Sci-fi is full of tales about people controlling machines with their minds, but sci-fi is already here, sorta. For example, earlier this year, during the Cognitive Technology exhibit at San Francisco's Exploratorium science museum, visitors could don electroencephalogram (EEG) headsets that read brainwave patterns to move a robotic arm. They could also create images based on their brainwave activity. There are many other examples of scientists or just DIY "neurohackers," jacking in to starting figuring out how the electric energy in their skulls can be deciphered.
A new tool for that, with roots in the U.S. government's Defense Advanced Research Projects Agency (DARPA), hit Kickstarter today, delivering a high-end EEG headset kit starting at $449. The Kickstarter aims to raise at least $80,000 and ship gear to backers in July 2016.
It's produced by a little Brooklyn startup called OpenBCI (brain-computer interface) created in 2014—not by neuroscientists, but by a tech-focused art professor, Joel Murphy, and his grad student, Conor Russomanno, from New York's Parsons School of Design. It's not the first EEG headset to hit Kickstarter, but the creators say that it's completely "open," meaning DIYers can modify the system and get access to the full raw data coming off the electrodes. It also has a higher data-sampling rate than other devices, making it "research-grade," says Russomanno.
I first met Russomanno in August during the NeuroTechNYC meetup at OpenBCI's Brooklyn lab. For about the first half hour, I was wearing what looked like a geodesic bicycle helmet. It was the third and just-made version of Russomanno's 3-D-printed Ultracortex headset; and I scarcely noticed the eight sets of metal pins poking lighting into my scalp. The whole time, I got to see readouts from various brainwave types and from muscle movements in my scalp on a big computer screen in front of me. This was a dry-electrode headset, meaning I didn't have to squirt conductive gel onto my scalp. The fact that I have a buzz cut made it especially easy (in fact, Russomanno had recently buzzed off his bushy brown hair), but it's not required.
Russomanno and Murphy are now getting ready to sell the fourth version of the Ultracortex (Mark IV), which they will offer in three sizes to Kickstarter backers for $350 (if backers 3-D print it themselves) or fully assembled for $550.
OpenBCI first hit Kickstarter in 2013 with a 32-bit circuit board that can be set up to process signals from up to 16 electrodes and pass the data to a user-friendly visualizing program for Windows, Mac, and Linux called OpenBCI GUI. It came out of work that Murphy was doing for a DARPA-funded project—part of President Obama's BRAIN Initiative—to develop a "High-Quality, Low-Cost, Multi-Channel EEG System for Non-Traditional Users." Murphy brought on Russomanno to help with the project. And while the team they worked for didn't make it to the second phase of the DARPA program, their board was a hit with visitors to Maker Faire 2013 in New York City, winning the Maker Faire Educator’s Choice Award. That encouraged them to hit Kickstarter, where they raised $215,438 against a $100,000 funding target.
The 32-Bit open BCI board can process electrical activity both from brainwaves and from the action of muscles. "I think at this point [OpenBCI] should be body-computer interface," says Russomanno, "because it can do muscle and heart and all that." One member of the OpenBCI group has used it to move a Hexbug toy robot around using basic brainwave patterns. Others are working on a prosthetic robotic hand controlled by input from forearm muscles.
Russomanno wrote to me in an email that "roughly 1500-2000 people" have bought the 32-bit BCI board, which is extremely powerful for the price. But at $550, the price is still pretty high for DIYers. And they didn't offer their own headset in the first Kickstarter. Today's Kickstarter features a new board, called Ganglion, that sells for $99. The compromise is that it supports just four electrodes.
Another goal of headsets and cheaper circuit boards is to collect as much data as possible for bigger projects aimed at deciphering brainwaves, such as Cloudbrain. It's an online repository of brainwave scans, applying an artificial-intelligence process called machine learning to look for recurring patterns that will provide a possible alphabet or vocabulary for the language of brainwaves. Marion Le Borgne, a senior software engineer at Bay Area artificial intelligence company Numenta, heads up Cloudbrain and also worked on the Cognitive Technology exhibit at the Exploratorium.
It's not clear yet how far this brainwave decoding can go. Russomanno is cautious in his predictions. "I don't think we'll ever be able to straight-up read thoughts with non-invasive BCI," he says, meaning, without sticking electrodes into the brain. "I think we'll be able to look at general brain states and patterns, and maybe have minimal levels of interactivity or like control over devices."
But as we are learning in this era of cheap computer storage and artificial intelligence, the only way to really find out what data means is to collect a whole lot of it and let the algorithms start poking around. Putting more headsets on people's noggins is a step toward that process. "With non-invasive BCI, we're a long way off having maximized their potential," says Russomanno.