V1 Neurons Research Articles

Visuomotor Mismatch Responses as a Hallmark of Explaining Away in Causal Inference.

How are visuomotor mismatch responses in primary visual cortex embedded into cortical processing? We here show that mismatch responses can be understood as the result of a cooperation of motor and visual areas to jointly explain optic flow. This cooperation requires that optic flow is not explained redundantly by both areas, meaning that optic flow inputs to V1 that are predictable from motor neurons should be canceled (i.e., explained away). As a result, neurons in V1 represent only external causes of optic flow, which could allow the animal to easily detect movements that are independent of its own locomotion. We implement the proposed model in a spiking neural network, where coding errors are computed in dendrites and synaptic weights are learned with voltage-dependent plasticity rules. We find that both positive and negative mismatch responses arise, providing an alternative to the prevailing idea that visuomotor mismatch responses are linked to dedicated neurons for error computation. These results also provide a new perspective on several other recent observations of cross-modal neural interactions in cortex.

Haploinsufficiency of Shank3 increases the orientation selectivity of V1 neurons

Autism spectrum disorder (ASD) is a neurodevelopmental disorder whose hallmarks are social deficits, language impairment, repetitive behaviors, and sensory alterations. It has been reported that patients with ASD show differential activity in cortical regions, for instance, increased neuronal activity in visual processing brain areas and atypical visual perception compared with healthy subjects. The causes of these alterations remain unclear, although many studies demonstrate that ASD has a strong genetic correlation. An example is Phelan–McDermid syndrome, caused by a deletion of the Shank3 gene in one allele of chromosome 22. However, the neuronal consequences relating to the haploinsufficiency of Shank3 in the brain remain unknown. Given that sensory abnormalities are often present along with the core symptoms of ASD, our goal was to study the tuning properties of the primary visual cortex to orientation and direction in awake, head-fixed Shank3+/− mice. We recorded neural activity in vivo in response to visual gratings in the primary visual cortex from a mouse model of ASD (Shank3+/− mice) using the genetically encoded calcium indicator GCaMP6f, imaged with a two-photon microscope through a cranial window. We found that Shank3+/− mice showed a higher proportion of neurons responsive to drifting gratings stimuli than wild-type mice. Shank3+/− mice also show increased responses to some specific stimuli. Furthermore, analyzing the distributions of neurons for the tuning width, we found that Shank3+/− mice have narrower tuning widths, which was corroborated by analyzing the orientation selectivity. Regarding this, Shank3+/− mice have a higher proportion of selective neurons, specifically neurons showing increased selectivity to orientation but not direction. Thus, the haploinsufficiency of Shank3 modified the neuronal response of the primary visual cortex.

Rapid development of motion-streak coding in the mouse visual cortex

Despite its importance, the development of higher visual areas (HVAs) at the cellular resolution remains largely unknown. Here, we conducted 2-photon calcium imaging of mouse HVAs lateromedial (LM) and anterolateral (AL) and V1 to observe developmental changes in visual response properties. HVA neurons showed selectivity for orientations and directions similar to V1 neurons at eye opening, which became sharper in the following weeks. Neurons in all areas over all developmental stages tended to respond selectively to dots moving along an axis perpendicular to their preferred orientation at slow speeds, suggesting a certain level of conventional motion coding already at eye opening. In contrast, at high speeds, many neurons responded to dots moving along the axis parallel to the preferred orientation in older animals but rarely after eye opening, indicating a lack of motion-streak coding in the earlier stage. Together, our results uncover the development of visual properties in HVAs.

A two-photon imaging study on the roles of macaque V1 and V4 in texture segregation

The segregation of visual scenes on the basis of texture differences is a fundamental process in early/middle vision. Previous studies are inconsistent on whether V1 neurons could signal texture based figures from grounds (Lamme, 1995), or simply detect local feature discontinuities through surround modulation (Rossi_et_al., 2001). Here we used two-photon calcium imaging to study neuronal activities associated with texture-based figure-ground stimuli in V1, as well as in V4, in awake fixating macaques. Responses to parafoveal texture stimuli were obtained from 1237 orientation-tuned V1 neurons in two macaques, and from 1049 orientation-tuned V4 neurons in two additional macaques. The figure-ground texture stimuli consisted of a 32x32 arcdeg array of oriented lines in 4 orientations (ground), within which a 4x4 arcdeg square patch (figure) was formed by orthogonally orientated line segments. The center position of the figure patch was varied relative to the pRF of the recording site, so that the pRF fell within the figure, on the figure-ground boundary, or within the ground. Responses to a single oriented bar (to identify orientation-tuned neurons and measure tuning functions) and the uniform textures were also recorded. V1 neurons’ responses to texture stimuli were suppressed compared to neurons’ peak responses to an oriented bar. Slightly less suppression was observed when the figure-ground boundary fell on the pRF, regardless of neurons’ orientation preferences. Similar effects were not observed in V4 neurons, likely because larger V4 RFs made it difficult to align with a specific part of the texture stimuli. However, V4 neurons that preferred higher SFs (≥4 cpd) responded strongly to the figure-ground boundary oriented in the neurons’ preferred orientation. These results suggest that V1 neurons may weakly signal the figure-ground boundary by detecting the local feature discontinuities, while V4 neurons participate in texture segregation by detecting the second-order figure-ground boundary.

Binocular viewing geometry shapes neural processing of slanted planes: Results from theoretical V1 modeling and human psychophysics

An important component of real-world scene perception is the estimation of 3D surface orientation (slant and tilt). However, neither the neural circuits that compute 3D orientation nor the supporting computations have been fully uncovered. To help address that gap, we apply a projective geometry framework to the study of both the neural processing and perception of 3D orientation, building on recent work that suggests the mapping of the 3D environment onto the two retinae shapes both the tuning of neurons in area MT to motion-in-depth and corresponding patterns of perceptual errors in human psychophysics (Bonnen et al., 2020). The projection of slanted planes results in interocular disparities in retinal orientation, and interestingly, binocular V1 neurons can have different monocular orientation preferences (Bridge & Cumming, 2001), which could make them sensitive to orientation disparities. Therefore, we have constructed an encoding model that predicts responses of binocular V1 neurons to the orientation disparities produced by a stimulus plane with varying slant, given each theoretical neuron’s monocular orientation preferences and ocular dominance level. The resulting slant tuning curves are noncanonical in shape, do not tile slant evenly, and are strongly affected by viewing distance. Notably, at greater viewing distances, extreme slants (beyond approximately +/- 70 degrees) are encoded by steep portions of the curves, but moderate and low slants are encoded by relatively flat portions. As expected based on those observations, our model population decoder performs best at extreme slants, and performance degrades as slant approaches zero and as viewing distance increases. We additionally observed these patterns in human perceptual performance on a slant discrimination task. These results support the theory that the neural processing of 3D scene components is shaped by projective geometry and represent a step toward advancing understanding of the computations supporting 3D orientation perception.

Determining how color and form are integrated within macaque V1 neurons through combined neurophysiology and computational modeling

Determining how color and form are integrated within macaque V1 neurons through combined neurophysiology and computational modeling

Long-term recordings from area V4 neurons and an accurately-predicting deep convolutional energy model reveal spatial, chromatic and temporal tuning properties under naturalistic conditions

Long-term recordings from area V4 neurons and an accurately-predicting deep convolutional energy model reveal spatial, chromatic and temporal tuning properties under naturalistic conditions

Strong tuning for stereoscopic depth indicates orientation-specific recurrent circuitry in tree shrew V1

Neurons in the primary visual cortex (V1) are tuned to specific disparities between the two retinal images, which form the neural substrate for stereoscopic vision. We show that V1 neurons in tree shrews, but not in mice, display highly selective responses to narrow ranges of disparity in random-dot stereograms. Surprisingly, V1 neurons in both species show similarly strong tuning to gratings of varying interocular phase differences. This stimulus-dependent dissociation of disparity tuning can be explained by a network model that combines both feedforward and recurrent connections. The features of the model connections are supported by cortical organizations specific to each species. We validate this model by identifying putative inhibitory neurons and confirming their predicted disparity tuning in both species. Together, our studies establish a foundation for using tree shrews in studying binocular vision and raise an exciting possibility of how cortical columns could be uniquely important in computing stereoscopic depth.

A data-based large-scale model for primary visual cortex enables brain-like robust and versatile visual processing.

We analyze visual processing capabilities of a large-scale model for area V1 that arguably provides the most comprehensive accumulation of anatomical and neurophysiological data to date. We find that this brain-like neural network model can reproduce a number of characteristic visual processing capabilities of the brain, in particular the capability to solve diverse visual processing tasks, also on temporally dispersed visual information, with remarkable robustness to noise. This V1 model, whose architecture and neurons markedly differ from those of deep neural networks used in current artificial intelligence (AI), such as convolutional neural networks (CNNs), also reproduces a number of characteristic neural coding properties of the brain, which provides explanations for its superior noise robustness. Because visual processing is substantially more energy efficient in the brain compared with CNNs in AI, such brain-like neural networks are likely to have an impact on future technology: as blueprints for visual processing in more energy-efficient neuromorphic hardware.

V1-bypassing suppression leads to direction-specific microsaccade modulation in visual coding and perception

Microsaccades play a critical role in refreshing visual information and have been shown to have direction-specific influences on human perception. However, the neural mechanisms underlying such direction-specific effects remains unknown. Here, we report the emergence of direction-specific microsaccade modulation in the middle layer of V2 but not in V1: responses of V2 neurons after microsaccades moved toward their receptive fields were stronger than those when microsaccades moved away. The decreased responses from V1 to V2, which are correlated with the amplitude of microsaccades away from receptive fields, suggest topographically location-specific suppression from an oculomotor source. Consistent with directional effects in V2, microsaccades function as a guide for monkeys’ behavior in a peripheral detection task; both can be explained by a dynamic neural network. Our findings suggest a V1-bypassing suppressive circuit for direction-specific microsaccade modulation in V2 and its functional influence on visual sensitivity, which highlights the optimal sampling nature of microsaccades.

Adaptive processing and perceptual learning in visual cortical areas V1 and V4

Neurons in visual cortical areas primary visual cortex (V1) and V4 are adaptive processors, influenced by perceptual task. This is reflected in their ability to segment the visual scene into task-relevant and task-irrelevant stimulus components and by changing their tuning to task-relevant stimulus properties according to the current top-down instruction. Differences between the information represented in each area were seen. While V1 represented detailed stimulus characteristics, V4 filtered the input from V1 to carry the binary information required for the two-alternative judgement task. Neurons in V1 were activated at locations where the behaviorally relevant stimulus was placed well outside the grating-mapped receptive field. By systematically following the development of the task-dependent signals over the course of perceptual learning, we found that neuronal selectivity for task-relevant information was initially seen in V4 and, over a period of weeks, subsequently in V1. Once the learned information was represented in V1, on any given trial, task-relevant information appeared initially in V1 responses, followed by a 12-ms delay in V4. We propose that the shifting representation of learned information constitutes a mechanism for systems consolidation of memory.

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.

Analysis based on neural representation of natural object surfaces to elucidate the mechanisms of a trained AlexNet model.

Analysis and understanding of trained deep neural networks (DNNs) can deepen our understanding of the visual mechanisms involved in primate visual perception. However, due to the limited availability of neural activity data recorded from various cortical areas, the correspondence between the characteristics of artificial and biological neural responses for visually recognizing objects remains unclear at the layer level of DNNs. In the current study, we investigated the relationships between the artificial representations in each layer of a trained AlexNet model (based on a DNN) for object classification and the neural representations in various levels of visual cortices such as the primary visual (V1), intermediate visual (V4), and inferior temporal cortices. Furthermore, we analyzed the profiles of the artificial representations at a single channel level for each layer of the AlexNet model. We found that the artificial representations in the lower-level layers of the trained AlexNet model were strongly correlated with the neural representation in V1, whereas the responses of model neurons in layers at the intermediate and higher-intermediate levels of the trained object classification model exhibited characteristics similar to those of neural activity in V4 neurons. These results suggest that the trained AlexNet model may gradually establish artificial representations for object classification through the hierarchy of its network, in a similar manner to the neural mechanisms by which afferent transmission beginning in the low-level features gradually establishes object recognition as signals progress through the hierarchy of the ventral visual pathway.

Ionic and morphological contributions to the variable gain of membrane responses in layer 2/3 pyramidal neurons of mouse primary visual cortex.

Many neuronal cell types exhibit a sliding scale of neuronal excitability in the subthreshold voltage range. This is due to a variable contribution of different voltage-gated ion channels, leading to scaling of input resistance (RN) as a function of membrane potential (Vm) and a voltage-dependent dynamic gain of neuronal responsiveness. In layer 2/3 pyramidal neurons within the primary visual cortex (V1), this response influences sensory processing by tightening neuronal tuning to preferred orientations, but the identity of the ionic conductances involved remains unknown. Here, we used in vitro physiological recordings in acute slices to identify the contributions of several voltage-dependent conductances to the dynamic gain of membrane responses in layer 2/3 pyramidal neurons in mouse primary visual cortex. We found that the steep voltage dependence of input resistance in these cells was mediated in part by a combination of persistent sodium, inwardly rectifying potassium, and hyperpolarization-activated nonselective cation channels. In addition, the steepness of the slope of the RN/Vm relationship was inversely correlated with the number of branches on the proximal apical dendrite. These data have uncovered physiological and morphological factors that underlie the scaling of membrane responses in L2/3 neurons of rodent V1. Regulation of these channels would serve as a mechanism of real-time neuromodulation of neuronal processing of sensory information.NEW & NOTEWORTHY Layer 2/3 pyramidal neurons in primary visual cortex scale subthreshold voltage responses with resting membrane potential because RN increases as Vm is depolarized. Here, we uncovered the voltage-dependent contributions of NaP, Kir, and HCN conductances toward this behavior, and we additionally demonstrated that the strength of the RN/Vm relationship is inversely correlated with proximal branching along the apical dendrite.

The neurons in mouse V1 show different degrees of spatial clustering

Higher mammals’ primary visual cortex exhibits columnar organization, where neurons with similar response preferences are clustered. In contrast, rodents are presumed to lack this fine-scale organization; their neurons appear to be randomly arranged, described as a salt-and-pepper map. However, recent studies suggested a weak but significant spatial clustering of tuning in the salt-and-pepper map, similar to columnar organization. Thus, the salt-and-pepper map possesses the characteristics of both columnar organization and random arrangement. This raises the question about whether this mixed organization is attributed to different types of neurons. Here, we examined the tuning of primary visual cortical neurons in awake mice with a two-photon calcium imaging dataset, which were released by Allen Institute MindScope Program. First, we demonstrated that neurons with similar response preferences were clustered by showing that neighboring neurons tended to have similar orientation and temporal frequency preferences. Then, we compared the clustering of tuning between simple cells and complex cells and found the clustering of tuning among simple cells was significantly more prominent than that among complex cells. Furthermore, the simple/complex cell classification correlated with the stability of neuronal response. Neurons with stable responses were arranged independent of their tuning similarity, whereas unstable neurons were clustered according to their tuning similarity. These findings might represent a balance between efficiency and robustness: relatively independent tuning among stable neurons represents visual information efficiently, whereas unstable neurons with similar response preferences are clustered to obtain a robust representation with population codes.

Early impairment of cortical circuit plasticity and connectivity in the 5XFAD Alzheimer’s disease mouse model

Genetic risk factors for neurodegenerative disorders, such as Alzheimer’s disease (AD), are expressed throughout the life span. How these risk factors affect early brain development and function remain largely unclear. Analysis of animal models with high constructive validity for AD, such as the 5xFAD mouse model, may provide insights on potential early neurodevelopmental effects that impinge on adult brain function and age-dependent degeneration. The 5XFAD mouse model over-expresses human amyloid precursor protein (APP) and presenilin 1 (PS1) harboring five familial AD mutations. It is unclear how the expression of these mutant proteins affects early developing brain circuits. We found that the prefrontal cortex (PFC) layer 5 (L5) neurons in 5XFAD mice exhibit transgenic APP overloading at an early post-weaning age. Impaired synaptic plasticity (long-term potentiation, LTP) was seen at 6–8 weeks age in L5 PFC circuit, which was correlated with increased intracellular APP. APP overloading was also seen in L5 pyramidal neurons in the primary visual cortex (V1) during the critical period of plasticity (4–5 weeks age). Whole-cell patch clamp recording in V1 brain slices revealed reduced intrinsic excitability of L5 neurons in 5XFAD mice, along with decreased spontaneous miniature excitatory and inhibitory inputs. Functional circuit mapping using laser scanning photostimulation (LSPS) combined with glutamate uncaging uncovered reduced excitatory synaptic connectivity onto L5 neurons in V1, and a more pronounced reduction in inhibitory connectivity, indicative of altered excitation and inhibition during VC critical period. Lastly, in vivo single-unit recording in V1 confirmed that monocular visual deprivation-induced ocular dominance plasticity during critical period was impaired in 5XFAD mice. Our study reveals plasticity deficits across multiple cortical regions and indicates altered early cortical circuit developmental trajectory as a result of mutant APP/PS1 over-expression.

New Evidence of Central Nervous System Damage in Diabetes: Impairment of Fine Visual Discrimination.

Diabetes can damage both the peripheral sensory organs, causing retinopathy, and the central visual system, leading to contrast sensitivity and impaired color vision in patients without retinopathy. Orientation discrimination is important for shape recognition by the visual system. Our psychophysical findings in this study show diminished orientation discrimination in patients with diabetes without retinopathy. To reveal the underlying mechanism, we established a diabetic mouse model and recorded in vivo electrophysiological data in the dorsal lateral geniculate nucleus (dLGN) and primary visual cortex (V1). Reduced orientation selectivity was observed in both individual and populations of neurons in V1 and dLGN, which increased in severity with disease duration. This diabetes-associated neuronal dysfunction appeared earlier in the V1 than dLGN. Additionally, neuronal activity and signal-to-noise ratio are reduced in V1 neurons of diabetic mice, leading to a decreased capacity for information processing by V1 neurons. Notably, the V1 in diabetic mice exhibits reduced excitatory neuronal activity and lower levels of phosphorylated mammalian target of rapamycin (mTOR). Our findings show that altered responses of both populations of and single V1 neurons may impair fine vision, thus expanding our understanding of the underlying causes of diabetes-related impairment of the central nervous system.

Task Context Modulates Feature-Selective Responses in Area V4.

Feature selectivity of visual cortical responses measured during passive fixation provides only a partial view of selectivity because it does not account for the influence of cognitive factors. Here we focus on primate area V4 and ask how neuronal tuning is modulated by task engagement. We investigated whether responses to colored shapes during active shape discrimination are simple, stimulus-agnostic, scaled versions of responses during passive fixation, akin to results from attentional studies. Alternatively, responses could be subject to stimulus-specific scaling, that is, responses to different stimuli are modulated differently, resulting in changes in underlying shape/color selectivity. Among 83 well-isolated V4 neurons in two male macaques, only a minority (16 of 83), which were weakly tuned to both shape and color, displayed responses during fixation and discrimination tasks that could be related by stimulus-agnostic scaling. The majority (67 of 83), which were strongly tuned to shape, color, or both, displayed stimulus-dependent response changes during discrimination. For some of these neurons (39 of 83), the shape or color of the stimulus dictated the magnitude of the change, and for others (28 of 83) it was the combination of stimulus shape and color. Importantly, for neurons with one strong and one weak tuning dimension, stimulus-dependent response changes during discrimination were associated with a relative increase in selectivity along the stronger tuning dimension, without changes in tuning peak. These results reveal that more strongly tuned V4 neurons may also be more flexible in their selectivity, and imbalances in selectivity are amplified during active task contexts.SIGNIFICANCE STATEMENT Tuning for stimulus features is typically characterized by recording responses during passive fixation, but cognitive factors, including attention, influence responses in visual cortex. To determine how behavioral engagement influences neuronal responses in area V4, we compared responses to colored shapes during passive fixation and active behavior. For a large fraction of neurons, differences in responses between passive fixation and active behavior depended on the identity of the visual stimulus. For a subgroup of strongly feature-selective neurons, this response modulation was associated with enhanced selectivity for one feature at the expense of selectivity for the other. Such flexibility in tuning strength could improve performance in tasks requiring active judgment of stimuli.

Model-based characterization of the selectivity of neurons in primary visual cortex.

Statistical models are increasingly being used to understand the complexity of stimulus selectivity in primary visual cortex (V1) in the context of complex time-varying stimuli, replacing averaging responses to simple parametric stimuli. Although such models often can more accurately reflect the computations performed by V1 neurons in more natural visual environments, they do not by themselves provide insight into V1 neural selectivity to basic stimulus features such as receptive field size, spatial frequency tuning, and phase invariance. Here, we present a battery of analyses that can be directly applied to encoding models to link complex encoding models to more interpretable aspects of stimulus selectivity. We apply this battery to nonlinear models of V1 neurons recorded in awake macaque during random bar stimuli. In linking model properties to more classical measurements, we demonstrate several novel aspects of V1 selectivity not available to simpler experimental measurements. For example, this approach reveals that individual spatiotemporal elements of the V1 models often have a smaller spatial scale than the neuron as a whole, resulting in nontrivial tuning to spatial frequencies. In addition, we propose measures of nonlinear integration that suggest that classical classifications of V1 neurons into simple versus complex cells will be spatial-frequency dependent. In total, rather than obfuscate classical characterizations of V1 neurons, model-based characterizations offer a means to more fully understand their selectivity, and link their classical tuning properties to their roles in more complex, natural, visual processing.NEW & NOTEWORTHY Visual neurons are increasingly being studied with more complex, natural visual stimuli, and increasingly complex models are necessary to characterize their response properties. Here, we describe a battery of analyses that relate these more complex models to classical characterizations. Using such model-based characterizations of V1 neurons furthermore yields several new insights into V1 processing not possible to capture in more classical means to measure their visual selectivity.

Figure-ground responsive fields of monkey V4 neurons estimated from natural image patches.

Neurons in visual area V4 modulate their responses depending on the figure-ground (FG) organization in natural images containing a variety of shapes and textures. To clarify whether the responses depend on the extents of the figure and ground regions in and around the classical receptive fields (CRFs) of the neurons, we estimated the spatial extent of local figure and ground regions that evoked FG-dependent responses (RF-FGs) in natural images and their variants. Specifically, we applied the framework of spike triggered averaging (STA) to the combinations of neural responses and human-marked segmentation images (FG labels) that represent the extents of the figure and ground regions in the corresponding natural image stimuli. FG labels were weighted by the spike counts in response to the corresponding stimuli and averaged over. The bias due to the nonuniformity of FG labels was compensated by subtracting the ensemble average of FG labels from the weighted average. Approximately 50% of the neurons showed effective RF-FGs, and a large number exhibited structures that were similar to those observed in virtual neurons with ideal FG-dependent responses. The structures of the RF-FGs exhibited a subregion responsive to a preferred side (figure or ground) around the CRF center and a subregion responsive to a non-preferred side in the surroundings. The extents of the subregions responsive to figure were smaller than those responsive to ground in agreement with the Gestalt rule. We also estimated RF-FG by an adaptive filtering (AF) method, which does not require spherical symmetry (whiteness) in stimuli. RF-FGs estimated by AF and STA exhibited similar structures, supporting the veridicality of the proposed STA. To estimate the contribution of nonlinear processing in addition to linear processing, we estimated nonlinear RF-FGs based on the framework of spike triggered covariance (STC). The analyses of the models based on STA and STC did not show inconsiderable contribution of nonlinearity, suggesting spatial variance of FG regions. The results lead to an understanding of the neural responses that underlie the segregation of figures and the construction of surfaces in intermediate-level visual areas.