Mar 26 2009

Brain on a Chip

Published by under Uncategorized
Comments: 24

Work continues to develop computers which more closely simulate brain activity. As I recently discussed, the brain is not a computer, and computers are not brains. But since brains operate by physical laws, there is no theoretical reason why we could not design a computer to function like a brain.

There are two types of approach to this – virtual brains and hardwired brain-like computers. The former approach uses conventional computer hardware to simulate a virtual brain. This project is far from achieving this goal, but programmers have already achieved significant milestone, such as the simulation of a cortical column of 10,000 neurons and 30 million connections.

The advantage of the virtual approach is that brain structure can be simulated directly without worrying about having to engineer functional neurons or their connections. The downside is that virtual brains require much more processing power than the brains they simulate. So virtual brains run incredibly slowly. It will likely still be 2-3 decades before raw computing power reaches the level where a desktop computer will be able to simulate a human brain in real time. (I am just talking about raw power here, not the knowledge to actually pull off the simulation.)

The advantage of the hardwired approach is that you end up with a processor that is much faster than a biological brain. The challenge, however, is in building something in silicon (or whatever) that acts like biological neurons in all the important ways.

An international team of researchers has taken a step in this direction.The project is called Fast Analog Computing with Emergent Transient States, or FACETS. Their latest product is a silicon chip that has 200,000 “neurons” which make 50 million synapses. This is still tiny compared to the human brain, which has 100 billion neurons. But they plan to connect a bunch of these wafers together into a superchip with 1 billion neurons and 10^23 synapses – still only 1% of the way to the power of the human brain.

The structure of these chips uses transistors and capacitors to recreate the same kinds of connections that neurons make. Project member Henry Markram has developed an algorithm for the synapses that he calls spike-timing dependent plasticity, which is designed to allow the device to learn and adapt. The chip is designed, like the brain, to be massively parallel. Traditional computers are serial – they can crunch numbers very quickly, but they have one stream of processing.

This chip design, it is claimed, combines the power of parallel and adaptive processing with the speed of silicon. They claim it can function 100,000 times faster than a human brain.

While the purpose of creating virtual brains is primarily to study the brain (although this line of research may result in artificial intelligence – AI), the purpose of this chip development is not to study the brain at all, but rather to develop more powerful computers and possibly computer AI.

Of course, while this is an important step, what FACETS is developing is not a silicon brain but a chip that may serve as a basic unit in such a brain. You cannot, in my opinion, just stack a bunch of these chips together and get AI.ย  A higher order of architecture is required. I guess, although I don’t know, that advancements in this technology – building silicon neurons – will advance more quickly than our understanding of brain architecture and functioning. The latter is progressing nicely, but the level of complexity is much greater.

So we may get to the point where we can build a massively parallel and learning silicon computer with the memory and processing power of a human brain, but not know how to put it together to create something we would call AI or, even more difficult, conscious.

I am curious as to which approach to designing the overall structure of such a silicon brain will take. The two basic approaches I am thinking of are 1 – duplicating the human brain, or 2 -designing it through trial and error from the ground up. Likely a combined approach will be taken, but we’ll see.

Also we have to consider what function we wish for the end result. Will it be placed in a robot which will serve as its interface with the outside world, or will it be on a desktop, or will it be the brains of a commercial jet or some other piece of equipment? Or – will it be implanted inside a human to serve as a supplemental brain?

I suspect that the approach that will emerge is to first design simple functions, like seeing, interacting with objects, controlling whatever physical form it is in, etc. And then progressively more complex “higher” functions will be added that control the more basic functions. This is exactly how the mammalian brain is organized. This process will be informed by our knowledge of how the brain is organized and functions, but will also have to learn through trial and error to fill in the gaps in our neurological knowledge.

And of course it is always fun to speculate about where this will lead – to Cylons, the Matrix, or to benign and obedient droids. Let’s hope it’s the latter.

24 responses so far

24 thoughts on “Brain on a Chip”

  1. Watcher says:

    And of course it is always fun to speculate about where this will lead – to Cylons, the Matrix, or to benign and obedient droids. Letโ€™s hope itโ€™s the latter.

    It’s just to soon after the end of BSG to make a point like, “Oh that’ll never happen, we’ll program them to be nice.” That ending with the baltar and caprica angels just hammered home how close we are (within the next 50 years) of producing a viable mechanized life-form. Pair that with the fact that many humans are too xenophobic to look past being a “toaster” let alone giving one rights, and then we have skynet, cylons, and the matrix.

    Although, maybe I just read too much sci-fi and watch to much TV ๐Ÿ˜›

  2. Watcher says:

    I should clarify “Viable mechanized life-form.”

    I’m thinking of a Roomba with the intelligence of a canine.

  3. tmac57 says:

    One of the more intriguing ideas that I read about years ago in a book about the quest for AI was that a computer with sufficient AI capabilities , much faster processing power that the brain, and unlimited access to all of our knowledge base, might be able to synthesize novel ideas (invent, discover) at a pace unattainable by even the most exceptional human brain (renaissance computer?) . Pretty wild idea perhaps, but who knows?

  4. Magnus says:

    Will we be able to cut back on many neurons because AIs won’t need to be able to control a human body?

  5. Magnus – if the AI is desktop, but not if it is in a robot. But likely it will need to take in and process sensory input in any case.

    It all depends on what we want it to do.

  6. barto says:

    Maybe you find interesting this: . I find some opinions very worth reading (especially Pinker’s and Hofstadter’s opinions)

    Dr. Novella, you are the expert here, so I just ask: how far we are from knowing the representation system of the brain?. We know what the electrons moving in a chip mean (and despite the hughe computational effort, we could draw direct lines between them and a high level language ), but do we know what the pulses between neurons mean?. If the answer is no, shouldn’t we know the representation system before comparing ‘computing power’ between a brain and a electronic circuit?

    (sorry for my english)

  7. “It all depends on what we want it to do.”

    Debate creationists.

  8. HHC says:

    I envision the brain on the chip as a useful add-on to the brain. For example, assisting brain damaged persons with information processing. The chip could provide feedback like a pacemaker for the human heart and keep brain waves regulated. I really don’t foresee future robots with a human consciousness. The mechanical device would have to have a duplicate of every biological system and organ in the human body.

  9. artfulD says:

    If it could be programmed to recommend actions based on choices that were in turn based on predictions formulated algorithmically from direct sensory inputs of sight and sound (receptors of which have already been been simulated), then by doing so it will experience a form of consciousness.

  10. tmac57 says:

    I’m not sure that human consciousness would be the best goal. Humans come with all sorts of baggage and requirements that could be dispensed with. Just don’t put it in charge anything dangerous please!

  11. artfulD says:

    Human consciousness by definition is restricted to entities that are human.

  12. Watcher says:

    Read a book awhile back called “CUSP” that had something similar to what HHC is talking about. Humans used a brain enhancement that basically input a slave/helper AI that was visualized only by the host. The AI could essentially run in the background when the host didn’t need it, but could be called up to do any number of things … including saving the host, by taking over the body, from a 10,000 foot free fall by continuously jumping off random falling debris. ๐Ÿ˜€

    Good book ๐Ÿ™‚

  13. Benjamin Lobato says:


    You mentioned (facetiously I think) that you hope these artificially brained beings will be obedient. But if these man-made brains are sufficiently sophisticated, couldn’t they have consciousness and emotions that are similar to humans? And if they do have these qualities, wouldn’t we have a moral obligation to treat them in a similar way as humans? Why should we expect them to be obedient?

    Perhaps it’s too early to be debating the moral implications of artificial intelligence, but whenever people talk about these possible new life-forms and imply that the obvious place for these beings is as our obedient slaves, it makes me uneasy.

  14. HHC says:

    The FACETS project will most likely be the artificial intelligence of the future which will be able to perform tests in space better than astronauts and withstand the temperatures and long hours of space travel in this solar system. This silicon project is more appealing than the French rat brain.

  15. Phospholipid says:

    I saw a presentation from Markrams group at a conference last year (can’t remember if it was SFN or EMBC) and the thing I still don’t get is what the point of the virtual cortical column really is. In essence, they’re trying to simulate Markrams multiple patch-clamp experiments at a much bigger scale, but what they constantly and unexpectedly find is that the simulation doesn’t correspond to reality. So they tweak it and start over. My problem is (and if someone can shed some light on this I’d be much obliged) that they always seem to be just far enough behind on the tweaking that, even when they get a better simulation than the last try, they can never make a prediction of anything that hasn’t allready been predicted by a simpler method (in their own lab, no less). What’s the point in Really Cool Science that doesn’t give us any new knowledge, other than the knowledge to do that specific Really Cool Experiment? Like you said in your post, it’s an artificial brain that requires far more computational power than that of the brain it’s simulating.

    On the other hand, even a very big and clunky computer that could Hello Dave me satisfactory would be pretty awesome.

  16. Phospholipid says:

    Did I say unexpectedly? I meant unsurprisingly. My bad.

  17. eiskrystal says:

    If we are going to build these up then it seems sensible to follow evolution. It has already been successful and it is currently being studied. The technology may also help fill in some of the voids we have about our ancestors.
    Going straight to humans will not i think reap as many benefits over the long run, fascinating as it is (and a serious slap in the face to religion).

    We don’t need more humans. Metal or otherwise.

  18. Ben – such silicon AI would only have the emotions we built into them. There is no reason to think that specific emotions will emerge out of consciousness itself. We have human emotions because human brains evolved to have them, and they are produced by specific structures and pathways in the brain.

    Silicon AI would not necessary have to have any survival instinct, or anger, need to be loved, or anything else.

    I do wonder if certain mental states, like boredom, do just emerge out of being conscious combined with a sense of time. AI will be fascinating because I think it will teach us a great deal about the nature of consciousness itself.

    Of course, this also brings up the notion of the 3 laws of robotics – we probably should build into the AI architecture basic safety features, like a reticence to harm humans.

  19. Watcher says:

    If we are going to build these up then it seems sensible to follow evolution. It has already been successful and it is currently being studied.

    Like self replicating robots? Or do you mean in another way?

  20. tmac57 says:

    What do you think about the possibility of an autonomously creative AI ; a robot inventor , so to speak?

  21. HHC says:

    The FACETS Project will teach us about programmed logical thought and AI processing. If the silicon chip is able to perceive and respond, this would be a limited form of consciousness. Perception and responses would be mechanized through programming.

  22. kvsherry says:

    I’m having trouble understanding how the FACETS project or others can lead us to create AI when we don’t fully understand our own consciousness. For it to be “true AI” wouldn’t it need to be more than just feelings we programmed?

  23. HHC says:

    FACETS project is for rapid computer processing. “Emotional” responses would not be useful in a computer generated decision.

  24. eiskrystal says:

    Like self replicating robots? Or do you mean in another way?

    Another way.

    I meant merely that putting money into studying and trying to replicate earlier and simpler creatures before tackling the more complicated would probably reap more benefits. I mentioned evolution as a ready made path of successes.

    I feel people get much more excited about human-type AI than the other research in this field.

Leave a Reply