Neuroscientists Are One Step Closer to Using AI to Read Minds (Sort Of)

The human brain is one of science’s final frontiers and in many ways we are far from understanding the mysteries of consciousness. But according to a paper published in the Journal of Neuroscience, we’ve come one step closer to mind reading.

Two neuroscientists from the University of Oregon, Brice Kuhl and Hongmi Lee, have figured out a way to use an AI program to take MRI readings of a subject’s brain activity and reconstruct the image of what the person is seeing.

In other words, in the experiment they devised, they had people sit in MRIs and stare at pictures of faces, and by analyzing MRI readings of their brains, an AI program was able to mock up a basic approximation of the faces the subjects were staring at. Kind of sounds like mind reading, no?

MRIs provide doctors and scientists with a basic understanding of the brain’s layout and functioning. They use magnets to measure changes in blood flow in areas of the brain, with changes in blood flow being used as a proxy for neurological activity.

Kuhl and Brice measured brain activity in the angular gyrus (ANG), which deals with vivid memories, and occipitotemporal cortex (OTA), which deals with visual inputs, of the human guinea pigs while they were staring at pictures of faces. After feeding their AI with the MRI readings as well as some basic details about the faces that the patients were staring at, the AI created a rough rendering of what it thought the face looked like.

This approach is far from perfect. See the results for yourself in the above image.

At the very least, these AI-generated images allow us to determine whether the faces that the subjects were staring at were happy or sad, light-skinned or dark-skinned, and male or female, which is a significant advance. And ultimately, it’s important to note that it’s not mind reading but a better understanding of the brain and what brain activity signifies that these scientists are after.