Abstract
Buetti, Xu and Lleras (in press) measured RTs when participants searched for a target that differed from distractors along a single feature, either color or shape (unidimensional search) and used it to predict RTs when targets differed from distractors along both color and shape (bidimensional search). Three models were compared: the Best Guidance model (the most distinctive feature determines performance), the Orthogonal Combination model (features are integral, Euclidian metric), and the Collinear Combination model (features are separable, city-block metric). The results favored the Collinear Combination model, indicating that shape and color are separable features and thus the visual distinctiveness along each dimension combined linearly to produce the overall distinctiveness of a target defined by color and shape. Here we present a new set of experiments studying combinations of two different feature dimensions. In Experiments 1-5, we explored how unidimensional search for texture (Experiment 1) and shape (Experiment 2) combine to determine the distinctiveness of the target when it differed from distractors along both texture and shape (Experiments 3-5). We predicted RTs in each of the 12 conditions that were run in each experiment (4 set sizes x 3 different distractor types), for a total of 36 separate predictions across three new groups of subjects. Results showed that the orthogonal combination model gave the best prediction (R^2= 90%, with a mean prediction error of 14 ms, with a prediction range of 190ms), indicating that texture and shape are integral features and the distinctiveness along these single dimensions combine with a Euclidian metric to determine the overall bidimensional distinctiveness. Furthermore, there was evidence these two dimensions coactivated: processing efficiency improved by 25%. Overall, this project provides a new framework for understanding how a target distinctiveness is computed by the visual system, as a function of each of the target’s defining visual dimensions.