Frequently asked question in India

In this post, I try to answer the three most frequently asked question we had to answer during Techfest. Most of the audience we had was totally naive about BCI technology. For this reason, I believe that at some point, those hearing or reading about BCI probably ask themselves the same questions…

Could a BCI damage the brain ?

The answer is no ! Actually BCI based on ElectroEncephaloGraphy (or EEG) are totally safe. The idea is to use amplifier in order to accurately measure very small electrical currents on the surface of the scalp. These small currents are the results of the neurons activity being propagated through the brain, through the skull and all over the skin. This is only measures, nothing is sent back to the brain. So yes it is totally safe. Actually, the worse thing in using EEG is that you must wash your head after using it. This is because of the gel that we have to put in order to create contact between the skin and the electrode. But dry electrodes are coming and I anticipate a big jump in the next 5 years on the sensors as mass market is currently being targeted by some video games companies (see for instance Emotiv products which already replace the gel by a more convenient saline solution)…

Is it or could it be adapted to be a lie detector ?

This question was probably asked because we showed demos based on P300 detection.

The short answer is probably no.

First of all, one should notice that it is currently not trivial to make such BCI work correctly when the user consents to use the system. It means that if the user wants to make the system fail, or even if he just does not focus enough, then the system falls down quite quickly.

Now suppose that there is a big jump on acquired signal quality and detection algorithms so that the computer is able to guess something without the consent of the user (which is what is likely to happen in the case of a lie detector, right ?). Then the user should be able to train to make the system fail. He could use neurofeedback in order to have a feedback of how close to detecting a lie the computer is, and consequently, train himself to regulate his brain activity in order to make this detection fail, being prepared for the real lie detection session. That means this lie detector could probably not be reliable.

Is it the same technology as Stephen Hawking uses ?

That was totally unexpected question and I must admit that even if I thought this was not similar, this question has been asked so many times that I ended up with a big doubt !

So I am sorry but no, this is not the same technology as what Stephen Hawking uses. Quoting Wikipedia :

… he appears to speak fluently through his synthesizer, but in reality, it is a tedious drawn-out process. Hawking’s setup uses a predictive text entry system, which requires only the first few characters in order to auto-complete the word, but as he is only able to use his cheek for data entry, constructing complete sentences takes time. His speeches are prepared in advance, but having a live conversation with him provides insight as to the complexity and work involved….

Actually, what Wikipedia does not say is how Pr Hawking writes the first few characters. And this is based on eye tracking. The computer tracks where the user is looking on screen in order to select some letters. I did not find much details about how his specific system works… By the way, most of the time, the user has to look at the letter he wants to spell for a couple of seconds in order to spell it. This is like a click… But there are more original and efficient ways to spell some text based on eye tracking. For instance, in the following video, the letters are presented to the user depending on the probability that they appear after what has already been spelled. If using an eye tracker, the user can chose the next letter to spell looking left/right and the chose writing speed looking up/down. The software demonstrated here is dasher.

Dasher poem…

Comments are closed.