Exploring the distinct patterns of pupil dilation between space- and feature-based attention in a visual search task

Top-down visual attention prioritized selected visual information such that deploying an appropriate selection mechanism in accordance with pre-cued information facilitates target search. For example, if the likely color of an upcoming target is known, observers may deploy feature-based attention (FBA) to improve the speed and/or accuracy with which the target is identified; if the likely location is known, observers may instead deploy space-based attention (SBA). Our goal here was to determine whether pupil size during the pre-cueing period and throughout visual search could distinguish between these selection mechanisms. We hypothesized that FBA deployment should be associated with a larger change in pupil size, given that (1) FBA is deployed globally across the visual filed, in contrast to SBA, and (2) pupil size tends to increase with broad attentional deployment. To test this hypothesis, we reanalyzed the pupil data from a previously published study in which participants were either presented with a central color or location pre-cue. Following the short cueing period (300ms), all aspects of the visual display and task design were matched between groups. As predicted, a time course time-locked to cue onset reveals significant differences in pupil size between selection mechanisms, with participants assigned to the FBA task showing on average a greater change in pupil size throughout the trial, as well as a higher mean amplitude. Visual inspection of individual trial data suggested consistent differences between selection mechanisms. Permutation tests further suggested that the differences in pupil diameter emerged as early as the pre-cue onset, suggesting a potential distinct pattern of pupil dilation between SBA and FBA even before stimulus presentation – and continued throughout visual search. This finding opens the possibility that pupillometry may be used to differentiate between FBA and SBA deployment during visual search, even before stimulus presentation or a behavioral response.