Predicting how color and shape combine in the human visual system to direct attention

Abstract

Objects in a scene can be distinct from one another along a multitude of visual attributes, such as color and shape, and the more distinct an object is from its surroundings, the easier it is to find it. However, exactly how this distinctiveness advantage arises in vision is not well understood. Here we studied whether and how visual distinctiveness along different visual attributes (color and shape, assessed in four experiments) combine to determine an object's overall distinctiveness in a scene. Unidimensional distinctiveness scores were used to predict performance in six separate experiments where a target object differed from distractor objects along both color and shape. Results showed that there is mathematical law determining overall distinctiveness as the simple sum of the distinctiveness scores along each visual attribute. Thus, the brain must compute distinctiveness scores independently for each visual attribute before summing them into the overall score that directs human attention.

Figure 1

Top panel: Illustration of the approach, using geometric shape stimuli. Performance is first evaluated in homogeneous displays containing only one type of distractor on any given trial. Performance in this task is then used to predict performance in displays simultaneously containing multiple types of items in varying proportions. Bottom panel: Observed Reaction Times in heterogeneous search tasks as a function of predicted Reaction Times based on Eq. 1 when geometric stimuli were tested (left panel, Lleras et al.) and when images of real-world objects were used as stimuli (right panel, Wang et al.).

Figure 2

Parallel search efficiency (i.e., logarithmic search slope) varies as a function of target-distractor similarity. The figures show the reaction times when searching for a red triangle among either blue circles, yellow triangles, or orange diamonds (left panel), for a cyan semicircle among the same set of distractors (middle panel), and for a teddy bear among either carrot-top dolls, gray reindeers, or car toys (right panel). High-similarity distractors lead to steeper logarithmic search slopes than low-similarity distractors, indicating longer processing times per item. The left and middle figures have been adapted from Buetti et al. and the right figure from Wang et al..

Figure 3

Illustration of the approach used in the present study. In Step 1A, parallel search efficiency was evaluated when participants reported the direction of the cyan semicircle, that was presented among 0, 1, 4, 9, 19, or 31 homogeneous distractors (orange, yellow, or blue) of identical shape. In Step 1B, parallel efficiency was evaluated when participants reported the direction of the gray semicircle, that was presented among a set of 0, 1, 4, 9, 19, or 31 homogeneous gray distractors (either diamonds, circles, or triangles). Steps 1A and 1B illustrate the experimental conditions used in Experiments 1A and 1B. In Step 2, parallel efficiency was evaluated when participants searched targets in displays that contained combinations of features from Steps 1 and 2 (compound stimuli as used in Experiment 2A–C).

Figure 4

Top: Observed search slopes (D parameters, Experiments 2A–C and 4A–C) plotted against the predicted search slopes for the three models considered. The dotted line, y = x, is also plotted for reference as it indicates a perfect prediction. Symbols in the figure closely represent the distractor types in the different experiments and are detailed in the Supplementary Text. Bottom: The left, middle, and right figures show the observed reaction times plotted against the predicted reaction times for the Best feature guidance model, the Orthogonal contrast integration model, and the Collinear contrast integration model, respectively. Error bars on each data point indicate the standard error of the observed reaction time for each specific condition.