Alex J. Hoogerbrugge¹, Jeremy M. Wolfe^2,3, Johan Hulleman¹; ¹University of Manchester, ²Brigham and Women's Hospital, ³Harvard Medical School

Wolfe’s Guided Search 6.0 (2021) posits that covert attention is deployed serially to individual items within a Functional Visual Field (FVF). In contrast, Hulleman & Olivers (2017) proposed an item-free model, where target presence within the FVF is detected without those serial shifts of covert attention. Here, we tested these contrasting viewpoints. In two online experiments (n=122 & n=120), participants searched for a T amongst Ls (Experiment 1: set sizes 6 and 8; Experiment 2: set size 18). Display durations were 67–333 ms. Participants reported either target presence (2-AFC), target location (m-AFC), or both. In an item-based search account, target detection implies knowledge of its location – it was specifically attended, after all. Therefore, across presentation durations, detection accuracy should roughly match localization accuracy, and both should increase in a linear fashion (e.g., one item per ~50 ms). In an item-free account, more search time allows further evidence accumulation, but this is spread across the FVF. Localization is a second step which may not be executed if insufficient time is available. Consequently, targets may be detected without their exact location being known. Across presentation durations, localization accuracy should therefore lag detection accuracy, with both increasing non-linearly. We found in both experiments that localization performance was worse than detection, especially for the shortest presentation durations. Although chance-level performance is inherently higher for 2-AFC detection than for m-AFC localization, none of our chance-correction methods completely removed the performance deficit for localization. Furthermore, no clear evidence about the (non-)linearity of performance increases was found in either experiment. These results seem to place the theoretical burden on an item-based theory to explain how observers might know that a target is present without them knowing where that target was located.

Acknowledgements: AJH and JH were supported by UKRI grant ES/X000443/1. JMW was supported by NIH-NEI: EY017001, NSF: 2146617, and NIH-NCI: CA207490.