Tuesday, March 05, 2013

Brain-to-Brain Interface Established

io9 reports: [edited]

Researchers have created an electronic link between the brains of two rats, and demonstrated that signals from the mind of one can help the second solve basic puzzles in real time — even when those animals are separated by thousands of miles.

Here's how it works. An 'encoder' rat in Natal, Brazil, trained in a specific behavioral task, presses a lever in its cage it knows will earn it a reward. A brain implant records activity from the rat's motor cortex and converts it into an electrical signal that is delivered via neural link to the brain implant of a second 'decoder' rat.

Rat number two is in North Carolina. The second rat's motor cortex processes the signal from rat number one and — despite being unfamiliar with the behavioral task the first rat has been conditioned to perform — uses that information to press the same lever.

Untrained decoder rats receiving input from a trained encoder partner only chose the correct lever around two-thirds of the time. That's definitely better than random odds, but still a far cry from the 95% accuracy of the encoder rats.

"These experiments demonstrated the ability to establish a sophisticated, direct communication linkage between rat brains," he said in a statement, "so basically, we are creating an organic computer that solves a puzzle."

1 comment:

brett jordan said...

From Conrad Gempf:

OK, look. I think this sounds spookier than it is. To work in basic mode, all that needs to happen is this:

It is possible to tell from a mouse's brain whether it is using the muscles in its left paw or its right paw, true? True.

Use that to trigger some more mundane event: If the electronic signals we detect are left paw, then play the first few bars of Beethoven's First Symphony; if the signals we detect are right paw, then play the first few bars of King Crimson's "21st Century Schizoid Man".

Have the music piped in to Decoder Rodent's house. It will be no surprise if, unless he's really crap at music, he learns to press left when he hears violins, right when he hears distortion through a massive Fender amp and chill when he hears no music.

Now just drop the music. What kind of electrical signal do you transmit instead? Same deal. The Decoder Rat isn't exactly responding to the other rat's brain, but to signals created by the machinery that detects electrical changes in the Encoder Rat's head. The electrode is the only thing that is reading anyone's mind. And the electrode turns the brain state into a signal that can be acted upon either by people with music collections or by other rats.

For the cooperation bit, you're just training the Encoder Rat that he tends to get more reward if he presses harder or whatever action, unknown to him, also makes it easier for the electrodes to report brain state. He's not thinking "Listen carefully, North Carolina Buddy, I'm thinking about my left paw for all I'm worth", he's thinking, "Last time, I got more when I kind of did this twisty thing with my tail at the same time."

It's worth noting that the electrode may be measuring the electronic impulses involved in the mechanics of using muscles, rather than thinking Left or Right. If we get to the stage where Decoder Mouse is making the right decisions based on Encoder Mouse's OBSERVATIONS of a third mouse pressing the bars, THEN perhaps we'd be reading mental processes rather than mechanical ones.