The Role of Color and Shape in Bottom-Up Processing with Realistic Stimuli
Undergraduate Just-In-Time Abstract
Poster Presentation 56.358: Tuesday, May 19, 2026, 2:45 – 6:45 pm, Banyan Breezeway
Session: Undergraduate Just-In-Time 3
Schedule of Events | Search Abstracts | Symposia | Talk Sessions | Poster Sessions
Sandra Park1 (), Alejandro Lleras2, Junming Yu3, Simona Buetti4; 1University of Illinois Urbana-Champaign
While substantial work exists on characterizing top-down guidance, much less systematic work has been done on bottom-up processing. It is not currently well understood how signals from different visual dimensions combine to capture attention and the eyes, nor how these signals operate with realistic stimuli. Here, we used naturalistic stimuli (apples) presented in a 9 x 9 grid with the center removed on a uniform gray background. Participants (n = 23) searched for a target that differed from distractors by color (green or yellow among red apples), shape (one or two bites taken from an apple among whole apples), or both features combined, while eye movements were recorded using an EyeLink 1000. Overall, color contrasts produce faster saccadic initiation times (188.84 ms vs. 233.79 ms) and more accurate saccades (guessing rate of 0.097% vs. 0.5583%) than shape contrasts. In the feature-combined conditions, we observed a cross-over interaction between shape and color contrasts: shape contrasts that were highly dissimilar from the background apples reduced the contribution of color contrasts in the color-similar (but not in the color-dissimilar) condition. Likely reflecting that the shape manipulation (taking two bites out of the apple) disproportionately reduced the ability of color in the remaining part of the apple to guide attention. This suggests that the two features do not always combine additively in peripheral vision to guide attention. Overall, these results indicate that color and shape contribute to bottom-up attentional guidance in somewhat different ways. Color produces faster and more reliable guidance while shape produces smaller guidance effects. Importantly, across all conditions, saccadic accuracy strongly predicted saccadic initiation time (R2 = 0.918), with faster initiation times being associated with smaller guessing rates. This final result suggests both feature dimensions contribute in analogous though unequal fashion to the overall saccade execution system.