Human identity verification and authentication relies strongly on biometric signals: we recognize familiar people by their faces, their voices, or the way they walk, and are fairly good at matching photographs to the actual faces of strangers. With computers and similar devices, however, passwords are still the most widely used authentication method. So even a smartphone, which may spend more time in close proximity to a person than any other human, will most likely rely on a password or personal identification number (PIN) to authenticate its use. Biometric measures are becoming more widely used, but are still limited in scope and reliability.
In their paper, the authors investigate the use of consumer grade brain-computer interfacing (BCI) devices such as the Emotiv EPOC (http://www.emotiv.com/epoc.php) or NeuroSky MindWave (http://store.neurosky.com/products/mindwave-1) as biometric identity verification systems. The main part of the paper consists of a discussion of experiments with the EPOC device where the mapping between PINs and corresponding brain signals is used as an identity verification measure.
While I found the overall idea interesting, the authors failed to convince me that this is a promising method. Foremost are the results: a 62 percent recognition rate for brain signal patterns from the same user indicates that one out of three attempts fail, while a 33 percent recognition rate for signals of different users gives access to one out of three unauthenticated users. In the introduction, the authors indicate that multi-modal measures are preferable, but it is unclear how this is reflected in the experiments. While there is a discussion of steps that use speech processing, the decision concerning authentication seems to be based only on the brain signals. And from a practical perspective, even more convenient headsets like the forthcoming Emotiv Insight (https://emotiv.com/insight.php), which does not require more than a dozen saline-hydrated sensors on one’s head, will likely be restricted to limited circumstances. After all, in most situations it’s much easier to type a few digits or characters than to strap on a headset. On the other hand, authentication through passwords for devices like Google Glass is quite cumbersome; maybe a future version will include a few brain wave sensors to recognize and authenticate a user based on thought patterns.
The paper also has a number of editorial and language issues, which to some degree makes it difficult to understand what the authors are trying to express.
