Search

Anterior regions of the ventral visual stream encode substantial information about object categories. Are top-down category-level forces critical for arriving at this representation, or can this representation be formed purely through domain-general learning of natural image structure? Here we present a fully self-supervised model which learns to represent individual images, rather than categories, such that views of the same image are embedded nearby in a low-dimensional feature space, distinctly from other recently encountered views. We find that category information implicitly emerges in the local similarity structure of this feature space. Further, these models learn hierarchical features which capture the structure of brain responses across the human ventral visual stream, on par with category-supervised models. These results provide computational support for a domain-general framework guiding the formation of visual representation, where the proximate goal is not explicitly about category information, but is instead to learn unique, compressed descriptions of the visual world., It is unknown whether object category learning can be formed purely through domain general learning of natural image structure. Here the authors show that human visual brain responses to objects are well-captured by self-supervised deep neural network models trained without labels, supporting a domain-general account.

Crowding is the interference by surrounding objects (flankers) with target perception. Low target-flanker similarity usually yields weaker crowding than high similarity (‘similarity rule’) with less interference, e.g., by opposite- than same-contrast polarity flankers. The advantage of low target-flanker similarity has typically been shown with attentional selection of a single target object. Here, we investigated the validity of the similarity rule when broadening attention to multiple objects. In three experiments, we measured identification for crowded letters (Experiment 1), tumbling Ts (Experiment 2), and tilted lines (Experiment 3). Stimuli consisted of three items that were uniform or alternating in contrast polarity and were briefly presented at ten degrees eccentricity. Observers reported all items (full report) or only the left, central, or right item (single-item report). In Experiments 1 and 2, consistent with the similarity rule, single central item performance was superior with opposite- compared to same-contrast polarity flankers. With full report, the similarity rule was inverted: performance was better for uniform compared to alternating stimuli. In Experiment 3, contrast polarity did not affect performance. We demonstrated a reversal of the similarity rule under broadened attention, suggesting that stimulus uniformity benefits crowded object recognition when intentionally directing attention towards all stimulus elements. We propose that key properties of crowding have only limited validity as they may require a-priori differentiation of target and context.

Funder: Alexander von Humboldt-Stiftung (Alexander von Humboldt Foundation); doi: https://doi.org/10.13039/100005156, Distinguishing animate from inanimate things is of great behavioural importance. Despite distinct brain and behavioural responses to animate and inanimate things, it remains unclear which object properties drive these responses. Here, we investigate the importance of five object dimensions related to animacy (“being alive”, “looking like an animal”, “having agency”, “having mobility”, and “being unpredictable”) in brain (fMRI, EEG) and behaviour (property and similarity judgements) of 19 participants. We used a stimulus set of 128 images, optimized by a genetic algorithm to disentangle these five dimensions. The five dimensions explained much variance in the similarity judgments. Each dimension explained significant variance in the brain representations (except, surprisingly, “being alive”), however, to a lesser extent than in behaviour. Different brain regions sensitive to animacy may represent distinct dimensions, either as accessible perceptual stepping stones toward detecting whether something is alive or because they are of behavioural importance in their own right.

Stone toolmaking is a human motor skill which provides the earliest archeological evidence motor skill and social learning. Intentionally shaping a stone into a functional tool relies on the interaction of action observation and practice to support motor skill acquisition. The emergence of adaptive and efficient visuomotor processes during motor learning of such a novel motor skill requiring complex semantic understanding, like stone toolmaking, is not understood. Through the examination of eye movements and motor skill, the current study sought to evaluate the changes and relationship in perceptuomotor processes during motor learning and performance over 90 h of training. Participants’ gaze and motor performance were assessed before, during and following training. Gaze patterns reveal a transition from initially high gaze variability during initial observation to lower gaze variability after training. Perceptual changes were strongly associated with motor performance improvements suggesting a coupling of perceptual and motor processes during motor learning., In order to evaluate changes and relationships in perceptual process during motor learning and motor performance over training, Bayani et al examine eye movements and motor skill in a paleolithic stone toolmaking task. They demonstrate the importance of understanding the emergence and evolution of perceptual-motor coupling not only in complex motor skill learning.

(This work has meanwhile been published in Scientific Reports: Lauer, T., Schmidt, F. & Võ, M.LH. The role of contextual materials in object recognition. Sci Rep 11, 21988 (2021). https://doi.org/10.1038/s41598-021-01406-z). While scene context is known to facilitate object recognition, little is known about whichcontextual “ingredients” are at the heart of this phenomenon. Here, we address the question ofwhether the materials that frequently occur in scenes (e.g., tiles in bathroom) associated withspecific objects (e.g., a perfume) are relevant for processing of that object. To this end, wepresented photographs of consistent and inconsistent objects (e.g., perfume vs. pinecone)superimposed on scenes (e.g., bathroom) and close-ups of materials (e.g., tiles). In Experiment1, consistent objects on scenes were named more accurately than inconsistent ones, while therewas only a marginal consistency effect for objects on materials. Also, we did not find anyconsistency effect for scrambled materials that served as color control condition. In Experiment2, we recorded event-related potentials (ERPs) and found N300/N400 responses – markers ofsemantic violations – for objects on inconsistent relative to consistent scenes. Critically, objectson materials triggered N300/N400 responses of similar magnitudes. Our findings show thatcontextual materials indeed affect object processing – even in the absence of spatial scenestructure and object content – suggesting that material is one of the contextual “ingredients”driving scene context effects.

Perception results from the interplay of sensory input and prior knowledge. Despite behavioral evidence that long-term priors powerfully shape perception, the neural mechanisms underlying these interactions remain poorly understood. We obtained direct cortical recordings in neurosurgical patients as they viewed ambiguous images that elicit constant perceptual switching. We observe top-down influences from the temporal to occipital cortex, during the preferred percept that is congruent with the long-term prior. By contrast, stronger feedforward drive is observed during the non-preferred percept, consistent with a prediction error signal. A computational model based on hierarchical predictive coding and attractor networks reproduces all key experimental findings. These results suggest a pattern of large-scale information flow change underlying long-term priors’ influence on perception and provide constraints on theories about long-term priors’ influence on perception., Priors learnt from lifetime experiences influence perception. The authors show that when perception is congruent with a long-term prior, there is increased top-down input in the ventral visual stream, whereas bottom-up input is enhanced when perception is incongruent with prior.

Recognising a potential visual-field (VF) defect in paediatric patients might be challenging, especially in children before the age of 5 years and those with developmental delay or intellectual disability. Visual electrophysiological testing is an objective and non-invasive technique for evaluation of visual function in paediatric patients, which can characterise the location of dysfunction and differentiate between disorders of the retina, optic nerve and visual pathway. The recording of electroretinography (ERG) and visual-evoked potentials (VEP) is possible from early days of life and requires no subjective input from the patient. As the origins of ERG and VEP tests are known, the pattern of electrophysiological changes can provide information about the VF of a child unable to perform accurate perimetry. This review summarises previously published electrophysiological findings in several common types of VF defects that can be found in paediatric patients (generalised VF defect, peripheral VF loss, central scotoma, bi-temporal hemianopia, altitudinal VF defect, quadrantanopia and homonymous hemianopia). It also shares experience on using electrophysiological testing as additional functional evidence to other tests in the clinical challenge of diagnosing or excluding VF defects in complex paediatric patients. Each type of VF defect is illustrated with one or two clinical cases.

To optimize visual search, humans attend to objects with the expected size of the sought target relative to its surrounding scene (object-scene scale consistency). We investigate how the human brain responds to variations in object-scene scale consistency. We use functional magnetic resonance imaging and a voxel-wise feature encoding model to estimate tuning to different object/scene properties. We find that regions involved in scene processing (transverse occipital sulcus) and spatial attention (intraparietal sulcus) have the strongest responsiveness and selectivity to object-scene scale consistency: reduced activity to mis-scaled objects (either unusually smaller or larger). The findings show how and where the brain incorporates object-scene size relationships in the processing of scenes. The response properties of these brain areas might explain why during visual search humans often miss objects that are salient but at atypical sizes relative to the surrounding scene., Lauren Welbourne et al. use functional magnetic resonance imaging to investigate the neural dynamics linked to how humans process object size in the environment. After showing participants a series of images with appropriately-sized or misscaled objects (such as a giant toothbrush on a bathroom sink), the authors observed that the temporal occipital sulcus and intraparietal sulcus were strongly responsive to normally-sized, but not misscaled, objects, suggesting that object representations in both brain regions incorporate the objects’ typical size relationships to the surrounding scene.

Encoding and recognising complex natural sequences provides a challenge for human vision. We found that observers could recognise a previously presented segment of a video of a hearth fire when embedded in a longer sequence. Recognition performance declined when the test video was spatially inverted, but not when it was hue reversed or temporally reversed. Sampled motion degraded forwards/reversed playback discrimination, indicating observers were sensitive to the asymmetric pattern of motion of flames. For brief targets, performance increased with target length. More generally, performance depended on the relative lengths of the target and embedding sequence. Increased errors with embedded sequence length were driven by positive responses to non-target sequences (false alarms) rather than omissions. Taken together these observations favour interpreting performance in terms of an incremental decision-making model based on a sequential statistical analysis in which evidence accrues for one of two alternatives. We also suggest that prediction could provide a means of providing and evaluating evidence in a sequential analysis model.

Reflectance, lighting and geometry combine in complex ways to create images. How do we disentangle these to perceive individual properties, such as surface glossiness? We suggest that brains disentangle properties by learning to model statistical structure in proximal images. To test this hypothesis, we trained unsupervised generative neural networks on renderings of glossy surfaces and compared their representations with human gloss judgements. The networks spontaneously cluster images according to distal properties such as reflectance and illumination, despite receiving no explicit information about these properties. Intriguingly, the resulting representations also predict the specific patterns of ‘successes’ and ‘errors’ in human perception. Linearly decoding specular reflectance from the model’s internal code predicts human gloss perception better than ground truth, supervised networks or control models, and it predicts, on an image-by-image basis, illusions of gloss perception caused by interactions between material, shape and lighting. Unsupervised learning may underlie many perceptual dimensions in vision and beyond., Storrs et al. train unsupervised generative neural networks on glossy surfaces and show how gloss perception in humans may emerge in an unsupervised fashion from learning to model statistical structure.

Convolutional neural networks (CNNs) are increasingly used to model human vision due to their high object categorization capabilities and general correspondence with human brain responses. Here we evaluate the performance of 14 different CNNs compared with human fMRI responses to natural and artificial images using representational similarity analysis. Despite the presence of some CNN-brain correspondence and CNNs’ impressive ability to fully capture lower level visual representation of real-world objects, we show that CNNs do not fully capture higher level visual representations of real-world objects, nor those of artificial objects, either at lower or higher levels of visual representations. The latter is particularly critical, as the processing of both real-world and artificial visual stimuli engages the same neural circuits. We report similar results regardless of differences in CNN architecture, training, or the presence of recurrent processing. This indicates some fundamental differences exist in how the brain and CNNs represent visual information., Convolutional neural networks are increasingly used to model human vision. Here, the authors compare the performance of 14 different CNNs and human fMRI responses to real-world and artificial objects to show some fundamental differences exist between them.

This project contains the stimuli, the raw MEG data and the analysis code used in 'A humanness dimension to visual object coding in the brain' by Contini, Goddard, Grootswagers, Williams and Carlson (2020) Neuroimage (https://doi.org/10.1016/j.neuroimage.2020.117139). Please refer to this work for full details of the methodology, and cite this paper if using any of this data in future work. - To use the high resolution, segmented object images from this study, download the 'STIMULI' folder. - To recreate the MEG data analyses from scratch, download the 'MEG_DATA', 'BEHAVIOURAL' and 'ANALYSIS_CODE' folders and start with the Matlab function 'CategoryAndPosition_MEG_Analysis.m' located in the 'ANALYSIS_CODE' directory. To skip any of the analysis steps, download the 'ANALYSED_DATA' folder. - To recreate the manuscript figures, download only the 'FIGURES' and 'STIMULI' folders and use the Matlab scripts in the 'FIGURES' folder. - In the 'ONLINE_EXP' folder, humanness-online.zip contains the raw data and experiment script for the online shape-similarity-rt part of the paper (also publicly available on https://gitlab.pavlovia.org/tijl/humanness).

The ability to maintain visual working memory (VWM) associations about the identity and location of objects has at times been found to decrease with age. To date, however, this age-related difficulty was mostly observed in artificial visual contexts (e.g., object arrays), and so it is unclear whether it may manifest in naturalistic contexts, and in which ways. In this eye-tracking study, 26 younger and 24 healthy older adults were asked to detect changes in a critical object situated in a photographic scene (192 in total), about its identity (the object becomes a different object but maintains the same position), location (the object only changes position) or both (the object changes in location and identity). Aging was associated with a lower change detection performance. A change in identity was harder to detect than a location change, and performance was best when both features changed, especially in younger adults. Eye movements displayed minor differences between age groups (e.g., shorter saccades in older adults) but were similarly modulated by the type of change. Latencies to the first fixation were longer and the amplitude of incoming saccades was larger when the critical object changed in location. Once fixated, the target object was inspected for longer when it only changed in identity compared to location. Visually salient objects were fixated earlier, but saliency did not affect any other eye movement measures considered, nor did it interact with the type of change. Our findings suggest that even though aging results in lower performance, it does not selectively disrupt temporary bindings of object identity, location, or their association in VWM, and highlight the importance of using naturalistic contexts to discriminate the cognitive processes that undergo detriment from those that are instead spared by aging.

Value-based decisions about alternatives we have never experienced can be guided by associations between current choice options and memories of prior reward. A critical question is how similar memories need to be to the current situation to effectively guide decisions. We address this question in the context of associative learning of faces using a sensory preconditioning paradigm. We find that memories of reward spread along established associations between faces to guide decision making. While memory guidance is specific for associated facial identities, it does not only occur for the specific images that were originally encountered. Instead, memory guidance generalizes across different images of the associated identities. This suggests that memory guidance does not rely on a pictorial format of representation but on a higher, view-invariant level of abstraction. Thus, memory guidance operates on a level of representation that neither over- nor underspecifies associative relationships in the context of obtaining reward. Open-Access-Publikationsfonds 2020 peerReviewed

Our conscious visual perception relies on predictive signals, notably in the periphery where sensory uncertainty is high. We investigated how such signals could support perceptual stability of objects’ size across the visual field. When attended carefully, the same object appears slightly smaller in the periphery compared to the fovea. Could this perceptual difference be encoded as a strong prior to predict the peripheral perceived size relative to the fovea? Recent studies emphasized the role of foveal information in defining peripheral size percepts. However, they could not disentangle bottom-up from top-down mechanisms. Here, we revealed a pure top-down contribution to the perceptual size difference between periphery and fovea. First, we discovered a novel Ebbinghaus illusion effect, inducing a typical reduction of foveal perceived size, but a reversed increase effect in the periphery. The resulting illusory size percept was similar at both locations, deviating from the classic perceptual difference. Then through an updating process of successive peripheral-foveal viewing, the unusual peripheral perceived size decreased. The classic perceptual eccentricity difference was restored and the peripheral illusion effect changed into a fovea-like reduction. Therefore, we report the existence of a prior that actively shapes peripheral size perception and stabilizes it relative to the fovea.

Human functional magnetic resonance imaging (fMRI) typically employs the blood-oxygen-level-dependent (BOLD) contrast mechanism. In non-human primates (NHP), contrast enhancement is possible using monocrystalline iron-oxide nanoparticles (MION) contrast agent, which has a more temporally extended response function. However, using BOLD fMRI in NHP is desirable for interspecies comparison, and the BOLD signal’s faster response function promises to be beneficial for rapid event-related (rER) designs. Here, we used rER BOLD fMRI in macaque monkeys while viewing real-world images, and found visual responses and category selectivity consistent with previous studies. However, activity estimates were very noisy, suggesting that the lower contrast-to-noise ratio of BOLD, suboptimal behavioural performance, and motion artefacts, in combination, render rER BOLD fMRI challenging in NHP. Previous studies have shown that rER fMRI is possible in macaques with MION, despite MION’s prolonged response function. To understand this, we conducted simulations of the BOLD and MION response during rER, and found that no matter how fast the design, the greater amplitude of the MION response outweighs the contrast loss caused by greater temporal smoothing. We conclude that although any two of the three elements (rER, BOLD, NHP) have been shown to work well, the combination of all three is particularly challenging.

Theoretical frameworks of human vision argue that object responses remain stable, or ‘invariant’, despite changes in viewing conditions that can alter object appearance but not identity. Here, in a major departure from previous approaches that have relied on two-dimensional (2-D) images to study object processing, we demonstrate that changes in an object’s appearance, but not its identity, can lead to striking shifts in behavioral responses to objects. We used inverse multidimensional scaling (MDS) to measure the extent to which arrangements of objects in a sorting task were similar or different when the stimuli were displayed as scaled 2-D images, three-dimensional (3-D) augmented reality (AR) projections, or real-world solids. We were especially interested in whether sorting behavior in each display format was based on conceptual (e.g., typical location) versus physical object characteristics. We found that 2-D images of objects were arranged according to conceptual (typical location), but not physical, properties. AR projections, conversely, were arranged primarily according to physical properties such as real-world size, elongation and weight, but not conceptual properties. Real-world solid objects, unlike both 2-D and 3-D images, were arranged using multidimensional criteria that incorporated both conceptual and physical object characteristics. Our results suggest that object responses can be strikingly malleable, rather than invariant, with changes in the visual characteristics of the stimulus. The findings raise important questions about limits of invariance in object processing, and underscore the importance of studying responses to richer stimuli that more closely resemble those we encounter in real-world environments.

When two different objects are sequentially presented at the same location, the viewer tends to misjudge them as identical (spatial congruency bias). The present study examined whether the spatial congruency bias would involve not only retinotopic but also non-retinotopic processing using the Ternus-Pikler illusion. In the experiments, two objects (central and peripheral) appeared in an initial frame. The target object was presented in the central area of the display, while the peripheral object was either on the left or right side of the target object. In the second frame, the target object was again presented in the central area, and the peripheral object was on the opposite side. Two kinds of inter-stimulus intervals were used. In the no-blank condition, the target object was perceived as stationary, and the peripheral object appeared to move to the opposite side. However, in the long-blank condition, the two objects were perceived to move together. Participants judged whether the target objects in the two frames were identical. As a result, the spatial congruency bias occurred irrespective of the ISI conditions. Our findings suggest that the spatial congruency bias is mainly based on retinotopic processing.

Extrastriate visual areas are strongly activated by image symmetry. Less is known about symmetry representation at object-level rather than image-level. Here we investigated electrophysiological responses to symmetry, generated by amodal completion of partially-occluded polygon shapes. We used a similar paradigm in four experiments (N = 112). A fully-visible abstract shape (either symmetric or asymmetric) was presented for 250 ms (t0). A large rectangle covered it entirely for 250 ms (t1) and then moved to one side to reveal one half of the shape hidden behind (t2, 1000 ms). Note that at t2 no symmetry could be extracted from retinal image information. In half of the trials the shape was the same as previously presented, in the other trials it was replaced by a novel shape. Participants matched shapes similarity (Exp. 1 and Exp. 2), or their colour (Exp. 3) or the orientation of a triangle superimposed to the shapes (Exp. 4). The fully-visible shapes (t0–t1) elicited automatic symmetry-specific ERP responses in all experiments. Importantly, there was an exposure-dependent symmetry-response to the occluded shapes that were recognised as previously seen (t2). Exp. 2 and Exp.4 confirmed this second ERP (t2) did not reflect a reinforcement of a residual carry-over response from t0. We conclude that the extrastriate symmetry-network can achieve amodal representation of symmetry from occluded objects that have been previously experienced as wholes.

Face-selective neurons are observed in the primate visual pathway and are considered as the basis of face detection in the brain. However, it has been debated as to whether this neuronal selectivity can arise innately or whether it requires training from visual experience. Here, using a hierarchical deep neural network model of the ventral visual stream, we suggest a mechanism in which face-selectivity arises in the complete absence of training. We found that units selective to faces emerge robustly in randomly initialized networks and that these units reproduce many characteristics observed in monkeys. This innate selectivity also enables the untrained network to perform face-detection tasks. Intriguingly, we observed that units selective to various non-face objects can also arise innately in untrained networks. Our results imply that the random feedforward connections in early, untrained deep neural networks may be sufficient for initializing primitive visual selectivity., Face-selective neurons are observed in the primate visual pathway and are considered as the basis of face detection in the brain. Here, using a hierarchical deep neural network model of the ventral visual stream, the authors suggest that face selectivity arises in the complete absence of training.

Early theories of efficient coding suggested the visual system could compress the world by learning to represent features where information was concentrated, such as contours. This view was validated by the discovery that neurons in posterior visual cortex respond to edges and curvature. Still, it remains unclear what other information-rich features are encoded by neurons in more anterior cortical regions (e.g., inferotemporal cortex). Here, we use a generative deep neural network to synthesize images guided by neuronal responses from across the visuocortical hierarchy, using floating microelectrode arrays in areas V1, V4 and inferotemporal cortex of two macaque monkeys. We hypothesize these images (“prototypes”) represent such predicted information-rich features. Prototypes vary across areas, show moderate complexity, and resemble salient visual attributes and semantic content of natural images, as indicated by the animals’ gaze behavior. This suggests the code for object recognition represents compressed features of behavioral relevance, an underexplored aspect of efficient coding., Visual recognition depends on the ability to extract specific shape and colour features from complicated natural scenes. Here, the authors show that neurons along the object-recognition cortical pathway encode information-concentrating features of moderate complexity and of behavioural relevance.

In order to better understand how the brain perceives faces, it is important to know what objective drives learning in the ventral visual stream. To answer this question, we model neural responses to faces in the macaque inferotemporal (IT) cortex with a deep self-supervised generative model, β-VAE, which disentangles sensory data into interpretable latent factors, such as gender or age. Our results demonstrate a strong correspondence between the generative factors discovered by β-VAE and those coded by single IT neurons, beyond that found for the baselines, including the handcrafted state-of-the-art model of face perception, the Active Appearance Model, and deep classifiers. Moreover, β-VAE is able to reconstruct novel face images using signals from just a handful of cells. Together our results imply that optimising the disentangling objective leads to representations that closely resemble those in the IT at the single unit level. This points at disentangling as a plausible learning objective for the visual brain., Little is known about the brain’s computations that enable the recognition of faces. Here, the authors use unsupervised deep learning to show that the brain disentangles faces into semantically meaningful factors, like age or the presence of a smile, at the single neuron level.

To assess the distribution of axial length as surrogate for myopia and its determinants in an old population, we performed the Ural Very Old Study as a population-based cohort study. Out of 1882 eligible individuals aged 85 + years, the Ural Very Old Study performed in an urban and rural region in Bashkortostan/Russia included 1526 (81.1%) individuals undergoing ophthalmological and medical examinations with sonographic axial length measurement. Biometric data were available for 717 (47.0%) individuals with a mean age of 88.0 ± 2.6 years (range 85–98 years; 25%). Mean axial length was 23.1 ± 1.1 mm (range 19.37–28.89 mm). Prevalences of moderate myopia (axial length 24.5–2 0.25) with taller body height (standardized regression coefficient beta:0.16;non-standardized regression coefficient B: 0.02; 95% confidence interval (CI) 0.01, 0.03; P P = 0.002), and lower corneal refractive power (beta: − 0.35; B: − 0.23; 95% CI − 0.28, − 0.18; P P = 0.007) and lower corneal refractive power (OR 0.77; 95% CI 0.63, 0.94; P = 0.01). In this old study population, prevalence of moderate axial myopia (6.6% versus 9.7%) was lower than, and prevalence of high axial myopia (1.4% versus 1.4%) was similar as, in a corresponding study on a younger population from the same Russian region. Both myopia prevalence rates were higher than in rural Central India (1.5% and 0.4%, respectively). As in other, younger, populations, axial length and moderate myopia prevalence increased with higher educational level, while high myopia prevalence was independent of the educational level.

We review and revisit the predictive processing inspired "Gestalts as predictions" hypothesis. The study of Gestalt phenomena at and below threshold can help clarify the role of higher-order object selective areas and feedback connections in mid-level vision. In two psychophysical experiments assessing manipulations of contrast and configurality we showed that: (1) Gestalt phenomena are robust against saliency manipulations across the psychometric function even below threshold (with the accuracy gains and higher saliency associated with Gestalts being present even around chance performance); and (2) peak differences between Gestalt and control conditions happened around the time where responses to Gestalts are starting to saturate (mimicking the differential contrast response profile of striate vs. extra-striate visual neurons). In addition, Gestalts are associated with steeper psychometric functions in all experiments. We propose that these results reflect the differential engagement of object-selective areas in Gestalt phenomena and of information- or percept-based processing, as opposed to energy- or stimulus-based processing, more generally. In addition, the presence of nonlinearities in the psychometric functions suggest differential top-down modulation of the early visual cortex. We treat this as a proof of principle study, illustrating that classic psychophysics can help assess possible involvement of hierarchical predictive processing in Gestalt phenomena. ispartof: SCIENTIFIC REPORTS vol:11 issue:1 ispartof: location:England status: published

Though the range of invariance in recognition of novel objects is a basic aspect of human vision, its characterization has remained surprisingly elusive. Here we report tolerance to scale and position changes in one-shot learning by measuring recognition accuracy of Korean letters presented in a flash to non-Korean subjects who had no previous experience with Korean letters. We found that humans have significant scale-invariance after only a single exposure to a novel object. The range of translation-invariance is limited, depending on the size and position of presented objects. To understand the underlying brain computation associated with the invariance properties, we compared experimental data with computational modeling results. Our results suggest that to explain invariant recognition of objects by humans, neural network models should explicitly incorporate built-in scale-invariance, by encoding different scale channels as well as eccentricity-dependent representations captured by neurons’ receptive field sizes and sampling density that change with eccentricity. Our psychophysical experiments and related simulations strongly suggest that the human visual system uses a computational strategy that differs in some key aspects from current deep learning architectures, being more data efficient and relying more critically on eye-movements.

3-D gait analysis is the gold standard but many healthcare clinics and research institutes would benefit from a system that is inexpensive and simple but just as accurate. The present study examines whether a low-cost 2-D motion capture system can accurately and reliably assess adaptive gait kinematics in subjects with central vision loss, older controls, and younger controls. Subjects were requested to walk up and step over a 10 cm high obstacle that was positioned in the middle of a 4.5 m walkway. Four trials were simultaneously recorded with the Vicon motion capture system (3-D system) and a video camera that was positioned perpendicular to the obstacle (2-D system). The kinematic parameters (crossing height, crossing velocity, foot placement, single support time) were calculated offline. Strong Pearson’s correlations were found between the two systems for all parameters (average r = 0.944, all p

The human brain has evolved specialised mechanisms to enable the rapid detection of threat cues, including emotional face expressions (e.g., fear and anger). However, contextual cues – such as gaze direction – influence the ability to recognise emotional expressions. For instance, anger paired with direct gaze, and fear paired with averted gaze are more accurately recognised compared to alternate conjunctions of these features. It is argued that this is because gaze direction conveys the relevance and locus of the threat to the observer. Here, we used continuous flash suppression (CFS) to assess whether the modulatory effect of gaze direction on emotional face processing occurs outside of conscious awareness. Previous research using CFS has demonstrated that fearful facial expressions are prioritised by the visual system and gain privileged access to awareness over other expressed emotions. We hypothesised that if the modulatory effects of gaze on emotional face processing occur also at this level, then the gaze-emotion conjunctions signalling self-relevant threat will reach awareness faster than those that do not. We report that fearful faces gain privileged access to awareness over angry faces, but that gaze direction does not modulate this effect. Thus, our findings suggest that previously reported effects of gaze direction on emotional face processing are likely to occur once the face is detected, where the self-relevance and locus of the threat can be consciously appraised.

Although the neuropeptide oxytocin (OT) is thought to regulate prosocial behavior in mammals, there is considerable debate as to how intranasal OT influences primate behavior. The aim of this study was to determine whether intranasal OT has a general anxiolytic effect on the performance of rhesus monkeys tasked with matching face stimuli, or a more selective effect on their behavior towards aversive facial expressions. To this end, we developed an innovative delayed match-to-sample task where the exact same trials could be used to assess either a monkey’s ability to match facial expressions or facial identities. If OT has a general affect on behavior, then performance in both tasks should be altered by the administration of OT. We tested four male rhesus monkeys (Macaca mulatta) in both the expression and identity task after the intranasal administration of either OT or saline in a within-subjects design. We found that OT inhalation selectively reduced a selection bias against negatively valenced expressions. Based on the same visual input, performance in the identity task was also unaffected by OT. This dissociation provides evidence that intranasal OT affects primate behavior under very particular circumstances, rather than acting as a general anxiolytic, in a highly translatable nonhuman model, the rhesus monkey.

When a material is stretched along a spatial axis, it is causally compressed along the orthogonal axis, as quantified in the Poisson effect. The present study examined how human observers assess this causality. Stimuli were video clips of a white rectangular region that was horizontally stretched while it was vertically compressed, with spatially sinusoidal modulation of the magnitude of vertical compressions. It was found that the Poisson’s ratio—a well-defined index of the Poisson effect—was not an explanatory factor for the degree of reported causality. Instead, reported causality was explained by image features related to deformation magnitudes. Comparing a material’s shape before and after deformation was not always required for the causality assessment. This suggests that human observers determine causality in the Poisson effect by using heuristics based on image features not necessarily related to the physical properties of the material.

Deep learning (DL) is a subset of artificial intelligence (AI), which uses multilayer neural networks modelled after the mammalian visual cortex capable of synthesizing images in ways that will transform the field of glaucoma. Autonomous DL algorithms are capable of maximizing information embedded in digital fundus photographs and ocular coherence tomographs to outperform ophthalmologists in disease detection. Other unsupervised algorithms such as principal component analysis (axis learning) and archetypal analysis (corner learning) facilitate visual field interpretation and show great promise to detect functional glaucoma progression and differentiate it from non-glaucomatous changes when compared with conventional software packages. Forecasting tools such as the Kalman filter may revolutionize glaucoma management by accounting for a host of factors to set target intraocular pressure goals that preserve vision. Activation maps generated from DL algorithms that process glaucoma data have the potential to efficiently direct our attention to critical data elements embedded in high throughput data and enhance our understanding of the glaucomatous process. It is hoped that AI will realize more accurate assessment of the copious data encountered in glaucoma management, improving our understanding of the disease, preserving vision, and serving to enhance the deep bonds that patients develop with their treating physicians.摘要: 深度学习 (Deep learning,DL) 是人工智能的一个分支, 通过模仿哺乳动物视皮层合成影像的能力, 建立多层神经网络模型。 这种技术会在青光眼领域起到变革的作用。自主DL算法能够最大化地收集眼底图像和OCT中包含的信息, 在疾病探查方面甚至能够超越眼科医生。其他的无监管算法例如主成分分析 (纵学习) 和原型分析 (角点学习) 有助于对视野结果进行解释, 并且与传统软件包相比能够更好地检测青光眼功能性进展, 并与非青光眼相鉴别。此外, 如卡尔曼滤波器等预测性工具可收录一系列影响因素后确定维持视力的目标眼压值, 从而彻底改变了青光眼的管理。DL算法通过处理青光眼数据可生成激活图, 引导我们关注高通量数据中嵌入的关键数据元素, 并加强我们对青光眼发展过程的理解。最后, 希望AI能够更加精准地评估青光眼治疗管理过程中的大量数据, 提高我们对青光眼、保护视力的认识, 成为患者与医生之间的深厚纽带。.

What is color vision for? Here we compared the extent to which memory modulates color appearance of objects and faces. Participants matched the colors of stimuli illuminated by low-pressure sodium light, which renders scenes monochromatic. Matches for fruit were not predicted by stimulus identity. In contrast, matches for faces were predictable, but surprising: faces appeared green and looked sick. The paradoxical face-color percept could be explained by a Bayesian observer model constrained by efficient coding. The color-matching data suggest that the face-color prior is established by visual signals arising from the recently evolved L-M cone system, not the older S-cone channel. Taken together, the results show that when retinal mechanisms of color vision are impaired, the impact of memory on color perception is greatest for face color, supporting the idea that trichromatic color plays an important role in social communication., What is the function of color vision? Here, the authors show that when retinal mechanisms of color are impaired, memory has a paradoxical impact on color appearance that is selective for faces, providing evidence that color contributes to face encoding and social communication.

A puzzle for neuroscience—and robotics—is how insects achieve surprisingly complex behaviours with such tiny brains. One example is depth perception via binocular stereopsis in the praying mantis, a predatory insect. Praying mantids use stereopsis, the computation of distances from disparities between the two retinal images, to trigger a raptorial strike of their forelegs when prey is within reach. The neuronal basis of this ability is entirely unknown. Here we show the first evidence that individual neurons in the praying mantis brain are tuned to specific disparities and eccentricities, and thus locations in 3D-space. Like disparity-tuned cortical cells in vertebrates, the responses of these mantis neurons are consistent with linear summation of binocular inputs followed by an output nonlinearity. Our study not only proves the existence of disparity sensitive neurons in an insect brain, it also reveals feedback connections hitherto undiscovered in any animal species., The praying mantis, a predatory insect, estimates depth via binocular vision. In this way, the animal decides whether prey is within reach. Here, the authors explore the neural correlates of binocular distance estimation and report that individual neurons are tuned to specific locations in 3D space.

The visual hierarchy of the ventral stream has been widely studied. However, it remains unclear how the hierarchical system organizes its functional coupling during top-down cognitive process. The present fMRI study investigated task-dependent functional connectivity along the ventral stream, while twenty-eight participants performed object recognition tasks that required different types of visual processing: i) searching or ii) memorizing visual objects embedded in natural scene images or iii) free viewing of the same images. Utilizing a seed-based approach that explicitly compared task-specific BOLD time-series, we identified task-dependent functional connectivity of the visual ventral stream, demonstrating different correlation structures. Searching for a target object manifested both correlated and anti-correlated structures, separating the visual areas V1 and V4 from the posterior part of the inferior temporal cortex (PIT). In contrast, the ventral stream structure remained correlated during memorizing objects, but increased the correlation between the right V4 and PIT. On the other hand, V1 and V4 showed task-dependent activation, whereas PIT was deactivated. These results highlight the context-dependent nature of the visual ventral stream and shed light on how the visual hierarchy is selectively organized to bias object recognition toward features of interest.

Spatial processing by receptive fields is a core property of the visual system. However, it is unknown how spatial processing in high-level regions contributes to recognition behavior. As face inversion is thought to disrupt typical holistic processing of information in faces, we mapped population receptive fields (pRFs) with upright and inverted faces in the human visual system. Here we show that in face-selective regions, but not primary visual cortex, pRFs and overall visual field coverage are smaller and shifted downward in response to face inversion. From these measurements, we successfully predict the relative behavioral detriment of face inversion at different positions in the visual field. This correspondence between neural measurements and behavior demonstrates how spatial processing in face-selective regions may enable holistic perception. These results not only show that spatial processing in high-level visual regions is dynamically used towards recognition, but also suggest a powerful approach for bridging neural computations by receptive fields to behavior., It is unknown whether spatial processing in the ventral (‘what’) stream contributes to high-level visual recognition. Here the authors show that spatial processing in face-selective regions directly contributes to whole face recognition behavior.

A central regularity of visual perception is the co-occurrence of objects in the natural environment. Here we use machine learning and fMRI to test the hypothesis that object co-occurrence statistics are encoded in the human visual system and elicited by the perception of individual objects. We identified low-dimensional representations that capture the latent statistical structure of object co-occurrence in real-world scenes, and we mapped these statistical representations onto voxel-wise fMRI responses during object viewing. We found that cortical responses to single objects were predicted by the statistical ensembles in which they typically occur, and that this link between objects and their visual contexts was made most strongly in parahippocampal cortex, overlapping with the anterior portion of scene-selective parahippocampal place area. In contrast, a language-based statistical model of the co-occurrence of object names in written text predicted responses in neighboring regions of object-selective visual cortex. Together, these findings show that the sensory coding of objects in the human brain reflects the latent statistics of object context in visual and linguistic experience., When people view an object, they can often guess the setting from which it was drawn and the other objects that might be found in that setting. Here the authors identify regions of the human visual system that represent this information about which objects tend to appear together in the world.

We effortlessly interact with objects in our environment, but how do we know where something is? An object’s apparent position does not simply correspond to its retinotopic location but is influenced by its surrounding context. In the natural environment, this context is highly complex, and little is known about how visual information in a scene influences the apparent location of the objects within it. We measured the influence of local image statistics (luminance, edges, object boundaries, and saliency) on the reported location of a brief target superimposed on images of natural scenes. For each image statistic, we calculated the difference between the image value at the physical center of the target and the value at its reported center, using observers’ cursor responses, and averaged the resulting values across all trials. To isolate image-specific effects, difference scores were compared to a randomly-permuted null distribution that accounted for any response biases. The observed difference scores indicated that responses were significantly biased toward darker regions, luminance edges, object boundaries, and areas of high saliency, with relatively low shared variance among these measures. In addition, we show that the same image statistics were associated with observers’ saccade errors, despite large differences in response time, and that some effects persisted when high-level scene processing was disrupted by 180° rotations and color negatives of the originals. Together, these results provide evidence for landmark effects within natural images, in which feature location reports are pulled toward low- and high-level informative content in the scene.