Disentangling the contributions of spatiotopic, retinotopic, and configural frames of reference to the filtering of probable distractor locations.

Human observers can allocate their attention to locations likely to contain a target and can also learn to avoid locations likely to contain a salient distractor during visual search. However, it is unclear which spatial frame of reference such learning is applied to. As such, our aim was to systematically disentangle the contributions of spatiotopic, retinotopic, and configural frames of reference to provide a comprehensive account of how the probabilistic distractor filtering effect comes about. We first demonstrate that the filtering effect is better determined by the probability of a salient distractor appearing at a relative location (i.e., in relation to one's eye position or an item's position in relation to other items within a display) rather than a fixed (spatiotopic) location, by varying the position of visual search arrays (along with fixation) across spatial contexts. We then separate retinotopic and configural reference frames by varying the configural but not retinotopic properties of biased (i.e., displays containing a probable distractor location) and unbiased visual search arrays and vice versa. In doing so, we find the filtering effect to be restricted to biased contexts when retinotopic positions are maintained, but configural properties are varied. In contrast, when the configural properties of visual search arrays are maintained, we show the transfer of the filtering effect across retinotopic positions. Thus, we demonstrate that probabilistic distractor filtering primarily emerges via a configural representation that codes the relative positions of items within search displays independent of spatiotopic and retinotopic coordinates. (PsycInfo Database Record (c) 2025 APA, all rights reserved).