It was previously believed that the way humans recognize their peers is through a set of nerve cells that specifically respond to certain faces. While this explanation for how the brain recognizes and responds to human faces makes sense, attributing facial recognition of all the people we encounter on a day to day basis would require a vast amount of neurons in order to process all of that information.
Now after 15 years of research, a pair of scientists from Caltech may have just cracked the code for human facial recognition, and they discovered this breakthrough with the help of macaques.
A Single Image From Separate Processes
Doris Y. Tsao and her associate Steven Le Chang found evidence that perhaps neurons do not respond to specific faces by merely identifying them, but in fact a group of face cells recognize — not faces — but separate facial features that eventually provide a complete picture of a face.
What researchers first did was to identify the 50 variables that represent the most complex differences in facial features between 200 face photos. From there, they extracted 2,000 faces and projected the photos onto a screen for two macaques to view.
Using a combination of brain imaging and single-neuron recording, the team recorded a total of 205 cells from the two macaques.
Specifically, cells in the anterior medial patch would process information about the appearance of faces such as the hairline, while cells in another group, the ones in the middle fundus and middle lateral patches would process the specific shapes in the face such as the contours of the lips and eyes. These two processes would then build a single image of a face based on the separately gathered information.
Essentially, these findings show that contrary to prior beliefs on facial recognition where certain faces are merely identified as a whole, the face patches evidently measure and recognize a face based on combined information of facial features and shapes gathered by face cells.
The study was published June 1 in the journal Cell.
Primate Facial Recognition
Tsao is not new to studying facial recognition. In the year 2000, she and an electrophysiologist colleague of hers, Winrich Freiwald, gathered intracranial recordings of monkeys as they viewed a slideshow of various objects and human faces.
What they found was that different neurons in the monkeys' middle face patch would be activated each time a human face flashed on the screen, a reaction that is absent when inanimate objects or other body parts were shown. What's more, further research showed that different human faces activated different neurons in the face patch each time.
What this showed then is that primates have the incredible ability to recognize and differentiate specific human faces. Further, at the time, this led to the belief that the different neurons in the face patches are merely sensitive to certain different faces.
While primate facial recognition continues to amaze scientists, this previous belief regarding facial recognition has now changed along with the latest study where they have essentially cracked the brain's facial recognition code.