Fingerprint scanning, iris scanning, and voice stress analyze are dated technology compared to the latest form of biometric authentication: the brain’s semantic memory. A team led by postdoctoral researcher Blair Armstrong of the Basque Center of Cognition, Brain, and Language in Spain, observed that the brain signals of 45 participants reading a list of 75 acronyms — such as FBI and DVD — react differently when processing the semantic memory, or memory of language, of each word.
The difference was so palpable, that the team’s software could identify the volunteer with up to 94 percent accuracy, implying an unmined potential for future security systems, especially in light of the fact that fingerprint scanning has proven to be susceptible to duplication with as little as a high-resolution imagine of the target’s hand. One clear advantage that semantic memory has over fingerprint recognition is that the signals it generates provide a continuous form authentication, rather than a one-time identification. Not only does this scenario heighten security, but it improves usability, eliminating the need to re-enter passwords each time an account is accessed.
Armstrong’s team is not the first to identify users based on electrical signals in the brain, but their approach significantly improves on past models by eliminating much of the noised typically associated with brain signals as a result of exclusively focusing on the segment of the brain required for reading and recognizing words.
The outputted semantic memory, or memory associated with words, subtly varies from person to person much like a mental fingerprint. What’s more, semantic memory doesn’t change over time the way episodic memory does, meaning that episodic memory — our recollection of past experiences — will affect our perception of a specific word if we’ve interacted with that subject. So, if you’ve been attacked by a dog, the word “dog” will fire off a different set neural relays in your episodic memory, but the semantic neuron output will remain the same.
The 94% accuracy is not enough of a guarantee full authentication, obviously, but it’s a start.
Source: Newscientists
Learn more about Electronic Products Magazine