Scientists create facial models using identity information stored in the brain

Scientists have worked out how to construct 3D facial models using the information stored in an individual's brain when recalling the face of a familiar person.

In a world first, neuroscientists from the University of Glasgow claim to have "cracked the code" of what defines visual identity, and have generated it with a computer program.

The discovery will lead to a greater understanding of how the brain identifies faces, and could have applications for AI, gaming technology and eyewitness testimony.

"It's difficult to understand what information people store in their memory when they recognise familiar faces," said Philippe Schyns, Professor of Visual Cognition at the Institute of Neuroscience and Psychology.

"But we have developed a tool which has essentially given us a method to do just that.

"By reverse engineering the information that characterises someone's identity, and then mathematically representing it, we were then able to render it graphically."

In the experiment, volunteers were asked to rate the resemblance between a remembered familiar face, and randomly generated faces that shared factors of sex, age and ethnicity.

From this, the researchers were able to work out what information is specific to the identity of an individual in someone else's memory.

They then used complex mathematical models to teach the computer to recreate the remembered face with fairly good accuracy.

They could also use the computer model to age the recreated face, and change its gender or ethnicity.

The paper, Modelling Face Memory Reveals Task-Generalizable Representations , is published today in the journal Nature Human Behaviour.

Source: Read Full Article