternative to eyes open rest in neuroimaging studies, particularly with pediatric and clinical populations prone to head motion. Although prior work has established that functional connectivity patterns during Inscapes closely resemble those during rest, no study has examined whether the two conditions differ in aperiodic neural activity, a broadband feature of the power spectrum linked to excitation/inhibition balance. Here we used magnetoencephalography (MEG) in 54 healthy adults to compare spectrally parameterised aperiodic and periodic measures between eyes open rest and Inscapes viewing (visual component only, without audio). At the sensor level, both the aperiodic exponent and offset were significantly higher during rest than during Inscapes across widespread frontoparietal and occipital distributions in both magnetometers and gradiometers. Source level analyses at both the parcellation and vertex levels largely supported these patterns. The pericalcarine cortex was a notable exception, where both aperiodic measures were higher during Inscapes than during rest, indicating a regionally specific reversal in primary visual cortex. These results demonstrate that Inscapes and eyes open rest produce distinct aperiodic spectral profiles, indicating that the two conditions are not interchangeable for analyses involving broadband spectral dynamics or excitation/inhibition balance estimation.

Neural decoders serve as probabilistic interfaces in co-control brain-to-text BCIs, where predicted uncertainty shapes hypothesis generation and language model integration, enabling decisions to be made safely under uncertainty. However, it remains unclear whether these decoders produce reliable and informative uncertainty, or how training objectives shape these properties. This work characterizes and improves uncertainty representations in brain-to-text decoding. We extend two metrics, calibration error (ECE) and resolution (RES), to evaluate sequential probabilistic predictions from frame-level phoneme estimates to word-level hypotheses, quantifying the reliability and informativeness of model uncertainty. Using this framework, we analyze neural decoders trained with connectionist temporal classification (CTC). To isolate the causal role of uncertainty independent of accuracy, we manipulate predicted probability distributions while holding predicted sequences fixed. Motivated by the observed failures, we further examine the role of the training objective and propose a two-stage cross-entropy (CE) formulation that decouples alignment inference from classification. We show that widely used CTC-trained neural decoders in brain-to-text BCIs produce systematically over-confident predictions, with high confidence persisting even when predictions are incorrect. Controlled manipulations of the prediction reveal that improved ECE and RES enhance hypothesis generation and language-model integration by promoting diverse alternatives and more effective re-ranking of hypotheses aligned with user intent. Mechanistically, CTC relies on over-confident predictions to resolve alignment ambiguity. Replacing CTC with CE loss yields significantly more reliable and informative probabilistic predictions without degrading decoding accuracy. Uncertainty emerges as a system-level design variable in brain-to-text interfaces. Calibrated uncertainty from neural decoders enables effective integration with independently trained language models and reliable error detection. This work reframes uncertainty from a passive output into an active control signal, identifies key components and evaluation criteria for probabilistic co-control, and outlines a pathway toward next-generation BCIs that supports increasingly complex interactions with the world.

Diabetic retinopathy (DR) is the leading cause of preventable blindness in working-age adults, affecting an estimated 103 million people worldwide. Standard deep learning classifiers treat fundus images as independent samples, ignoring latent inter-patient relational structure that is most informative at clinically ambiguous intermediate severity levels. We propose a topology-aware, graph-based deep learning framework combining three complementary components: (i) an EfficientNet-B3 convolutional backbone for high-level visual feature extraction; (ii) persistent homology descriptors (H0 and H1) derived from morphologically skeletonised retinal vascular networks, characterising global vascular topology in a noise-robust manner; and (iii) a GraphSAGE graph neural network propagating disease-related information across a population-level similarity graph, refining representations through inductive neighbourhood aggregation. The similarity graph combines cosine similarity on visual features with 2-Wasserstein distance between persistence diagrams. Evaluated on three public benchmarks, the framework achieves 95.5% accuracy on Kaggle DR, 96.1% on Messidor-2, and 94.6% on APTOS 2019, consistently outperforming a strong CNN baseline by 1.5-2.3 percentage points across accuracy, Quadratic Weighted Kappa, and macro-F1. Ablation experiments confirm synergistic contributions of topological feature augmentation and relational graph learning. One-way ANOVA (F > 80, p < 0.001) confirms that DR progression is reflected in global vascular topology across all five severity stages, providing quantitative biological grounding for the framework design. Code and data are publicly available at https://github.com/Nader-BelHadj/plosene.

Functional localizer scans have long served as the classic method for mapping individualized functional topographies, but they require dedicated scan time and can be difficult to implement in neuropsychological populations. Previous work has shown that individualized functional topographies can be estimated with high fidelity in typical participants using hyperalignment, but it remains unknown whether this approach generalizes to populations with functional deficits. Here, we tested this question in developmental prosopagnosia (DP), a neuropsychological condition characterized by severe face recognition impairments. Using two independent datasets that included both DP and control participants, we estimated individualized category-selective functional topographies from independent participants using hyperalignment derived from either a task-based scan or a naturalistic movie-viewing scan. Across datasets, whole-brain correlations and searchlight analyses showed that predicted topographies were highly similar to topographies estimated from participants own localizer data, especially in cortical areas with strong category-selective responses. Hyperalignment successfully recovered idiosyncratic features of category-selective topographies and consistently outperformed anatomical alignment. Importantly, predictions generalized across groups, such that individualized topographies in DPs could be estimated from control participants and vice versa. In addition, predicted topographies preserved the reduced face selectivity in DPs previously reported in the literature. These findings support a hyperalignment-based framework for estimating individualized functional topographies in neuropsychological populations without requiring separate localizer scans, and provide a foundation for integrating existing datasets to study the underlying neural basis in DP and other atypical populations.

Practice enhances motor acuity, enabling movement execution with greater speed and accuracy. However, the learning principles underlying improvements in speed, accuracy, and efficiency remain less understood than those supporting motor skill acquisition and adaptation. Here, we examined motor execution in a skill-based practice task to characterize learning, retention, and generalization of motor acuity. Using a gamified two-dimensional racing task, right-handed participants controlled a stylus-driven car along a curved track as quickly and accurately as possible. Across two studies (N = 83 total, 54 females), participants completed 300 training laps on Session 1 and returned for Session 2 to assess retention and generalization to novel track configurations: one with altered spatial configuration (rotated track) and one requiring movement in the opposite direction of training (reverse track). Movement speed improved rapidly and showed robust, though incomplete, retention across sessions. Speed improvements generalized substantially to both novel tracks. Accuracy was high at training onset and showed strong retention. However, we do not observe offline gains between sessions. Notably, accuracy declined transiently for the novel track configurations, suggesting interference from prior training. Movement efficiency, indexed by path length, was retained and generalized to the rotated track. However, reversing movement direction impaired efficiency, revealing a movement direction effect. This effect persisted when training direction was reversed in a second study, with counterclockwise movements remaining slower and less efficient than clockwise movements. These findings show that practice produces durable and broadly transferable motor execution improvements, while inherent movement direction biases constrain how improvements generalize across contexts. New & NoteworthyThe learning principles underlying improvements in motor acuity remain less well understood than those governing other forms of motor learning. Prior work suggests that motor execution improvements show limited generalization. In contrast, the present findings demonstrate that execution-based practice can produce robust, transferable gains, while also revealing a key constraint: inherent movement direction biases that limit generalization. By characterizing learning, retention, and generalization, this work provides new insight into how motor acuity improvements compare with skill acquisition and adaptation.

Background: Despite strong evidence from controlled trials, uncertainty remains about the real-world use of 0.05% atropine in patients with lighter irises due to tolerability concerns, and predictors of treatment response are poorly understood. Here, we evaluated the effectiveness, tolerability, and early biometric response to 0.05% atropine in clinical practice among patients with predominantly light irises. Methods: This prospective cohort study included 33 patients treated with 0.05% atropine (82% with light irises). Cycloplegic spherical equivalent refraction (SER) was measured at baseline and 3-month intervals. Axial length (AL), photopic pupil diameter, accommodation amplitude, and subjective side effects were monitored more frequently initially. Results: Median age at treatment initiation was 11.97 years, SER -5.38 D, and AL 25.42 mm. Over 12 months, SER changed by -0.078 {+/-} 0.349 D (mean {+/-} SD), and AL increased by 0.052 {+/-} 0.115 mm. Eighty-eight percent of participants had a SER change of <0.5 D, and 91% had axial elongation of <0.2 mm, indicating clinically limited myopia progression. Photopic pupil diameter was larger, and accommodation amplitude was reduced throughout follow-up. Early in treatment, side effects, including photophobia and near-work difficulties, were common but minimally disruptive. Their incidence decreased rapidly and rarely required treatment modification. In exploratory analyses, early AL changes predicted 12-month AL outcomes, with associations detectable as early as 1 week and strengthening over time. Conclusions: 0.05% atropine was well tolerated and effective in this population with light irises. Early AL changes may predict 12-month treatment response. These findings support the implementation of 0.05% atropine in routine clinical practice in populations with light irises and highlight the potential for early AL monitoring to guide timely treatment adjustments.

To determine whether achieving normal corrected visual acuity independently influences myopia progression in school-aged children wearing single-vision spectacles. In a one-year real-world cohort study, 9-year-old myopic children were classified into three groups: uncorrected, adequately corrected (normal visual acuity), and under-corrected (subnormal visual acuity). One-to-one propensity score matching was used to balance baseline characteristics, and annual axial length growth was compared. The adequately corrected group showed the slowest axial elongation (0.23 {+/-} 0.14 mm/year), significantly less than both the under-corrected (0.35 &{+/-} 0.14 mm/year) and uncorrected groups (0.37 &{+/-} 0.16 mm/year) (all P < 0.001). Although the under-corrected group exhibited marginally slower progression than the uncorrected group, this minimal benefit was not sustained in semiannual analyses and lacked clinical relevance. Simply prescribing spectacles is insufficient for myopia control; achieving normal corrected visual acuity is essential to meaningfully slow axial elongation.