Individual recognition is associated with viewpoint-independent face recognition in a species-specific way.

Identifying three-dimensional signals (e.g. faces) can be challenging because the signals appear different when seen from different viewpoints. One simple solution is to always view signals from a particular viewpoint. A more flexible but also more cognitively challenging solution is viewpoint-independent recognition, where receivers can identify signals from multiple viewing angles. Here, we use same/different concept learning to test viewpoint-independent recognition for conspecific and heterospecific faces in two species of Polistes paper wasps that have three-dimensional visual signals. P. fuscatus use conspecific facial patterns for individual recognition, while P. dominula use conspecific facial patterns as a signal of fighting ability. Previous work has shown that P. fuscatus are able to identify novel viewpoints of conspecific faces through extrapolation. Here, we show that P. fuscatus and P. dominula differ in their capacity for viewpoint-independent recognition. P. fuscatus exhibit viewpoint-independent recognition for both P. fuscatus and P. dominula faces. In contrast, P. dominula do not have viewpoint-independent recognition for conspecific or heterospecific faces. These results suggest that viewpoint-independent recognition through extrapolation may be an adaptive strategy to facilitate individual face recognition across a wide range of taxa.