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 <i>Polistes</i> paper wasps that have three-dimensional visual signals. <i>P. fuscatus</i> use conspecific facial patterns for individual recognition, while <i>P. dominula</i> use conspecific facial patterns as a signal of fighting ability. Previous work has shown that <i>P. fuscatus</i> are able to identify novel viewpoints of conspecific faces through extrapolation. Here, we show that <i>P. fuscatus</i> and <i>P. dominula</i> differ in their capacity for viewpoint-independent recognition. <i>P. fuscatus</i> exhibit viewpoint-independent recognition for both <i>P. fuscatus</i> and <i>P. dominula</i> faces. In contrast, <i>P. dominula</i> 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.