Viewpoint-independent face recognition via extrapolation in paper wasps

Abstract

ABSTRACT Visual recognition of three-dimensional signals, such as faces, is challenging because the signals appear different from different viewpoints. A flexible but cognitively challenging solution is viewpoint-independent recognition, where receivers identify signals from novel viewing angles. Here, we used same/different concept learning to test viewpoint-independent face recognition in Polistes fuscatus, a wasp that uses facial patterns to individually identify conspecifics. We found that wasps use extrapolation to identify novel views of conspecific faces. For example, wasps identify a pair of pictures of the same wasp as the ‘same’, even if the pictures are taken from different views (e.g. one face 0 deg rotation, one face 60 deg rotation). This result is notable because it provides the first evidence of view-invariant recognition via extrapolation in an invertebrate. The results suggest that viewpoint-independent recognition via extrapolation may be a widespread strategy to facilitate individual face recognition.