Imaging vision : functional mapping of intermediate visual processes in man

Abstract

The human visual system is highly adept at recognizing objects independent of changes in viewing conditions, such as viewpoint, illumination, and partial occlusions by other objects in the visual environment. This invariance to viewing conditions is believed to be achieved by generating invariant shape representations from the variable visual input. The neuronal mechanisms which generate such representations belong to the category of intermediate visual processes, an aspect of visual processing which is not well understood. In this thesis 1 have investigated the neuronal mechanisms of a number of intermediate visual processes relating to the generation of invariant shape representations. Three main issues were addressed: generation of invariant contours, occlusion invariance, and size invariance.

First, the neuronal mechanisms underlying perception of illusory contours, an important type of invariant contour representations, were investigated. The results showed that illusory and real (luminance) contours are processed by spatially overlapping mechanisms in early visual areas, but that illusory contour perception is associated with stronger activation of a region of ventral extrastriate cortex anterior to V2. This area has been implicated in the processing of complex visual information. Furthermore, illusory contour perception was associated with a decoupling of the activity in V1 and V2.

Second, based on previous studies it was hypothesized that a shape discrimination task involving perceptual segregation of overlapping shapes would activate the extrastriate visual cortex more strongly than a task in which shapes do not overlap. It was found that perceptual segregation was associated with increased activity in a region of the posterior extrastriate cortex at c anterior border of V2, which was proposed to correspond to V4v based on location and functional properties.

Third, the mechanisms underlying size invariant recognition were studied. In a psychophysical study the influence of spatial attention on size invariant recognition was investigated. Spatial attention was found to influence size transformation processes in partial accordance with a proposed role for spatial attention as the mechanism underlying size invariance. Using FMRI the neuronal correlates of visual size transformation were investigated directly. Size transformation was found to be associated with increased activity in the intraparietal cortex, particularly its posterior part, and the activation was proportional to the magnitude of the size change. The activation was specific for size transformation and could not be a result of spatial attention shifts alone. Together, these results indicate that while spatial attention may influence processes generating size invariance, it is unlikely to be the sole mechanism underlying size invariant recognition. The activation of the posterior parietal cortex by a shape matching task adds to the body of evidence suggesting that aspects of object shape are processed in the parietal cortex.

In addition, a method for automated extraction of anatomically and topologically accurate cortical surfaces was developed and shown to be practically applicable. Because this method, unlike previous approaches, does not require any user intervention it promises to be a useful tool for cortical surface-based analysis in future studies.