Brainwaves Can Be Passwords, Scientists Explain How EEG Authentication Works

Brainwave authentication is one of the many new biometric measures that are being proposed as an alternative to passwords. The idea is for authentication to be done with electroencephalogram (EEG) readings.

According to New Scientist, for instance, instead of using a password, a computer can display a series of words and then measure the response via an EEG headset. These signatures are unique and more complex, meaning they are more difficult to hack than a standard password.

However, New Scientist reports that recent research suggests that EEG readings can be able to authenticate someone's identity with accuracy rates around 94-percent. There are some factors, however, including if we have had too many drinks.

Tommy Chin, a security researcher from cyber security firm Grimm and Peter Muller, a Rochester Institute of Technology graduate student, tested this theory by analyzing the brainwaves of people before and after drinking Fireball shots.

Chin says brainwaves can be manipulated by external influenecs such as opioids, caffeine and alcohol. These make it a significant challenge to verify the authenticity of the user because they drank a lot of alcohol.

Chin and Muller initially presented their findings at a security conference in Washington DC this month. They said brainwave authentication accuracy may fall to 33 percent in drunk users. They recruited more participants at SchmooCon to gather more data.

However, John Chuang at the University of California, Berkeley, said the problem is not just confined to drinks and drugs. He recently published the impact of exercise on EEG authentication and found that accuracy degrades immediately after a workout, though it's easily recovered. According to New Scientist, he suggests that other factors such as fatigue, stress and hunger can reduce reliability.

It could be possible to collect "templates" for a user by mapping their EEG signatures when drunk, tired and so on in order to measure accuracy in different conditions. Chin and Muller also found it possible to tweak the EEG data analysis using machine learning.