9631573 WATANABE Visual perception is often considered a passive and simple event. This view derives partly from the general assumption that visual information goes only from the retina to higher-level areas in the brain, and not in the other direction. However, recent anatomical and physiological findings indicate that there are massive feedback pathways in the brain along which information is carried from these higher-level brain areas back to low-level areas, including those processing visual information. It is not yet well known what role these feedback projections play in visual information processing in the brain. This research is concerned with the role of attention, one of the most important cognitive functions originating in higher brain areas. It will examine what level of visual processing is the lowest that can be influenced by attention. Answering this question will lead to a better understanding of the role of feedback projections from high-level brain areas, where cognitive functions likely originate, to lower level visual processing. In a series of experiments, people will watch a moving object on a computer display and will be instructed to pay attention only to a local contour of the object. One minute after the onset of the moving object, each person's sensitivity to motion in several directions, such as upward and rightward, will be measured. Motion processing is known to have at least two stages: In the first stage, the direction of motion of the whole object is not detected by the visual system. Only the many local motion components perpendicular to the local contours of a moving object are detected. This stage begins in the primary visual cortex in which very low-level visual processing occurs. In the second stage, the signals from detection of the directions of these local motion components are integrated, so that the true direction of motion of the entire object can be calculated. In these experiments, if attention influences the first stage of motion processing, sensitivity to the originally attended local motion should decrease. On the other hand, if attention does not influence such a low-level stage of processing, sensitivity to the direction of motion should not be influenced by the allocation of attention. A preliminary experimental result using only two people indicated that the former is the case. If this result is confirmed with additional people, it will indicate that attention can influence even very primitive stages of visual information processing, leading to the conclusion that feedback projections play a very important role in making visual perception active and influenced by high-level cognitive functions, rather than entirely passive and automatic. These results should have important implications for the design of computer vision systems, a very difficult computer problem. ***
