Preception of depth from geometrical cues differs between monkeys and humans

We have previously studied monkey 3D shape perception using a depth adjustment task (ARVO 2000). In this task stimuli consisted of computer generated 3D objects like the Necker cube that were presented in stereoscopic or monocular view. Subjects were required to align a stereoscopic cursor with one of the object edges. Unlike human subjects, monkeys do not seem to perceive depth reversals of the Necker cube in monocular view suggesting that their depth perception may differ from that of humans. When matching the depth of these displays monkey subjects never produced the bimodal distributions that were typical for human subjects. For the three monkeys that were studied extensively, Monkey B set the cursor depth consistently near zero while the other two produced scattered settings in the 'near' range. To eliminate the possibility that monkeys had difficulty adjusting a binocular cursor to a monocular display we used a binocular priming paradigm. In this task, the test objects were initially displayed binocularly and then switched to monocular when the monkey started to adjust the cursor. Under these conditions Monkey B consistently produced settings that varied according to the primed disparity. Moreover, the monkey produced random settings in control trials in which the monocular display was replaced by a blank. Thus, these results show that monkeys are capable of aligning the cursor with a monocular object and are able to perceive the two modes of the Necker cube. We conclude that some displays that produce compelling monocular depth perception for humans may look flat to monkeys. Our findings suggest that monkeys rely more strongly on binocular stereopsis than do humans.