Slashdot Mirror
Mind Over Machine
broKenfoLd writes "Monkeys moving robotic arms by manipulating a cursor on a computer screen, simply by thinking about it? Mice who cause their water tube to dispense some refreshing H2O just by wishing it? Signal processing and decoding has long been a dream of Matrix fans and lazy system administrators for years, and science is amazingly keeping up! Popular Science's Carl Zimmer has written a fascinating piece documenting recent progress in decoding brain signals and interpreting commands issued from thoughts alone. If you heard a single violin playing Beethoven's 5th, you would be able to tell what piece of music was being played even though the rest of the orchestra was not heard. In the same way, by monitoring a relatively few neurons, computers can recognize patterns and allow programming based on these patterns to say, know if a mouse is thinking about pushing his water lever.
You can pass the time waiting for Matrix-style video games and motionless system adminstration/utilization by reading the full article."
But can it be used for channel surfing. That's the ultimate goal.
Monsters, John! Monsters... from the id!
There's a growing sense that even if The Future comes,
most of us won't be able to afford it.
-- Lemmy
Can you imagine what might happen when a "hottie" walks though the office?
I call it sleeping.
My water tube can dispense a refreshing H20-based substance just by wishing.
the best of these kind of devices (devloped in nasa labs) can only do 95% accuracy. sure that might be fine for say playing a video game(unless its fast pace), but if you tried to walk with 95% accuracy, you'd be the but of more jokes then the "how do you get bob dole out of a tree, wave to him" jokes.
but over all its really cool that they are even able to do this at all.
"Mice who cause their water tube to dispense some refreshing H2O just by wishing it?"
Uncanny! Just this morning I caused by "water tube" to dispense liquid just by wishing it too!
And I ran him over with my SCV?
"I didn't run him over!"
"Did you THINK about running him over?"
*long pause*
I didn't run him over!
I mean, the rat can think about water, and get some water. But a rat's mind is way simpler.
Have you ever thought about suicide? Now imagine if when you thought about it a machine would come and kill you. Also I don't know about you, but I can't control my mind completely, sometimes I have thoughts that are completely unrelated with what I am doing... I really don't think I could trust a machine to make my thoughts come true, I'm sure in the future machines will be able to interpret the signals in your brain with a 99% precision, I just can't trust my own mind.
The best part is that this computer can be mass produced with unskilled labor!
But that would require thinking, and that hurts :(
If you read 'The Light of Other Days' by Arthur C Clarke & Stephen Baxter you'll get a good insight into the possible consequences. While the book is centred around the idea that wormholes can be used to view anyone at any time, knowing what people think would have the same effect of first causing terrible unrest but eventually destroying barriers and allowing everyone to work together. It's a very interesting read and I feel that every day we get closer to that reality.
This does not answer how brain works at all :-) As a motor control neuroscientist by trade, I can tell that finding out what the brain actually doing has little to do with these neat things.
:-)
This is not to say that it's not important -- all kinds of prosthetic devices can be made to help people with disabled limbs or other parts of the motor control system -- so it's a great benefit to those people. The important thing is that these devices are still controlled by the human brain, and nobody has a good idea how.
The fact is, you can probably hook up whatever device to whatever portion of the brain (e.g. an artificial arm to you toenail brain area) and after some practice the subject will learn now to move it. So when they say "we don't see the brain as a mysterious organ anymore" they are telling you a bold-face lie.
The mystery would be demonstrated to be solved when we can build a computer with massively parallel and slow (up to 1kHz) elements that can match human performance in tasks like tracking, reaching, as well as learning those tasks.
So far, all the beatiful performance of the cool gadgets is accomplished by super-fast feedback and super-fast computing elements. Our neurons are ways slower, but they do much better. Therefore, the whole essense and mystery of the brain is how to connect 10^10 shitty elements into a great learnable machine. Algorythms and parallelism are still the mystery of the brain, even if the popular science magazines claim otherwise
They talk about using this for people with disabilities. This would be great for sure but ... interpret though.
;-)
they say in the article that they need to "train" the computer beforehand (no pun intended)
before it can
My question, which was not answered in the article, is: Are every brain emmitting the same signals for the same action
or do they need to "train" the program for every new user (monkey)? I would think that every individual have a somewhat
unique "brain signature" and if it is the case, how can a totally impaired person train a computer to use an artificial arm or
leg or whatever if anyway he isn't able to move a "joystick"?
Can the computer associated anything as an input to compare with the brain activity?
Could (let's say ) S. Hawking program the system by blowing in a tube harder or smoother for example?
Am I clear?
I'd rather be sailing...
But a rat's mind is way simpler.
Allow me to introduce you to our management team.
KFG
FIRST POST! FIRST POST!
(Damn, didn't work)
Natalie Portman in my cube.
(Nope, still doesn't work)
I want to be overworked and overpaid, but still have time to surf slashdot from work.
(Woohoo! It works!)
It is easier to build strong children than to repair broken men. -Frederick Douglass
The real exciting part isn't about the machine learning what the brain is doing, but rather the brain learning how the machine works. Near the end of the article, he talks about a cluster of neurons that grew in the monkey brains after the implants, and would fire only when the implants were active. The monkey's brain, in effect, sensed a new presence and adapted to it within minutes of its arrival.
If you've ever tried learning an activity that instinctive reflexes like skateboarding or ice skating or even playing the piano, you realize that no matter how much instruction someone gives you, at some point you feel like once you've done it enough, you just "get it". It's the whole muscle memory thing, how your brain encounters something new and just adapts, learning exactly which neurons to fire at the right moments to get the desired affect. Seeing neuron's grow and cluster especially for the robot arm is indicative that the monkey's brain can assimilate the arm and treat it as a natural extension as opposed to a external tool with an awkward interface. In geekspeak, it's like a kernel that, on detecting a new device, can probe it, learn the API, and build its own device driver automatically, without ever knowing anything other than that it's something on the other end of a bus.
Extending that line of thought, who's to say that if the signal processing and classification algorithms advanced far enough to classify even our thoughts, our brains wouldn't be able to instinctively learn how the mind-readers worked and retaliate in return?
Pleeeeeeease don't let MS get their hands on this one, mister!