| This dissertation presents a biologically plausible computational model for solving the visual feature binding problem. The feature binding problem appears due to the distributed nature of visual processing in the brain, and the gradual loss of spatial information along processing hierarchies. Starting from experiments on primates, this dissertation puts forward the proposal that by using multiple passes of the visual processing hierarchy, both bottom-up and top-down, we can explain many of the different behaviours that primate vision exhibits. The study uses computational modelling to demonstrate visual feature binding, with a strong emphasis on matching the performance and timing characteristics of primate vision. As an application of binding in computer vision, a novel approach to object recognition is presented, integrating fast, parallel detection and categorization, and recognition without interference from distracters. The biological relevance of the proposed model is highlighted by showing that it is able to account for many apparently contradictory and puzzling aspects of primate vision, and by suggesting a number of predictions. One particular prediction regarding the effect of inter-stimulus distance manipulations on visual detection is investigated in detail, and it is demonstrated that the computational model has the same behaviour as humans tested in a psychophysics experiment. |
