Researchers reconstruct facial images locked in a viewer's mind

Researchers reconstruct facial images locked in a viewer’s mind
Credit: Alan Cowen

Using only data from an fMRI scan, researchers led by a Yale University undergraduate have accurately reconstructed images of human faces as viewed by other people.

"It is a form of mind reading," said Marvin Chun, professor of psychology, cognitive science and neurobiology and an author of the paper in the journal Neuroimage.

The increased level of sophistication of fMRI scans has already enabled scientists to use data from scans taken as individuals view scenes and predict whether a subject was, for instance, viewing a beach or city scene, an animal or a building.

"But they can only tell you they are viewing an animal or a building, not what animal or building," Chun said. "This is a different level of sophistication."

One of Chun's students, Alan S. Cowen, then a Yale junior now pursing an advanced degree at the University of California at Berkeley, wanted to know whether it would be possible to reconstruct a human face from patterns of brain activity. The task was daunting, because faces are more similar to each other than buildings. Also large areas of the brain are recruited in the processing of human faces, a testament to its importance in survival.

"We perceive faces in a much greater level of detail than we perceive other things," Cowen said.

Working with funding from the Yale Provost's office, Cowen and post doctoral researcher Brice Kuhl, now an assistant professor at New York University, showed six subjects 300 different "training" faces while undergoing fMRI scans. They used the data to create a sort of statistical library of how those brains responded to individual faces. They then showed the six subjects new sets of faces while they were undergoing scans. Taking that fMRI data alone, researchers used their statistical library to reconstruct the faces their subjects were viewing.

Cowen said the accuracy of these facial reconstructions will increase with time and he envisions they can be used as a research tool, for instance in studying how autistic children respond to .

Chun said the study shows the value of funding research ambitions of Yale undergraduates.

"I would never have received external funding for this, it was too novel," Chun said.

add to favorites email to friend print save as pdf

Related Stories

Dogs recognize familiar faces from images

Dec 18, 2013

So far the specialized skill for recognizing facial features holistically has been assumed to be a quality that only humans and possibly primates possess. Although it's well known, that faces and eye contact ...

Brain differences seen in depressed preschoolers

Jul 01, 2013

(Medical Xpress)—A key brain structure that regulates emotions works differently in preschoolers with depression compared with their healthy peers, according to new research at Washington University School ...

Recommended for you

Emotional adjustment following traumatic brain injury

4 hours ago

Life after a traumatic brain injury resulting from a car accident, a bad fall or a neurodegenerative disease changes a person forever. But the injury doesn't solely affect the survivor – the lives of their spouse or partner ...

New ALS associated gene identified using innovative strategy

Oct 22, 2014

Using an innovative exome sequencing strategy, a team of international scientists led by John Landers, PhD, at the University of Massachusetts Medical School has shown that TUBA4A, the gene encoding the Tubulin Alpha 4A protein, ...

User comments

Adjust slider to filter visible comments by rank

Display comments: newest first

Captain Stumpy
5 / 5 (1) Mar 25, 2014
pure speculation, mind you, but IMHO there seems to be definite possibilities for this technology in the future law enforcement/psychology/profiling etc arena, starting with lie detection and expanding into criminal and abnormal psychology (think rape, pedophilia, serial killers)
imaging and being able to "see" what the offender/suspect see's during the recount of the criminal act (or during an interview) would give researchers/criminal psychologists/profilers/law enforcement etc quite a lot to work with
then there is the possibilities for victims as well... an image from the victim (non invasive fMRI) would be better than a traumatic recount where a sketch artist tries to reconstruct a face... just take it directly from the victim

just my opinion mind you

Mayday
5 / 5 (1) Mar 25, 2014
Once the contents of our minds is no longer considered sacrosanct, all is lost.
XavieX
not rated yet Mar 25, 2014
1984
TheKnowItAll
5 / 5 (1) Mar 25, 2014
Once they get more sophisticated I'll do the world a favor and let them be entertained by some of my crazy dreams lol
RobertKarlStonjek
3 / 5 (1) Mar 26, 2014
Why would faces appear anywhere in the brain in the same form as seen in a photograph? This is the problem of infinite regress (the homunculus problem) in that some downstream process must now look at the reconstructed face. But why the extra step? Why not just skip the reconstruction phase and go directly to breaking down the image for interpretation?

As there are more than ten times as many neurons innovating the earliest visual processing area than coming from the retina it is likely that any 'recognised' aspect of the image is replaced with one from memory leaving only totally mysterious, novel or new features of the face.

This is the process most probably intercepted by the fMRI scan. If this is the case then faces of familiar people will be altered far more than faces of strangers (because recognition cues/feedback will be more elaborate). For instance a person that is always smiling may appear in the fMRI scan to be smiling even if the photograph shows no smile.