Long before we learn anything, we learn to process and recognise faces. At as early as four months, a Stanford research shows, infants’ brains do this with near-adult competence. May be this is why it is generally easy to recall a face than a name or any other particular, and to pick out a familiar face in a crowd. How do we do it, though?
Face recognition technologies are based on algorithms that identify unique “landmarks” of a subject’s face—say, the relative size, shape and position of her eyes or nose or jaws—and associate it with any recorded photographic or video data. Findings of a study published in the journal, Cell, say our brain uses pretty much the same method. Previously, all sorts of theories on face recognition have been propounded—and demolished—including one that held that individual neurons store information regarding individual faces. Neuroscientists Doris Tsao and Le Chang at the California Institute of Technology conducted a study involving rhesus monkeys and found that each ‘face cell’, in a group of hundred of neurons, pays attention to specific ranked combinations of facial features and the cluster, working in tandem, signals the correct identification.
Tsao and Chang, to identifying the locus of the animals’ face cells, showed two monkeys various images, of faces, bodies, fruit and random patterns and used MRI to see brain regions were active when the monkeys saw a face. Then, the animals were shown 2,000 pictures of human faces with varying features and implanted electrodes in their brains to study responses of individual neurons to facial distinctions. The response of each individual neuron in a cluster of 205 in each animal corresponded with a specific combination of some of the facial features.
To test whether each neuron was better tuned to a particular or a few particular features within a specific combination of features, the researchers tried to recreate the faces the monkeys were shown on the basis of each neuron’s response to its lot of characteristics. Though the recreations nearly mirrored the real faces, when the monkeys were shown faces that varied according to the features that a particular neuron was indifferent to, the face cell didn’t fire. Basically, in order to recollect a face, all face cells need to fire together, with each picking up a specific set of visual cues. This is important to understanding—and perhaps alleviating—prosopoagnosia, the dysfunction (because of developmental/congenital reasons or acute brain damage) which impairs face recognition in humans. The brain’s fusiform gyrus—the region also associated with dyslexia—helps humans process faces with greater detail than inanimate objects with similarly complex features. With prosopoagnosia, the patient must rely on the inferior ‘object recognition system’ of the brain.