Experimental Evidence for Top-Down Attentional Selection in the Selective Tuning Model of Visual Attention

To overcome limited processing capacity, our visual system facilitates information that relates to the task at hand while inhibiting irrelevant information via selective attention. Among various attention models and theories, the Selective Tuning model of visual attention (ST) is a computation model of visual processing that is based on biological mechanisms. This model emphasizes the role of top-down feedback processing in visual perception and has predicted its unique consequences, such as an attentional surround suppression in which the attentional focus is accompanied by an inhibitory surround. The previous studies have experimentally validated STs predictions, indicating that the components in ST do reflect actual visual processing in the brain. Nevertheless, many aspects of ST still need to be elaborated and several predictions and assumptions remain untested. The series of works in this dissertation investigate different aspects of top-down feedback processing in visual perception that ST has proposed to corroborate this model and to broaden our understanding of visual attention. The first study examined whether top-down feedback processing is necessary for an attention-demanding, fine-grained visual localization (Chapter 2). The subsequent two studies focused on the properties of different types of the attentional surround suppression, the end-result of top-down feedback processing. The second study suggested the interplay between the location-based and feature-based surround suppression and tested the potential factors that could manipulate the spatial extent of the location-based suppressive surround (Chapter 3). The last study demonstrated feature-based surround suppression in motion processing and its neurophysiological mechanism (Chapter 4). Collectively, this work reinforces functional significance of top-down, attention-mediated feedback for visual processing and supports the validity of ST as well.