Mind-Reading Experiment Sends Thoughts Over the Internet

University of Washington graduate student Jose Ceballos, the experiment’s “respondent,” wears a cap connected to an electroencephalography (EEG) machine that records brain activity and sends a response to a second participant, the “inquirer,” over the Internet. University of Washington

Researchers have managed to send thoughts over the Internet, using a direct brain-to-brain connection and a big magnet.

It’s not very efficient — the volunteers could only send “yes” or “no” answers to one another. But using EEG (electroencephalographs) to read brain signals, and a big magnet on the other end to transmit it, their players got the correct answer 72 percent of the time.

"This is the most complex brain-to-brain experiment, I think, that's been done to date in humans," said Andrea Stocco, an assistant professor of psychology at the University of Washington.

"They have to interpret something they're seeing with their brains," added Stocco’s colleague Chantel Prat, associate professor of psychology. "It's not something they've ever seen before."

"This is the most complex brain-to-brain experiment, I think, that's been done to date in humans."

The volunteers played a version of the “20 questions” game. One person looked at a word on a screen — in one test it was the name of an animal — and the other, sitting in a lab a mile away, had three chances to guess what the animal was. To make it simple, the questions were pre-arranged: “Can it fly?”, “Is it a mammal?” and “Is it a pet?”

There were eight possible answers: shark, turtle, bear, dog, vulture, parakeet, bat and sugar glider. That last one was added to make the experiment more entertaining, Stocco said.

The volunteer in the know, called the respondent, could look at either a “yes” or “no” flashing light on a computer screen. Her thought patterns were recorded as EEG activity and transmitted over the internet to a transcranial magnetic stimulation (TMS) device attached to the back of the head of the “questioner”.

University of Washington postdoctoral student Caitlin Hudac, the “inquirer,” wears a cap that uses transcranial magnetic stimulation (TMG) to deliver brain signals from the other participant, the “respondent.” University of Washington

The questioner sent one of the three questions to the respondent via computer, the respondent looked at the “yes” or “no” light, and that signal went back to the TMS machine. A “yes” answer would elicit a strong TMS pulse, which creates a signal perceived by the brain as a pulse, a blob or a line called a phosphene.

A “no” answer sent a weaker signal, too low to create a phosphene.

There's a video here.

The 10 volunteers got the final answer right at least 72 percent of the time, the researchers report in the Public Library of Science journal PLoS ONE. When they guessed by chance, they got the answer right 18 percent of the time. (It’s less than 50 percent because there were more than two answers).

The next step is to try to send more complex information, Stocco says.

“Imagine that we can record from the visual part of the brain."

“It all depends on what we can transmit,” he said. An image, for example.

“Imagine that we can record from the visual part of the brain,” he said. “Right now, we are starting from something very simple — yes or no.” But other groups have managed to decode signals from the brain’s visual cortex and understand what someone is looking at, he said.

And a yes-or-no system is simply a binary system — the same language that computers use.

His group’s next step is to try to send a brain state —awake, or relaxed, for instance. That could be read via brain waves, which are well-characterized from asleep to alert states. An EEG is well suited for measuring such information.

Justin Abernethy, a research assistant at the UW Institute for Learning & Brain Sciences, positions a magnetic coil that is activated by the brain signal from the other respondent. A “yes” answer from the respondent will cause the inquirer to see a flash of light known as a phosphene. University of Washington