Douglas Addleman^1,2 (), Viola Stoermer²; ¹Gonzaga University, ²Dartmouth College

Knowledge about non-spatial features of targets can speed visual search when either explicitly provided by an external cue or incidentally learned through experience. Whether these search advantages are due to attentional guidance towards the relevant feature or to post-selective processes is not well understood, particularly for learned feature-based attention. Here, we manipulated visual search set size to test whether incidental learning (Exp. 1) and explicit cues (Exps. 2 & 3) guide attention towards likely target features. Participants searched for a target C (left or right gap) among either three (set size 4) or seven (set size 8) distractor Cs (top or bottom gaps). In Exp. 1 (N=64), targets–unbeknownst to participants–appeared more often in one of the four potential item colors (66% of trials). We found overall search advantages for the frequent target color, replicating previous work, but this advantage did not interact with set size (p = .207), indicating that learning about target colors did not reliably guide attention. Exp. 2 (N=63) used the same paradigm but explicitly cued participants with the likely target color before each trial. There was evidence for guidance from a modest interaction between set size and cue condition (p < .001, ηp2 = .11). Exp. 3 (N=25) tested whether more predictive cues would elicit stronger guidance. Participants completed alternating search blocks containing either 100% predictive color cues or no cues. Here, cues clearly guided attention, with shallower search slopes on cue trials (slope = 54ms/item) than non-cued trials (slope=66ms/item; p < .001, ηp2 = .41). Thus, the amount of guidance was modulated by the validity of the cue and was only reliably present during cued, but not learned attention. Overall, this suggests that different processes underlie how incidental learning and explicit cueing benefit feature-based attention, with cueing more readily affecting attentional guidance than experience.

Acknowledgements: This research was supported in part by NSF Grant BCS-1850738.