Grant: Discovery Grants Program - Individual

When we look at an object it appears to us as a single thing, when in fact it is encoded at our retinae by thousands or even millions of photoreceptors that each code a small part. One of the main jobs of visual cortex is to perceptually organize this pointillist representation into a spatially coherent representation that corresponds to our reality. The proposed project is focused specifically on the organization of local orientation responses (edges) into closed contours that represent the boundaries of objects. This leap from 1D to 2D structure is achieved by a cooperative circuit in object pathway (V1 to LOC) that has yet to be elucidated. Understanding this circuit, at both computational and neural levels, is a central question in visual neuroscience. Based upon prior computational and psychophysical investigations, we have developed a hierarchical model for this process that yields superior results while corresponding well with what we know of cortical organization. However, the model is incomplete in that it does not embody a well-specified model for 2D (planar) shape. In the proposed work I intend to investigate a novel shape representation that I have tentatively called the warplet model. In this model, complex shapes are represented by deforming the planar space in which they are embedded. Crucially, these deformations can be selected so as to preserve the topology of the contour. Four computational projects will develop and objectively evaluate this theory of shape perception, and integrate it within our current hierarchical model. Five psychophysical projects will probe aspects of perceptual organization and shape perception, and test crucial predictions of the theory. The project will train a strong team of experimental psychologists and computer scientists in interdisciplinary vision research.