Abstract
Some studies have shown evidence that various animals can recognize partly occluded objects in the same manner as humans. However, there is no positive evidence for amodal completion in pigeons. In this study, we re-investigated the perception of amodal completion together with perceptual transparency in pigeons.
Stimulus was a combination of one horizontal and two vertical rectangles. One of the vertical rectangles was placed in front and the other was placed behind the horizontal rectangle. Four pigeons were separated into two groups (Front and Back group) and trained in 2AFC tasks. The subjects in the Front group were trained to peck a vertical rectangle that was placed in front of the horizontal rectangle, while subjects in the Back group were trained to peck a vertical rectangle lying behind the horizontal one. After subjects reached the training criterion, they proceeded to a test phase. A test session consisted of 80 baseline trials of trained stimuli and 10 trials of probe stimuli randomly inserted in the baseline trials. The probe stimuli differed from the trained stimuli in the luminance of the area where the vertical and horizontal rectangles overlapped, and induced perceptual transparency while maintaining the depth relations among the rectangles.
We analyzed the number of peckings at the probe stimuli. The pecking behavior to trial stimuli was analogous to that to trained stimuli. Subjects in the Front group responded significantly more to the rectangle that was seen transparent in front of the horizontal rectangle. On the other hand, the Back group responded significantly more to the rectangle that was placed behind the horizontal transparent rectangle. The results indicate that the pigeons in Back group didn't perceive the occluded vertical rectangle as two fragments, but as a unitary rectangle in training sessions. These results suggest that pigeons are capable of perceiving both transparency and amodal completion in the same way as humans.