Efficient coding of natural images in the mouse visual cortex.

How the activity of neurons gives rise to natural vision remains a matter of intense investigation. The mid-level visual areas along the ventral stream are selective to a common class of natural images-textures-but a circuit-level understanding of this selectivity and its link to perception remains unclear. We addressed these questions in mice, first showing that they can perceptually discriminate between textures and statistically simpler spectrally matched stimuli, and between texture types. Then, at the neural level, we found that the secondary visual area (LM) exhibited a higher degree of selectivity for textures compared to the primary visual area (V1). Furthermore, textures were represented in distinct neural activity subspaces whose relative distances were found to correlate with the statistical similarity of the images and the mice's ability to discriminate between them. Notably, these dependencies were more pronounced in LM, where the texture-related subspaces were smaller than in V1, resulting in superior stimulus decoding capabilities. Together, our results demonstrate texture vision in mice, finding a linking framework between stimulus statistics, neural representations, and perceptual sensitivity-a distinct hallmark of efficient coding computations.

Mouse visual cortex as a limited resource system that self-learns an ecologically-general representation.

Studies of the mouse visual system have revealed a variety of visual brain areas that are thought to support a multitude of behavioral capacities, ranging from stimulus-reward associations, to goal-directed navigation, and object-centric discriminations. However, an overall understanding of the mouse's visual cortex, and how it supports a range of behaviors, remains unknown. Here, we take a computational approach to help address these questions, providing a high-fidelity quantitative model of mouse visual cortex and identifying key structural and functional principles underlying that model's success. Structurally, we find that a comparatively shallow network structure with a low-resolution input is optimal for modeling mouse visual cortex. Our main finding is functional-that models trained with task-agnostic, self-supervised objective functions based on the concept of contrastive embeddings are much better matches to mouse cortex, than models trained on supervised objectives or alternative self-supervised methods. This result is very much unlike in primates where prior work showed that the two were roughly equivalent, naturally leading us to ask the question of why these self-supervised objectives are better matches than supervised ones in mouse. To this end, we show that the self-supervised, contrastive objective builds a general-purpose visual representation that enables the system to achieve better transfer on out-of-distribution visual scene understanding and reward-based navigation tasks. Our results suggest that mouse visual cortex is a low-resolution, shallow network that makes best use of the mouse's limited resources to create a light-weight, general-purpose visual system-in contrast to the deep, high-resolution, and more categorization-dominated visual system of primates.

Gain, not concomitant changes in spatial receptive field properties, improves task performance in a neural network attention model.

Attention allows us to focus sensory processing on behaviorally relevant aspects of the visual world. One potential mechanism of attention is a change in the gain of sensory responses. However, changing gain at early stages could have multiple downstream consequences for visual processing. Which, if any, of these effects can account for the benefits of attention for detection and discrimination? Using a model of primate visual cortex we document how a Gaussian-shaped gain modulation results in changes to spatial tuning properties. Forcing the model to use only these changes failed to produce any benefit in task performance. Instead, we found that gain alone was both necessary and sufficient to explain category detection and discrimination during attention. Our results show how gain can give rise to changes in receptive fields which are not necessary for enhancing task performance.

What has vision science taught us about functional MRI?

In the domain of human neuroimaging, much attention has been paid to the question of whether and how the development of functional magnetic resonance imaging (fMRI) has advanced our scientific knowledge of the human brain. However, the opposite question is also important; how has our knowledge of the brain advanced our understanding of fMRI? Here, we discuss how and why scientific knowledge about the human and animal visual system has been used to answer fundamental questions about fMRI as a brain measurement tool and how these answers have contributed to scientific discoveries beyond vision science.

Texture-like representation of objects in human visual cortex.

The human visual ability to recognize objects and scenes is widely thought to rely on representations in category-selective regions of the visual cortex. These representations could support object vision by specifically representing objects, or, more simply, by representing complex visual features regardless of the particular spatial arrangement needed to constitute real-world objects, that is, by representing visual textures. To discriminate between these hypotheses, we leveraged an image synthesis approach that, unlike previous methods, provides independent control over the complexity and spatial arrangement of visual features. We found that human observers could easily detect a natural object among synthetic images with similar complex features that were spatially scrambled. However, observer models built from BOLD responses from category-selective regions, as well as a model of macaque inferotemporal cortex and Imagenet-trained deep convolutional neural networks, were all unable to identify the real object. This inability was not due to a lack of signal to noise, as all observer models could predict human performance in image categorization tasks. How then might these texture-like representations in category-selective regions support object perception? An image-specific readout from category-selective cortex yielded a representation that was more selective for natural feature arrangement, showing that the information necessary for natural object discrimination is available. Thus, our results suggest that the role of the human category-selective visual cortex is not to explicitly encode objects but rather to provide a basis set of texture-like features that can be infinitely reconfigured to flexibly learn and identify new object categories.

Increasing neural network robustness improves match to macaque V1 eigenspectrum, spatial frequency preference and predictivity.

Task-optimized convolutional neural networks (CNNs) show striking similarities to the ventral visual stream. However, human-imperceptible image perturbations can cause a CNN to make incorrect predictions. Here we provide insight into this brittleness by investigating the representations of models that are either robust or not robust to image perturbations. Theory suggests that the robustness of a system to these perturbations could be related to the power law exponent of the eigenspectrum of its set of neural responses, where power law exponents closer to and larger than one would indicate a system that is less susceptible to input perturbations. We show that neural responses in mouse and macaque primary visual cortex (V1) obey the predictions of this theory, where their eigenspectra have power law exponents of at least one. We also find that the eigenspectra of model representations decay slowly relative to those observed in neurophysiology and that robust models have eigenspectra that decay slightly faster and have higher power law exponents than those of non-robust models. The slow decay of the eigenspectra suggests that substantial variance in the model responses is related to the encoding of fine stimulus features. We therefore investigated the spatial frequency tuning of artificial neurons and found that a large proportion of them preferred high spatial frequencies and that robust models had preferred spatial frequency distributions more aligned with the measured spatial frequency distribution of macaque V1 cells. Furthermore, robust models were quantitatively better models of V1 than non-robust models. Our results are consistent with other findings that there is a misalignment between human and machine perception. They also suggest that it may be useful to penalize slow-decaying eigenspectra or to bias models to extract features of lower spatial frequencies during task-optimization in order to improve robustness and V1 neural response predictivity.

Voxel-Wise Linearity Analysis of Increments and Decrements in BOLD Responses in Human Visual Cortex Using a Contrast Adaptation Paradigm.

The linearity of BOLD responses is a fundamental presumption in most analysis procedures for BOLD fMRI studies. Previous studies have examined the linearity of BOLD signal increments, but less is known about the linearity of BOLD signal decrements. The present study assessed the linearity of both BOLD signal increments and decrements in the human primary visual cortex using a contrast adaptation paradigm. Results showed that both BOLD signal increments and decrements kept linearity to long stimuli (e.g., 3 s, 6 s), yet, deviated from linearity to transient stimuli (e.g., 1 s). Furthermore, a voxel-wise analysis showed that the deviation patterns were different for BOLD signal increments and decrements: while the BOLD signal increments demonstrated a consistent overestimation pattern, the patterns for BOLD signal decrements varied from overestimation to underestimation. Our results suggested that corrections to deviations from linearity of transient responses should consider the different effects of BOLD signal increments and decrements.

Population Models, Not Analyses, of Human Neuroscience Measurements.

Selectivity for many basic properties of visual stimuli, such as orientation, is thought to be organized at the scale of cortical columns, making it difficult or impossible to measure directly with noninvasive human neuroscience measurement. However, computational analyses of neuroimaging data have shown that selectivity for orientation can be recovered by considering the pattern of response across a region of cortex. This suggests that computational analyses can reveal representation encoded at a finer spatial scale than is implied by the spatial resolution limits of measurement techniques. This potentially opens up the possibility to study a much wider range of neural phenomena that are otherwise inaccessible through noninvasive measurement. However, as we review in this article, a large body of evidence suggests an alternative hypothesis to this superresolution account: that orientation information is available at the spatial scale of cortical maps and thus easily measurable at the spatial resolution of standard techniques. In fact, a population model shows that this orientation information need not even come from single-unit selectivity for orientation tuning, but instead can result from population selectivity for spatial frequency. Thus, a categorical error of interpretation can result whereby orientation selectivity can be confused with spatial frequency selectivity. This is similarly problematic for the interpretation of results from numerous studies of more complex representations and cognitive functions that have built upon the computational techniques used to reveal stimulus orientation. We suggest in this review that these interpretational ambiguities can be avoided by treating computational analyses as models of the neural processes that give rise to measurement. Building upon the modeling tradition in vision science using considerations of whether population models meet a set of core criteria is important for creating the foundation for a cumulative and replicable approach to making valid inferences from human neuroscience measurements.

Context effects on probability estimation.

Many decisions rely on how we evaluate potential outcomes and estimate their corresponding probabilities of occurrence. Outcome evaluation is subjective because it requires consulting internal preferences and is sensitive to context. In contrast, probability estimation requires extracting statistics from the environment and therefore imposes unique challenges to the decision maker. Here, we show that probability estimation, like outcome evaluation, is subject to context effects that bias probability estimates away from other events present in the same context. However, unlike valuation, these context effects appeared to be scaled by estimated uncertainty, which is largest at intermediate probabilities. Blood-oxygen-level-dependent (BOLD) imaging showed that patterns of multivoxel activity in the dorsal anterior cingulate cortex (dACC), ventromedial prefrontal cortex (VMPFC), and intraparietal sulcus (IPS) predicted individual differences in context effects on probability estimates. These results establish VMPFC as the neurocomputational substrate shared between valuation and probability estimation and highlight the additional involvement of dACC and IPS that can be uniquely attributed to probability estimation. Because probability estimation is a required component of computational accounts from sensory inference to higher cognition, the context effects found here may affect a wide array of cognitive computations.

Humans Perceive Binocular Rivalry and Fusion in a Tristable Dynamic State.

Human vision combines inputs from the two eyes into one percept. Small differences "fuse" together, whereas larger differences are seen "rivalrously" from one eye at a time. These outcomes are typically treated as mutually exclusive processes, with paradigms targeting one or the other and fusion being unreported in most rivalry studies. Is fusion truly a default, stable state that only breaks into rivalry for non-fusible stimuli? Or are monocular and fused percepts three sub-states of one dynamical system? To determine whether fusion and rivalry are separate processes, we measured human perception of Gabor patches with a range of interocular orientation disparities. Observers (10 female, 5 male) reported rivalrous, fused, and uncertain percepts over time. We found a dynamic "tristable" zone spanning from ∼25-35° of orientation disparity where fused, left-eye-, or right-eye-dominant percepts could all occur. The temporal characteristics of fusion and non-fusion periods during tristability matched other bistable processes. We tested statistical models with fusion as a higher-level bistable process alternating with rivalry against our findings. None of these fit our data, but a simple bistable model extended to have three states reproduced many of our observations. We conclude that rivalry and fusion are multistable substates capable of direct competition, rather than separate bistable processes.SIGNIFICANCE STATEMENT When inputs to the two eyes differ, they can either fuse together or engage in binocular rivalry, where each eye's view is seen exclusively in turn. Visual stimuli have often been tailored to produce either fusion or rivalry, implicitly treating them as separate mutually-exclusive perceptual processes. We have found that some similar-but-different stimuli can result in both outcomes over time. Comparing various simple models with our results suggests that rivalry and fusion are not independent processes, but compete within a single multistable system. This conceptual shift is a step toward unifying fusion and rivalry, and understanding how they both contribute to the visual system's production of a unified interpretation of the conflicting images cast on the retina by real-world scenes.

A flexible readout mechanism of human sensory representations.

Attention can both enhance and suppress cortical sensory representations. However, changing sensory representations can also be detrimental to behavior. Behavioral consequences can be avoided by flexibly changing sensory readout, while leaving the representations unchanged. Here, we asked human observers to attend to and report about either one of two features which control the visibility of motion while making concurrent measurements of cortical activity with BOLD imaging (fMRI). We extend a well-established linking model to account for the relationship between these measurements and find that changes in sensory representation during directed attention are insufficient to explain perceptual reports. Adding a flexible downstream readout is necessary to best explain our data. Such a model implies that observers should be able to recover information about ignored features, a prediction which we confirm behaviorally. Thus, flexible readout is a critical component of the cortical implementation of human adaptive behavior.

Computing Social Value Conversion in the Human Brain.

Social signals play powerful roles in shaping self-oriented reward valuation and decision making. These signals activate social and valuation/decision areas, but the core computation for their integration into the self-oriented decision machinery remains unclear. Here, we study how a fundamental social signal, social value (others' reward value), is converted into self-oriented decision making in the human brain. Using behavioral analysis, modeling, and neuroimaging, we show three-stage processing of social value conversion from the offer to the effective value and then to the final decision value. First, a value of others' bonus on offer, called offered value, was encoded uniquely in the right temporoparietal junction (rTPJ) and also in the left dorsolateral prefrontal cortex (ldlPFC), which is commonly activated by offered self-bonus value. The effective value, an intermediate value representing the effective influence of the offer on the decision, was represented in the right anterior insula (rAI), and the final decision value was encoded in the medial prefrontal cortex (mPFC). Second, using psychophysiological interaction and dynamic causal modeling analyses, we demonstrated three-stage feedforward processing from the rTPJ and ldPFC to the rAI and then from rAI to the mPFC. Further, we showed that these characteristics of social conversion underlie distinct sociobehavioral phenotypes. We demonstrate that the variability in the conversion underlies the difference between prosocial and selfish subjects, as seen from the differential strength of the rAI and ldlPFC coupling to the mPFC responses, respectively. Together, these findings identified fundamental neural computation processes for social value conversion underlying complex social decision making behaviors.SIGNIFICANCE STATEMENT In daily life, we make decisions based on self-interest, but also in consideration for others' status. These social influences modulate valuation and decision signals in the brain, suggesting a fundamental process called value conversion that translates social information into self-referenced decisions. However, little is known about the conversion process and its underlying brain mechanisms. We investigated value conversion using human fMRI with computational modeling and found three essential stages in a progressive brain circuit from social to empathic and decision areas. Interestingly, the brain mechanism of conversion differed between prosocial and individualistic subjects. These findings reveal how the brain processes and merges social information into the elemental flow of self-interested decision making.

Inverted Encoding Models Reconstruct an Arbitrary Model Response, Not the Stimulus.

Probing how large populations of neurons represent stimuli is key to understanding sensory representations as many stimulus characteristics can only be discerned from population activity and not from individual single-units. Recently, inverted encoding models have been used to produce channel response functions from large spatial-scale measurements of human brain activity that are reminiscent of single-unit tuning functions and have been proposed to assay "population-level stimulus representations" (Sprague et al., 2018a). However, these channel response functions do not assay population tuning. We show by derivation that the channel response function is only determined up to an invertible linear transform. Thus, these channel response functions are arbitrary, one of an infinite family and therefore not a unique description of population representation. Indeed, simulations demonstrate that bimodal, even random, channel basis functions can account perfectly well for population responses without any underlying neural response units that are so tuned. However, the approach can be salvaged by extending it to reconstruct the stimulus, not the assumed model. We show that when this is done, even using bimodal and random channel basis functions, a unimodal function peaking at the appropriate value of the stimulus is recovered which can be interpreted as a measure of population selectivity. More precisely, the recovered function signifies how likely any value of the stimulus is, given the observed population response. Whether an analysis is recovering the hypothetical responses of an arbitrary model rather than assessing the selectivity of population representations is not an issue unique to the inverted encoding model and human neuroscience, but a general problem that must be confronted as more complex analyses intervene between measurement of population activity and presentation of data.

Optimality and heuristics in perceptual neuroscience.

The foundation for modern understanding of how we make perceptual decisions about what we see or where to look comes from considering the optimal way to perform these behaviors. While statistical computation is useful for deriving the optimal solution to a perceptual problem, optimality requires perfect knowledge of priors and often complex computation. Accumulating evidence, however, suggests that optimal perceptual goals can be achieved or approximated more simply by human observers using heuristic approaches. Perceptual neuroscientists captivated by optimal explanations of sensory behaviors will fail in their search for the neural circuits and cortical processes that implement an optimal computation whenever that behavior is actually achieved through heuristics. This article provides a cross-disciplinary review of decision-making with the aim of building perceptual theory that uses optimality to set the computational goals for perceptual behavior but, through consideration of ecological, computational, and energetic constraints, incorporates how these optimal goals can be achieved through heuristic approximation.

A quantitative framework for motion visibility in human cortex.

Despite the central use of motion visibility to reveal the neural basis of perception, perceptual decision making, and sensory inference there exists no comprehensive quantitative framework establishing how motion visibility parameters modulate human cortical response. Random-dot motion stimuli can be made less visible by reducing image contrast or motion coherence, or by shortening the stimulus duration. Because each of these manipulations modulates the strength of sensory neural responses they have all been extensively used to reveal cognitive and other nonsensory phenomena such as the influence of priors, attention, and choice-history biases. However, each of these manipulations is thought to influence response in different ways across different cortical regions and a comprehensive study is required to interpret this literature. Here, human participants observed random-dot stimuli varying across a large range of contrast, coherence, and stimulus durations as we measured blood-oxygen-level dependent responses. We developed a framework for modeling these responses that quantifies their functional form and sensitivity across areas. Our framework demonstrates the sensitivity of all visual areas to each parameter, with early visual areas V1-V4 showing more parametric sensitivity to changes in contrast and V3A and the human middle temporal area to coherence. Our results suggest that while motion contrast, coherence, and duration share cortical representation, they are encoded with distinct functional forms and sensitivity. Thus, our quantitative framework serves as a reference for interpretation of the vast perceptual literature manipulating these parameters and shows that different manipulations of visibility will have different effects across human visual cortex and need to be interpreted accordingly. NEW & NOTEWORTHY Manipulations of motion visibility have served as a key tool for understanding the neural basis for visual perception. Here we measured human cortical response to changes in visibility across a comprehensive range of motion visibility parameters and modeled these with a quantitative framework. Our quantitative framework can be used as a reference for linking human cortical response to perception and underscores that different manipulations of motion visibility can have greatly different effects on cortical representation.

Task-dependent enhancement of facial expression and identity representations in human cortex.

What cortical mechanisms allow humans to easily discern the expression or identity of a face? Subjects detected changes in expression or identity of a stream of dynamic faces while we measured BOLD responses from topographically and functionally defined areas throughout the visual hierarchy. Responses in dorsal areas increased during the expression task, whereas responses in ventral areas increased during the identity task, consistent with previous studies. Similar to ventral areas, early visual areas showed increased activity during the identity task. If visual responses are weighted by perceptual mechanisms according to their magnitude, these increased responses would lead to improved attentional selection of the task-appropriate facial aspect. Alternatively, increased responses could be a signature of a sensitivity enhancement mechanism that improves representations of the attended facial aspect. Consistent with the latter sensitivity enhancement mechanism, attending to expression led to enhanced decoding of exemplars of expression both in early visual and dorsal areas relative to attending identity. Similarly, decoding identity exemplars when attending to identity was improved in dorsal and ventral areas. We conclude that attending to expression or identity of dynamic faces is associated with increased selectivity in representations consistent with sensitivity enhancement.

A Switching Observer for Human Perceptual Estimation.

Human perceptual inference has been fruitfully characterized as a normative Bayesian process in which sensory evidence and priors are multiplicatively combined to form posteriors from which sensory estimates can be optimally read out. We tested whether this basic Bayesian framework could explain human subjects' behavior in two estimation tasks in which we varied the strength of sensory evidence (motion coherence or contrast) and priors (set of directions or orientations). We found that despite excellent agreement of estimates mean and variability with a Basic Bayesian observer model, the estimate distributions were bimodal with unpredicted modes near the prior and the likelihood. We developed a model that switched between prior and sensory evidence rather than integrating the two, which better explained the data than the Basic and several other Bayesian observers. Our data suggest that humans can approximate Bayesian optimality with a switching heuristic that forgoes multiplicative combination of priors and likelihoods.

Inverted Encoding Models of Human Population Response Conflate Noise and Neural Tuning Width.

Channel-encoding models offer the ability to bridge different scales of neuronal measurement by interpreting population responses, typically measured with BOLD imaging in humans, as linear sums of groups of neurons (channels) tuned for visual stimulus properties. Inverting these models to form predicted channel responses from population measurements in humans seemingly offers the potential to infer neuronal tuning properties. Here, we test the ability to make inferences about neural tuning width from inverted encoding models. We examined contrast invariance of orientation selectivity in human V1 (both sexes) and found that inverting the encoding model resulted in channel response functions that became broader with lower contrast, thus apparently violating contrast invariance. Simulations showed that this broadening could be explained by contrast-invariant single-unit tuning with the measured decrease in response amplitude at lower contrast. The decrease in response lowers the signal-to-noise ratio of population responses that results in poorer population representation of orientation. Simulations further showed that increasing signal to noise makes channel response functions less sensitive to underlying neural tuning width, and in the limit of zero noise will reconstruct the channel function assumed by the model regardless of the bandwidth of single units. We conclude that our data are consistent with contrast-invariant orientation tuning in human V1. More generally, our results demonstrate that population selectivity measures obtained by encoding models can deviate substantially from the behavior of single units because they conflate neural tuning width and noise and are therefore better used to estimate the uncertainty of decoded stimulus properties.SIGNIFICANCE STATEMENT It is widely recognized that perceptual experience arises from large populations of neurons, rather than a few single units. Yet, much theory and experiment have examined links between single units and perception. Encoding models offer a way to bridge this gap by explicitly interpreting population activity as the aggregate response of many single neurons with known tuning properties. Here we use this approach to examine contrast-invariant orientation tuning of human V1. We show with experiment and modeling that due to lower signal to noise, contrast-invariant orientation tuning of single units manifests in population response functions that broaden at lower contrast, rather than remain contrast-invariant. These results highlight the need for explicit quantitative modeling when making a reverse inference from population response profiles to single-unit responses.

Adaptable history biases in human perceptual decisions.

When making choices under conditions of perceptual uncertainty, past experience can play a vital role. However, it can also lead to biases that worsen decisions. Consistent with previous observations, we found that human choices are influenced by the success or failure of past choices even in a standard two-alternative detection task, where choice history is irrelevant. The typical bias was one that made the subject switch choices after a failure. These choice history biases led to poorer performance and were similar for observers in different countries. They were well captured by a simple logistic regression model that had been previously applied to describe psychophysical performance in mice. Such irrational biases seem at odds with the principles of reinforcement learning, which would predict exquisite adaptability to choice history. We therefore asked whether subjects could adapt their irrational biases following changes in trial order statistics. Adaptability was strong in the direction that confirmed a subject's default biases, but weaker in the opposite direction, so that existing biases could not be eradicated. We conclude that humans can adapt choice history biases, but cannot easily overcome existing biases even if irrational in the current context: adaptation is more sensitive to confirmatory than contradictory statistics.