Jun-Ming Yu¹ (), Zoe Jing Xu^1,2, Alejandro Lleras¹, Buetti Simona¹; ¹University of Illinois, Urbana Champaign, ²University of Washington

The visual attention system is thought to be highly adaptive to the ecology of the environment. Here, we attempted to quantify how the visual system prioritizes various visual features as a function of target-distractor featural contrast. Specifically, we quantified the contributions of color and shape to attentional guidance in conditions where shape and color dimensions varied in terms of their relative contrast. We first measured featural contrasts, indexed by logarithmic search efficiency, between targets and a wide range of homogeneous distractors in unidimensional searches. In Experiment 1, participants searched for a target differing from distractors only in shape (10 shapes). In Experiment 2, they searched for a target differing from distractors only in color (8 colors). In Experiments 3-4, the target differed from distractors along both color and shape and we selected shape and color features so that one dimension had larger contrasts than the other. In Experiment 3 (color contrasts > shape contrasts), the color logarithmic search slopes were 60, 68, 93, whereas the shape slopes were 276, 175 and 150. In Experiment 4 (color contrasts < shape contrasts), the color search slopes were 249, 144 and 140, whereas the shape slopes were 87, 75, 52. Contrary to our predictions, the results of Experiment 3 showed that participants relied more on the shape dimension, even though relying more on color dimension would have led to better performance. In Experiment 4, participants relied to equal extents on both dimensions, even though relying more on shape would have led to better performance. Overall, the results suggests that participants might not always rely on the most efficient feature to search. A follow-up study using more efficient search conditions for the larger contrast dimension was run to investigate at which point participants choose to preferentially attend to the most useful feature.