Single-unit studies of visual motion processing in cat extrastriate areas

Abstract

Motion vision has high survival value and is a fundamental property of all visual systems. The old Greeks already studied motion vision, but the physiological basis of it first came under scrutiny in the late nineteenth century. Later, with the introduction of single-cell (single-unit) recordings around 1950, the cellular basis of motion perception could be explored. It became clear, that the mammalian visual brain consists of specialized regions for processing different kinds of visual information. The existence of specialized motion pathways from the retina through several subcortical and cortical areas is nowadays indisputable. The primary visual area is the first cortical stage in the mammalian brain, where direction selective neurons have been found. However, this area is a major relay station for virtually all other visual attributes as well. More specialized motion processing occurs subsequently in the so-called extrastriate brain areas, of which little is known yet. In this thesis, I study so-called complex cells in two extrastriate areas of the cat that are involved in motion processing, area 18 and PMLS. Complex cells are able to couple 'corresponding' elements in subsequent images (to solve the correspondence problem), which implies a basic role in motion detection. In PMLS I find, that complex cells possess quite elaborate receptive field structures, which suggests that they also play a role in the analysis of higher order motion information. I therefore examine the basic spatial and temporal motion processing properties of complex cells as well as their higher order temporal interactions and compare results for the two extrastriate cortical areas. Processing of motion information by the analyzed cells proves to occur in parallel in PMLS and area 18 (chapter 2). Contrary to complex cells in PMLS, those in area 18 favor non-smooth motion (chapter 3). The second order temporal interactions differ markedly for cells in the two areas (chapter 4). In addition, area 18 complex cells prove to be velocity tuned (chapter 1), with a sharp tuning for step-size and a broad tuning for step-delays. The similarities and differences of cell responses in area 18 and PMLS are discussed in detail, together with the general significance of these findings for motion information processing in the cat.