Slashdot Mirror
A Skeptical Reaction To IBM's Cat Brain Simulation Claims
kreyszig writes "The recent story of a cat brain simulation from IBM had me wondering if this was really possible as described. Now a senior researcher in the same field has publicly denounced IBM's claims."
More optimisticaly, dontmakemethink points out an "astounding article about new 'Neurogrid' computer chips which offer brain-like computing with extremely low power consumption. In a simulation of 55 million neurons on a traditional supercomputer, 320,000 watts of power was required, while a 1-million neuron Neurogrid chip array is expected to consume less than one watt."
Think about it. Think about it like a cat.
Somehow my pet parrot now seems oddly... delicious. :O
Wouldn't power consumption grow more than linearly with neuron count? I would think the number of connections is the dominant factor - so the comparison of two data points of power consumption vs neuron count is meaningless.
All those neurons using less than 1 watt?
I know some people like that.
No brain, no pain.
The cat's brain is made up of 1 BILLION neurons and 10 trillion synapses. So with the nuerogrid chips, it will require at least a kilowatt to simulate.
so tuck your head between your legs and wait for the Universe to explode! ~:-)
If you have custom silicon to do each neuron then you are going to be hugely more power efficient that a general purpose processor simulating a neuron in software. There is nothing new there and anyone who thinks otherwise is just clueless. Given IBM have the facilities and resources to fabricate some custom silicon I fail to see the issue.
From the original FA: "The simulation, which runs 100 times slower than an actual cat's brain, is more about watching how thoughts are formed in the brain and how the roughly 1 billion neurons and 10 trillion synapses in a cat's brain work together."
So the most bad-ass computer simulation, assuming it worked, which this guy is saying it probably didn't, was still 100 times slower than a real cat's brain. A real cat's brain also fits inside a tiny furry space the size of a baseball... and it runs on a once-daily small bowl of cat food. We have a long ways to go.
Except that individual neurons have tens of thousands of possible connections to other neurons, and continually morph and change those connections. That's impossible to do on a rigid piece of hardware.
I bet they just based their simulation on Simon's Cat which, to be honest, is a pretty accurate representation.
Summation 2
I don't really see how they would have verified that they were able to simulate a cat's brain. AFAIK, we don't have single-neuron level imaging, and the resolution on FMRI and EEG put those right out. Looking at macro level behavior would be pretty absurd- I too, can write a program that will decide to play with yarn. Unless there's something I'm missing, IBM seems to have made a claim it can't support.
So according to this guy rant letter, the "cat-brain simulation" was nothing more than the simulation of a ANN wiht X number of neurons with X equal to the average number of neurons in a cat.
However, it seems the /complexity/ of the simulated neurons is not remotely similar to that of the neurons of a real cat.
With that view, yes it seems less breakthrough. The experiment reminds me of AI researchers that thought that we could get intelligent machines using a brute-force kind of approach; this by adding /enough/ knowledge-rules, /enough/ processing power, etc...
Ubuntu is an African word meaning 'I can't configure Debian'
This IBM announcement was just ridiculous. To cite only one argument, the brain does not consist only of neurons. It contains at least as many other cells which are also involved in signal processing. Mohda would be laughed at in any neuroscience conference and he certainly doesn't help the cause of theoreticians in the neuroscience field by making such stupid announcements. Eugene Izhikevich who designed the neuron model being used for these simulations had a PNAS paper not too long ago modeling the entire human brain and he did not claim that he successfully modeled the human brain. Plus no one has any clue how the brain computes really so making a claim about the formation of thoughts is just nonsense.
VINDICATION!
Finally had enough. Come see us over at https://soylentnews.org/
IBM has a known history of making overblown claims. This is what happens when you let your PR mesh with your technical research. Deep Blue was a giant PR stunt, and they had humans retooling the code in between matches. What a crock. When they get a robot that catches mice, purrs, and jumps on the table to eat my burger when I leave the room for 2 seconds, maybe then I'll believe it.
I saw that story earlier and dismissed it for the crap that it was. I'd like to thank Henry Markram for vindicating my snap judgment with his flame email.
My research recently took me to some of Markram's work - the guy is brilliant and REALISTIC. His research goals are simple and attainable and any claims of success he has are *well* within the real world. He's incrementally worked his way up from a few neurons - the way a *real* scientist works; and to him, the simplest "brain simulation" of any sort is definitely possible, but far off in the future.
Think about it. Think about it like a cat.
In block-capital Papyrus on top of a humourous cat photo.
In a survey of 100 programmers, 111111 thought that duck-typing was a good idea.
http://intranet.cs.man.ac.uk/apt/projects/SpiNNaker/
It seems that for quite a lot of folks toying with topology and interconnects is a promising approach.
One that hath name thou can not otter
Think about it. Think about it like a cat.
SMBC explained why cat translation products fail. Although there are financial endeavors to decode dog.
My work here is dung.
Has anybody claimed/verified that the cat in question is alive?
I have simulated a Conservative Republican's brain with only two points: 1. Listen to Rush Limbaugh. 2. Mindlessly repeat what Rush says.
Can one of those cats run linux?
"I used to think that the brain was the most wonderful organ in my body. Then I realized who was telling me this."
Faith is a willingness to accept something w/o complete proof and to act on it. Reason allows you to correct that faith.
Brains use quantum computing plus new physics. You can't simulate that in classical digital efficiently. The computational power of the quantum computer scales exponentially, whereas computational power of digital computers scale linearly. The advantages of quantum are somewhat offset by increased error handling problems, but that doesn't make it much easier to simulate.
Having said that, with our digital computing power increasing exponentially with time, they could make a digital analog to a brain, and thereby prove that the brain is more than electrical circuits, because that digital version doesn't function properly.
These cat-brain deniers are making me SICK! idiots!
The debate about cat brain simulation is OVER!
We now know what AC does for living. Or at least that it has something to do with supercomputers and brain simulation, possinbly biologics too... This certainly explains why he seems to be able to claim expertise in nearrly any given subject!
It's hard to verify anything cause the machine just sits there and ignores everyone.
It must have been something you assimilated. . . .
Actually, I'd say you are being unduly optimistic about human nature, and unduly pessimistic about how things are going.
When was the golden age you imply once existed when most individuals worked diligently and wisely for the benefit of all?
The human animal is still the same animal that it has been since before recorded history. We're selfish. There is quite literally no limit to what we will do towards our own selfish ends.
We've done pretty well for ourselves since we started making selfish work, and even though some times are better than others we're still doing pretty well.
It's not nearly as bad as it seems.
The media doesn't understand what fair and balance is. They assume every opinion are equal and as valid as facts. They are not.
Media generates controversy and then display it for all to see. Hence, the perception is that it's all a fight and confusion. This is generally incorrect.
Science marches and and continues to deliver the goods.
The Kruger Dunning explains most post on
Seriously, you haven't posted in 4-5 years, and you jump out to post now? Let me guess, you work in his lab...
Please help metamoderate.
Until it can piss on my briefcase because it thinks I've been ignoring it we have no way of confirming that it is actually simulating a real cat's brain.
I've abandoned my search for truth; now I'm just looking for some useful delusions.
I think people are missing the obvious potential here. I mean, if you could engineer a computer to accurately simulate a cat's brain, then you could implant that computer in a sexy gynoid body, and have a robot-girl with the mind of a cat!
Bow-ties are cool.
If pressed, Dr. Boahen himself would contradict the Discover article and say the chips were not "brain like" at all. He's working from the same place Karl Pribram worked from 50 years ago, and Karl still can't say he knows how the brain works. Simulating a process that's assumed to be a part of brain function because it can produce results more effectively and/or efficiently that brute force digital computing does not make it "brain like". The comparison/contrast done on power consumption doesn't make a case for similarity to brain function either. It's even more misleading because the sole source of power for the brain is metabolism of glucose, with no consideration of Ohm's law and such to be taken into account.
Some of the most primitive neural networking devices were capable of learning to a depth and at a speed far outstripping the brain when the number of neurons/switched are taken into account. To make a chip more "brain like" is to negate that benefit. There are 7 billion human brains running around loose, and way too many cat brains if you ask the SPCA. We don't need to build more. At least not the hard way. What we need are devices that can out perform the brain in specialized functions. Mimicking a process in order to mimic a result does not produce an efficient process or even an efficient mimicking. Building in "noise" when one doesn't even understand what the noise is for (here taken in the signal processing sense: anything other than a defined signal is noise) doesn't improve the situation. Even more salient, building a digital simulation of an analog process introduces error that compounds the longer it runs. No, not even neurons "firing" are digital in nature. They require a voltage curve specific to the neuron type that in general follows a specific pattern of hyperpolarization and depolarization which is itself analog. Furthermore, the voltage measured is a change over time either inside or outside the cell membrane; taken together the internal and external voltages balance as to the respective changes. Digital signals are one voltage throughout the channel.
The continual insistence at trying to satisfy the descriptor "brain like" when one doesn't have an accurate description of that the brain itself is like, only makes Edsger Dijkstra's quote more relevant: "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim."
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
They did manage to simulate a cat brain... but they failed to mention it was a dead cat.
Whoa. Déjà vu.
So God does play dice. And we call it "free will".
I think we want a system that we can ask to do a complex task in natural language, and which will perform the task, only asking for further instruction when what we've told it is sufficiently ambiguous.
I suspect consciousness will be a byproduct in such a system (as it is in us), but to me, consciousness is not the goal. In fact, if we could achieve it without consciousness, that would be better, since a whole swath of ethical issues in AI go away.
Which reminds me of something else I thought yesterday regarding this: is anyone considering the ethical issues?
I don't think this simulation approaches the trouble spot yet, but at some point we have a good enough simulation of a brain that we're essentially maintaining a sentient creature in an environment with very limited stimuli (a torture in itself) with a half-functioning brain. I'm sure we'd decide that what we learn is worth it, but we should at least acknowledge the issue.
Eventually, it's going to be a simulation of a human we're torturing for science.
It's nonlinear for small numbers of neurons since you need to count connections to second and third nearest neighbors as well as first nearest neighbors. But once you get past the length scale of the longest connection, it scales linearly from there.
It's like the road system. A city with a bunch of intersections will have more road segments between the intersections than there are intersections themselves. But a second city won't build roads from each of its intersections back to each intersection in the first city. If you want to get from one to the other, you'll traverse some local intersections and then choose one of the roads spanning the rural area between. So the cost to repave the roads scales linearly with the area of the nation.
Our world increasingly looks like Fredrick Pohl's story "The Marching Morons"...
Not saying this because I know better, but because your mention of the story intrigued me and I hoped to find it or at least find out more about it. It appears it was written by Cyril M. Kornbluth, a contemporary and good friend of Pohl's.
link
I think I must find this story, as the premise of "Idiocracy" was interesting but the execution seemed, to me, quite flawed.
Karma: Excellent, but still won't get you laid.
So you're saying that a cat queues up its activities for the hour of wakefulness by planning during its hours of sleep? Kind of like the Mars rovers get commands issued from Earth, move a little, and then wait around for another batch? And therefore cats can't respond to any new stimuli during their wakeful hour until they have another sleep cycle to process the new information? Fascinating.
HAHA OH YAY ANOTHER CAT BRAIN ARTICLE WITH TWO COMMENTS ABOUT NEUROSCIENCE AND TWO HUNDRED 'LOL CATS R FUNNY' COMMENTS. Nobody fucking CARES if your cat 'acts funny'.
Filter error: Don't use so many caps. It's like YELLING.Filter error: Don't use so many caps. It's like YELLING.Filter error: Don't use so many caps. It's like YELLING.Filter error: Don't use so many caps. It's like YELLING.
Matching the neuron count and connection count of a cat brain is clearly not sufficient to simulate the functionality. Neurons in a mammal brain are not randomly connected. A great level of organization happens during the growth of the brain cells and connections starting from the embryonic stage. Much of the functionality is "hardwired" as result of this organized growth process which has evolved over hundreds of millions of years, and for higher level mammal like a cat a lot of the functionality is wired (learned) during early "kittenhood". Without reverse engineering some of this "schematic diagram", I am not sure how useful it is to simulate a random set of neurons that are wired randomly unless the object is to create a high-grade white noise generator.
Better get it, now! You don't know what those lolsimcats will do without their cheezburger.
John Connor! Please save us!
Actually, I thought it seems more like :
a poor imitation of a "badly designed fish" (sci-fi quote).
The T.M.W. Laboratory Complex has already solved the problem of modeling cat thoughts with this contraption:
http://www.thismodernworld.org/arc/1990/90cats.gif
The real problem is that matching the wiring diagram of a simulated brain to that of a cat probably will not result in a computer that thinks like a cat, because a brain is a dynamic entity. All sorts of things change with brain activity, including numbers of receptors, their properties, their distribution, the shape of synaptic spines, etc., etc. We don't know how much of this is really critical to the major functions of the brain, and how much of this is minor evolutionary tweaks, or even fortuitous "spandrels" that are present for incidental reasons but are not crucial to function. So perhaps a simple rule for how synaptic coupling changes over time will suffice, or perhaps it is necessary to carry the simulation down to the level of individual receptors moving around on the cell membrane.
So simply setting out to simulate a relatively large mammal like a cat is pretty much a waste of time, because it is virtually certain that the first attempt won't begin to reproduce the function of the animal brain. A model would be very useful for hypothesis testing, but it makes sense to start small. I'd be happy to see a simulated mouse brain that could do simple things like finding its way around a room or remembering where food is hidden. Even that is probably too ambitious to start with. How about a simulated fly or cockroach that reproduces the behavior (in a simulated space) of the actual insect?
Won't somebody PLEASE think of the cats!?
The MAFIAA is a bunch of mindless jerks who will be the first up against the wall when the revolution comes
Mohda would be laughed at in any neuroscience conference and he certainly doesn't help the cause of theoreticians in the neuroscience field by making such stupid announcements.
Yes. I am not a neuroscientist, but applied linguistics, at the technical end (where I live) does borrow from that field from time to time, and even with that relatively small understanding of the way the brain works, I immediately called "bullshit" on this crazy claim from IBM. We just don't really know how a brain works yet; how could we model it?
In my own field, the same nonsense continually pops up from computer scientists blathering about machine translation, which, despite all the rosy claims, does not actually work. The reason is that the computer loads up on rules and objects (words and phrases), and then just links them up with another set of them from a different language. The computer doesn't "speak" either of these languages, and doesn't know what the words mean, and doesn't know what they connote (emotional response), and cant tell when it's spitting out nonsense. The problem is always approached as though languages were rule systems, when those rules are made up after the fact, and most psycholinguists believe that basically there is very little processing involved in language production--just a lot of spitting out crap we've heard before that is tied to specific feelings and concepts that we are trying to recreate in the other. Basically, we don't know how it "works," because it is an innately human activity, which is the product of the human brain and the evolutionary pressures of 2 million years plus. The computer can't do it right because it's not human.
I'm not saying "never," because I believe, and hope, that we'll be able to crack it one day--at least to the point of being useful for something. But computer scientists always overstate their successes, because they don't know enough about the end point. No one does.
I was almost-audibly cheering as I read TFA. Spot on.
the long-pondered question of why humans only use 1-15% of their brain is largely a matter of power consumption
This old wives' tale is just plain nonsense. If you really only "used" such a small fraction of an organ, natural selection would quickly have reduced it down to a more manageable size. Your brain is a wonderfully heterogeneous organ composed of somewhere between 300-500 sub-organelles, many of which are permanently "on" to regulate such basic functions as temperature, corneal reflex, appetite, serum CO2 level and O2 levels, oculovestibular vestibular tracking, swallowing, gagging, and so on.
Another way of thinking about it is to look at the biochemistry of glucose takeup in the brain compared to the body. Unlike organs whose energy requirements and input/output profile vary dramatically (such as skeletal muscle with GLUT4), the brain has no insulin-regulated GLUT cell membrane transporters. And unlike, say, the liver, which uses high-capacity, low-affinity GLUT2 transporters, neurons tend to use GLUT3 (high affinity, low capacity), while the brain's endothelial barrier tends to use the basal-rate GLUT1. The brain is simply not an organ designed to rapidly up- or down-regulate its energy consumption.
Da Blog
This is where skepticism has value... Many people think Michael Shermer is the leading skeptic of our time, but is he really? Think about. Shermer defends mainstream, widely accepted and widely defended positions, rather than positions that are controversial.
No, where we need skepticism is in areas like this where the claims are too good to be true and the evidence is a bit murky.
I've been hearing about artificial intelligence since the 1970's, and while it is very interesting, the actual progress that has been made pales in comparison to the claims of its most vociferous proponents who say that human like intelligence is just 10 years away.
Its like the Darwinist who points to variation in the sizes of Finch beaks as compelling evidence for "goo to you" evolution.