Unraveling brain interactions in vision: the example of crowding

In visual crowding, the presence of neighboring elements impedes perception a target. Crowding is traditionally explained with feedforward, local models. However, increasing number can decrease i.e., lead to uncrowding, which demonstrates inadequacy classic feedforward explanation. Global models are needed, but behavioral experiments alone cannot discriminate between them. Here, we used fMRI study effects (un)crowding on BOLD response and effective connectivity regions V1 V4 lateral occipital complex (LOC). We tested three experimental conditions: no crowding. First, following standard approach crowding studies, extracted percent signal change (PSC) for each condition in area. replicated previous results attenuation beginning V2 persisting up hierarchy. uncrowding further attenuated response, suggests that PSC not (monotonically) related level as commonly assumed. then dynamic causal modeling (DCM) Bayesian model comparison. Specifically, contrasted top-down, bottom-up recurrent Recurrent fit data best all conditions, even simplest condition. Our explain discrepancies investigations crowding: hierarchy, effect theoretically be detected at any stage. Beyond demonstrate need data-driven like DCM understand processing presumably underlies general. The framework allows us only compare architectures also estimate computational details form connection strengths regions, inform theoretical