Abstract
When a spiky object is occluded, we have no trouble believing that the hidden portions are also spiky. Whether this process occurs automatically in perception, or requires cognitive reasoning is unclear. Here, we devised a novel “free-choice” paradigm to reveal what monkeys see behind an occluded display. On 80% of the trials, denoted as unambiguous trials, two items (sample and test) were shown with an intervening delay, and monkeys were trained to make a “same” or “different” response. These trials consisted of unoccluded displays with two spatially separated shapes. On the remaining 20% of the trials, denoted as free-choice trials, monkeys were rewarded for making either response. These free-choice trials contained an occluded display as sample, and an unoccluded display as test. For each occluded target display, we included three test displays containing separated items: (1) A mosaic display in which the target shape is moved apart without any completing contours; (2) A “likely completion” shape in which the statistics of the hidden portions are matched to the visible portions; and (3) An “unlikely completion” shape in which the statistics of the hidden portions were mismatched. Since monkeys were never explicitly trained on occluded displays, their responses represent an unbiased account of what they see in the occluded display. Our main findings are as follows. Both monkeys were highly accurate in the unambiguous trials (88.5% across monkeys M1 and M2). Importantly, on the free choice trials, they responded “same” more frequently when they saw likely completions compared to mosaic and unlikely completions (% same response for likely, unlikely and mosaic: 53%, 34% and 3%, p < 0.005 on a chi-squared test for likely vs unlikely & likely vs mosaic). Taken together, our results confirm that monkeys indeed see what we see when shapes are occluded.