There’s lots of evidence that humans have a specialized mechanism for identifying and responding to faces; for example, people with a condition called prosopagnosia have difficulty recognizing faces but not other objects.  A few years ago, researchers showed that individual paper wasps of the species Polistes fuscatus recognize each other’s faces; the same team has now gone on to show that, like humans, P. fuscatus accomplishes this via a specialized mechanism for facial recognition rather than through general shape or pattern recognition.  This story is an excellent example of a  complex cognitive ability being exhibited by a creature with a relatively simple nervous system.

Sheehan & Tibbetts studied the recognition abilities of paper wasps using a T-shaped maze with pictorial cues at the intersection.  Wild-caught adult female wasps were introduced into the maze and chose to go down one arm or the other; the entire floor of the maze was electrified except for a “safe zone” in one arm which was consistently associated with one of a pair of images of wasp faces.  Each wasp was tested 40 times with a pair of images; depending on how well the wasp could distinguish and recognize the two images, she would learn to go down the arm with the “safe” image.  The researchers used changes in the speed and accuracy of this decision to measure the wasps’s ability to learn.  In order to compare facial recognition with other kinds of discrimination, the researchers also used paired images of geometric patterns, caterpillars (the wasps’ prey) or wasp faces that had been digitally manipulated (either rearranged or antenna-less).

The researchers found that Polistes fuscatus females were quicker and more accurate at learning to distinguish pairs of faces than paired patterns or paired caterpillars.  The wasps also had trouble with antenna-less or rearranged faces, learning to recognize them about as well as they did the patterns or caterpillars.  This suggests that the digitally altered faces were being recognized and learned by the same general process, but that these wasps have a specialized mechanism geared specifically towards facial recognition, allowing them to more quickly and accurately learn faces.

By contrast, wasps of another species (P. metricus) were unable to learn to recognize images of faces; after 40 trials, they still performed no better than chance.  This isn’t due to a general difference in visual learning, since P. metricus learned to recognize patterns and caterpillars about as well as P. fuscatus did.  It’s also unlikely to be due to a difference in visual systems; in fact, the researchers suggest that P. metricus should have more acute vision than P. fuscatus (based on morphological measurements).  The difference in performance seems to result from P. metricus lacking a specific facial recognition system.

P. fuscatus faces are more variable than those of P. metricus, so it’s possible that these results are due to how recognizable the individual images are rather than to a cognitive difference in recognition ability.  In order to test for this, the researchers tested the ability of individuals of each species to recognize faces of the other species.  P. fuscatus learned to recognize individuals of either species more quickly than P. metricus did.  Interestingly, P. metricus individuals were eventually able to learn to recognize P. fuscatus faces, despite being unable to distinguish individuals of their own species.  However, since they recognized the faces about as well as they could recognize caterpillars and did just as well even when the antenna were digitally removed, it’s unlikely that they were using a specific facial recognition mechanism; it may be that the more variable P. fuscatus faces are easier to distinguish with a general pattern recognition mechanism.

The researchers suggest that the difference in recognition abilities may be because P. fuscatus are social wasps, unlike P. metricus.  P. fuscatus nests are established by a co-operative group of queens; there is a strict dominance hierarchy determining reproduction, making it important to be able to recognize other individuals.  By contrast, P. metricus usually nests alone, meaning there isn’t a similar pressure to evolve individual recognition.

In mammals, facial recognition involves several brain regions and even specialized neurons.  Wasps have a much simpler nervous system, yet this research shows that they have been able to evolve a similar facial recognition ability.  The neurological mechanism behind this ability isn’t known and the authors highlight this as an avenue for further research.   It’s also interesting that complex cognitive abilities often seem to have evolved in response to the needs of social interaction in animals as diverse as bees, ravens and dolphins.  Evolution is remarkably effective at generating solutions to a problem; maybe results like these should serve as a reminder not to appraise other organisms and their abilities on the basis of things like neural complexity.

Ref: Sheehan, M., & Tibbetts, E. (2011). Specialized Face Learning Is Associated with Individual Recognition in Paper Wasps Science, 334 (6060), 1272-1275 DOI: 10.1126/science.1211334