The discrimination of structure: III. Representation of spatial relationships.

Pigeons received a discrimination in which the spatial relationship between 2 adjacent rectangles filled with different colors signaled the trial outcome. Test trials then involved the same rectangles separated horizontally by a gap. The tests in Experiment 1 disrupted the discrimination more when the rectangles were tall and thin than when they were short and wide. Experiment 2 revealed that the width of the rectangles rather than their height determined the extent to which separating them would disrupt the original discrimination. The results are explained in terms of a template-matching account of pattern recognition with the additional assumption, supported by Experiment 3, that the size of a template can be altered to improve its match with a test pattern.