Abstract
Mirror image confusion refers to the phenomenon whereby views of visual objects reflected about a central axis are easily confused. Behavioral studies in various animal species have led to the conclusion that lateral mirror images are inherently more confusable than vertical mirror images. None of these studies, however, used large batteries of complex, unfamiliar visual patterns as test stimuli. In this study we trained two monkeys to perform a delayed match-to-sample task in which the non-match item was sometimes either a lateral or vertical mirror reflection of the sample. To eliminate any contribution of prior experience, we developed a stimulus generation method that ensured that novel objects containing no axis of symmetry could be used on a trial-by-trial basis. Across a large number of test sessions, we found no evidence that monkeys are more likely to confuse horizontally reflected mirror images than their vertical mirror image counterparts. This trend was seen throughout many days of experimental testing, during which both monkeys showed steady overall task improvement. We attribute the lack of exaggerated lateral versus vertical mirror image confusion to the use of both novel and complex stimuli, and suggest that previously observed biases in mirror image confusion are likely a result of either attention or experience, but not inherent perceptual biases.