Scientists Demonstrate Thought-Controlled Computer

Da Massive writes with a link to ComputerWorld coverage of a unique gadget shown at this past week's CeBit show. The company g.tec was showing off a brain/computer interface (BCI) in one corner of the trade hall. The rig, once placed on your head, detects the brain's voltage fluctuations and can respond appropriately. This requires training, where "the subject responds to commands on a computer screen, thinking 'left' and 'right' when they are instructed to do so ... Another test involves looking at a series of blinking letters, and thinking of a letter when it appears." Once the system is trained, you can think letters at the machine and 'type' via your thoughts. Likewise, by thinking directions you can move objects around onscreen. The article provides some background on the history of g.tec's BCI, and suggests possible uses for the technology in the near future.

Re:Type thoughts?

[linguizic]

If we knew where each phoneme was stored in the brain and we could stimulate the computer via activity in this area all we would have to do is think about speaking the word. But things like allophones and the ranges of phonemes in different dialects would throw this off. No matter how dumb people may think southern speech is, it is actually hardwired into their nervous system via language acquisition and would have to be accounted for physiologically. It would probably be easier to have the nodes hooked up to the part of our brains that's responsible for controlling our fingers when we type, the amount of variability found in dialects gets reduced quite a bit just by removing the vocal element of language, even if you're Cletus.

Re:Just what we need

[logixoul]

Other way round ;)

Re:Nothing so new about that... [url correction]

[QuickFox]

Don't worry, you're not lame. Posts can't be edited on Slashdot.

faster and easier to wiggle a finger

[HuguesT]

Current research papers put a EEG-based BCI throughput limit at about 30bpm. This is bits per minute. 18bpm has actually been achieved. This is because it is rather hard to alter one's beta waves : one need to concentrate for about 2s or more to make a change (flip a bit) reliably. EEG is what the linked article talks about.

With this kind of throughput one can compose no more than a couple of sentences a day. Clearly this is not going to replace typing for most people anytime soon. Even if one is severely impaired by some brain damage (e.g. a stroke) even a little bit of retained mobility is better. There was for instance this man who manage to write a whole book (the diving bell and the butterfly [doyletics.com]) through his fluttering eyelid.

However different techniques are being developed. The best in terms of throughput and quality of data make use of f-MRI and other advanced techniques, or are very invasive (actual electrodes in the brain), and clearly this is not going to be possible as a usable tool for most people anytime soon either.

Check back in a few years. Right now BCI is definitely pie-in-the-sky, although it does sound cool.

Not Quite....

[ThePopeLayton]

I saw this exact device in action at the Annual Society for Neuroscience meeting last October in Atlanta Georgia. I spent about an hour talking to the group that were displaying the device. It uses EEG technology to detect voltage potentials across the skin (caused by inputs into layer 4 of the cortex). The tech who explained the device to me told me that current EEG analysis is not good enough to detect what a person is thinking about, rather it can detect IF a person is thinking.

The device does not recognize thoughts about specific letters, rather it recognizes general thought. The person has a grid of electrodes on the scalp that are measuring the voltage. The person then looks at a computer screen that displays groups of letters.

A band like "A D T E R K" is displayed and the person is instructed to count every band that appears that contains the desired letter. So if the person wants to type an "S" then upon seeing the band "S T V W K N" they would register having seen the S and the process of counting produces a large enough EEG signal that it is logged by the computer. The computer then displays separate bands that contain no more then one letter from the first band. Bands like " T D E I M" or "S B C X Z" might appear and as the second band contains an S the person would count it and produce the EEG signal. The computer then looks for the common elements between the bands and as S is the only common element the letter S is typed.

So again the computer isn't reading specific thoughts, rather just general thinking. The subject doesn't think "K" and then K is typed rather the computer displays a K and the person confirms the choice by thinking.

This display process is very fast (about 1 band a second) but it is rather a slow process to write. It takes around 5 or 6 minutes to write a sentence. It isn't as great as the article makes it seem, but it certainly is a step in the right direction.

They must have hired

[Anonymous Coward]

They must have hired ex-atari engineers to get his higly advanced 1970's tech

http://www.atarihq.com/museum/2678/mindlink.html [atarihq.com]

Re:Type thoughts?

[zsau]

If we are looking at the phones themselves then this can't be the case.

Um... I know that most parts are in the same general area between people, but I was of the impression that there are still differences: The parts of my brain which are responsible for control of my left hand/tongue/etc will be in basically the same place, but might be relatively larger/smaller than for you. Am I wrong on this? Otherwise surely we'd need to have brains that are exactly the same.

there is the conscious knowledge of the sounds of speech--without which we would not be able to learn to read and write.

This I am even more dubious of. People have a conscious knowledge of syllables and rhythm, and we who've learnt to read and write and analyse phonetically and so forth have a conscious knowledge of the sounds of speech—but I remain unconvinced that knowledge of them is at all relevant to literacy and even to speech. For instance, I'm informed there's a Swedish dialect which distinguishes open and close e (Standard Swedish doesn't), but there's no indication of this in the orthography: Although speakers of this dialect produce and use this distinction, they're unaware of it. (I had a similar difficulty learning the distinction between voiced and voiceless th.)

Now subconciously there's obviously some means of recording the distinction, but again, I don't really know that phonemes enter into it...

I never completely bought OT, though that might be the result of who taught it to me, and the text she chose to teach from. I'm hard pressed to say that realized forms are the product of garbage going down the chute and getting sieved into grammatical constructions. If this is an unfair generalization please feel free to make a better generalization and make me a better informed individual on this matter.

Umm... Your analogy is apt. I don't know what I could say to change your opinion; you might make more concrete objections. For best results, present a viable alternative, if you know of one :)

I would love to keep going but now I have to put my children to bed.

On the internet, no-one cares (or knows) if you take half an hour (or half a day) longer to reply :)