Emma Mueller¹ (), Guangsheng Liang^1.2, Miranda Scolari¹; ¹Texas Tech University, ²Lingnan University

Given competing visual information, top-down attention flexibly prioritizes relevant over irrelevant input to accommodate behavioral goals. This is demonstrated by manipulating the validity of pre-cues that provide distinguishing information about an upcoming target. Participants may deploy space-based attention (SBA) following a location cue, where any visual object in the cued location is afforded an attentional benefit compared to uncued locations. Conversely, following a color cue, they may deploy feature-based attention (FBA) to objects depicted in the cued color regardless of location. Thus, although both SBA and FBA operate on sensory stimuli, they differ in how selection is distributed. Here, we investigate whether temporal changes in pupil size are associated with the distribution of selection. Participants were given either a location (N = 19) or color (N = 19) pre-cue (80% valid) before a visual search display containing one target and three distractors. Participants then made a two-alternative-forced-choice response about an orthogonal feature of the target. Importantly, all visual aspects were well-matched between conditions to allow us to draw conclusions about cognitive systems from pupillometry, and all analyses were restricted to correct trials. Among the location-cue group, pupillary responses were significantly greater on invalid compared to valid trials at 1,010-1,810 ms from search display onset; no significant differences were detected for the color-cue group. These findings likely reflect differences in attentional distribution: Successful shifts from one invalidly cued location to multiple uncued locations result in wider selected regions, whereas successful shifts from an invalidly cued to uncued colors result in no such spatial change, consistent with the global deployment of FBA. If this explanation is correct, manipulating the validity of a location pre-cue should modulate the effect, where an unreliable cue should elicit broad selection for both valid and invalid trials. A follow-up experiment was designed to confirm this hypothesis.