Isaac Savelson¹, Tzu-Yao Chiu¹, Brandon M. Turner¹, Andrew B. Leber¹; ¹The Ohio State University

The visual world is full of irrelevant but distracting information that must be efficiently handled to support effective visual processing. Fortunately, distracting stimuli often exhibit regularities, which can be exploited to minimize their interference. One such example is humans’ ability to quickly learn to ignore spatial locations where distractors commonly appear. However, despite extensive empirical research on the subject, the mechanisms behind learning and implementing spatial suppression are still poorly understood. Here, we sought to enhance our understanding of distractor suppression mechanisms using a computational evidence accumulation model. To this end, we developed a model of the latent decision process associated with overt attentional selection. The model accounts for visual search behavior through a combination of attentional priority values for the target stimulus, salient distractors, and frequent distractor locations. The attentional priority levels of specific stimulus properties were estimated as a set of attentional weights, which determined the rate at which evidence for an attentional shift accumulated toward candidate display locations. When fit to saccadic response behavior collected during an additional singleton task, our model accounted for individuals’ choice-response time distributions remarkably well. To capture the inter-trial dynamics of learned suppression we integrated a reinforcement learning mechanism into our basic diffusion model, allowing us to estimate changes in attentional weights across trials. The model output showed a rapidly learned down-weighting of the frequent distractor location across trials consistent with the literature’s assumptions about learned suppression. Our approach demonstrates the utility of computational modeling techniques in investigating the latent mechanisms underlying attentional capture and distractor suppression.