The broad goals of this proposal are to increase understanding of the functional organization of object selective visual cortex and the representations used in recognition of faces and objects with particular focus on how neural signals in visual cortex reflect within category similarity as defined both by psychophysical judgments and simple image based metrics. This kind of research falls under the NEI's stated mission of understanding visual function.
Our first aim i s to explore how similarity between faces changes across the visual hierarchy. Using high resolution imaging techniques developed at Stanford and an adaptation paradigm we will assess how the visual system keeps track of both identity and physical changes in faces. We expect that adaptation in lower visual areas will be proportional to image based similarity metrics, while adapation in later visual areas will reflect subjects'judgments of similarity.
Our second aim i s to collect mean responses from individual voxels in response to the same stimuli as in 1 and use the responses as input to a non-metric MDS algorithm. The low dimensional spatial representation of the similarity between the neural response will be compared to mds outputs based on psychophysical judgments and image based metrics. We expect to find a correspondence between the psychological space and the neural space generated from face-selective voxels drawn from the FFA, with earlier regions reflecting local image properties. This would provide support for a computational theory of recognition, which argues that recognition works by projecting visual input onto a low dimensional space comprised by a set of reference shapes (or faces). We will also compare the results from this pattern based approach to the adaptation approach in aim 1, which uses the average signal across voxels.
Our final aim i s to use the design in aim 2 while employing novel parameterized objects as a stimuli. We expect that MDS generated spatial representations of the neural data will come to correspond more closely to psychophysical spaces as voxels are chosen from later areas. The results can also be used in conjunction with those from aim two to examine how neural regions and representations for face and object processing differ. An important step in understanding the visual deficits exhibited by people with disabilites such as autism and dyslexia is characterizing how visual recognition works in the normal visual system. In both autism and dyslexia, early diagnosis can be key, and with reliable measurements of how the normal visual cortex handles simple discrimination task, it may be possible to devise simple imaging paradigms which can be used to identify atypical brain responses.
