Abstract
Many earlier studies have failed to find evidence that pigeons complete partially occluded objects, although positive results have been reported for other birds. We investigated amodal completion in pigeons by analyzing the errors they committed while learning a 3-alternative simultaneous discrimination task. Pigeons were trained to discriminate among occluded, complete, and incomplete stimuli. The occluded stimulus comprised two colored shapes, one of which occluded the other (e.g., blue triangle occluded by red circle); the complete and incomplete stimuli comprised a single shape that had been partially covered in the corresponding occluded stimulus (e.g., whole triangle, or partial triangle, respectively). The correct response was to select the occluded stimulus and the dependent measure was the proportion of errors made to the complete stimulus. Pigeons were trained until they reached criterion with one set of stimuli; novel sets of the stimuli were repeatedly introduced.
At the beginning of training, the proportion of errors to the complete stimuli was equal to the incomplete stimuli and did not differ from chance (50%). After repeated training with different shapes, the proportion of errors to the complete stimulus increased above chance. Moreover, when the pigeons were re-exposed to the initial set of training stimuli, they committed 70% of their errors to the complete stimulus. These results suggest that pigeons came to view the complete stimulus as being more similar to the occluded stimulus than the incomplete stimulus. Extensive experience with two-dimensional images may facilitate amodal completion of partially occluded objects in pigeons.
supported by NIMH grant R01-MH047313