Neural Code for Facial Recognition

In the article, “Neural correlates of individuation and categorization of other-species faces in infancy" by Dixon et al., researchers examine the use of familiarization effects on individuation and subordinate-level categorization in infants. At 6 months, infants are equally likely to differentiate between human faces and other species’ faces. By 9 months, infants have a lesser ability to distinguish novelty and pre-exposed species’ faces in non-humans when compared to human faces. The researchers predict that this is due to infants having much more exposure to a variety of human faces than any other species. How exactly does the brain process facial features in order to be able to make these conclusions? Researchers at the California Institute of Technology conducted a study that provided scientists with a systematic model for how the brain perceives faces by finding the firing rate of each cell in a face patch, or region of cells in the temporal lobe specialized to respond to faces, corresponding to an individual facial feature. In this study, human mugshots with variations in 50 features were shown to monkeys whose electrical activity was measured from individual neurons in three separate face patches. These features included face shape, skin tone, and distance between the eyes. They found that each neuron responded to only a single feature and each patch processed complementary information. The information from different cells in these patches was then integrated to provide a comprehensive image. They then tested their mathematical model of encoding by showing a novel face to the monkey and used the neural activity to recreate the image. The recreated image was practically indistinguishable from the image shown to the monkey. Monkeys have a similar cerebral machinery to humans and it has been shown that our face patches respond in the same way as theirs through fMRI studies (Sheikh, 2017). Given this, we expect humans to have a highly comparable neural code for facial recognition. The data from this study provides a framework for decoding how the brain processes facial information, and could potentially extend to non-facial object recognition as well. 


References
Dixon, K. C., Reynolds, G. D., Romano, A. C., Roth, K. C., Stumpe, A. L., Guy, M. W., &  Mosteller, S. M. (2019). Neural correlates of individuation and categorization of other- species faces in infancy. Neuropsychologia, 126, 27–35. https://doi-org.flagship.luc.edu/ 10.1016/j.neuropsychologia.2017.09.037

Sheikh, K. (2017). Saving Face. Scientific American, 317(2), 12–14. https://doi- org.flagship.luc.edu/10.1038/scientificamerican0817-12


https://doi-org.flagship.luc.edu/ 10.1016/j.neuropsychologia.2017.09.037
https://doi-org.flagship.luc.edu/10.1038/scientificamerican0817-12