How Neuroscientists Are Learning To Read Our Minds
New study uses resting brain scans to predict how individual brains would respond to specific tasks
Neuroscientists are not mind-readers, but they’re coming close according to a new study published in the journal Science. With little more than a functional MRI scan, scientists have managed to program a computer model that can predict the specific path that that individual’s brain will likely take when performing a given task. This tool could eventually be used to conduct top-level neuroscience with infants and people with disabilities.
“We show that we can essentially predict how people will use their brains,” coauthor Saad Jbabdi of Oxford University told The Scientist.
Arguably the next big project in science is mapping out the human neural connectome—the network between our brain regions that activates when we perform specific tasks, and then leaves its unique pathway etched in our brains long after. Prior studies have shown that functional MRI brain scans can highlight these pathways, but Jbabdi and colleagues wondered whether the vague outlines of the connectome could provide enough information to help predict how a resting brain would act if faced with a particular task.
For the study, Jbabdi’s team analyzed fMRI scans from 98 individuals from the Human Connectome Project. The scans showed their brains at rest, while engaged in no particular activity. But the precise layouts of their brains were still quite different from one another, presumably because “subjects may use different strategies or cognitive processes that involve different brain circuits” to accomplish the same tasks—leaving behind unique etches, even at rest.
“We investigated the possibility that individual differences in brain activation are inherent features of individuals and, to a large degree, independent of volatile factors,” the authors explain. In other words, all of the subjects knew how to read, but not all of them mustered their brains for the task of reading in the exact same way. And it showed on their fMRIs—even when they weren’t reading.
Armed with this information, the researchers painstakingly flipped through fMRIs, assembled individual connectome profiles for each subject and taught a computer program to identify those faint pathways. The model then produced what a theoretical fMRI scans should look like for each individual when he or she was reading, gambling or performing a mathematical calculation, tailored to their resting brain scans. Since the Human Connectome Project conveniently provided actual fMRIs of each subject performing those very tasks, the researchers were able to check their work, so to speak, and see if their predicted fMRIs matched reality.
Remarkably, it worked. The researchers were able to predict the difference between what one individual’s brain looked like while reading, gambling or calculating, as distinct from how another participants’ brains would appear while engaged in those same tasks. The authors say that their findings could have important implications in medical settings, where it is not always possible to predict brain function—especially among paralyzed patients or infants.
“Our brains have elaborate networks,” Jbabdi told The Scientist. “We can now learn what these networks are for every individual.”