Search

Pigeons discriminated the length of a bar located between two context lines. Responses to one key were reinforced when the bar was longer than a predetermined length, and those to the other key were reinforced when the bar was shorter. The inclination of the context lines was systematically varied from 54.6 degrees (converging upward) to 125.4 degrees (converging downward). Five out of 6 subjects tended to report 'long' when the bars were located near the apex of the context lines, regardless of whether the context lines were oriented upward or downward. The magnitude of the illusion varied almost linearly with the ratio of the length of the stimulus bar to the gap between the bar and the context lines. This relationship held equally for upward- and downward-converging context lines.

Three experiments were performed to address factors contributing to the Ponzo illusion. In Experiment 1, pigeons learned to peck at the longer of two bars in varying line contexts. When these lines converged, the birds had difficulty in learning several patterns in which a Ponzo illusion would reduce the perceived difference. In Experiment 2, the subjects chose one response if a stimulus bar was longer than a predetermined length and a second response for shorter bars. The subjects were more likely to choose 'long' as the stimulus bar approached the apex of converging lines. These results suggested that pigeons experience the Ponzo illusion. In Experiment 3, the effects fo contexts that did and did not form a texture gradient were compared. The magnitude of the illusion did not differ according to context lines. This result failed to support a perspective theory of the Ponzo illusion.