Neural and Behavioral Analyses of the Representation of Shape

点击翻译按钮获取中文摘要

In a fraction of a second a human observer can easily comprehend an image of an object or scene that has never been encountered before. Knowledge of the neural basis of this remarkable ability has been advanced by the discovery of a cortical area that is activated by the shape of intact objects but not by scrambled versions of these same images. This brain region, termed the lateral occipital complex (LOC), is activated in functional magnetic resonance imaging (fMRI) studies by both familiar and unfamiliar objects. The LOC is also activated by both photographs and line drawings of the same object, indicating that the surface qualities are represented elsewhere. And, bilateral lesions to LOC have been shown to produce a complete inability to recognize objects on the basis of their shape. LOC thus represents physical shape to which, presumably, semantic information, including a name, can be associated. With support from the National Science Foundation, Dr. Irving Biederman and colleagues at the University of Southern California will conduct a series of experiments that aim to better determine how shape is represented in the LOC. One project will test whether parts and relations have distinguishable neural loci. Other projects will assess whether the coding of shape in the LOC is accomplished in terms of local image features, parts, global shape or concepts, and the locus and extent to which invariant coding of shape - which allows an object to be recognized as unchanged even when it is viewed at a different position in the visual field, at a different size, at a different orientation, or lit from a different direction - is manifested in LOC. This research will fill a critical gap between human psychophysical research on object recognition and work on single-neuron recordings in response shape variations that have been performed in non-human primates. Although the homologues between human and non-human primate primary sensory and motor cortices are well established there are significant questions concerning the homologues of later, perceptual association areas. Perhaps the most important potential impact of the proposed research is that it could allow us to understand the neural representations that subserve not only object recognition but cognition more generally.