A facial recognition coding system in the brains of primates (humans).

When you look at a face photo, the brain immediately identifies who the person in the photo is or whether you've seen it before.
years, neuroscientists have been trying to figure out how the brain recognizes and perceives faces.
, published in the journal Cell, showed that scientists already know the face recognition coding system in the brains of primates.
" we found that the coding system was very simple.
Found that when a monkey sees a face, only 205 neurons in its brain are involved in the coding process," said Doris Tsao, a professor of bioengineering at the California Institute of Technology.
earlier study, the Tsao team used functional nuclear resonance imaging to identify areas of the brain responsible for identifying faces in humans and other primates.
they called the six areas located in the lower temporal corty as face patches.
further studies have shown that these areas are crammed with special nerve cells.
new study, Tsao and his postdoctoral researcher Steven Le Chang found that each facial recognition cell represents a specific axis in a multidisory space.
researchers call this space facial space.
and red, blue, green and light combine in different ways to produce different colors, and these axes are combined in different ways to produce each possible face.
researchers say a face can be broken down into 50 dimensions.
25-dimensional is related to shape, such as the distance between the eyes or the width of the hairline, while the other 25-dimensional is independent of shape and mainly represents features such as skin tone and texture.
Tsao team inserted electrodes into the macaque's brain to record signals from facial recognition cells in the face patch area.
found that when a face is cast into a single axis within 50-dimensional facial space, each cell is stimulated proportionally.
, the researchers developed an algorithm to decode the neural response of facial recognition.
results showed that the macaque's two facial patches had enough cells to reconstruct the face -- 106 cells in one area and 99 in one area.
, the results are not only of great significance to the study of neurocoding, but also applied to the field of artificial intelligence.
.