Mapping the Brain's Neural Network

Ponca City, We Love You writes "New technologies could soon allow scientists to generate a complete wiring diagram of a piece of brain. With an estimated 100 billion neurons and 100 trillion synapses in the human brain, creating an all-encompassing map of even a small chunk is a daunting task. Only one organism's complete wiring diagram now exists: that of the microscopic worm C. elegans, which contains a mere 302 neurons. The C. elegans mapping effort took more than a decade to complete. Research teams at MIT and at Heidelberg in Germany are experimenting with different approaches to speed up the process of mapping neural connections. The Germans start with a small block of brain tissue and bounce electrons off the top of the block to generate a cross-sectional picture of the nerve fibers. They then take a very thin slice, 30 nanometers, off the top of the block. 'Repeat this [process] thousands of times, and you can make your way through maybe the whole fly brain,' says the lead researcher. They are training an artificial neural network to emulate the human process of tracing neural connections to speed the process about 100- to 1000-fold. They estimate that they need a further factor of a million to analyze useful chunks of the human brain in reasonable times."

Comments

Missing tag

[2.7182]

Vaporbiology

Re:Missing tag

[Feminist-Mom]

Not to mention - and I a sure I'll get modded down here - is that neural networks aren't very effective. They get a lot of hype in the media, and people who don't work in optimization like them. They sound appealing and cool, but there are other methods that are much better. If you look at the liturature, there is not that much hard science behind them. Some statistical mechanics people have some results, but they really are just a fad, like fractals or catastrophe theory. But bear in mind LOTS of people in applied areas immediately jump to the conclusion that neural networks are great, especially computer vision and robotics people (where I used to work.)

Re:Missing tag

[Space cowboy]

Perhaps you didn't get much out of neural networks, but my PhD thesis was on the similarities between a kohonen network and relaxation-labelling equations. Part of it is up on my blog (I haven't actually got as far as that bit yet, but the groundwork is there).

A neural network (well, anything more complex than the single-layer perceptron anyway) is an arbitrary classifier. I'm curious as to why other methods are "much better". Unless you do an exhaustive search of the feature-space, all classifier methods are subject to the same limitations - local maxima/minima (depending on the algorithm), noise effects, and data dependencies. All of the various algorithms have strengths and weaknesses - in pattern recognition (my field) NN's are pretty darn good actually.

It's also a bit odd to just say 'neural networks' - there are many many variants of network, from Kohonen nets through multi-layer perceptrons, but focussing on the most common (MLP's), there's a huge amount of variation (Radial-basis function networks, real/imaginary space networks, hyperbolic tangent networks, bulk-synchronous parallel error correction networks, error-diffusion networks to name some off the top of my head), and many ways of training all these (back-prop, quick-prop, hyper-prop, batch-error-update, etc. etc.) I guess my point is that you're tarring a large branch of classification science with a very broad brush, at least IMHO.

Not to mention that this is all the single-network stuff. It gets especially interesting when you start modelling networks of networks, and using secondary feature-spaces rather than primary (direct from the image) features. Another part of my thesis was these "context" features - so you can extract a region of interest, determine the features to use to characterise that region, do the same thing for surrounding regions, and present a (primary) network with the primary region features while simultaneously(*) presenting other (secondary) networks with the features for these surrounding regions and feeding the secondary network results in at the same time as the primary network gets its raw feature data. This is a similar concept (if different implementation) to the eye's centre-surround pattern, and works very well.

If you work through the maths, there's no real difference between a large network and a network of networks, but the training-time is significantly less (and the fitness landscape is smoother), so in practice the results are better, even if in theory they ought to be the same. I was using techniques like these almost 20 years ago, and still (very successfully, I might add) use neural networks today. If it's a fad, it's a relatively long-running one.

Simon.

(*) In practice, you time-offset the secondary network processing from the primary network, so the results of the secondary networks are available when the primary network runs. Since we still run primarily-serial computers, the parallelism isn't there to run all of these simultaneously. This is just an implementation detail though...

Re:Missing tag

[schnipschnap]

The pig go. Go is to the fountain. The pig put foot. Grunt. Foot in what? ketchup. The dove fly. Fly is in sky. The dove drop something. The something on the pig. The pig disgusting. The pig rattle. Rattle with dove. The dove angry. The pig leave. The dove produce. Produce is chicken wing. With wing bark. No Quack.

Re:Missing tag

[Moralpanic]

How is it not effective? Could the problem be, like you alluded to, that there isn't much science behind it atm... and thus perhaps we don't fully understand how it exactly works, so we can't replicate it 'effectively?'

For something that's a couple of pounds, and doesn't require a lot of energy, it's pretty damn effective if you ask me.

Re:Missing tag

[Tablizer]

If you work through the maths, there's no real difference between a large network and a network of networks, but the training-time is significantly less (and the fitness landscape is smoother), so in practice the results are better, even if in theory they ought to be the same.

Almost every artificial organizational technique (computers, militaries, governments, etc.) seems to partition up complexity into semi-independent units. Perhaps there's in inherent advantage of such.

Also, could you please recommend an intro book that helps one get a natural "feel" for how NN's work? I read Maureen Caudill's "Naturally Intelligent Systems" (et al), but it jumped from super-trivial to complex too fast for me, becoming a mere catalog of techniques instead of an explainer of them. For example, I got a feel for how basic image recognition can work, but not for say identifying image patterns of different orientation (rotated) or scalings.

Mellanie Mitchell did a fine job on her Genetic Algorithm intro book; I wish she also wrote a NN book in the same style. She's such a gifted writer for such topics that she should have made a series of AI-related intro books.

Re:Missing tag

[Tablizer]

When Haiku fans snap, news at 11.

Re:Missing tag

[Maian]

I figure that inherent advantage of partitioning up complexity is parallelism and the efficiency that results. Depending on what's being parallelized, it increases throughput, execution times, etc. Heck, you could consider human productivity itself one massively parallel machine. Ever heard of the term "man-hours"?

Re:Missing tag

[mesterha]

A neural network (well, anything more complex than the single-layer perceptron [wikipedia.org] anyway) is an arbitrary classifier. I'm curious as to why other methods are "much better". Unless you do an exhaustive search of the feature-space, all classifier methods are subject to the same limitations - local maxima/minima (depending on the algorithm), noise effects, and data dependencies. All of the various algorithms have strengths and weaknesses - in pattern recognition (my field) NN's are pretty darn good actually.

Actually, more recent methods don't have local maxima/minima. Something like a support vector machine optimizes an objective function. Of course, this is somewhat of a tangent, in that the objective function might not be a useful metric for performance, but people have shown that the minimum objective function value of a SVM does relate to its generalization performance. It's a little disconcerting that a NN has an objective function but that it can find it's minimum or that the minimum doesn't give good performance on test data (over-fitting)...

Of course, part of the NN's problem stems from the fact that it is an arbitrary classifier. It's hard to give generalization results for an algorithm that has an infinite VC dimension. (There are techniques to restrict the size of the weights to give some guarantees.) However, this doesn't mean NNs can't perform well in practice. It probably means that the current theoretical analysis is somewhat flawed in relation to the real world.

So have you ever compared your NN algorithms with the popular algorithms of the day such as SVMs with kernels or boosting algorithms. Also, are your NN algorithms generic or do you heavily customize and tweak to get good performance.

Re:Missing tag

[Space cowboy]

As far as generalisation goes, with the stuff I do there is generally an abundance of training data - it's easy to get imagery, and it's easy to generate the input data from that imagery. As such, using some of the data as training data, and the remainder as generalisation-test data is a good (and pretty standard) approach. All you do is train on (say) 2/3 of the data while simultaneously checking the results against the remaining 1/3 of the data. You'll get two error-curves, which initially approximate each other, decreasing exponentially with time. After a while, however, the net will start to model the specifics of the training data (and thus continue to reduce the error for the training-data), but the generalisation-test data is different data, so *that* error curve will start to rise. To get the best generalisation from the network, all you need do is take the weights-state at the point of minimum error on the generalisation-test data.

I've used some linear discriminant techniques (primarily Principal Component Analysis and Canonical Discriminant Analysis) to get some information about the separability of the network input data, but not actually used an SVM as a classifier.

Generally I create networks using the centre-surround approach I detailed above. Each network is automatically trained with N runs over M configurations (for both hidden nodes and hidden layers), with the best result being selected. Networks are then trained to handle the fusion of these first-order networks into a new layer of networks - think of it as a multi-resolution approach. At the bottom level you have nodes, which combine to make networks, which combine to make networks of networks. I've also used network topologies where each classification-label has a single network representing that label rather than N outputs, one per classification, as well as hybrid systems.

Once the first-pass training has been completed, the confusion matrix (*) is determined, and is used to highlight typical misclassifications (where 'sky' and 'water' are being classified as the same thing, for example). This is where PCA and CDA come in, and I also have a derivation for the actual information content of any given input-data feature, expressed in terms of the output classification accuracy. These are used to split the conflicting categories and add network(s) that have a better chance of discriminating the problem categories, just prior to the fusion stage. All subsequent networks in the decision tree are retrained, and if the result is better, the next most-confusing pair is chosen and targeted. If the results are not better, the new network is discarded. This repeats until the desired error-rate is achieved on the generalisation-test data, or a set number of runs through the whole system are complete.

So, the answer to your last question is: both :) All the above is automatic, I just have to review the results. My main decisions lie in making sure the system is still generic at the end of its development, and if the error-target wasn't reached, doing more research on input-data that is more-likely to give a good result.

Simon.

(*) A confusion matrix is simply a plot of "what the network classified as" vs "what the truth is". What you want to see is high numbers on the leading diagonal, and preferably zeros everywhere else.

Re:Missing tag

[mi]

They sound appealing and cool, but there are other methods that are much better.

Much better for what? If, in 30 years, this technology allows full mapping of the entire human brain, I'll be quite happy... It will mean, that my conscience may live after the mortal flesh dies.

It may take another hundred years or more for full reconstruction of a new brain (and the rest of the body) to become possible, but in the mean time I'll be preserved just as I was, when I died and my brain was scanned.

Even better, an electronic "simulator" may allow me to continue to exist even if without the body. Many of the physical experiences will be beyond my reach, but I will be able to enjoy my profession — without interruptions for sleep, food and related discharges — communicate with friends, and even post to Slashdot.

Come to think of it, there are very few activities, that I enjoy, which require a traditional body — and even fewer of them are good for it. There is a bigger part of life, which requires interfacing with computers — and that's where most humans are currently rather handicapped...

Today a large part of population of developed countries is employed in taking care of the elderly. As we develop further, the younger generation will be busy with taking care of the computers and with reproduction — simply copying digital images of the same people will, likely, be frowned upon, or even illegal.

Jobs as diverse as stock-trading to traffic control will be performed by "virtual" minds...

No GPS Needed

[pipingguy]

In my case it'd be trivial: brain cell 1 ---> brain cell 2. Done!

Did I get that right?

[UbuntuDupe]

They are training an artificial neural network to emulate the human process of tracing neural connections to speed the process about 100- to 1000-fold.

So, they're training a neural network to automate the process of mapping a neural network, in the hopes of creating an intelligence that they can train to automate other processes?

My brain hurts...

Re:Did I get that right?

[AndrewBuck]

Hey I think you might be on to something here. If we can just layer enough neural networks decoding neural networks on top of each other we can follow the pain through our brain an map the connections that way.

-Buck

Re:Did I get that right?

[Air-conditioned cowh]

"My brain hurts..." Please hold still. It's being mapped. Thanks.

Re:Did I get that right?

[Anonymous Coward]

There's a big difference: Artificial Neural networks are trained for specific purposes, while the normal one is self-configures in an environment that is difficult to get access to.

Re:Did I get that right?

[Anonymous Coward]

There's a big difference: Artificial Neural networks are trained for specific purposes...

Maybe that's a good thing.

...while the normal one is self-configures...

[move cursor] [highlight results] [beer] [copy to clipboard] [reply to boss' stupid question] [paste into report] [ummm, looks nice outside] [pr0n] [gotta' get that VPN tunnel working] [beer] [wow, looka' the tight sweater on that bird]

(boss) "Did you hear what I just said?" "How long are you going to keep messing with that report?"

the hardware is not the software

[Anonymous Coward]

just because you have a network, doesn't mean you have a system

what about the wetware?

Re:the hardware is not the software

[Anonymous Coward]

If you know the state of all fuses in an FPGA, don't you know all there is to know about it?

Re:the hardware is not the software

[Anonymous Coward]

Minds are not FPGAs, neurons are not fuses

your state of mind is not your mind state

Re:the hardware is not the software

[Anonymous Coward]

There are probably a bunch of chemicals involved, too, but unless you think that the brain is just a placeholder for a supernatural mind, the neural network is the hardware and the software, just not the "register contents".

Re:the hardware is not the software

[mikael]

The Halting Problem

Re:the hardware is not the software

[Anonymous Coward]

What about it? We're not trying to figure out if the brain halts, we're just trying to make a copy.

Re:the hardware is not the software

[onemorechip]

Indeed. We already know the brain halts.

Re:the hardware is not the software

[Anonymous Coward]

No, we only know that for the brains which we've looked at so far.

What constitutes a "map" here differs from elegans

[jpfed]

One reason we could do the aforementioned mapping with C. elegans is that the worm's neurons are always laid out the same way from worm to worm. This is not the case for humans, and probably not the case for any vertebrate.

Interesting

[rm999]

Sounds interesting, but is a map enough to understand the brain? I know in artificial neural networks, the actual structure isn't as important as the weights on the nodes. Will hitting the brain with electrons be enough to give us an understanding of these "weights", or just the connections between them?

Re:Interesting

[bazald]

Or more to the point, is a neural network really the correct interpretation of a brain's structure? I suppose we might find out soon, if they can get the "weights" "right".

Re:Interesting

[Anonymous Coward]

Yeah, good point, let's just go with "God did it."

Re:Interesting

[Anonymous Coward]

A neural network would only be fraction of a model for the brain, and the edges are merely the transmissions. We still have to decypher the chemical data stored in each node, as well as whether the location of the node adds context to the data contained, or at least observe enough discrete patterns to work out what is doing what on its own.

Better than a poke in the eye with a pointed stick however.

Re:Interesting

[melted]

Weights are trained and they'll be different depending on the connections and on the training. The rough structure of long-range connections is actually fundamental to how the brain works. This is essential to explain long and short term memory, fear, involuntary reactions to stimuli, perception of the world as a coherent audio/visual picture as opposed to a set of disparate inputs, etc.

Re:Interesting

[unixfan]

Hell, it will be fun to see what they figure out to be the capacity of the brain, as far as how much information it can store.

I know visually we are looking at least at something around 24 frames per second. The eye is supposed to have a resolution of around 1000 dpi. Not sure how to measure the viewing area. But let's say it is lesser and lesser resolution the higher the angle. Let's say, just to have a number, that we have a 16:9 viewing ratio at two feet distance. Lets say it's three feet wide. That should add up to something around 36"x20". At 1000dpi that would be something like 729,000 dots. Time 24 per second becomes 17,496,000 dots per second.

Though I think people who have dissected eyes and the stem to the brain would have a hard time quite understanding how that dpi reaches the brain.

On a daily basis, if we don't count 8 hours sleeping, we still come up with 280 million pieces of data in 16 hours.

If an average brain is 1400 cubic centimeters or 85 cubic inches, how many cells could it have if we say it is solid. Best case scenario.

I see numbers of 9,350 cells per cubed millimeter which is 93,500 per cube centimeter. With the above brain size we are looking at 130,900,000 cells in a solid brain.

To complicate things further, how many days of memory do you have? Most people have problems remembering all details but then some people have photographic memory. Which as far as I can see means that all of us has the potential to have it.

The running question is how much info is stored in each cell.

Of course that's not including all the other senses and impressions that are stored.

Re:Interesting

[no-body]

Will hitting the brain with electrons be enough to give us an understanding of these "weights", or just the connections between them?

Nope - it will be just the wire connections, just as if computer hardware without any software will have it's electrical circuits traced in order to understand better what a program running on the screen does - about in that magnitude, probaly much higher.

The "software" on a human brain is programmed from before birth and constantly changed.
Just the computing power of keeping all the muscles (multi-mode, contracting, brake-extending, stiff-blocking, spring-storing) coordinated and oriented in space to keep a human upright is enourmous.
The human (or any animated animal) body is an n-power parallel processor system with multi-media (pdf doc) parallel input/output processing.

Re:Interesting

[Anonymous Coward]

Just a quick FYI: The eye does not have a frame rate and the eye does not have a constant resolution. You would be surprised how little you see and how much of the image that you perceive is "fiction" (a good guess based on what you know and have seen).

Re:Interesting

[diskis]

The eye is suprisingly simple. Your brain doesn't get all that much information. A lot is filled in by your brain, based on interpolation and assumptions. You are aware that you have a huge hole in your vision right?
Read up on it. http://en.wikipedia.org/wiki/Blind_spot_(vision)

And the raw video feed from your eye is not stored. It is interpreted and discarded.

Memories are the same. Not that much information. Memory is suprisingly volatile, and your brain does not retrieve information from a built in hard drive. It remembers some, and rebuilds the scene according to your memories. I have a personal example of this, I can bring up a very lifelike picture/memory from when I was a kid. I'm in a dressing room, looking at myself in the mirror. However, I do have a scar on me, the same scar I got years later. So I know for a fact that the memory or image my brain digs out is a fake, a composite of some kind.

Re:Interesting

[unixfan]

Actually it's easy to see that the whole nervous system does run at a frequency.

Wave your hand rapidly in front of you and tell me if you see one hand moving or a series of snapshots. If you see a series of snapshots then you have exceeded the rate by which the eye sees.

Ditto, if you know anything about resonating frequencies in object you'd notice how your nervous system is responding to certain frequencies. It clearly runs at a frequency.

I'm not going to tell you about your vision, and I agree on the low volume of input from the eyes. It's the conclusion I don't agree with. My brain does not fill in what it is extrapolating because I would have probably been dead if I had lived part of my life with "guesses."

The "guessing" comes from not being able to come up with a satisfactory answer, having seen how poorly the eyes sees.

Re:Interesting

[Anonymous Coward]

If you can see snapshots, it is because you have shitty lighting, you are waving your hand in front of a monitor or your eyes are tracking the moving object for a split second. The photo receptors in your eye fire a signal when the "sensor is saturated". It's the rate of these signals which determines the brightness. The eye does not measure the amount of light which reaches its sensors in a fixed amount of time. The process does not need to be clocked and is not clocked, but it is discrete. You can think of it as a kind of pulse width modulation with a fixed on-time and variable off-time.

Your brain does indeed fill in a lot of information. You "see" things which are not really there and it's easy to prove. Another reply to your comment already mentioned the blind spot, but that's not all. The resolution of our eyes is indeed impressive, but only in a very small area at the center of your vision. The rest is pretty much a blur and only there to detect change and motion. You think you see the rest of the picture, but you only see it because you've looked there before and it hasn't changed enough to make you look again. Much of what you see, and even more so what you remember to have seen, is simply what you expect to see and not the result of actual sensory input. It works because we've learned to look at change. What simply "is" isn't so important and is filled in from prior information, and that is essentially a guess, an informed guess but still a guess. That's how people can have something in their field of view and literally not see it. It was in their view, but they just didn't look at it because their attention was captured by something else.

Re:Interesting

[smallfries]

When you pull figures from your arse, you don't actually add anything to the discussion. All that we've learned is that your arse is rather large, and you are used to removing things from it.

Even if your (incorrect assumptions) were correct, 36" x 20" at 1000dpi would be 36000 pixels x 20000 pixels = 720M pixels. Clue: dpi is a scalar measure rather than area.

Of course, the human eye does not work anything like that. Rather than farting numbers I spent 10 seconds on Google to find this which looks into the question of Visual Acuity. The "high-res" part of the eye is a very small circle with about 120 "dots" across its diameter.

As we do not resolve entire "frames" in a single go, the concept of a frame-rate is completely ludicrous. Your argument earlier in the thread about observing skipping when seeing a high speed stimuli doesn't show evidence of a *periodic* frame rate. It just shows that there is a *minimum* temporal resolution. One does not imply the other, especially when the eye is processing asychronous input (from rods and cones).

Although you don't believe that the brain fills in the missing images with educated guesswork, we've already established that what you believe is shit. Most (if not all) neuroscientists have accepted that the high resolution continuous visual imagery that we see is mostly an illusion produced by the mind. There are many well reported experiments that provide evidence of this. You should look for anything on Visual Illusions - there are far too many decent results in peer reviewed journals for me to spend time looking for you. Change Blindness is a related phenomena.

Finally you've cooked up some stupid figures for the number of cells in a brain. Why do you feel the need to demonstrate how stupid you are? The actual numbers (which you get wrong by 3 fucking orders of magnitude) are in the summary of the article! How hard is it to read the 100 billion neurons at the top of the page.

So next time you feel the need to pontificate needless about something that you don't know anything about. Don't. You, sir, are a thief of oxygen and your pointless ramblings have made everyone reading this article collectively dumber.

PS Feel free to mod me flamebait, as I am clearly annoyed. But when you do so remember that the everything the parent poster wrote was incorrect and that I have pointed out to him where he is wrong.

Re:Interesting

[unixfan]

Shitty lighting outside in the sun!?

Tell a fencer they are guessing what is coming in. I know this is the best explanation that brain diggers can come up with. I'm just not that narrow minded.

Re:Interesting

[Anonymous Coward]

Shitty lighting or one of the other reasons. You don't see fast moving objects as anything but a blur unless they're illuminated by a strobe light (shitty fluorescents), in front of a strobing background (monitor) or you're tracking them (which is mostly involuntary.)

A fencer's eye senses motion, not a rapier with a tip, a handle and a hand that's holding it. The fencer knows that that is what the moving object is, so that's what he "sees".

Re:Interesting

[E 99]

Sounds interesting, but is a map enough to understand the brain? I know in artificial neural networks, the actual structure isn't as important as the weights on the nodes. Will hitting the brain with electrons be enough to give us an understanding of these "weights", or just the connections between them?

To me, the bigger question than the "weights" and other mediating factors of the brain's network, is where do the network's outputs go? Artificial Neural Networks, like the biological kinds, are indeed great at processing a large collection of input to form a unified impression, but that means there are inputs and then outputs. So the millions of "pixel" inputs from the retina go into a huge neural network in the visual cortex, to output a unified 3D visual impression... so where does that 3D visual impression data get sent to so that we can actually consciously experience the visual impression?

Those who believe that the neural networks in our brains are themselves conscious, would, I suppose, say they go nowhere -- that the solution is computed and then the signal essentially dead-ends, that we somehow experience it without any actual mechanism to experience it. Personally, I find it an irrational belief.

However, if the brain is merely an interface to the conscious mind, where in the brain provides that interface and how? What kind of arrangement could allow matter to interact with mind?

Re:Interesting

[steelfood]

A) The eye compresses images even before it gets sent to the brain.

B) We don't memorize the entirety of whole scenes (not even those with photographic memories, though they're close). We use pattern recognition. That's why you can tell coke can is a coke can given from a slightly dented coke can to a crushed coke can, and can tell that even a crushed coke can isn't a crushed coke bottle. You memorize the patterns that make up a particular concept, match any given object to your set of memorized patterns (reducing the result space of the object with every recognized pattern), and if there are no exact matches, you deduce what would be the most likely match given how important certain patterns are to the object's identity.

c) Memory is not perfect. In fact, we fabricate memories to fill in the gaps that inadvertently appear. We use a combination of what we think we remember, what we logically deduce, and what we want to know, to create these gaps. How well we remember things is dependent on the strength of the former two, and how susceptible we are to the latter.

Re:Interesting

[Tablizer]

Actually it's easy to see that the whole nervous system does run at a frequency. Wave your hand rapidly in front of you and tell me if you see one hand moving or a series of snapshots. If you see a series of snapshots then you have exceeded the rate by which the eye sees.

I believe this is an artifact of artificial lighting acting as kind of a strobe. Because of the variance in the sample rate of the cells of our eyes, we generally see each "frame" as a blurred hand such that it is hard to see the actual brain frame-rate using such a technique.

Re:Interesting

[unixfan]

We simply don't share the same views. What's true for you is not true for me.

Re:Interesting

[Maian]

Dude...this is science. If you don't share the same view, please provide evidence against it rather than spout some relativistic crap. It's stupid to take the relativistic high ground, because in scientific debates, we must assume a shared scientific philosophy.

Re:Interesting

[unixfan]

"Filled in by your brain" eh?

I agree, if you look at my first post, about the poor eye vision. But not it being filled in in any way. I know from my active life in the military, martial arts, racing etc what is real and not. If you drive at 150mph or have a fast kick flying at you, you cannot do deal with it on a guess, or filled in best guesses. Not that the people who come up with these ideas are usually from that part of life anyway.

When you can discuss all sorts of details with someone who has a photographic memory, you quickly see this is not interpreted.

You think you are a brain. That's fine with me. Western philosophies says so. I know I'm not a brain since I was a little kid where I certainly practiced drowning a number of times. I remember things as a newborn that my mother confirmed, wondering how I knew. To you it is just my brain hallucinating or someone telling me. However only me and my mother knew and she had never told anyone this silly detail.

My life however does not subscribe to that point of view, to partly quote Sting. The brain "facts" does not add up. Too many interesting coincidents that does not fit.

As far as volatile memory, yeah it can certainly appear so. However throughout my life I've done experiments that clearly shows that we never forget anything. It's all there. It just is not something that can be easily explained if the only possibility is that you are a brain, a piece of meat.

Simple experiments like standing in a crowd and looking at a person who is not at all looking at you. Very shortly after I really start looking with interest, that person just turns around and looks right at me. Or the other way around with me finding someone looking intently at me.

I can tell you that happens to me all the time.

Someone I know, who I was visiting, was being burglerized. They were gone for the weekend and I walked in through the garage on the second floor. Instantly I felt the presence of someone who did not belong. It was something threatening. My neck hair stood straight up and I knew someone uninvited was there or had been there. When I made it down to the first floor I discovered things were not in it's normal orderly self and a broken french door. Someone had busted the glass door and spent time looking for something, valuable probably.

A guy in a company I was helping went to the owner and tried to have me fired one evening. That night I suddenly felt as if someone was attacking me, and when I walked in the door the next morning I asked the owner what was going on. He told me about the other guy's attempts. He had however never said or hinted anything to me, but been inspired from something the day before.

I know, I know, I'm severely hallucinating. If it never happened to you then it is impossible. Which is OK. For many people, even in the West, and most people in the East this is not strange. They simply smile when you tell them, as if having discovered a secret everyone knows.

I've never been able to explain life through these brain philosophies. Yet, I know what I know very clearly. I see and feel things that brains are not easily the answer to.

Look at the guy who call my views "bullshit". It clearly is so for him. I have however met a decent percentage of people who does recognize what I'm talking about. But most people, again in the West, talk quietly about their experiences.

Since being a child I've always held this simplistic view of life which is that there is more to life than meets the eye, and I don't think it's found in some firing synaptic.

Re:Interesting

[unixfan]

Thank you for proving my point. There are not enough eye "resolution to explain our quality vision. You should befriend and talk to some neurosurgents who actually practice vs someone full of theory, and discover how much they profess not to understand. Some of things that happens they simple can't match.

Yes, I rather rapidly threw together numbers of the net, screwed up my math, and yet there you came and made my point stronger.

The whole subject is best guesswork, which is following the same thread of thinking. The blind leading the blind.

You should however not call me sir, as you would have no idea of what that means.

Re:Interesting

[unixfan]

Haha, this is not science! This is /.

If I wanted a scientific discussion do you really think I would have started it here? Not to say that there could not be people capable of it, but rather the quantity that is not. Never mind those you don't even want to talk to.

Nothing said here could seriously be considered, well... serious. You go here to read interesting links and see some arbitrary views with some occasional value thrown in. I think the best value would be that it sometimes makes people look at and think about uncommon things.

Re:Interesting

[diskis]

Go look at the wikipedia link. There is a little experiment that exactly shows you the hole in your vision.
And about martial arts, humans detect motion over detail. You automatically evade something that comes your way, without actually seeing what it is. You see something is coming and do not assess if it's dangerous or not, you evade/block just in case.

An what you explain of noticing things, that should not be noticed, that is explained by your subconscious. Read up on how people missing most of their brains *can* be as intelligent as when they had a complete brain. So where does that excess capacity go? Your subconscience. If you'd ever tried LSD, you'd know what I'm talking about. On LSD you notice things, differences in human postures, smells, changes in voices. You can read a human way better than normally. You can track your surroundings, you know better than usual what's going on around you. Though, this is extremely tiring to the brain. You can't really think for a few days after a LSD trip. That's why your brain filters inputs. Most go to the subconscience and trashcan, some to your higher conscience. All this information that has been fed downstairs, will manifest in creepy feelings, weird ideas that something is about to happen.

Like with your burglar example. You could sensed a smell of a strange human, change in temperature due to the door being broken. You ignore the signs, but your subconsciense not. It informs you via a creepy feeling. Same with the firing. If a guy wanted you fired, he doesn't like you, and that's visible in his posture/tone of sound/way of looking when talking to you.

Re:Interesting

[Crazyswedishguy]

I see numbers of 9,350 cells per cubed millimeter which is 93,500 per cube centimeter.
Going from millimeter to centimeter in 1d is a factor of 10. Going from cubed millimeter to cubed centimeter (3 dimensions) is a factor of 1,000.

So you'd have 9,350,000 cells per cubed centimeter if your numbers are correct.

Re:Interesting

[unixfan]

I see you don't do martial arts. You have to see what is coming your way or you will buy the faint and be set up for another attack. The normal urge is to avoid what comes your way. But in order to be good, you train to only respond to real threats, not faints.

No I've never done LSD or any drugs, simply did not interest me. I've always been too interested in life and my senses to take chances of limiting them or adding vias.

Though I find it interesting how you want to explain it away with the subconscious. It is a popular why for things that cannot be explained. Yet, if you are not in your body there is no brain to aid you. The guy had not been in the garage, valuables were still there. The door to it was closed, there were no change in temperature as the garage has outdoor temperature and I came in from the outside. (And frankly that door had been open many times when I got there.) To you it's a few odd incidents, to me it's normal life.

The guy who tried to have me fired had not behaved differently as he had not even decided against me. Which is a totally silly explanation. I felt it when he attacked me, not before or after, while. (All three of us became very good friends.) What I pick up on always matches what happened.

My wife can be in a sudden need to talk to me (friends invited us away for the weekend) she must get hold of me and I'll get it and contact her. It goes on and on. After a while you run out of excuses. But you have not lived my life and cannot be expected to know. And I can tell you there's more to my life than being a brain. You can always be different. We don't have to be the same.

I know whatever I feel will mean something different to you. You are totally into the brain and I've noticed it doesn't do much.

Re:Interesting

[Anonymous Coward]

Don't waste your time. He's not into rationality. He senses and feels stuff. If he senses that there are 93500 cells per cubed centimeter, then that is his reality. Agree to disagree. It's probably the best you can do with that kind of a nut case.

Re:Interesting

[smallfries]

We both agree that your numbers weakened your point as they were clearly wrong.

While I don't practice neurobiology myself, my girlfriend's PhD was in psycho-physics and how to exploit the "compression" in the human visual system. Her research was very much at the practical end of the field.

Where we disagree is on whether or not your point stands. You claimed that despite the lack of bandwidth between the eye and the brain, the brain was *not* responsible for synthesizing the majority of what we think that we see. You are wrong on that aspect, and nothing that I have said backs you up on this at all. If you do believe this then feel free to try and argue your (incorrect) point. If you do not believe this then you have cleared messed up in what you have said, as this was your explicit claim that I quoted.

Re:Interesting

[Anonymous Coward]

Your "280 million pieces of data" never reach the brain. Most of it gets compressed and serialized in the neuron layers of the retina, so the actual input via optic nerve is much much less. The compression is way lossy. In the brain, the algorithms used to decide which pieces of the already-compressed data can be safely discarded and to do edge/motion detection and stuff like that are... surprisingly primitive, to say the least (google "optical illusion" for neat examples of implementation flaws).

Re:Interesting

[yet another coward]

RTFA, rm999 and moderators. This very point comes up at the end.

Re:Interesting

[Crazyswedishguy]

Wow, crazy that I'm going back to this discussion, but I've been laughing so hard at everything you've said, I love it.

I have a number of comments:

I. I'm not sure how good you are at martial arts, but I've practiced different martial arts for a number of years (around 7 or 8). In every one of them, I've been taught to look at a person's torso, in order to see two things:
1. motion at the periphery, both legs and arms. If you look at his face or into his eyes, you will likely not see his legs. If you look at his legs you won't see his arms. When he does strike, you see the motion, which leads your brain to interpret the actual body part. You don't have any close detail of that arm, the way you would when looking at it, but you've probably seen it before so you can guess what it looks like. I'm not saying your parry is guesswork, I'm just saying that you don't need to see a scar on his arm to know what his arm looks like and where it is. Your brain actually does extrapolate to make it look like the arm.
2. feints: yeah, since you're so good at martial arts, you will know that you can usually tell a feint by a guy's posture, just looking at his torso. That's because a hit won't have any power unless the guy shifts his weight and uses his whole body.

II. Did you try that link that was posted in a previous comment? Find your blind spot flash game
You'll be surprised that were there was a black spot, it is now "filled in" by the red background... Still don't believe it?

III. I'm sure you've counted all those times that guy turned around to look at you because you were staring really hard at him. Have you counted all the times when the guy didn't turn around to look at you? The idea that our selective memory tends to believe in supernatural things like that is quite documented. For instance, I often think "this is going to happen in about 2 minutes". It happens about 1 time out 10, but when it does, my reaction is "wow, I predicted it once again! Just like last time!" And I always remember the times it works, not the times it doesn't work. Same way, sometimes I can "feel" there's something wrong. Sometimes I'm right, and I could explain that I picked up on subconscious clues. But then again, sometimes I'm just wrong.

I'm sure it makes you feel special to believe these things. And yeah, we all are special in our own ways. But you probably haven't understood how complex the brain actually is, and how much work it's doing that you're not seeing (or in this case, that you think you're really seeing).

Re:Interesting

[unixfan]

Then you could have had the opportunity to discover that it is not needed with weapons, provided they are not heavy. As I'm sure you know it does not take much body motion to use many light weapons.

Don't know what's so new about the blind spot. No, I did not bother.

No, I did not refer to those times when someone is constantly looking around. Rather those where it's really unique to see them turn around at all. Those are the times that stand out and make it really odd.

Good that it got you laughing, I quite enjoyed your replies. And like you I think yours are just as crazy.

You mean ANNs ...

[Pinky's Brain]

Dunno about your NN, but mine is pretty effective ... come to think of it, way more effective than it should be on a Saturday night :(

Helpful for computer technology

[Heliogabalus]

Our computer technologies have yet to achieve the complexity of most biological brains. I'd love to see these new informations derive a new form of super-computer. Of course...We have to watch out for iRobot scenarios...

Re:Helpful for computer technology

[Fourier404]

How the heck do you program something that complex though?

Re:Helpful for computer technology

[Heliogabalus]

I'm not saying that we should program a brain, but possibly take some of the ideas of how brains work and implement them. There's been an estimate that by 2013 we will see a super computer that will exceed the power of the human mind.

Re:Helpful for computer technology

[canuck57]

Our computer technologies have yet to achieve the complexity of most biological brains. I'd love to see these new informations derive a new form of super-computer. Of course...We have to watch out for iRobot scenarios...

Don't hold your breath for an iRobot.

If each of the 100 billion neurons managed the 1000 or so synapses, and say a modern day PC with a quad processor could computationally handle say 100 neurons, you would need 1 billion PCs. Since 1 billion PCs would find it difficult to walk, the old adage of: "computers are as dumb as nails" still applies.

Now lets say usable processing power doubles every 5 years, and it shrinks to something small enough it can walk into our living rooms. That would be at least 150 years from now.

You could argue Moore's law is faster, but 2 issues. First is can it continue, not likely. Second is can we has humans program something this complex, not likely. Each generation of computers we get less and less value out of for the increases in CPU. You could say hello in 10-12 bytes in the '70's and early 80's, now it takes well over 1,000,000 bytes and 20 times longer to do the same thing on today's computers. Go figure.

So iRobot, Cherry 2000 and Commander Data are going to have to wait quite some time long after our great-great-great-great-great-great-great-great-great-great-great-great grandchildren have passed on. Say half the time to get a muppet.

Re:Helpful for computer technology

[multi io]

Now lets say usable processing power doubles every 5 years, and it shrinks to something small enough it can walk into our living rooms. That would be at least 150 years from now.

That's only as long as you simulate neurons in software, which is probably as inefficient as it gets, as opposed to building and connecting artificial neurons in hardware directly. The fact that the brain manages to cramp the intellectual differences between reptiles and humans into a few cubic centimeters should tell you something.

Re:Helpful for computer technology

[E 99]

If each of the 100 billion neurons managed the 1000 or so synapses, and say a modern day PC with a quad processor could computationally handle say 100 neurons, you would need 1 billion PCs.

You're assuming that a neuron is simple, which it is not. To simulate just one neuron, or any other kind of cell, you may need a billion PCs.

Re:Helpful for computer technology

[bodan]

You're assuming that the neuron's complexity must necessarily be simulated exactly. Artificial heart valves, hip replacements and even blood (this one is still in prototype phase) are much less complex then the organs they replace (they don't even have cell-scale features, let alone imitate those of the replaced organ) but can simulate their function very well in many respects.

So it's reasonably probable that a neuron's functions can be simulated with useful accuracy without going into nasty details like protein synthesis, electron transport, the quantum aspects of synapse chemistry and whatever else you were thinking of.

Re:Helpful for computer technology

[Metasquares]

Ah, but can such an approximation truly result in intelligence or would it just be a very slow and complex classifier? Or both? :)

Are all of the functions of a neuron required to produce intelligent behavior? If not, which can we omit? How will we even know when a system is behaving intelligently? Even humans take years to learn how to communicate and rationalize. Could we provide even a perfect simulation of the human brain the proper environment to train in to ensure these results? Once we have such a model, can we extend it or are we limited to the level of artificial people? Is it ethical to perform questionable experiments on a model with an essentially human brain in attempts to extend it?

These are some of the questions we need to confront before taking this route (another should always be "do we really want to do this?"). It's not enough to just say "connect lots of wires and we have intelligence". We have to understand at least some of the theory linking the low and high level functionality before we can hope to construct an intelligent agent based on this model, and I suspect we'll need a much more complete understanding to (a) make it computationally tractable and (b) generalize the system to above human-level intelligence.

Re:Helpful for computer technology

[Anonymous Coward]

Soooo...you want to study human brains? Then go for the simplest cases with the fewest neurons, etc.

Say, politicians, and/or lawyers...should be a cinch! :-)

Re:Helpful for computer technology

[bodan]

The "do we really want to do this?" question is for another discussion, as are the other ethical ones.

As for the rest: As I see it, there are two possibilities: Either (1) every little detail, down to the quantum behavior of atoms inside the neuron, are _necessary_ for intelligent behavior, or (2) there is one higher-level description of a neuron that is sufficient, and the lower-level (quantum, chemical, proteic, etc) behavior is just implementation detail. I'm rather sure (2) is true, mostly because of the huge variability of neurons, which implies that many variations can in fact be abstracted away.

Note that this doesn't mean that a "sufficient" approximation of a neuron is simple (e.g., a biased, weighted dot-product of the inputs, like a perceptron). It can be a very complex function, I'm just sure that it's much simpler than a physical neuron. Which supports the assumption at the thread's beginning, i.e. that we can probably implement it in hardware that is complex, but still much simpler than a simulation of the entire biological neuron, what with the life-support details and all. In short, it's probably hard, but many orders of magnitude easier than you'd think by looking at the physical complexity of a neuron.

On the other hand, I don't agree with the "before taking this route" part: in fact, I think we're taking this route precisely to help answer these questions. We know that the brain is, at a certain level of approximation, a very complex network of neurons. We have never built anything like that, and it's not currently feasible to study it dynamically (ie, while a brain is working, at synapse-level of detail). The obvious path is to study it statically (eg, what the article describes) and build a network that simulates the _structure_. Once we do that (either with a 300-neuron worm nervous system or with something bigger), the differences of behavior are caused by the _nodes_, i.e. neurons. So then we can focus on them.

(Of course, this doesn't mean that we can't study it in the other direction, e.g. by studying individual neurons, which is also done in other research.)

The summary of this part is that we don't need to know the _necessary_ complexity of the neuron simulation (as per my third paragraph) _before_ this research, but rather it will be the _result_ of the research.

The hope is that these two approaches (and others) will at one point converge and give us something that is reasonably intelligent. Which brings us to your other question, "how do we know when a system is behaving intelligently?" Well, we don't know that yet for every system (there's no good definition) but there are cases where we can tell. For instance, as humans we generally agree that most other humans act intelligently (not smart, just intelligently). And we can do that, despite the fact that we don't know how we work. Which, of course, leaves us with some variant of the Turing test.

(Note that the Turing test was designed for human-level intelligence. But it can be extended for some other cases. For instance, if we replace the brain of a worm/fly with an artificial one, and then we can't tell that worm/fly from a natural one just from their behaviors, we can reasonably argue that the artificial brain simulates the organic one well enough. This doesn't necessarily mean that we can easily extend that to a human brain, but it's already a great advancement.)

Re:Helpful for computer technology

[walshy007]

I understand your point that things are becoming bloated etc etc, but I'm going to go off topic a bit here

the overwhelming majority of that size is due to convenience and lazyness, your example, hello world, assuming just a c single line printf with the stdio include, you get an executable of 4.7kb,

now, I just went and re-wrote that in assembly, using system function calls instead of the c library, got a 725 byte binary with symbol info etc, 320byte when stripped, that still isn't terribly efficient considering it just says hello, however most of that size is due to the elf header which is on the order of a couple hundred bytes anyway.

now c is still pretty lean compared to some other languages, for instance, c# and basic, once you start getting into those realms you even have more waste unless you try to be very efficient with your resources, or need that functionality.

oh and here's the assembly program if any person is interested in seeing what it looks like for some strange reason. .equ STDOUT, 1 .equ SYS_EXIT, 1 .equ SYS_WRITE, 4 .equ SYSCALL, 0x80 .section .data
string: .ascii "hello" .section .text .globl _start

_start:
        movl $SYS_WRITE, %eax
        movl $STDOUT, %ebx
        movl $string, %ecx
        movl $5, %edx
        int $SYSCALL
        movl $SYS_EXIT, %eax
        movl $0, %ebx
        int $SYSCALL

when compiling use -nostdlib or it will die.

Need More Input

[Plasmagrid]

So if we can map a small portion will that mean we will be able to upload as well ??

Horse, push cart.

[zippthorne]

Science is all about the baby steps. You can't talk about determining the weights before you know what the connections are.

Computational Complexity

[Yahma]

It's quite interesting that these German researchers are mapping pieces of the brain; however, even if they were to map the entire human brain, first, we still do not know how to perfectly simulation the biological processes occuring in the brain. Yes, we are able to simulate a single neuron, or small clumps of neurons; however, the dynamics of simulating billions of interconnected neurons is not fully understood.

Second, even if we were able to map the entire human brain and run a perfect simulation, the computing power of today cannot handle this complex task in real time.

Re:Computational Complexity

[Renraku]

Lets say that we could, in fact, emulate an entire human brain in real time.

It would be very difficult to get something useful out of it. Answers wouldn't always be the same due to the semi-random effect brains have a tendency to produce. You would spend a lifetime putting something in and watching it come back differently than it did a few minutes ago. It wouldn't be too unlike a network connection, if packets were voltage gradients and various neurotransmitters. Since there are only a few ways each can function...good luck.

Re:Computational Complexity

[Fourier404]

If you emulate a human brain in real time, as well as connections to emulated eyes, ears, and mouths, you just have to talk to it for a couple years and it'll learn English. =D

Re:Computational Complexity

[bmo]

"It would be very difficult to get something useful out of it. Answers wouldn't always be the same due to the semi-random effect brains have a tendency to produce. You would spend a lifetime putting something in and watching it come back differently than it did a few minutes ago."

We call this "raising children"

--
BMO

Re:Computational Complexity

[vertinox]

Second, even if we were able to map the entire human brain and run a perfect simulation, the computing power of today cannot handle this complex task in real time.

To be fair, how much of the brain's processing power is devoted to involuntary issues (heart, lung, pain receptors), voluntary motion systems (muscles), and then how much left over for intelligence?

If you stripped away the brain needed for everything except cognitive thought, how much processing would you need? I suppose that is why its best to start out small.

Re:Computational Complexity

[Maian]

Not only should this be modded funny but insightful as well. Because what he's saying is very true.

People tend to think of computers as something that should just work out of the box, but really, not all computers (and software) can be like that. A human child can be considered as a computer that needs years to train. If robots ever become popular, I expect that "pet" robots would have to mimic these learning capabilities and "grow" with its master over the years.

Re:Computational Complexity

[rbarreira]

It may not be possible to do so without changing the behavior of the cognitive part. Investigating if that was the case or not would already be an extremely important milestone!

Re:Computational Complexity

[bmo]

Yeah, unfortunately I got modded "redundant" and "funny." I also got modded "redundant" for my Beck joke. How is it redundant when I was the only one to come up with that? What...ever. I got more karma than I know what to do with. Go ahead, waste your mod points on me.

If we truely come up with computers that mimic the human brain, we're going to see the same problems that we have "programming" children, maybe even worse, because human children have inherited behaviors that make teaching easier, and electronic neural networks won't unless we build them that way.

It's one thing to build a neural net. It's another to make it "want" to learn.

--
BMO

Re:Computational Complexity

[Renraku]

First you'd have to have the IO systems connected properly. If you just grow a brain, this doesn't happen. For all practical purposes, the spine is part of the brain. The optic nerves are part of the brain. The cranial nerves are brain too. They all grow out of the brain when the brain is given chemical signals at the right time. Or more specifically, the stem cells are. If its not given, they don't grow. The brain will be a closed system.

The best idea for that would be to get the brain, cranial nerves, spinal nerves, and optic/auditory nerves. Without all of those, there's not going to be a whole lot of IO going on.

Re:Computational Complexity

[Anonymous Coward]

Anyone ever play NERO? The teaching aspect applied to a simulated neural AI in order to get it to do something is rather true. Likewise it's not always predictable. It's neat teaching game bots to do something, even if it takes time. To think that instead of some developer coming up with some carefully balanced alogrithm to induce a behavior, it's your "reward" and "punishment" of the simulated robots during training mode that makes their eventual behavior in the game.

I just hope the developer of this project can keep the ball rolling. Definitely cool stuff, but it still crashes on my machine too much to really get into it just yet.

Re:Computational Complexity

[Anonymous Coward]

"Yes, we are able to simulate a single neuron" ...

Are you kidding me? I am very deep in the field of cell biology, and let me assure you: we still know very very little about the cell. If anyone thinks they can "simulate" a neuron, they are only fooling themselves.

CCortex

[Colin Smith]

http://www.ad.com/

An attempt to emulate a brain on a network of computers.

Re:CCortex

[Anonymous Coward]

LTSpice doesn't have a acetylcholine neuron model, so I modeled the brain using a series of RC networks. Just by adding a few vacuum triodes to the frontal cortex will turn the brain into a super 50W push-pull amp!

Plasticity

[DynaSoar]

That's what they call the brain's ability to change. By the time they complete a wiring diagram, it'll have changed. Also, knowing the wiring and connections is not enough. Knowing which connections are excitatory and which are inhibitory is necessary, and then tracking down loops of excitatory against excitatory resulting in inhibitory, etc. It's all fine and well to have a map, but that doesn't tell you squat about what anything does. A useful map would have to be dynamic, and the complexity of that is far more than just what they're considering for a wiring diagram.

Re:Plasticity

[Anonymous Coward]

step 1 : map the brain wiring and figure what theories are incorrect.
step 2 : find a way to weed out more theories.
step 3 : DESTROY ALL HUMANS ! (except that one : he's funny)

Re:Plasticity

[Morkano]

Yes, this won't solve all of our problems, but it's definitely a good thing to have. It makes it a lot easier to talk about what is excitatory and what is inhibitory if you have an base for how they're all connected.

Re:Plasticity

[earlymon]

Spot on, but there's more if the ideas in Neural Darwinism are true - and I think that they are.

Edelman showed that 1) individual circuits are not restricted to a single function, 2) that the operation of any brain circuit has a propagation rate dependent upon electrolytic characteristics at the time the circuit is activated, 3) a signal through the circuit, possible by the electrolytes, changes those electrolytes, 4) the next loop through that circuit will have a different propagation rate than the previous (or next) loop.

Here's the obligatory Wiki teaser: http://en.wikipedia.org/wiki/Neural_Darwinism

Edelman was the guy that first explained the brain as a complex wiring network. He realized his mistake and published to correct it. His frustration was endless, as he was constantly quoted with his own earlier - and wrong - work in ignorance that it was HIS earlier work.

He was the original proponent for the idea that thoughts are not infinite because brain wiring is finite. He later learned that even though the wiring is finite at any one instant, it's not just like the wiring like an electrical circuit - once made the propagation through a copper circuit is predictable and constant (I generalize - it's late, ok?) - but rather brain "wiring" propagation flexible.

It's not hardware, it's not a software neural net equivalent - it's wetware. The closest analogy most /.ers are too young to know about - it's closer to self-modifying code than anything else.

I always love to be proven wrong, but I think the whole exercise of trying to map brain "wiring" is an absolute, total and complete waste of time and is being performed without respect to prior art, and is instead stumblefucking around with the mantra of "Gee Whiz, we know so little about the brain....! We can try anything!"

Reminds me of the clown bitch trying to figure out Stradivarius magic by doing static resonance testing by placing a violin on top of a vibrating cone driver, and seeing what bounced around depending on what frequency she dialed on a function generator. Sadly hilarious. Saw it on PBS or something years ago. She was part of cutting edge research because no had thought of doing that to a violin before - and her excuses were - just like these brain wiring guys and other posters here - we know so little, we have to start somewhere..... In fact, the reason "no one had thought to do that" was because it was totally idiotic.

I wish researchers like this would just start in the unemployment line and free the money for people who stand a chance of accomplishing something.

It's completely true that research HAS to try whacky and novel things. It's also true that research is to search and search again - and these bozos just ain't doing that, as evidenced by a body of work pointing to what's wrong with the whole approach that didn't turn up in their searches - if indeed there were any.

OK, I'm getting worked up and I need to stop. Sloppy work passing as research gets to me that way.

Re:Plasticity

[Anonymous Coward]

They mentioned doing the mapping steps between 30nm slices. I am pretty confident the dead brain they're slicing would not be changing much.

Feedforward NNs versus biological NNs

[benhocking]

Many engineers have studies feed-forward neural networks and found them to be far inferior to other solutions. Of course, our brains use recurrent neural networks, which, unfortunately for engineers, are very difficult to analyze. There are many secrets yet to be teased out of these neural networks, but much progress has already been made. Researchers in our lab, for example, have demonstrated how introducing random synaptic failures improves not only energy efficiencies, but also the cognitive abilities of simulated neural networks. I'm currently researching the effects of variable activity (as measured in a biological neural network by an EEG), and I dare say there's a lot more that we don't know about these networks than we do.

Re:Feedforward NNs versus biological NNs

[Feminist-Mom]

our brains use recurrent neural networks

How do you know this ?

Re:Feedforward NNs versus biological NNs

[Anonymous Coward]

" our brains use recurrent neural networks

How do you know this ?"

Recurrently

computer wiring

[Anonymous Coward]

There is a well known analogue: Imagine you have a complete wiring diagram of the computer you now sit in front of. This will actually give you very little information about how it shows you the shlashdot.com, which you are reading now. Besides wiring you need information about software and current memory state. In addition to that you can take a different computer. Despite being very different in wiring and software they it show you the slashdot.com almost the same way.

Same thing with the brain. Just wiring information, even if you can actually get it, has very little value to understand how the brain works.

Google

[Anonymous Coward]

Can't they just google it?

they got the complete neural map of C.Elegans

[Z80a]

and it's only like 302 neurons,so,it's possible to write a simulator of it?

Re:they got the complete neural map of C.Elegans

[E 99]

and it's only like 302 neurons,so,it's possible to write a simulator of it?

You could write some kind of rough analog of the neural network. However, an actual simulator of a C. elgans would be impossibly complex, as a simulator of an actual cell would be impossibly complex.

Re:they got the complete neural map of C.Elegans

[lazybratsche]

Well, that's the hope, and a major source of appeal for the humble nematode. Unfortunately, that's still far beyond what we know right now. The physical map of every neuron and their connections has been complete for decades. Still, despite a whole lot of effort, researchers are still working to piece together small functional circuits for the simplest of behaviors. A lot of complexity arises in neural circuits -- one physical circuit can contain several independent functional circuits, depending on the types of inputs.

The best current knowledge of C. elegans neurophysiology involves qualitative descriptions of small circuits, involving a few dozen neurons. Unfortunately, while you can do a lot of good behavioral studies and other experiments, it's impossible to directly record the activity of specific neurons. Also, it turns out that some "neural" functions are actually performed by other cells. For example, one pattern generator in the digestive tract actually resides in intestinal cells instead of neurons -- my lab is working on the genetics involved.

This shit gets complicated, fast.

IAAUCER
I am an (undergrad) C. Elegans researcher

Re:they got the complete neural map of C.Elegans

[Tablizer]

[worm has] only like 302 neurons,so,it's possible to write a simulator of it?

while (! dead) {
    if (leftTenticalSensesFood()){wiggle(left);}
    if (rightTenticalSensesFood()){wiggle(right);}
    if (frontTenticalSensesFood()){munch();}
    if (femaleWormEncountered()){fuckTheWigglyMamma();}
}
end();

Re:they got the complete neural map of C.Elegans

[Anonymous Coward]

while (! dead) {if (leftTenticalSensesFood()){wiggle(left);} if (rightTenticalSensesFood()){wiggle(right);} if (frontTenticalSensesFood()){munch();} if (femaleWormEncountered()){fuckTheWigglyMamma();}
}

Hey, that's *my* life! I call prior art.

Re:they got the complete neural map of C.Elegans

[Anonymous Coward]

Yes. And no (sorry!).

The reason is that we still have a lot to learn about the internal workings of the neurons and the gila. So those are important, too. Seeing it from a reverse engineering perspective, though, it may or may not be that important to know what the implementation is. We shall find out soon

Re:they got the complete neural map of C.Elegans

[somasynth]

it's impossible to directly record the activity of specific neurons. It appears there are definitions of impossible I am unfamiliar with, I've read a number of papers that reference to not only directly measuring impulses of individual neurons, but also sending impulses to one or even multiple neurons with nanowire arrays. A qualitative description of functional circuits is unnecessary when it comes to testing it's 'transfer function' with any arbitrary input, or even giving it a full simulated environment. That is, an understanding of the network is unnecessary for simulating it's function. Understanding the weighting and wiring tendencies is of course required, but the real limitation for a full simulation appears to by physiological.

Re:they got the complete neural map of C.Elegans

[lazybratsche]

I was talking about C. elegans specifically. Certainly, there have been plenty of direct recordings of neurons for decades. These have tended to be of very specific model systems however. With the giant squid axon or other preparations, you can stick tiny electrodes in different parts of the neuron to record exactly what's going on. And there are the nanowire arrays you mention -- though these are a more indirect technique.

These techniques tend to involve either some in vitro preparation, where you have a collection of isolated neurons in a dish (which is occasionally done with C. elegans), or minimally intrusive but limited recordings of a rat or larger animal. At that size, it's possible to stick some recording device in the head of the test subject and still have it behave mostly normally.

The nematodes are far to small for the second approach. You might be able to get some recordings with inserted electrodes, but you'll have to immobilize and nearly eviscerate the poor worm to do so, so you won't be dealing with normal behaviors.

I'm sure you can also teach some other arbitrary neural network to simulate the nematode's behavior, but that's not very biologically relevant.

Factor of a million

[MarsDefenseMinister]

That's roughly 20 to 30 years of general purpose computer dev time. They could definitely get their factor of a million much faster with more cores, or specialized hardware.

In other words, it won't take long before the actual hardware is in place for this analysis to happen. Shortly after that will be the robot insurrection. I've seen movies about this. It doesn't end well.

Re:Factor of a million

[Dark_MadMax666]

Shortly after that will be the robot insurrection. I've seen movies about this. It doesn't end well. Depends on your point of view :) I for one welcome our robotic overlords :)

Lets not "say" (NS)

[Pinky's Brain]

Nuff Said.

what does a neuron map look like?

[wikinerd]

The neuron map of the average human brain looks like this. More evidence for the existence of him!

100 trillion synapses?

[SolusSD]

That's simply not true. High estimates for the total number of cells in the human body are 100 trillion, a quick google search yields this: "Science NetLinks, a resource for science teachers, stated that there are approximately "ten to the 14th power" (that's 100 trillion) cells in the human body." maybe they meant 100 billion? I remember from one of my AI classes that there are up to 10,000 synapses per neuron, while other have fewer than 100.

Re:100 trillion synapses?

[Layth]

A synapse isn't a cell.

Re:100 trillion synapses?

[Anonymous Coward]

Each neuron has many synapses. Neurons are cells. Synapses are not. So there's about 100 billion neurons and about 100 trillion synapses.

Re:100 trillion synapses?

[Tablizer]

Each neuron has many synapses. Neurons are cells. Synapses are not. So there's about 100 billion neurons and about 100 trillion synapses.

And I still cannot remember where I put the f&&&ing car keys!

Oblig. Futurama

[Dachannien]

Farnsworth: Lie down here and we'll do some tests. If Fry is out there, then Leela's brain could be acting as a five-pound Ouija board.
Leela: Is this some sort of brain scanner?
Farnsworth: Some sort, yes. In France, it's called a guillotine.
Leela: Professor! Can't you examine my brain without removing it?
Farnsworth: Yes, easily!

Neural Net as LAN for nanocomputing;Re:Missing tag

[Anonymous Coward]

Yours is the best comment yet on this thread. I did NN stuff at U.Massachusetts/Amherst 1973-1974 when NN was temporarily un-cool, due to the Minsky-Papert Perceptrons book. I shifted to genetic algorithms 1974-1977 while beta-testing John Holland's book. One conclusion I reached was that memory and cognition are nanomolecular processes, and the neural network is merely the Local Area Net that connects distant regions of nancomputers.

-- Prof. Jonathan Vos Post

Me!

[SEWilco]

Map my brain!
...reads article...
When I'm done with it!

Your post are so full of bullshit..

[aliquis]

.. I don't know where to start.

Repeat after me:
YOUR EYES AND BRAIN AREN'T DIGITAL FINALSPEC1.0 DESIGN.

MOVIES use 24 fps with motion blur because it gets an acceptable quality, your eyes can still notice things happing during 1/200 of a second.
In our digital graphics class the teacher mentioned something like 20 million cells in the eyes which registred data but only one million nervs to the brain.
YOU HAVE SHIT POOR QUALITY OUTSIDE ONLY A FEW PERCENT OF THE CENTRUM OF YOUR VISION.
I guess your dpi esimate may be somewhat close, 1" of what distance thought? As close as possible but still sharp? Not close to that on the non-central parts thought.

"I see numbers of 9,350 cells per cubed millimeter which is 93,500 per cube centimeter." Holy shit! You must be american! Try 9,350,000, times 1400 is 13 billion.

I don't see what memory have to do with vision thought, it's not like you can recall the perfect quality of any vision you have seen anytime.
Not that anyone would expect that you or someone else could estimate these things correcly, but anyway.

Re:Missing tag - The pig go. Go is to the fountain

[Anonymous Coward]

Not to mention - and I a sure I'll get modded down here - is that neural networks aren't very effective.

In the worst case they can always teach it to do poetry :p - http://worsethanfailure.com/Articles/No,_We_Need_a_Neural_Network.aspx

Perhaps

[reboot246]

they should start with an easier project. What's CowboyNeal doing for the next few days?

Re:Perhaps

[Tablizer]

they should start with an easier project. What's CowboyNeal doing for the next few days?

Modding you waaaay down

Re:Perhaps

[reboot246]

Who cares?

Weights are encoded in the structure

[chkn0]

'Weight' is an abstraction used in NN simulations.

In biology, the strength of interaction between two neurons is determined primarily by the number of synapses between them -- how many of Neuron B's dendrites lie on Neuron A's axon. This is exactly the information contained in the wiring diagram.

Of course, half the fun of any advancement is learning all the new things that are *not* explained or turned out to be more complicated than previously thought. With good wiring diagrams, we can make predictions about what kind of firing patterns we expect to see based on various assumptions about synapse behavior, etc., and in this way learn about other brain mechanisms.

Re:Neural Net as LAN for nanocomputing;Re:Missing

[Anonymous Coward]

"merely" =)

Maybe it's not all that bad ...

[ScrewMaster]

keep in mind that huge swaths of the human brain are for data storage, motor control, sensory processing, autonomic functions and other elements not directly related to sentience. If your intent is to use this information to produce a viable synthetic human intelligence, you wouldn't necessarily need to model the who shebang.

Re:Maybe it's not all that bad ...

[Anonymous Coward]

Who says motor control, sensory processing, and autonomic functions aren't directly related to sentience? In my experience at least, feeling and perceiving are pretty damn connected to my body.

But I suppose thats why I'm in the embodied cognition camp.

Feels good

[Tablizer]

I hooked one of these up to my brain and so far its working just fine. The only side effect is a slight tingly fe

Yes, but ...

[PPH]

... when the map is done, will anyone ever be able to re-fold it?

Re:What constitutes a "map" here differs from eleg

[Anonymous Coward]

Does this mean that we can program a virtual C. elegans that behaves in a manner similar to naturally occuring ones for a similar history of stimuli?

Glial cells are key; Re:100 trillion synapses?

[Anonymous Coward]

When I started graduate school in 1973, they said that there were 10^10 brain cells in a human being. By the time I left graduate school in 1977, they joked that there were 10^10 brain cells in a human being, of which 10^11 were glial cells.

The glial cells do more RNA to protein information processing than the other neurons. That's part of why I hypothesized that most of the information content in neurons was in proteins, not electrochemical patterns or nucleic acids. If you quantify how much information is needed to transcribe RNA to polypeptides, then orient the tertiary-structured protein and stick it on a membrane, you get about a billion bits per second per cell.

If the memory is mostly in glial cells (consistent with RNA transcription inhibitors causing amnesia) then why not the thinking, also? If so, the neural network is not where the action is -- and becomes merely the Local Area Network for a distributed molecular nanocomputer.

-- Jonathan Vos Post

Brain mapping in berkeley, 1961

[Anonymous Coward]

Brain mapping experiments were ongoing in Berkeley in 1961... So anything up to fifteen+ years later should be public info now. Nice photo link here:

http://www.4freeimagehost.com/uploads/6f01bf4e37a1.jpg

(used this for a local band's DVD cover)

You are a simmulant

[goombah99]

Bad news. You are a simulation on a playstation 9, not a real human.

No really. it's overwhelmingly probable you are a simulation.

According to the article above that are 100 trillion neurons to simulate. Even if they were multi-state that's approaching trivial by computational standards. And if you are willing to run the simulation at sub real time you could do it now.

So according to the anthropic principle, either 1) the human race goes extinct in the near future before this level of simmulation is possible at the level of a wrist watch computer 2) or there eventually must be more simulated humans than real ones. And of course once you start simulating humans the future time to have simulations grossly exceeds all human history.

So since we can now forsee this is just down to the details, it seems unlikely you exist.

Also when you consider that the biggest entertainment industry, bigger than hollywood by some measures, is the MPORG human role playing simmulations there ample reason that once it becomes possible to do so, people will be simmulating humans.

You are most likely a nintendo pet human.

As is often remarked after religiously incongruous tragedy happing to good people ,"god works in mysterious ways". Apparently this is actually true..since god is a 7 year old boy pushing the "smite" button.

Therefore your best stategy is to look around for the obvious "real" humans playing the avatars in the game and get to know them. That is, be a groupie to the stars or post on slashdot. Everything else is waste of life apparently.

That explains George Bush

[Anonymous Coward]

Could there be any other explanation of why congressmen and presidents are out of control sex maniacs or start wars?

Beat this - "Brainbow"

[cytoP7]

http://www.wired.com/science/discoveries/multimedia/2007/10/gallery_fluorescentneurons

Save Game

[Anonymous Coward]

It'd be good if it was possible to scan and store a snapshot the current state of your brain (without destroying it) so that when you die, it could be uploaded into some device (e.g a robot) so that your consciousness could "live" on from where you saved it.

Bruce McCormick

[eric76]

A professor at Texas A&M University, Bruce McCormick, was pushing for this for years.

Check out Welcome to the Brain Networks Laboratory at Texas A&M University!.

The idea is to use a knife-edge scanning microscope to make images of very thing slices from brains.

I'm curious if Dr. McCormick has retired. His web page last list courses he taught in 2002.

Re:What constitutes a "map" here differs from eleg

[Anonymous Coward]

Ahhhh, thank you for saying that. It eases my pain to hear someone state something obvious that lots of other people don't get.

Now if only somebody pointed out obvious reasons against brain region locations and right/left-brained thinking.

Saying, "If I was born with my eyes on the back of my head, the occipital lobe and other regions would be arranged differently", doesn't make sense to anyone else.

If the regions were fixed in place permanently, then losing a sense wouldn't cause it's corresponding region to be absorbed by the others.
Since normal people have the same physical traits, in relatively the same locations, the flexible nature of a brain is observed to be more constant than it is. My hypothesis: "The ideal arrangement and proportion of brain regions reflects the presence, arrangement and usage of their corresponding physical traits". (If I ever heard a science teacher say something like this, I'd probably die) As opposed to ideals, reality probably can't handle every physical variation (maybe even slight ones) and would result in regions being fit together badly. Such a thing might even be the cause of disorders like autism.

Right/left-brained thinking probably has more to do with which side of the brain has the best supply of oxygen from vein/artery distribution. Hence being the dominant side as a consequence of physiology and not caused by or causing personality traits. Compare the veins between your two arms to see that they don't match each other perfectly. (or somebody elses arms, if you can't see your veins)

Hope this wasn't too offtopic.

Global optimum for more than 2 classes?

[Pinky's Brain]

Pretty sure that is NP.

Re:What constitutes a "map" here differs from eleg

[Crazyswedishguy]

They could start with a football player to make the task easier!

Recurrent neural networks

[benhocking]

I'm writing my doctoral dissertation on the CA3 region of the hippocampus, although this is also true for the entire brain. In short, I've done a lot of reading on how the brain is wired in an attempt to reproduce it in silico. There's still a lot for us to learn about how the brain has accomplished certain feats.

like Lie groups!

[onion_joe]

Now I am starting to realizr the importance of bioinformatics in neuroscience. This sort of multiple inputs produce differing behavior is reminiscent of Lie groups in mathematics. Its basically the same thing: a large (huge?) system of inputs, operators, and outputs, but a limited set at that (though limited is used with a grain of salt here ;-)

Map out the structure, determine the operators, and combine the inputs with the outputs! Bang, easy as pie!*

*joke :-)

One has to start somewhere...

[LifesABeach]

Taking off one layer of tissue at a time to expose the neural net is a bit caustic. Maybe a method should be considered where the host is left functionally intact? Also, in order to "map" a human brain, one is going to need something on the order of about 2 to 4 peta bytes of storage, minimum. It is wonderful that we are heading in the direction of "downloading" our brains. I can see a time when forgetting the wife's birthday will become a footnote in the history books.

Schizophrenia

[neuro512]

Another example of crappy reporting: (from TFA) ...they could shed light on disorders thought to originate from faulty wiring, such as autism and schizophrenia.

Schizophrenia is NOT a result of faulty wiring! It is result of an unnatural levels of dopamine in the brain. Wiring may change after becoming mentally ill, but schizophrenia does not originate from abnormal brain wiring. More half-assed scientific reporting. :(

In Soviet Russia...

[madbawa]

...the brain maps you!

Re:What constitutes a "map" here differs from eleg

[MeditationSensation]

Yes.

Re:What constitutes a "map" here differs from eleg

[threelegduck]

Although we have the wiring diagram for a single c. elegans, we do not have a realistic model for how it crawls (or associatively learns to avoid pain, or seek food for that matter). Right now, one might come up with models and potential functions based on the efficiency of certain wirings (w/ exceptions being the more interesting as usual) - but really, we should be optimizing for the ability to crawl. Once we physically understand how the worm crawls (which we do not understand), we still have to optimize the nanoscopic spatial distribution of voltage sensitive and insensitive ion channels, plasticity machinery (with whatever learning rules come along with them), membrane capacitance, resistivity, the distribution/dynamics of these parameters through time, etc, etc etc. And to top it all off, the ionic conductance levels are all noisy, and are not (even in the simple worm) consistent from class of neuron to class of neuron or from animal to animal. When we study such a "simple"system", what is the question we that we are trying to answer anyway? Are we asking what the actual solution is, or are we asking what is the set of possible solutions?