Excerpt from the article in WIRED:
When it comes to black boxes, there is none more black than the human brain. Our gray matter is so complex, scientists lament, that it can’t quite understand itself.
But if we can’t grok our own brains, maybe the machines can do it for us. In the latest issue of Nature Communications, researchers led by University of Pennsylvania psychologist Michael Kahana show that machine learning algorithms—notoriously inscrutable systems themselves—can be used to decode and then enhance human memory. How? By triggering the delivery of precisely timed pulses of electricity to the brain.
Researchers, in other words, can use one black box to unlock the potential of another. Which on one hand sounds like a rather elegant solution to an absurdly difficult problem, and on the other sounds like the beginning of a techno-pocalypse horror flick.
When it comes to brain measurements, the best recordings come from inside the cranium. But people—and institutional review boards—aren’t usually amenable to cracking open skulls in the name of science. So Kahana and his colleagues collaborated with 25 epilepsy patients, each of whom had between 100 and 200 electrodes implanted in their brain (to monitor seizure-related electrical activity). Kahana and his team piggybacked on those implants, using the electrodes to record high-resolution brain activity during memory tasks.
Read the full article in WIRED...
