The functional geometry of local and horizontal connections in a model of V1.

A mathematical model of interacting hypercolumns in primary visual cortex (V1) is presented that incorporates details concerning the geometry of local and long-range horizontal connections. Each hypercolumn is modeled as a network of interacting excitatory and inhibitory neural populations with orientation and spatial frequency preferences organized around a pair of pinwheels. The pinwheels are arranged on a planar lattice, reflecting the crystalline-like structure of cortex. Local interactions within a hypercolumn generate orientation and spatial frequency tuning curves, which are modulated by horizontal connections between different hypercolumns on the lattice. The symmetry properties of the local and long-range connections play an important role in determining the types of spontaneous activity patterns that can arise in cortex.     

A spherical model for orientation and spatial-frequency tuning in a cortical hypercolumn

A theory is presented of the way in which the hypercolumns in primary visual cortex (V1) are organized to detect important features of visual images, namely local orientation and spatial-frequency. Given the existence in V1 of dual maps for these features, both organized around orientation pinwheels, we constructed a model of a hypercolumn in which orientation and spatial-frequency preferences are represented by the two angular coordinates of a sphere. The two poles of this sphere are taken to correspond, respectively, to high and low spatial-frequency preferences. In Part I of the paper, we use mean-field methods to derive exact solutions for localized activity states on the sphere. We show how cortical amplification through recurrent interactions generates a sharply tuned, contrast-invariant population response to both local orientation and local spatial frequency, even in the case of a weakly biased input from the lateral geniculate nucleus (LGN). A major prediction of our model is that this response is non-separable with respect to the local orientation and spatial frequency of a stimulus. That is, orientation tuning is weaker around the pinwheels, and there is a shift in spatial-frequency tuning towards that of the closest pinwheel at non-optimal orientations. In Part II of the paper, we demonstrate that a simple feed-forward model of spatial-frequency preference, unlike that for orientation preference, does not generate a faithful representation when amplified by recurrent interactions in V1. We then introduce the idea that cortico-geniculate feedback modulates LGN activity to generate a faithful representation, thus providing a new functional interpretation of the role of this feedback pathway. Using linear filter theory, we show that if the feedback from a cortical cell is taken to be approximately equal to the reciprocal of the corresponding feed-forward receptive field (in the two-dimensional Fourier domain), then the mismatch between the feed-forward and cortical frequency representations is eliminated. We therefore predict that cortico-geniculate feedback connections innervate the LGN in a pattern determined by the orientation and spatial-frequency biases of feed-forward receptive fields. Finally, we show how recurrent cortical interactions can generate cross-orientation suppression.     

Parsimony,Exhaustivity and Balanced Detection in Neocortex

The layout of sensory brain areas is thought to subtend perception. The principles shaping these architectures and their role in information processing are still poorly understood. We investigate mathematically and computationally the representation of orientation and spatial frequency in cat primary visual cortex. We prove that two natural principles, local exhaustivity and parsimony of representation, would constrain the orientation and spatial frequency maps to display a very specific pinwheel-dipole singularity. This is particularly interesting since recent experimental evidences show a dipolar structures of the spatial frequency map co-localized with pinwheels in cat. These structures have important properties on information processing capabilities. In particular, we show using a computational model of visual information processing that this architecture allows a trade-off in the local detection of orientation and spatial frequency, but this property occurs for spatial frequency selectivity sharper than reported in the literature. We validated this sharpening on high-resolution optical imaging experimental data. These results shed new light on the principles at play in the emergence of functional architecture of cortical maps, as well as their potential role in processing information.     

Top-Down Inputs Enhance Orientation Selectivity in Neurons of the Primary Visual Cortex during Perceptual Learning

Perceptual learning has been used to probe the mechanisms of cortical plasticity in the adult brain. Feedback projections are ubiquitous in the cortex, but little is known about their role in cortical plasticity. Here we explore the hypothesis that learning visual orientation discrimination involves learning-dependent plasticity of top-down feedback inputs from higher cortical areas, serving a different function from plasticity due to changes in recurrent connections within a cortical area. In a Hodgkin-Huxley-based spiking neural network model of visual cortex, we show that modulation of feedback inputs to V1 from higher cortical areas results in shunting inhibition in V1 neurons, which changes the response properties of V1 neurons. The orientation selectivity of V1 neurons is enhanced without changing orientation preference, preserving the topographic organizations in V1. These results provide new insights to the mechanisms of plasticity in the adult brain, reconciling apparently inconsistent experiments and providing a new hypothesis for a functional role of the feedback connections.     

Spatial and Feature-Based Attention in a Layered Cortical Microcircuit Model

Directing attention to the spatial location or the distinguishing feature of a visual object modulates neuronal responses in the visual cortex and the stimulus discriminability of subjects. However, the spatial and feature-based modes of attention differently influence visual processing by changing the tuning properties of neurons. Intriguingly, neurons'' tuning curves are modulated similarly across different visual areas under both these modes of attention. Here, we explored the mechanism underlying the effects of these two modes of visual attention on the orientation selectivity of visual cortical neurons. To do this, we developed a layered microcircuit model. This model describes multiple orientation-specific microcircuits sharing their receptive fields and consisting of layers 2/3, 4, 5, and 6. These microcircuits represent a functional grouping of cortical neurons and mutually interact via lateral inhibition and excitatory connections between groups with similar selectivity. The individual microcircuits receive bottom-up visual stimuli and top-down attention in different layers. A crucial assumption of the model is that feature-based attention activates orientation-specific microcircuits for the relevant feature selectively, whereas spatial attention activates all microcircuits homogeneously, irrespective of their orientation selectivity. Consequently, our model simultaneously accounts for the multiplicative scaling of neuronal responses in spatial attention and the additive modulations of orientation tuning curves in feature-based attention, which have been observed widely in various visual cortical areas. Simulations of the model predict contrasting differences between excitatory and inhibitory neurons in the two modes of attentional modulations. Furthermore, the model replicates the modulation of the psychophysical discriminability of visual stimuli in the presence of external noise. Our layered model with a biologically suggested laminar structure describes the basic circuit mechanism underlying the attention-mode specific modulations of neuronal responses and visual perception.     

An explanation of contextual modulation by short-range isotropic connections and orientation map geometry in the primary visual cortex

Recent experimental studies on the primary visual cortex have revealed complicated nonclassical neuronal activities. Contextual modulation on orientation-contrast is one typical example of nonclassical neuronal behavior. This modulation by surrounding stimuli in a nonclassical receptive field is mainly thought to be mediated by short- and long-range horizontal connections within the primary visual cortex. Short-range connections are circularly symmetrical and relatively independent of orientation preferences, while long-range connections are patchy, asymmetrical, and orientation specific. Although this modulation can be explained by long-range specific connections qualitatively, recent studies suggest that long-range connections alone may be insufficient with respect to the balance between two types of connections. Here, in order to clarify the role of short-range connections in the process of contextual modulation, we propose a model of the primary visual cortex with isotropic short-range connections and a geometric orientation map. Computational simulations using the model have demonstrated that contextual modulation can be explained by short-range connections alone. This is due to the interaction between the spatial periodicity of orientation domains and the excitatory-inhibitory regions arising from the propagation of activities.Acknowledgement We gratefully acknowledge useful conversations with Hiromichi, Sato. The present work was partly supported by Grant-in-Aid for Scientific Research on Priority Areas (C) Advanced Brain Science Project from the Japanese Ministry of Education, Science, Sports, and Culture.     

Neural network model of the primary visual cortex: From functional architecture to lateral connectivity and back

The role of intrinsic cortical dynamics is a debatable issue. A recent optical imaging study (Kenet et al., 2003) found that activity patterns similar to orientation maps (OMs), emerge in the primary visual cortex (V1) even in the absence of sensory input, suggesting an intrinsic mechanism of OM activation. To better understand these results and shed light on the intrinsic V1 processing, we suggest a neural network model in which OMs are encoded by the intrinsic lateral connections. The proposed connectivity pattern depends on the preferred orientation and, unlike previous models, on the degree of orientation selectivity of the interconnected neurons. We prove that the network has a ring attractor composed of an approximated version of the OMs. Consequently, OMs emerge spontaneously when the network is presented with an unstructured noisy input. Simulations show that the model can be applied to experimental data and generate realistic OMs. We study a variation of the model with spatially restricted connections, and show that it gives rise to states composed of several OMs. We hypothesize that these states can represent local properties of the visual scene. Action Editor: Jonathan D. Victor     

Pre-attentive segmentation in the primary visual cortex

The activities of neurons in primary visual cortex have been shown to be significantly influenced by stimuli outside their classical receptive fields. We propose that these contextual influences serve pre-attentive visual segmentation by causing relatively higher neural responses to important or conspicuous image locations, making them more salient for perceptual pop-out. These locations include boundaries between regions, smooth contours, and pop-out targets against backgrounds. The mark of these locations is the breakdown of spatial homogeneity in the input. for instance, at the border between two texture regions of equal mean luminance. This breakdown causes changes in contextual influences, often resulting in higher responses at the border than at surrounding locations. This proposal is implemented in a biologically based model of VI in which contextual influences are mediated by intra-cortical horizontal connections. The behavior of the model is demonstrated using examples of texture segmentation, figure-ground segregation, target-distractor asymmetry, and contour enhancement, and is compared with psychophysical and physiological data. The model predicts (1) how neural responses should be tuned to the orientation of nearby texture borders, (2) a set of qualitative constraints on the structure of the intracortical connections, and (3) stimulus-dependent biases in estimating the locations of the region borders by pre-attentive vision.     

The Caenorhabditis elegans Six/sine oculis class homeobox gene ceh-32 is required for head morphogenesis

Repulsion plays a fundamental role in the establishment of a topographic map of the chick retinotectal projections. This has been highlighted by studies demonstrating the role of opposing gradients of the EphA3 receptor tyrosine kinase on retinal axons and two of its ligands, ephrin-A2 and ephrin-A5, in the tectum. We have analyzed the distribution of these two ephrins in other retinorecipient structures in the chick diencephalon and mesencephalon during the period when visual connections are being established. We have found that both ephrin-A2 and ephrin-A5 and their receptors EphA4 and EphA7 are expressed in gradients whose orientation is consistent with the topography of the nasotemporal axis of the respective retinofugal projections. In addition, their distribution suggests that receptor-ligand interactions may be involved in the organization of connections between the different primary visual centers and, thus, in the topographic organization of secondary visual projections. Interestingly, where projections lack a clear topographic representation, a uniform expression of the Eph-ephrin molecules was observed. Finally, we also show that a similar patterning mechanism may be implicated in the transfer of visual information to the telencephalon. These results suggest a conserved function for EphA receptors and their ligands in the elaboration of topographic maps at multiple levels of the visual pathway.     

Improvement and impairment of visually guided behavior through LTP- and LTD-like exposure-based visual learning

Cellular studies have focused on long-term potentiation (LTP) and long-term depression (LTD) to understand requirements for persistent changes in synaptic connections. Whereas LTP is induced through high-frequency intermittent stimulation, low-frequency stimulation evokes LTD. Because of the ubiquitous efficacy of these protocols, they are considered fundamental mechanisms underlying learning. Here we adapted LTP/LTD-like protocols to visual stimulation to alter human visually guided behavior. In a change-detection task, participants reported luminance changes against distracting orientation changes. Subsequently, they were exposed to passive visual high- or low-frequency stimulation of either the relevant luminance or irrelevant orientation feature. LTP-like high-frequency protocols using luminance improved ability to detect luminance changes, whereas low-frequency LTD-like stimulation impaired performance. In contrast, LTP-like exposure of the irrelevant orientation feature impaired performance, whereas LTD-like orientation stimulation improved it. LTP-like effects were present for 10 days, whereas LTD-like effects lasted for a shorter period of time. Our data demonstrate that instead of electrically stimulating synapses, selective behavioral changes are evoked in humans by using equivalently timed visual stimulation, suggesting that both LTD- and LTP-like protocols control human behavior but that the direction of changes is determined by the feature incorporated into the stimulation protocol.     

Considering the Spatial Layout Information of Bag of Features (BoF) Framework for Image Classification

The spatial pooling method such as spatial pyramid matching (SPM) is very crucial in the bag of features model used in image classification. SPM partitions the image into a set of regular grids and assumes that the spatial layout of all visual words obey the uniform distribution over these regular grids. However, in practice, we consider that different visual words should obey different spatial layout distributions. To improve SPM, we develop a novel spatial pooling method, namely spatial distribution pooling (SDP). The proposed SDP method uses an extension model of Gauss mixture model to estimate the spatial layout distributions of the visual vocabulary. For each visual word type, SDP can generate a set of flexible grids rather than the regular grids from the traditional SPM. Furthermore, we can compute the grid weights for visual word tokens according to their spatial coordinates. The experimental results demonstrate that SDP outperforms the traditional spatial pooling methods, and is competitive with the state-of-the-art classification accuracy on several challenging image datasets.     

The Gaussian derivative model for spatial-temporal vision: I. Cortical model

How do we see the motion of objects as well as their shapes? The Gaussian Derivative (GD) spatial model is extended to time to help answer this question. The GD spatio-temporal model requires only two numbers to describe the complete three-dimensional space-time shapes of individual receptive fields in primate visual cortex. These two numbers are the derivative numbers along the respective spatial and temporal principal axes of a given receptive field. Nine transformation parameters allow for a standard geometric association of these intrinsic axes with the extrinsic environment. The GD spatio-temporal model describes in one framework the following properties of primate simple cell fields: motion properties, number of lobes in space-time, spatial orientation. location, and size. A discrete difference-of-offset-Gaussians (DOOG) model provides a plausible physiological mechanism to form GD-like model fields in both space and time. The GD model hypothesizes that receptive fields at the first stage of processing in the visual cortex approximate 'derivative analyzers' that estimate local spatial and temporal derivatives of the intensity profile in the visual environment. The receptive fields as modeled provide operators that can allow later stages of processing in either a biological or machine vision system to estimate the motion as well as the shapes of objects in the environment.     

Neuronal connections of ocular dominance columns in the cortex of monocularly deprived cats

We have investigated the developmental changes of intrahemispheric neuronal connections of the areas 17 & 18 ocular dominance columns in monocularly deprived cats. Single cortical columns were microiontophoretically injected with horseradish peroxidase and 3D reconstruction of retrogradely labelled cells' region was done. Ocular dominance of injected columns and their coordinates in the visual field map were determined. In area 17 it was shown that for non-deprived eye the connections of columns that are driven via the crossed pathways were longer than connections of columns driven via uncrossed ones, and in both cases they were longer than connections in intact cats. The connections of deprived eye columns are significantly reduced. We have observed some changes in the spatial organization of long-range connections in area 17 for columns driven by the non-deprived eye (more rounded shape of regions of labeled cells, non-uniform distribution of cells within it). Maximal length of such connections did not exceed the length of connections in strabismic cats. We speculate that the length of cell axons providing for the horizontal connections of cortical columns has some intrinsic limit that does not depend on visual stimulation during the critical period of development.     

The relationship between the Gabor elementary function and a stochastic model of the inter-spike interval distribution in the responses of visual cortex neurons

In a previously reported study (Berger et al. 1990) we analyzed distributions of interspike intervals recorded extracellularly from cat visual cortex under four stimulus conditions. Stimuli were gratings differing in orientation and spatial frequency. The probability density function of first passage time for a random walk with drift process, which is defined by its barrier height and drift coefficient, was used to characterize the generating process of axonal discharge under resting and stimulus conditions. Drift coefficient and barrier height were derived from the sample mean and standard deviation of the measured inter-spike intervals. For cells with simple receptive fields, variations in spatial frequency produced changes only in drift coefficient. Variations in barrier height were produced only by changes in orientation of the stimulus. Currently, the method used to analyze these data was implemented in a simulation which displayed the relationship between the interval distribution of impulses, the random walk which represents the time series characteristic of the spike train model and the Gabor filter function which represents the geometry of the receptive field process.     

Interhemispheric connections of ocular-dominance columns in cats with binocular vision impairment

We have investigated the interhemispheric connections of areas 17 and 18 in cats with impaired binocular vision (monocular deprivation, uni- and bilateral strabismus). Monosynaptic neuronal connections were studied using microionophoretic injections of horseradish peroxidase in the single cortical columns and analsys of spatial distribution of retrogradely labelled callosal cells was performed. In the cases of monocular deprivation and strabismus, the spatial asymmetry and eye-specificity of interhemispheric connections are retained. Quantitative changes of connections are more pronounced in strabismic cats. In cats with binocular vision impairments, as well as in control ones, the width of callosal-recipient zone is larger than of the callosal cells zone. This may indicate that interhemispheric connections are non-reciprocal in the areas of cortex that are more distant from the projection of vertical meridian of visual field. We expect that there should be morpho-functional in the cells that are providing connections in opposite directions.     

On the origin of the functional architecture of the cortex

The basic structure of receptive fields and functional maps in primary visual cortex is established without exposure to normal sensory experience and before the onset of the critical period. How the brain wires these circuits in the early stages of development remains unknown. Possible explanations include activity-dependent mechanisms driven by spontaneous activity in the retina and thalamus, and molecular guidance orchestrating thalamo-cortical connections on a fine spatial scale. Here I propose an alternative hypothesis: the blueprint for receptive fields, feature maps, and their inter-relationships may reside in the layout of the retinal ganglion cell mosaics along with a simple statistical connectivity scheme dictating the wiring between thalamus and cortex. The model is shown to account for a number of experimental findings, including the relationship between retinotopy, orientation maps, spatial frequency maps and cytochrome oxidase patches. The theory's simplicity, explanatory and predictive power makes it a serious candidate for the origin of the functional architecture of primary visual cortex.     

Synaptic integration fields and associative plasticity of visual cortical cells in vivo

Two major constraints in connectivity decide the spatial extent of visual cortical receptive fields, both during development and adult functioning: 1) feedforward input, extrinsic to visual cortex, is organized in an orderly projection to form a point-to-point mapping of the retina onto the cortical mantle and constitutes the core of the minimal discharge field (MDF) after amplification by local intracortical circuits; and 2) a second type of connectivity consists of long distance horizontal intracortical connections which are thought to favor the binding of local visual operations occurring simultaneously in different parts of the visual field. We review here possible factors, intrinsic to the considered cortical cell, that may control the effectiveness of post-synaptic integration. Their expression during sensory recognition could depend on the spatio-temporal distribution of active inputs onto the target cell dendrite and on their interplay with non-linearities of the membrane properties. On the basis of intracellular recordings in cat area 17, we have observed that peripheral responses (excitatory and inhibitory) could be boosted by coincident postsynaptic depolarization. This effect is lost if the response and the postsynaptic depolarization are mismatched by 1000 ms, suggesting that temporal correlation of central and peripheral responses could regulate in a non-linear manner the gain of center-surround interactions. This temporal selectivity is compatible with up-regulation of composite potentials by the transient voltage-dependent activation of slowly inactivating conductances in visual cortical neurons. A direct consequence of these different considerations is that the receptive field (RF) of neurons in visual pathways should not be considered as a static hardwired window probing the outer environment, but as an active filter which may continuously adapt and be updated as a function of global context and past experience.     

Integration of local features into global shapes: monkey and human FMRI studies

The integration of local image features into global shapes was investigated in monkeys and humans using fMRI. An adaptation paradigm was used, in which stimulus selectivity was deduced by changes in the course of adaptation of a pattern of randomly oriented elements. Accordingly, we observed stronger activity when orientation changes in the adapting stimulus resulted in a collinear contour than a different random pattern. This selectivity to collinear contours was observed not only in higher visual areas that are implicated in shape processing, but also in early visual areas where selectivity depended on the receptive field size. These findings suggest that unified shape perception in both monkeys and humans involves multiple visual areas that may integrate local elements to global shapes at different spatial scales.     

Oscillatory network with self-organized dynamical connections for synchronization-based image segmentation

An oscillatory network of columnar architecture located in 3D spatial lattice was recently designed by the authors as oscillatory model of the brain visual cortex. Single network oscillator is a relaxational neural oscillator with internal dynamics tunable by visual image characteristics - local brightness and elementary bar orientation. It is able to demonstrate either activity state (stable undamped oscillations) or "silence" (quickly damped oscillations). Self-organized nonlocal dynamical connections of oscillators depend on oscillator activity levels and orientations of cortical receptive fields. Network performance consists in transfer into a state of clusterized synchronization. At current stage grey-level image segmentation tasks are carried out by 2D oscillatory network, obtained as a limit version of the source model. Due to supplemented network coupling strength control the 2D reduced network provides synchronization-based image segmentation. New results on segmentation of brightness and texture images presented in the paper demonstrate accurate network performance and informative visualization of segmentation results, inherent in the model.     

A specialized area in limbic cortex for fast analysis of peripheral vision

In primates, prostriata is a small area located between the primary visual cortex (V1) and the hippocampal formation. Prostriata sends connections to multisensory and high-order association areas in the temporal, parietal, cingulate, orbitofrontal, and frontopolar cortices. It is characterized by a relatively simple histological organization, alluding to an early origin in mammalian evolution. Here we show that prostriata neurons in marmoset monkeys exhibit a unique combination of response properties, suggesting a new pathway for rapid distribution of visual information in parallel with the traditionally recognized dorsal and ventral streams. Whereas the location and known connections of prostriata suggest a high-level association area, its response properties are unexpectedly simple, resembling those found in early stages of the visual processing: neurons have robust, nonadapting responses to simple stimuli, with latencies comparable to those found in V1, and are broadly tuned to stimulus orientation and spatiotemporal frequency. However, their receptive fields are enormous and form a unique topographic map that emphasizes the far periphery of the visual field. These results suggest a specialized circuit through which stimuli in peripheral vision can bypass the elaborate hierarchy of extrastriate visual areas and rapidly elicit coordinated motor and cognitive responses across multiple brain systems.