Element and compound matching-to-sample performance in pigeons: the roles of information load and training history.

Pigeons were trained to match color and line orientation element or compound samples in a symbolic matching-to-sample task. In subsequent test sessions with element and compound samples, there was an initial superiority of element matching for the element-trained group and of compound matching for the compound-trained group. This difference persisted over the course of 100 test sessions for the element-trained group, whereas element- and compound-matching accuracy converged for the compound-trained group. In a second experiment, in which sample duration was manipulated, element-matching accuracy was superior to compound-matching accuracy for both groups. Thus, element-matching accuracy was superior to compound-matching accuracy under conditions that rule out generalization decrement and training history as explanations. The data are interpreted as supporting the view that the dimensions of visual compound stimuli compete for a limited cognitive resource.