4

I've seen advertisements for headsets such as the OCZ Neural Impulse Actuator, and recently there is a NeuroSky MindWave. Are these devices really capable of beating a keyboard and mouse for either high throughput or low latency?

Edit: For a simple example, just check the selling page for the OCZ NIA.

Hone Your Reflexes
Where others have to practice reaction times, you can use reflex-based game play to get the better of your opponents by cutting your reaction time by 50%.

  • 1
    There is no doubt about latency, but not sure what you mean by throughput in that case? – vartec Nov 14 '11 at 12:16
  • 1
    We want to focus our attention on doubtful claims that are widely held or are made by notable people. Please provide some references to places where this claim is being made. – Oddthinking Nov 14 '11 at 12:31
  • I've seen some advertisements for this on various web sites over the past few days, absent any mention of side effects. It seems to be a new technology: scientificamerican.com/… – Randolf Richardson Nov 14 '11 at 17:06
  • @Randolf, oh I know the systems exist. (Declaration: A friend used to work on these at Emotiv.) However, the question title suggests that someone, somewhere is claiming that they are faster than a keyboard/mouse. I have never seen anyone make that claim. – Oddthinking Nov 15 '11 at 6:42
  • I believe it when Pro-SC2 players start using this and win. – Jonas Mar 13 '12 at 3:52
8

Plausible reflex rates for limited functions. But won't replace keyboard and mouse.

A real answer could only come if someone tests the product in a laboratory and writes a solid review. But here's my educated guess:

Many non-invasive brain-computer interfacing (BCI) systems are based on measuring P300 signals with EEG, which take about 200-600ms. So even if that amount of time gave you enough information, and you had efficient enough feature processing algorithms, you're not going to get much better than sub-second categorization of brainwaves -- so, a command rate of a dozen words per minute. With current state of the art, looking at these three papers, subjects can only select about 4 or 5 characters per minute with P300-based BCI, or about 25 bits per minute.

However, the rates they are claiming for "reflexes" are plausible, because the OCZ NIA is not a brainwave reader: it's primarily based not on electroencephalography (EEG), but electromyography (EMG), which measures slight muscle movements.

EMG is frequently used in prosthetic limbs, and the signal is trivial to process compared to brainwaves. For instance, this review describes how to use wavelet transforms to preprocess data (read: efficient) and classify motor commands with neural networks. I've also seen talks that describe doing the classification with Linear Discriminant Analysis (read: easy peasy).

So they could be spot-on with their claim, if we assume that twitching a face muscle takes less time than moving a mouse. But it's more like strapping an extra button to your forhead and saying "do this when I twitch my eyebrow" than having it read your brainwaves and predict what you're trying to do -- so the romance is all but gone: With those limitations, we're not eliminating the keyboard and mouse any time soon!

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .