This submission's title is misleading, and the original work could be better summarized as "improving at one task while doing another". The authors first determined a pattern of brain activation (indexed by blood flow) associated with the presentation of a target (visual) stimulus for a given subject. Then, they gave subjects real-time feedback about their brain activation while they performed a superficially different task. The trick is that the feedback on the second, different task was based on a subject's ability to cause the same pattern of brain activation (again, indexed by blood flow) that the experimenters first observed when the target stimulus was presented in the earlier session. Finally, subjects were tested on their ability to identify the target stimulus and others like it, and they showed a statistically reliable improvement for the (unknowingly) trained target, but not for other, similar items. It's fascinating, but suggesting that this is anything like what's been depicted in science fiction movies (yes, like The Matrix) is silly.
It's fascinating, but suggesting that this is anything like what's been depicted in science fiction movies (yes, like The Matrix) is silly.
It's certainly a big leap beyond what's actually been demonstrated here. But it is a very interesting first step along that path, and if something like Matrix-style learning was ever going to be possible, I'd imagine this would be the most fruitful approach to follow.
Just to be clear: what's going on here is essentially that they've set up an apparatus to display an error function that measures the difference between the target and actual brain states in a person, and asked the person to minimize that error. The idea would be that by trying to match brain states you could shortcut the learning process that lead to that brain state, and instead skip directly to the result. Pretty standard optimization problem, except that it involves the brain.
The difficulties come in two parts:
1) Is it possible to efficiently "navigate" your own brain state based on feedback towards a target based on an error function, and do so well enough to jump out of crappy local maxima? More generally, is it possible to reliably navigate to a target brain state quickly enough so that it's more efficient than actually going through the learning process that the target state was arrived at from?
2) Can we even measure "brain state" accurately enough so that successful completion of 1) would be useful at all?
1) is a biological question, 2) is technical. This particular research doesn't offer much on either front, sadly, because the task they picked was so simple that it would clearly have been better achieved by direct learning (they spent several days training people on the "error function", and I can't imagine it took that long to recognize an orientation of an image for the training group...).
I'd be very skeptical about how useful low resolution views of brain activity will be when it comes to higher level understanding rather than simple visual recognition tasks. And I'd also be skeptical about whether we'd still be able to effectively navigate our brains through the fitness landscapes if we ever did achieve resolution good enough to help with more difficult tasks.
But I don't think it's a wash. It's pretty likely that even low-res views of brain activity for certain tasks can be helpful - if you could match your brain activity even at a rough scale to the way Richard Feynman's looked when he was thinking about physics, it would probably put you in a better state of mind to do physics than you might have been otherwise, and that could still be useful, even if it couldn't directly transfer his knowledge about QED to your brain.
If we ever get to a point where scanners are much better, I imagine there will be quite the industry in picking out which low-res reductions of high-res brain activity are the best to use as training targets...
I don't really understand why this is so significant or perhaps I don't understand the experiment. You are fed some training set and a certain brain activity pattern is identified as the "target." Then you show a whole bunch of other stimuli and try to teach the brain to display the target activity pattern for a different stimulus by having this proxy indicator that tells you how close you are (green disk).
Don't we do this all the time? If I'm teaching a child the letters of the alphabet then I would show an example, have them try to guess it and then give them some indication of how close they are (yes you got it, you are pronouncing it weird, no that's the wrong letter). The only difference here is that you are using proxy whereby you identify a target brain activity pattern first. It seems pretty much impossible, however, to get a target for an individual without first seeing what the target is. Certainly everyone's "target" for the same problem must look quite different.
I suppose this could be pretty useful for brain-computer interfaces but I don't know if I'd go as far as saying this is matrix-like learning.
Sometimes HN makes me sad. I thought that was a great joke.
I know HN doesn't want to devolve into a morass of stupid one-dimensional humor, but I sometimes find a place that bans humor pretty much entirely, equally miserable.
This looks either very amazing, or complete BS. I want examples. How did their experiments work? I don't care what it is that I'm learning, I wanna see a demo. Is there something I can buy/fund/donate to? Shut up and take my money.
This is perceptual learning we are talking about, at the very lowest level - in this case just recognizing the orientation of a bar. And that's all the results apply to - not 'Matrix-like learning'.
And it's not passive, either. Subjects were asked to "somehow regulate activity in the posterior part of the brain to make the solid green disc that was presented 6 s later as large as possible (the maximum possible size corresponds to the outer green circle)."
That's effortful, although yes, they may not realize what effect this has on patterns of activity in their brain (patterns identified by fmri that the experimenters are trying to get the subjects to reinforce).
As with all technologies, there is use and there is misuse. These sort of fears are why I don't work for the government, but it doesn't stop things from being discovered. Indeed, discovery isn't evil, only what these discoveries and advancements may be used for.
Might as well take the chance to go slightly off-topic: anyone here a skill acquisition nerd? What methods do you utilize to gain more "I know Kung Fu" moments in life? :)
I'm kind of an interested dabbler in such brain hacking... one principle I try to follow is to use multiple parts of (that conglomerate called) my mind to learn things. So for example, take programming techniques. Using data-directed programming techniques lets me use the visual parts of my mind more directly... so in Lisp or Javascript, you'd make a pretty datastructure which you'd code plumbing to interpret. Same with drawing pictures to help visualize, such as graphs. (The old _Data and Reality_ mentions we frequently use graphs in representing concepts on the computer, as opposed to in data processing.)
And using a REPL is very interactive, and uses my kinesthetic parts. (Some emacs chords may have that effect too, like paredit's alt-shift-(, which surrounds the next form in a parenthesis bubble.)
Alan Kay mentions the study of like 50 mathematicians, how they thought. Surprisingly few thought in terms of symbols. Most were primarily visual thinkers. Some were also kinesthetic. (To lend credibility, people always mention Einstein was in this last kinesthetic group.)
Another is critical learning. So you try to question why something Is The Way It Is, why someone believes what they believe... (Being confused is a very good sign; explore the confusion. Confusion over very simple things led to revolutions in our knowledge.) This takes you from the role of passive learner to co-creator. (Of course, time is limited, and taking this to its logical extreme would mean you'd do multiple ten-year research projects... Might be great for humanity's knowledge, but few of us have this luxury.)
