Yasuhiro Hatori

Early representation of shape by onset synchronization of border-ownership-selective cells in the V1-V2 network

Construction of surface is a crucial step toward the representation of shape through the integration of local information. Physiological studies have reported that the primary visual cortex (V1) codes the medial axis (MA) that is a skeletal structure equidistant from nearby contours, suggesting the early representation of surface in V1. Although the neural basis of surface construction has not been clarified, the onset synchronization of border ownership (BO)-selective cells is a plausible candidate for the generation of surface. We investigated computationally the representation of surface in a biophysically detailed model of primate V1-V2 networks. The simulation results showed that the simultaneous arrival of signals from BO-selective cells evoked strong responses of V1 cells located around the MA. The simulation results lead to a prediction that the perception of the direction of figure (DOF) depends on the degree of synchronous presentation of contour. We conducted a psychophysical experiment and showed that the perception of the DOF is biased toward a highly synchronized contour. These results suggest a crucial role of the onset synchronization of BO-selective cells for the construction of early representation of shape.

Neural construction of 3D medial axis from the binocular fusion of 2D MAs

The perceptual constancy of shape, including view invariance, is an amazing property of the visual system. Cortical representation by the medial axis (MA) is an attractive candidate for maintaining the constancy of a wide range of arbitrary shapes. Recent physiological studies have reported that neurons in the primary visual cortex (V1) show a response to two-dimensional (2D) MAs, and those in the inferior temporal cortex (IT) are selective to three-dimensional (3D) MAs. However, little is known about the neural mechanisms underlying the transformation of 2D to 3D MAs. As a first step toward investigating the cortical mechanism, we have proposed as a hypothesis that a pair of monocular 2D MAs is fused to generate a 3D MA. We examined the computational plausibility of the hypothesis; specifically, whether an energy-based fusion model is capable of generating 3D MAs. We generated blob-like, physiologically plausible 2D MAs, and used a standard energy model to detect the disparity between a pair of 2D MAs. The model successfully generated 3D MAs for a variety of objects that included typical shape characteristics. A reconstruction test showed that the computed 3D MAs captured the essential structure of the objects with reasonable accuracy and view invariance. These results indicate that the fusion of monocular blob-like 2D MAs is capable of generating a reasonable 3D MA within the framework of the energy model.

Perceptual representation and effectiveness of local figure–ground cues in natural contours

A contour shape strongly influences the perceptual segregation of a figure from the ground. We investigated the contribution of local contour shape to figure–ground segregation. Although previous studies have reported local contour features that evoke figure–ground perception, they were often image features and not necessarily perceptual features. First, we examined whether contour features, specifically, convexity, closure, and symmetry, underlie the perceptual representation of natural contour shapes. We performed similarity tests between local contours, and examined the contribution of the contour features to the perceptual similarities between the contours. The local contours were sampled from natural contours so that their distribution was uniform in the space composed of the three contour features. This sampling ensured the equal appearance frequency of the factors and a wide variety of contour shapes including those comprised of contradictory factors that induce figure in the opposite directions. This sampling from natural contours is advantageous in order to randomly pickup a variety of contours that satisfy a wide range of cue combinations. Multidimensional scaling analyses showed that the combinations of convexity, closure, and symmetry contribute to perceptual similarity, thus they are perceptual quantities. Second, we examined whether the three features contribute to local figure–ground perception. We performed psychophysical experiments to judge the direction of the figure along the local contours, and examined the contribution of the features to the figure–ground judgment. Multiple linear regression analyses showed that closure was a significant factor, but that convexity and symmetry were not. These results indicate that closure is dominant in the local figure–ground perception with natural contours when the other cues coexist with equal probability including contradictory cases.

Robust detection of medial-axis by onset synchronization of border-ownership selective cells and shape reconstruction from its medial-axis

There is little understanding on representation and reconstruction of object shape in the cortex. Physiological studies with macaque suggested that neurons in V1 respond to Medial-Axis (MA). We investigated whether (1) early visual areas could provide basis for MA representation, and (2) we could reconstruct the original shape from its MA, with a physiologically realistic computational model consisting of early- to intermediate-level visual areas. Assuming the synchronization of border-ownership selective cells at stimulus onset, our model was capable of detecting MA, indicating that early visual area could provide basis for MA representation. Furthermore, we propose a physiologically plausible reconstruction algorithm with the summation of distinct gaussians.

Representation of Medial Axis from Synchronous Firing of Border-Ownership Selective Cells

The representation of object shape in the visual system is one of the most crucial questions in brain science. Although we can perceive figure shape correctly and quickly, without any effort, the underlying cortical mechanism is largely unknown. Physiological experiment with macaque indicated the possibility that the brain represents a surface with Medial Axis (MA) representation. To examine whether early visual areas could provide basis for MA representation, we constructed the physiologically realistic, computational model of the early visual cortex, and examined what constraint is necessary for the representation of MA. Our simulation results showed that simultaneous firing of Border-Ownership (BO) selective cells at the stimulus onset is a crucial constraint for MA representation.

Sparse coding generates curvature selectivity in V4 neurons

The cortical area V4 produces a representation of curvature as the intermediate-level representation of an objects shape. We investigated whether sparse coding is the principle driving the generation of the spatial properties of the receptive field in V4 that exhibit curvature selectivity. To investigate the role of sparseness in the construction of curvature representations, we applied component analysis with a sparseness constraint to the activity of model V2 neurons that were responding to shapes derived from natural images. Our simulation results showed that single basis functions with medium degrees of sparseness (0.7-0.8) produced curvature selectivity, and their population activity produced acute curvature bias. The results support the hypothesis that sparseness plays an essential role in the construction of curvature selectivity in V4.

Surface-Based construction of curvature selectivity from the integration of local orientations

Recent physiological studies have reported that neurons in the cortical area V4 are selective to curvature along object contour. The neurons are capable of discriminating convexity and concavity, and indicating the direction of the curvature with respect to the contour projected onto their cRF. We propose that surface representation plays a crucial role in constructing the selectivity for curvature because convexity/concavity cannot be determined without the construction of object surface. To test the proposal, we developed a computational-model of V1-V4 networks that computes spatiotemporal activities of single cells, and carried out the simulations with the stimuli used by the physiological studies. The model neurons reproduced the selectivity for specific curvature and direction. Population of the model cells showed a bias toward convex curvature as consistent with V4 population in vivo. These results support that the representation of surface is crucial for the construction of the selectivity in V4.

Sparseness Controls the Receptive Field Characteristics of V4 Neurons: Generation of Curvature Selectivity in V4

Physiological studies have reported that the intermediate-level visual area represents primitive shape by the selectivity to curvature and its direction. However, it has not been revealed that what coding scheme underlies the construction of the selectivity with complex characteristics. We propose that sparse representation is crucial for the construction so that a sole control of sparseness is capable of generating physiological characteristics. To test the proposal, we applied component analysis with sparseness constraint to activities of model neurons, and investigated whether the computed bases reproduce the characteristics of the selectivity. To evaluate the learned bases quantitatively, we computed the tuning properties of single bases and the population, as similar to the physiological reports. The basis functions reproduced the physiological characteristics when sparseness was medium (0.6-0.8). These results indicate that sparse representation is crucial for the curvature selectivity, and that a sole control of sparseness is capable of constructing the representation.