quinta-feira, 22 de julho de 2010

Um vislumbre do futuro

Tan Le apresenta-nos a sua impressionante tecnologia: um interface para a cabeça que "lê" as ondas cerebrais e que através de um feedback neuronal nos permite manipular certos objectos e movimentos. A não perder. Absolutamente. 

Um comentário:

jellyfish disse...

how come i missed the time when you posted this? i've just seen it the other day but i am rather dissapointed. here are my two cents and although i am no eeg expert, i read some opinions and debates on this in the last few days, because it was part of a discussion on twitter.

not only is the name of the device hilarious but also the claim that it can learn to diversify patterns or even understand emotion. eeg measures summed activity at the scalp level, i doubt there can be any subtle diversification in patterns in response to different actions and certainly it does not reach subcortical structures.

also, she cannot rule out the possiblity that the activity that she is measuring comes from artifacts like blinking etc, i.e. the electrodes measure muscle movement. this is a common problem in eeg and i hardly believe anyone could consciously control their brain waves as she has demonstrated. it takes A LOT of training.

highly anticipated games that are based on this (like mindflex) have been found out to be fake and even the release of tan le's device was postponed due to problems.

she overgeneralizes and the last part of the talk is just ridiculus, in my opinion. we are far far away from input devices that are consciously controllable to that extent only by the mind.

the only notable efforts in this direction (to my knowledge) are:
- in general any neurofeedback-based projects, like brain pong or even brain machine interfaces (like the eeg controlled wheelchair)
- the "mind reading" Honda robot (eeg and nirs): skip the intro, it's a far strech, but the rest is impressive: http://www.youtube.com/watch?v=RuxAV0S71qU
(note: nirs is sort of magnetic resonance, and i believe if we are making any baby steps in mind reading, than it's not eeg but fmri we should be looking at