Berkeley researchers may have the perfect authentication device — assuming you’re okay with a computer analyzing your thoughts.
The safest way for UC Berkeley professor John Chuang to log on to his computer is to put on a headset that reads his EEG signals and then silently think of the soaring strains of the Star Wars theme.
Imagining the song activates the temporal lobe of his brain in a certain pattern, which can be detected by a consumer-grade EEG headset that wirelessly connects to a computer. Chuang, an electrical engineer, has developed an algorithm that recognizes that pattern and unlocks his computer. That makes the Star Wars tune Chuang’s “passthought” — kind of like a password in that it’s secret knowledge that’s easy for him to remember. But it’s much stronger than a password, because each brain shows up on an EEG with a telltale electrical signature that’s most likely unique. So Chuang doesn’t mind that you know his passthought. It’s even published on a UC Berkeley Web page. Because if you sat at his computer and thought of the same song, you still couldn’t log in.
This is not just a proof of concept. “I firmly believe that passwords will inevitably become obsolete in a post-screen, post-keyboard world,” Chuang says. “And passthoughts will quickly gain traction once we become accustomed to wearing computers on our heads or faces. This is actually not as far down the road as we may think.”
As they test the passthought on themselves and research subjects, Chuang and his students in the UC Berkeley BioSENSE Lab are not just trying to perfect a better method of authentication. They’re also racing to address tricky questions that will be posed by computers that can get inside our heads, thanks to improvements in EEG devices like the kinds used by neurogamers and to brain-machine interfaces yet to come. Is a passthought a privacy nightmare, or is it a privacy enhancement if it makes your computer much harder to hack? If a passthought or some other brain-scanning technology doesn’t work when you’re drunk, over-caffeinated, or under mental duress, are you ready for machines to decide when you are in your right mind?
The passthought works because, Chuang says, “brainwave signals are like our fingerprints, they are biometric. The argument is that even identical twins will produce different EEG signals, because of the way their cerebral cortex — the outer layer of our brain — is folded. That modulates the EEG signals.” Sarah Laszlo, a psychologist at Binghamton University, was one of two lead authors on a 2016 study that achieved 100 percent accuracy in using EEG scans as an identifier. Though it would take a scan of every person in the world to definitively prove the uniqueness of each brain, Laszlo says Chuang is making a “reasonable assumption” by using the EEG as a biometric.
Chuang’s software merely matches the telltale frequencies of any particular person’s passthought to data in a database. “It has no semantic knowledge of what you’re thinking,” says Noura Howell, a Ph.D. student in the BioSENSE Lab. But to the user, it might feel like a machine has intruded on the inmost sanctum of the brain and is recognizing secret thoughts. “People get creeped out when we put the thing on their head,” Nick Merrill, another Ph.D. student in the BioSENSE Lab, acknowledges.
The BioSENSE researchers have tried passthoughts with roughly 100 people in the lab, but now they’re fanning out to test the technology with headsets and ear-mounted EEG units out and about in the world. One big question will be how long a passthought remains viable before the brain changes too much.
Chuang says it will be up to someone else to commercialize passthoughts. When you’re a researcher, any malfunction or shortcoming of passthoughts is an opportunity to write a kick-ass paper — not a near-death experience for your business. “If we discover downsides, vulnerabilities, privacy and security implications, we want to be the first to point them out,” Chuang says. “If I have some venture capitalist behind me, I may be tempted to withhold.”