Orientation-cue invariant population responses to contrast-modulated and phase-reversed contour stimuli in macaque V1 and V2

Macaque V1 responses to 2nd-order contrast-modulated stimuli and the possible subcortical and cortical contributions

Natural images comprise contours and boundaries defined by 1st-order luminance-modulated (LM) cues that are readily encoded by V1 neurons, and 2nd-order contrast-modulated (CM) cues that carry local, but not over-the-space, luminance changes. The neurophysiological foundations for CM processing remain unsolved. Here we used two-photon calcium imaging to demonstrate that V1 superficial-layer neurons respond to both LM and CM gratings in awake, fixating, macaques, with overall LM responses stronger than CM responses. Furthermore, adaptation experiments revealed that LM responses were similarly suppressed by LM and CM adaptation, with moderately larger effects by iso-orientation adaptation than by orthogonal adaptation, suggesting that LM and CM orientation responses likely share a strong orientation-non-selective subcortical origin. In contrast, CM responses were substantially more suppressed by iso-orientation than by orthogonal LM and CM adaptation, likely suggesting stronger orientation-specific intracortical influences for CM responses than for LM responses, besides shared orientation-non-selective subcortical influences. These results thus may indicate a subcortical-to-V1 filter-rectify-filter mechanism for CM processing: Local luminance changes in CM stimuli are initially encoded by orientation-non-selective subcortical neurons, and the outputs are half-wave rectified, and then summed by V1 neurons to signal CM orientation, which may be further substantially refined by intracortical influences.

Retinal and Callosal Activity-Dependent Chandelier Cell Elimination Shapes Binocularity in Primary Visual Cortex

In mammals with binocular vision, integration of the left and right visual scene relies on information in the center visual field, which are relayed from each retina in parallel and merge in the primary visual cortex (V1) through the convergence of ipsi- and contralateral geniculocortical inputs as well as transcallosal projections between two visual cortices. The developmental assembly of this binocular circuit, especially the transcallosal pathway, remains incompletely understood. Using genetic methods in mice, we found that several days before eye-opening, retinal and callosal activities drive massive apoptosis of GABAergic chandelier cells (ChCs) in the binocular region of V1. Blockade of ChC elimination resulted in a contralateral eye-dominated V1 and deficient binocular vision. As pre-vision retinal activities convey the left-right organization of the visual field, their regulation of ChC density through the transcallosal pathway may prime a nascent binocular territory for subsequent experience-driven tuning during the post-vision critical period.

Contrast-modulated stimuli produce more superimposition and predominate perception when competing with comparable luminance-modulated stimuli during interocular grouping

Interocular grouping (IOG) is a binocular visual function that can arise during multi-stable perception. IOG perception was initiated using split-grating stimuli constructed from luminance (L), luminance-modulated noise (LM) and contrast-modulated noise (CM). In Experiment 1, three different visibility levels were used for L and LM (or first-order) stimuli, and compared to fixed-visibility CM (or second-order) stimuli. Eight binocularly normal participants indicated whether they perceived full horizontal or vertical gratings, superimposition, or other (piecemeal and eye-of-origin) percepts. CM stimuli rarely generated full IOG, but predominantly generated superimposition. In Experiment 2, Levelt's modified laws were tested for IOG in nine participants. Split-gratings presented to each eye contained different visibility LM gratings, or LM and CM gratings. The results for the LM-vs-LM conditions mostly followed the predictions of Levelt's modified laws, whereas the results for the LM-vs-CM conditions did not. Counterintuitively, when high-visibility LM and low-visibility CM split-gratings were used, high-visibility LM components did not predominate IOG perception. Our findings suggest that higher proportions of superimposition during CM-vs-CM viewing are due to binocular combination, rather than mutual inhibition. It implies that IOG percepts are more likely to be mediated at an earlier monocular, rather than a binocular stage. Our previously proposed conceptual framework for conventional binocular rivalry, which includes asymmetric feedback, visual saliency, or a combination of both (Skerswetat et al. Sci Rep 8:14432, 2018), might also account for IOG. We speculate that opponency neurons might mediate coherent percepts when dissimilar information separately enters the eyes.

Levelt's laws do not predict perception when luminance- and contrast-modulated stimuli compete during binocular rivalry

Incompatible patterns viewed by each of the two eyes can provoke binocular rivalry, a competition of perception. Levelt's first law predicts that a highly visible stimulus will predominate over a less visible stimulus during binocular rivalry. In a behavioural study, we made a counterintuitive observation: high visibility patterns do not always predominate over low visibility patterns. Our results show that none of Levelt's binocular rivalry laws hold when luminance-modulated (LM) patterns compete with contrast-modulated (CM) patterns. We discuss visual saliency, asymmetric feedback, and a combination of both as potential mechanisms to explain the CM versus LM findings. Competing orthogonal LM stimuli do follow Levelt's laws, whereas only the first two laws hold for competing CM stimuli. The current results provide strong psychophysical evidence for the existence of separate processing stages for LM and CM stimuli.

Beyond Rehabilitation of Acuity, Ocular Alignment, and Binocularity in Infantile Strabismus

Infantile strabismus impairs the perception of all attributes of the visual scene. High spatial frequency components are no longer visible, leading to amblyopia. Binocularity is altered, leading to the loss of stereopsis. Spatial perception is impaired as well as detection of vertical orientation, the fastest movements, directions of movement, the highest contrasts and colors. Infantile strabismus also affects other vision-dependent processes such as control of postural stability. But presently, rehabilitative therapies for infantile strabismus by ophthalmologists, orthoptists and optometrists are restricted to preventing or curing amblyopia of the deviated eye, aligning the eyes and, whenever possible, preserving or restoring binocular vision during the critical period of development, i.e., before ~10 years of age. All the other impairments are thus ignored; whether they may recover after strabismus treatment even remains unknown. We argue here that medical and paramedical professionals may extend their present treatments of the perceptual losses associated with infantile strabismus. This hypothesis is based on findings from fundamental research on visual system organization of higher mammals in particular at the cortical level. In strabismic subjects (as in normal-seeing ones), information about all of the visual attributes converge, interact and are thus inter-dependent at multiple levels of encoding ranging from the single neuron to neuronal assemblies in visual cortex. Thus if the perception of one attribute is restored this may help to rehabilitate the perception of other attributes. Concomitantly, vision-dependent processes may also improve. This could occur spontaneously, but still should be assessed and validated. If not, medical and paramedical staff, in collaboration with neuroscientists, will have to break new ground in the field of therapies to help reorganize brain circuitry and promote more comprehensive functional recovery. Findings from fundamental research studies in both young and adult patients already support our hypothesis and are reviewed here. For example, presenting different contrasts to each eye of a strabismic patient during training sessions facilitates recovery of acuity in the amblyopic eye as well as of 3D perception. Recent data also demonstrate that visual recoveries in strabismic subjects improve postural stability. These findings form the basis for a roadmap for future research and clinical development to extend presently applied rehabilitative therapies for infantile strabismus.

Revealing Detail along the Visual Hierarchy: Neural Clustering Preserves Acuity from V1 to V4

How primates perceive objects along with their detailed features remains a mystery. This ability to make fine visual discriminations depends upon a high-acuity analysis of spatial frequency (SF) along the visual hierarchy from V1 to inferotemporal cortex. By studying the transformation of SF across macaque parafoveal V1, V2, and V4, we discovered SF-selective functional domains in V4 encoding higher SFs up to 12 cycles/°. These intermittent higher-SF-selective domains, surrounded by domains encoding lower SFs, violate the inverse relationship between SF preference and retinal eccentricity. The neural activities of higher- and lower-SF domains correspond to local and global features, respectively, of the same stimuli. Neural response latencies in high-SF domains are around 10 ms later than in low-SF domains, consistent with the coarse-to-fine nature of perception. Thus, our finding of preserved resolution from V1 into V4, separated both spatially and temporally, may serve as a connecting link for detailed object representation.

More superimposition for contrast-modulated than luminance-modulated stimuli during binocular rivalry

Luminance-modulated noise (LM) and contrast-modulated noise (CM) gratings were presented with interocularly correlated, uncorrelated and anti-correlated binary noise to investigate their contributions to mixed percepts, specifically piecemeal and superimposition, during binocular rivalry. Stimuli were sine-wave gratings of 2 c/deg presented within 2 deg circular apertures. The LM stimulus contrast was 0.1 and the CM stimulus modulation depth was 1.0, equating to approximately 5 and 7 times detection threshold, respectively. Twelve 45 s trials, per noise configuration, were carried out. Fifteen participants with normal vision indicated via button presses whether an exclusive, piecemeal or superimposed percept was seen. For all noise conditions LM stimuli generated more exclusive visibility, and lower proportions of superimposition. CM stimuli led to greater proportions and longer periods of superimposition. For both stimulus types, correlated interocular noise generated more superimposition than did anti- or uncorrelated interocular noise. No significant effect of stimulus type (LM vs CM) or noise configuration (correlated, uncorrelated, anti-correlated) on piecemeal perception was found. Exclusive visibility was greater in proportion, and perceptual changes more numerous, during binocular rivalry for CM stimuli when interocular noise was not correlated. This suggests that mutual inhibition, initiated by non-correlated noise CM gratings, occurs between neurons processing luminance noise (first-order component), as well as those processing gratings (second-order component). Therefore, first- and second-order components can contribute to overall binocular rivalry responses. We suggest the addition of a new well to the current energy landscape model for binocular rivalry that takes superimposition into account.

Organization of spatial frequency in cat striate cortex

Primary visual cortex, the first cortical stage of visual information processing, is represented by diverse functional maps that demonstrate the selectivity for specific visual features such as spatial frequency (SF). Although the local organization of SF maps in cat area 17 (A17) has been largely investigated, the global arrangement remains elusive. To address this unclear aspect, we evaluated the organization of SF maps within A17 by intrinsic signal optical imaging and extracellular electrophysiological recording. Our results explicitly showed that SF organization in cat A17 displayed a global asymmetrical unimodal distribution. In particular, we found the highest SF preference within the global distribution concentrated around the horizontal meridian. These results significantly contribute to a more comprehensive understanding of the SF organization in visual cortex.

The divisive normalization model of V1 neurons: a comprehensive comparison of physiological data and model predictions

The physiological responses of simple and complex cells in the primary visual cortex (V1) have been studied extensively and modeled at different levels. At the functional level, the divisive normalization model (DNM; Heeger DJ. Vis Neurosci 9: 181-197, 1992) has accounted for a wide range of single-cell recordings in terms of a combination of linear filtering, nonlinear rectification, and divisive normalization. We propose standardizing the formulation of the DNM and implementing it in software that takes static grayscale images as inputs and produces firing rate responses as outputs. We also review a comprehensive suite of 30 empirical phenomena and report a series of simulation experiments that qualitatively replicate dozens of key experiments with a standard parameter set consistent with physiological measurements. This systematic approach identifies novel falsifiable predictions of the DNM. We show how the model simultaneously satisfies the conflicting desiderata of flexibility and falsifiability. Our key idea is that, while adjustable parameters are needed to accommodate the diversity across neurons, they must be fixed for a given individual neuron. This requirement introduces falsifiable constraints when this single neuron is probed with multiple stimuli. We also present mathematical analyses and simulation experiments that explicate some of these constraints.

Nonlinear Y-Like Receptive Fields in the Early Visual Cortex: An Intermediate Stage for Building Cue-Invariant Receptive Fields from Subcortical Y Cells

Many of the neurons in early visual cortex are selective for the orientation of boundaries defined by first-order cues (luminance) as well as second-order cues (contrast, texture). The neural circuit mechanism underlying this selectivity is still unclear, but some studies have proposed that it emerges from spatial nonlinearities of subcortical Y cells. To understand how inputs from the Y-cell pathway might be pooled to generate cue-invariant receptive fields, we recorded visual responses from single neurons in cat Area 18 using linear multielectrode arrays. We measured responses to drifting and contrast-reversing luminance gratings as well as contrast modulation gratings. We found that a large fraction of these neurons have nonoriented responses to gratings, similar to those of subcortical Y cells: they respond at the second harmonic (F2) to high-spatial frequency contrast-reversing gratings and at the first harmonic (F1) to low-spatial frequency drifting gratings ("Y-cell signature"). For a given neuron, spatial frequency tuning for linear (F1) and nonlinear (F2) responses is quite distinct, similar to orientation-selective cue-invariant neurons. Also, these neurons respond to contrast modulation gratings with selectivity for the carrier (texture) spatial frequency and, in some cases, orientation. Their receptive field properties suggest that they could serve as building blocks for orientation-selective cue-invariant neurons. We propose a circuit model that combines ON- and OFF-center cortical Y-like cells in an unbalanced push-pull manner to generate orientation-selective, cue-invariant receptive fields.

SIGNIFICANCE STATEMENT

A significant fraction of neurons in early visual cortex have specialized receptive fields that allow them to selectively respond to the orientation of boundaries that are invariant to the cue (luminance, contrast, texture, motion) that defines them. However, the neural mechanism to construct such versatile receptive fields remains unclear. Using multielectrode recording, we found a large fraction of neurons in early visual cortex with receptive fields not selective for orientation that have spatial nonlinearities like those of subcortical Y cells. These are strong candidates for building cue-invariant orientation-selective neurons; we present a neural circuit model that pools such neurons in an imbalanced "push-pull" manner, to generate orientation-selective cue-invariant receptive fields.

Mice Can Use Second-Order, Contrast-Modulated Stimuli to Guide Visual Perception

Visual processing along the primate ventral stream takes place in a hierarchy of areas, characterized by an increase in both complexity of neuronal preferences and invariance to changes of low-level stimulus attributes. A basic type of invariance is form-cue invariance, where neurons have similar preferences in response to first-order stimuli, defined by changes in luminance, and global features of second-order stimuli, defined by changes in texture or contrast. Whether in mice, a now popular model system for early visual processing, visual perception can be guided by second-order stimuli is currently unknown. Here, we probed mouse visual perception and neural responses in areas V1 and LM using various types of second-order, contrast-modulated gratings with static noise carriers. These gratings differ in their spatial frequency composition and thus in their ability to invoke first-order mechanisms exploiting local luminance features. We show that mice can transfer learning of a coarse orientation discrimination task involving first-order, luminance-modulated gratings to the contrast-modulated gratings, albeit with markedly reduced discrimination performance. Consistent with these behavioral results, we demonstrate that neurons in area V1 and LM are less responsive and less selective to contrast-modulated than to luminance-modulated gratings, but respond with broadly similar preferred orientations. We conclude that mice can, at least in a rudimentary form, use second-order stimuli to guide visual perception.

SIGNIFICANCE STATEMENT

To extract object boundaries in natural scenes, the primate visual system does not only rely on differences in local luminance but can also take into account differences in texture or contrast. Whether the mouse, which has a much simpler visual system, can use such second-order information to guide visual perception is unknown. Here we tested mouse perception of second-order, contrast-defined stimuli and measured their neural representations in two areas of visual cortex. We find that mice can use contrast-defined stimuli to guide visual perception, although behavioral performance and neural representations were less robust than for luminance-defined stimuli. These findings shed light on basic steps of feature extraction along the mouse visual cortical hierarchy, which may ultimately lead to object recognition.

Breaking cover: neural responses to slow and fast camouflage-breaking motion

Primates need to detect and recognize camouflaged animals in natural environments. Camouflage-breaking movements are often the only visual cue available to accomplish this. Specifically, sudden movements are often detected before full recognition of the camouflaged animal is made, suggesting that initial processing of motion precedes the recognition of motion-defined contours or shapes. What are the neuronal mechanisms underlying this initial processing of camouflaged motion in the primate visual brain? We investigated this question using intrinsic-signal optical imaging of macaque V1, V2 and V4, along with computer simulations of the neural population responses. We found that camouflaged motion at low speed was processed as a direction signal by both direction- and orientation-selective neurons, whereas at high-speed camouflaged motion was encoded as a motion-streak signal primarily by orientation-selective neurons. No population responses were found to be invariant to the camouflage contours. These results suggest that the initial processing of camouflaged motion at low and high speeds is encoded as direction and motion-streak signals in primate early visual cortices. These processes are consistent with a spatio-temporal filter mechanism that provides for fast processing of motion signals, prior to full recognition of camouflage-breaking animals.

Deep Neural Networks as a Computational Model for Human Shape Sensitivity

Theories of object recognition agree that shape is of primordial importance, but there is no consensus about how shape might be represented, and so far attempts to implement a model of shape perception that would work with realistic stimuli have largely failed. Recent studies suggest that state-of-the-art convolutional ‘deep’ neural networks (DNNs) capture important aspects of human object perception. We hypothesized that these successes might be partially related to a human-like representation of object shape. Here we demonstrate that sensitivity for shape features, characteristic to human and primate vision, emerges in DNNs when trained for generic object recognition from natural photographs. We show that these models explain human shape judgments for several benchmark behavioral and neural stimulus sets on which earlier models mostly failed. In particular, although never explicitly trained for such stimuli, DNNs develop acute sensitivity to minute variations in shape and to non-accidental properties that have long been implicated to form the basis for object recognition. Even more strikingly, when tested with a challenging stimulus set in which shape and category membership are dissociated, the most complex model architectures capture human shape sensitivity as well as some aspects of the category structure that emerges from human judgments. As a whole, these results indicate that convolutional neural networks not only learn physically correct representations of object categories but also develop perceptually accurate representational spaces of shapes. An even more complete model of human object representations might be in sight by training deep architectures for multiple tasks, which is so characteristic in human development.

Shape plays an important role in object recognition. Despite years of research, no models of vision could account for shape understanding as found in human vision of natural images. Given recent successes of deep neural networks (DNNs) in object recognition, we hypothesized that DNNs might in fact learn to capture perceptually salient shape dimensions. Using a variety of stimulus sets, we demonstrate here that the output layers of several DNNs develop representations that relate closely to human perceptual shape judgments. Surprisingly, such sensitivity to shape develops in these models even though they were never explicitly trained for shape processing. Moreover, we show that these models also represent categorical object similarity that follows human semantic judgments, albeit to a lesser extent. Taken together, our results bring forward the exciting idea that DNNs capture not only objective dimensions of stimuli, such as their category, but also their subjective, or perceptual, aspects, such as shape and semantic similarity as judged by humans.