Parallel processing of efficient, heterogeneous visual search with real-world objects

Previous work in our lab has demonstrated that efficient visual search has a reaction time by set size function that is best characterized by logarithmic curves, and the steepness of these logarithmic curves is determined by the similarity between target and distractor items (Buetti et al., in press). This thesis presents a theoretical account of these phenomena, emphasizing that a parallel, unlimited capacity, exhaustive processing system must be underlying such data. Two experiments were conducted to expand our findings to a set of real-world stimuli, in both homogeneous and heterogeneous search displays. Based on our current theory and numerical simulations, we were able to very accurately predict RT performance in heterogeneous search using parameters from homogeneous search tasks of a different group of subjects. By examining the systematic deviation of our predictions from observed data, we concluded that early visual processing for individual items were not independent. Instead, items in homogeneous displays seemed to facilitate each other's processing by a multiplicative factor. These results challenge previous accounts of heterogeneity effects in visual search, and demonstrate the explanatory and predictive power of our current theory of visual search