Skip to main navigation Skip to main content
The University of Southampton
The Hummingbird Project

Super-recognisers, sheep, and chimpanzees (oh my?)

Published: 12 March 2019
Searching for faces

A summary of recent articles that caught our eye. This time, we're exploring the strategies used by super-recognisers - people with extraordinary face recognition ability - and how they compare to those used by the rest of us. Also included is a critique of a study claiming that sheep have comparable facial recognition abilities to humans, and an interesting application of facial recognition technology to help fight illegal wildlife trade.

Super-recognisers

At the Hummingbird Project, we are interested in whether human strategies for face and voice recognition can be built into computer algorithms in order to improve performance. Therefore, one of the questions that we get asked is why we study the strategies used by ordinary humans to recognise faces and voices. Should we not be studying those with extraordinary face recognition ability, known as super-recognisers? Fair point. But actually if we look into the research on super (face) recognisers, we see that the strategies they use are not substantially different from those used by average face recognisers.

Use of face information varies systematically from developmental prosopagnosics to super-recognizers

A recent study by Tardif and colleagues illustrates this point. In an eye-tracking study of 112 people with a range of face processing ability – from prosopagnosia to super-recognition – the authors found that the eyes, eyebrows, and mouth are the most important regions for face identification across the face processing spectrum. Thus, the difference between super-recognisers and those with prosopagnosia is one of quantity (the amount of information used) rather than quality (the type of information used).

Tardif, J., Morin Duchesne, X., Cohan, S., Royer, J., Blais, C., Fiset, D., ... & Gosselin, F. (2018). Use of face information varies systematically from developmental prosopagnosics to super-recognizers. Psychological science, 0956797618811338.

The limits of super recognition: An other-ethnicity effect in individuals with extraordinary face recognition skills

Further evidence that super-recognisers are not all that different from the rest of us (just better) comes from a recent study by Sarah Bate and colleagues. The authors studied the well-known other-ethnicity effect, in which people are worse at recognising faces of ethnicities different to their own, in a group of super-recognisers. Across 3 studies of face matching and memory, Caucasian super-recognisers performed worse for other-ethnicity faces compared to same-ethnicity faces. These findings are important because they suggest that super-recognisers have the same biases in facial recognition as average face processors. Therefore, when seeking to identify other-ethnicity faces, super-recognisers may not be much better than average face processors of the same ethnicity.

Bate, S., Bennetts, R., Hasshim, N., Portch, E., Murray, E., Burns, E., & Dudfield, G. (2018). The limits of super recognition: An other-ethnicity effect in individuals with extraordinary face recognition skills. Journal of Experimental Psychology: Human Perception and Performance.

Cognitive and neural markers of super-recognisers’ face processing superiority and enhanced cross-age effect

Super-recognisers also exhibit cross-age effects, with better face recognition performance for adult compared to infant faces. However, this effect appears to be modulated to some extent by the level of exposure, as one super-recogniser who worked in a paediatric ward showed no advantage for adult over infant faces.

Belanova, E., Davis, J. P., & Thompson, T. (2018). Cognitive and neural markers of super-recognisers’ face processing superiority and enhanced cross-age effect. Cortex, 108, 92-111.

Taken together, these studies suggest that face processing ability is a spectrum, with super-recognition at one end and prosopagnosia at the other. Super-recognisers and average face processors rely on the same type of information to support recognition and are subject to the same biases.

Other Interesting Items

Are face recognition abilities in humans and sheep really ‘comparable’?

Although most studies of human face recognition use human participants, some studies have explored how other species recognize human faces. In a recent study, Knolle and colleagues claimed that sheep had face recognition abilities ‘comparable with those of humans.’ However, a group of human face recognition researchers lead by Alice Towler argue that this claim is unjustified. While the study by Knolle and colleagues did show that sheep can learn to recognise human faces that were paired with reward (cereal-based pellets), this is not enough to suggest that sheep have comparable face recognition abilities to those of humans. For one, there is no study in human that used an equivalent task to that completed by the sheep. Secondly, sheep performance was worse than would be expected for humans under similar conditions. Finally, when presented with a picture of their handler (a highly familiar person), sheep did not perform much better than for the newly learned faces. This stands in contrast to the human literature on face recognition, which shows significantly better performance for familiar compared to newly learned faces. Therefore, Towler and colleagues suggested instead that sheep and humans rely on fundamentally different mechanisms to recognise human faces. For links to both the original paper and comment by Towler et al., follow the links below:

Knolle, F., Goncalves, R. P., & Morton, A. J. (2017). Sheep recognize familiar and unfamiliar human faces from two-dimensional images. Royal Society open science, 4(11), 171228.

Towler, A., Kemp, R. I., Bruce, V., Burton, A. M., Dunn, J. D., & White, D. (2019). Are face recognition abilities in humans and sheep really ‘comparable’?. Royal Society Open Science, 6(1), 180772.

ChimpFace algorithm aims at fighting illegal wildlife trade

In other animal-related news, (chimpanzee) facial recognition may soon be used to help fight illegal animal trafficking. A new computer algorithm has been developed that will search through social media posts for faces of trafficked animals and flag these posts to human experts who can take action. For more about the project: https://conservationx.com/project/id/8

Privacy Settings