Primate face recognition
Jun 21, 2017
Our brains have evolved to recognize and remember faces ever since we were infants, since the first thing we learn is to look at the faces of those around us and recognize them, and later instantly identify them. Brain-imaging studies revealed that we have the so-called face patches in our temporal lobe in our bran that specialize in responding to faces. However, until recently, neuroscientists were still not sure how exactly the cells in face patches work.
It seems that this has changed, since biologist Doris Tsao and her colleagues at Caltech used a combination of brain imaging and single-neuron recording in macaque monkeys to crack the neural code for face recognition. They found out that the firing rate of each cell corresponds to separate facial features along an axis, and the cells are fine-tuned to bits of information like a set of dials, and they can channel these bits together in different combination to create an image of every possible face. They were able to recreate a face that the monkey saw by simply tracking the electrical activity of its face cells.
They only needed readings from a small set of neurons for their algorithm to accurately recreate the faces monkeys were viewing – 106 cells in one patch and 99 in another were enough – which emphasizes how compact and efficient primate feature-based neural code is.
Want to learn more? Read about our face-recognition software and try it out.
- Chang, Le et al. (2017): The Code for Facial Identity in the Primate Brain. Cell 169.6: 1013-1028.