9631132 ROBERTSON Objects and scenes are normally coherent and integrated in visual perception. Yet, biological and behavioral evidence has demonstrated visual modularity; different visual areas within the mammalian brain respond to particular features such as color, motion, shape, size, or location. The way in which these features get bound together in awareness to produce normal coherence is not known. One major cognitive theory has proposed that attention to spatial locations provides the basis for feature binding. It follows that damage to areas of the brain that represent spatial information should produce severe problems in binding features together to produce perceptual wholes. This research will explore the relationships between selective attention, spatial representations, feature binding, and object individuation, using evidence from the breakdown of performance in individuals with damage to brain areas that disrupt spatial perception. The main hypothesis is that interactions between neural areas that represent object features and those that represent space achieve accurate visual percepts and produce rapid and accurate visual search, especially when feature binding is required. Thus, the spatial problems observed in individuals with selected types of neurological damage should be accompanied by problems with specific types of object perception that require integration. Individuals with selected spatial deficits will be tested in both object perception and spatial tasks to look for hypothesized relationships between the two. Other studies will attempt to determine the type of spatial information that is critical for feature binding and that which is not. Behavioral measures will include error data and reaction time. Biological measures will include lesion location and volume, as determined by reconstructed 3-D structural MRI. Although physiological and metabolic techniques for measuring cortical function in neurologically normal individu als have improved dramatically in the recent past, observing the effects of lesions remains a major means to determine whether or not a particular region of the brain is both sufficient and necessary for a particular function. Evidence derived from humans with lesions to particular areas provides informative data to build hypotheses about function and interactions between functions that then might be observed using other techniques, such as single unit recording in animals or functional imaging in normal individuals. The data may contribute at least a partial solution to the binding problem and will certainly suggest where to look for correlated neural activity that may support accurate feature integration. ***
