Slashdot Mirror
A New Look At Brain Control
one_neuron_two_neuron writes "Researchers at Harvard have taken a new look at how electricity can make neurons fire in the brain. The scientists found some surprising things: if you stick an electrode in the brain and apply current, you don't just make a small group of neurons fire — many neurons fire a long way away from the electrode. That's probably because instead of activating the cell bodies of the neurons, their axons fire. Those axons are the wiring of the brain. Your cerebral cortex is something like a big pile of unwound yo-yos — if you stick an electrode into the cortex, you're much more likely to hit the strings (the axons), and the yo-yo connected to the string can be really far away. So, how will you ever hook up a computer to your brain? This data shows that we need to rethink how to do that with electrical current. If you stick an electrode in one place, neurons in a totally different place will fire. New optogenetic methods (e.g. using viral delivery of proteins) might work. Or possibly we will figure out how to make the brain learn to interpret these sparse, widespread electrical patterns. New optical techniques have made a dramatic impact on neuroscience recently, and this study uses pulsed-laser-scanning microscopy (two-photon microscopy) to take pictures of neurons deep inside the living brain. The academic paper (PDF) is available on the author's site."
Without a computer hooked onto your brain, how else would you be able to relate to a female computer and have sex with it? Duh.
Most likely, it will be firewire. Also, hooking an electromechanical computer to a biochemical computer might turn into something fairly awesome.
With some type of ray gun or beam weapon and hardened carapaces.
[http://it-tastes-so-good.blogspot.com] Are you hungry?
Who wants to? I can't think of anything dumber.
With the power of a computer attached to your mind, you would be able to think of many things that are dumber, several million times per second!
So, how will you ever hook up a computer to your brain?
I thought the keyboard already solved this problem... It's not quite direct but it seems to work.
Your cerebral cortex is something like a big pile of unwound yo-yos
WTF? Why don't they just say "the brain is a big pile of neurons and axons"? It would be more helpful than this bizarre analogy.
... and then they built the supercollider.
So, how will you ever hook up a computer to your brain?
Who wants to? I can't think of anything dumber.
People have always said that about new technology. Eventually someone comes up with a killer application and it takes off.
Here I sit, all broken hearted.
Came to poop, but only farted.
Most likely, it will be firewire.
So, it burns?
... and then they built the supercollider.
It makes me sad to think that some day, I'll be stumbling around with my brain machine interface, creating thought streams one at a time and accidentally thinking of something else halfway through, losing my work, and asking the kids for help.
And they'll look me in the eye and feign sympathy as they blast high-frequency shorthand thoughts back and forth to one another, mocking my generation for being so dumb that its members can't even work a brain port properly.
Or it doesn't, and it fades into obscurity like countless non-notable technologies.
The World Wide Web is dying. Soon, we shall have only the Internet.
the mouse.
dildo.
anal beads.
nasal spray containers.
gerbil.
Toilet paper is common tech and the application is most certainly direct.
Everything in the Universe sucks: It's the law!
Though, in this case, the killer application is going to be heavier on the "transhuman cyborg killbot" and light on the "visicalc"...
If you want to get as specific as neural stimulation, then cochlear implants, depending on how common you mean. But I'm glad to be in a world that has that technology. It's also very highly analogous and can benefit from the same research, as it involves direct electrical stimulation to a population of neurons. One of the limiting factors in cochlear implants' success is that it is currently (no pun intended) impossible to stimulate anything close to a narrow, precise range of auditory neurons; thus frequency resolution in CI users suffers dramatically. We have something like 30,000 auditory neurons along on almost continuum of frequencies; studies tend to show that with a CI you can have up to a maximum of 8 separate frequency channels before performance saturates.
Or it doesn't, and it fades into obscurity like countless non-notable technologies.
Applications of the technology may fail to catch on and fade into obscurity, but that doesn't prevent someone else from rediscovering using the technology in a new application later on.
Here I sit, all broken hearted.
Came to poop, but only farted.
From the sophomoric language in the summary (my apologies to real sophomores, like the ones in high school) and the fact the paper hasn't been submitted anywhere and so who else would have access, I conclude that it is the first author (a student), not the second (a post doc that did the matlab work) or third (an MD/PHD) that posted this here. Calling yourself a researcher and not making your student status clear is a failure of full disclosure. It also works against you in that people will forgive a students a lot of mistakes that they'll end up holding against you otherwise.
In your hurry to promote your ideas you gloss over or ignore a large body of research that disproves your assertions.
- We've known for a long time about sparse networks and distal activation. The distribution is due to Hebbian assemblies. Hebb, Donald O.
- We can and do stimulate and record nearby spatially and temporally. Use the stimulation pulse as the deactivation signal on the recording probe. We can even do them simultaneously with high impedance, high speed data collection equipment and adjusting the output Y scale to logarithmic. Ask Vince to explain this last part. We also use magnetic stimulation and shielded electrical probes.
- Inadequate and poorly stated background material is a sure way to not only get rejected but also get remembered by those editors and rejected in the future. You've got to assume that not only will some of them know about the field, some of those reading your submission may be among those being misrepresented or those who you fail to represent but should have.
- The above goes double for trying to make your idea sound good by making others sound bad. It pisses people off, and you may be wrong (you are) which just gets your manuscript returned.
- You fail to show that your method can be generalized to normal neuron activation, and thus be useful for (the inadequately described) computer interface, or anything else that requires 'normal' operation. And not being able to show normal operation means not being able to tell if and/or when your method causes incorrect operation or damage.
- The background theory and history shortcomings call into question your theory, the need for your technique, and the usefulness of your large volume of supporting data and graphics. This last part looks like argumentation by bafflement - flooding the reader with overly technical details in order to bullshit them into accepting your other points. Even if it's not that, it looks like it.
Does Dr. Reid know this was posted here?
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
Not quite direct?
It's like trying to get water from one bucket to another, but instead of using a pipe, having to turn it into steam and use a tennis racquet to force it into its new bucket.
Think of all the unnecessary processing that goes into the hand and eye coordination for data entry into a computer, or for the opposite case: all the visual processing that brain needs to do to in order to recognise symbols on a screen, form words and subsequently string those into coherent thoughts. Its utterly and completely inefficient. The only reason we find it acceptable is because at the moment it's the best system we have.
If we could either directly interpret thoughts accurately or feed information into the brain (a task of incredible complexity) we could drastically speed up all sorts of information processing tasks.
Mechanical data entry is probably much easier than a direct linkup though. Unless you implant the connections in a child way before birth - ie before the brains wires itself for the mechanical way of doing things exclusively. How would you learn to use a direct linkup with a computer? Not a direct concept at all. Easier to grow a child with it already in place so they can learn to use it over a decade or so - just like we learn to use our hands so well.
sudo mount --milk --sugar
You learn by doing. Much the same way you learn to type. You start off chicken pecking and a with some work you start to type faster and faster and faster. You type faster because your brain adapts and no longer needs visual input to know where the keys are, your finger muscles get stronger so you can move your hands faster.
Another example, games. I used to only be able to play first person shooters on the PC. I was one of those people who said "you cant get better then a mouse and keyboard" and "there is no way you can play a fps on a console". Then Halo 2 came out while I was in college, and everyone had to play it. I did not want to at first cause I though I would suck. And I did suck. I could bearly look and move at the same time. However, after playing it for a month I started to get used to moveing and aiming with a controller. I started likeing the fact I could go from a slow walk to a full run just by tilting the stick more. Today I can pick up a console shooter and play it as naturaly as I can a shooter on the PC.
With the brain interface I dont think it will be much different. At first you will really have to focus on each letter or word to get it to display but after a while it would get eaiser and eaiser for words to just appear on screen and before you know it you are makeing fully formed paragraphs with no more effort then simply thinking about what you want it to say.
I do that every day and I don't need no stinking computers. Sigh.
Your cerebral cortex is something like a big pile of unwound yo-yos
Yeah, great. Thanks for clearing that one up Slashdot.
either directly interpret thoughts accurately or feed information into the brain
That is a call for i/o engineers right ?
It also requires extensive (and dangerous) surgery, has a high rate of failure due to poor education techniques, and the machines themselves have a high rate of failure. They're extremely rare and tend to be uncomfortable, causing rashes every once in a while. They're also uncomfortable to keep on.
A keyboard. How quaint. -cracks knuckles-
I am officially gone from
If you start improving cochlear implants significantly (better fidelity etc) and attaching computers to them, I'm sure the RIAA and gang will have a word with you, and "convince" you to include DRM.
So will it be a penny for your ("their") thoughts? I'm thinking they might charge more.
Particularly the language "totally different place."
Neurons in a "totally different place" will fire....this kind of language doesn't comfort me about the prospect of sticking electrodes into my brain. "We're not totally sure which neurons we're firing. It could be the ones we're touching with this little wire or it could be ones in a totally different place.. I guess we'll find out, huh?!?!"
ï
If my neurosurgeon said that to me as he started drilling into my skull I would be wanting to strongarm my way out of the OR....
ïIf only I could get my head out of this damn Clockwork Orange device...
So true. It's too bad that once something is invented it's impossible for it to ever be improved.
The technology has improved, it still remains slapping a computer in your cochlea, the discomfort will not go away, the rates of failure have not gone away, and the surgery is still done in a ridiculously sensitive part of the body.
This is so patently obvious that to call it a new discovery is just silly. Who conducted this study at Harvard; the accounting department?
Anyone involved in brain science knows how interconnected the brain is and how remotely connected neurons are. Why would anyone assume that forcing one neuron to fire from an electrical stimulus would only affect neurons locally?
you don't just make a small group of neurons fire -- many neurons fire a long way away from the electrode. That's probably because instead of activating the cell bodies of the neurons, their axons fire. [...] and the yo-yo connected to the string can be really far away.
Was this ever not blatantly obvious?
I'm always shocked by how unobvious the whole concept of the brain is to even the very "scientists" who work on it.
They seem to always be caught in some tiny box and/or focusing on details that are only artifacts of the foundational rules.
I'm sorry, but trying to identify what a specific part of the brain does (except for some really specific parts like the cerebellum), is a pointless exercise, that shows that you don't really understand the point of the way a neural network works.
It is like looking at the digital data inside a memory chip, instead of the wires themselves, to understand how the chip itself works. You can gain something from it. But why use such a bad point of view?
And this is even worse: I mean, who here seriously did not think that when you stick an electrode in your brain, the electricity will run down the axons it touches, to land anywhere that axon ends (Which sometimes can be more than 2 or 3 feet away somewhere down/up the spine!)
What did they think before? That it only touches the cells themselves? That's like only touching the fruits inside of a rolled up raspberry bush.
Oh, and by the way: Yo-yos? Really? That's the worst analogy ever. Or did you do that to foster "Yo dawg, [...] yo momma and yo yo ma's yo-yo [...]" jokes?
Any sufficiently advanced intelligence is indistinguishable from stupidity.
If we could either directly interpret thoughts accurately or feed information into the brain (a task of incredible complexity) we could drastically speed up all sorts of information processing tasks.
Why do you believe that?
You insist that we "think of all the unnecessary processing" that goes on, but give us no reason to believe that it is unnecessary other than, presumably, the fantasy that it can be dispensed with. Why do you believe that it can be dispensed with? The most likely case seems to me that it can be externalized, so that a computing device will interface with our neural networks and neuro-chemical systems somewhat further upstream than it does now.
In particular, I have no idea at all what you mean by "directly interpret thoughts". What is a "thought" in this context? There is a huge amount of stuff going on in the brain at any given time, and no reason to believe that any of it can be mapped onto an isolated entity that could reasonably be labelled a "thought" by any means other than looking at the operational outputs of our conventional interfaces: our words (spoke, written, or internally recited) and actions.
You are bringing an huge and utterly unjustified load of assumptions to your argument, and if some doesn't buy into your peculiar model of the brain, which I don't, then what you're arguing for it just silly and your conclusion that somehow externalizing the transformation from internal neuro-chemical and neuro-electrical behaviour to operational effects could "drastically speed up all sorts of information processing tasks" is entirely unjustified.
The brain is primarily a neuro-chemical machine--the neuro-electrical aspects of it that computer people tend to focus on are a very small layer on top of the underlying neuro-chemistry. And the chemistry has time constants associated with it that are adequate for information processing using the operational inputs and outputs we have: there is no reason to believe that bypassing those systems would result in any speedup at all. Indeed, it would be evolutionarily bizarre if the brain were capable of doing much more than its current speed, because there cannot ever have been any evolutionary pressure of any kind to select for that.
Your argument is like saying that if we put a couple of jet engines on a steam locomotive it would be able to break the speed of sound, ignoring things like the mechanical limitations of th wheels and the tracks: you're focused on one obvious limiting factor in the body/brain identity's performance, and supposing that because you can imagine the rest of the system is capable of much faster speeds that it actually is.
Blasphemy is a human right. Blasphemophobia kills.