The EU is sinking €1.2bn (and the US is proposing to spend more, $3 billion) into a colossal project to build a supercomputer simulation of the human brain. To which I say, "What the hell? We aren't even close to building such a thing for a fruit fly brain, and you want to do that for an even more massive and poorly mapped structure? Madness!" It turns out that I'm not the only one thinking this way: European scientists are exasperated with the project.
"The main apparent goal of building the capacity to construct a larger-scale simulation of the human brain is radically premature," Peter Dayan, director of the computational neuroscience unit at UCL, told the Guardian.
"We are left with a project that can't but fail from a scientific perspective. It is a waste of money, it will suck out funds from valuable neuroscience research, and would leave the public, who fund this work, justifiably upset," he said.
There is a place for Big Science. I'd suggest that when you're at the preliminary exploratory stage, as we are with human brain function, it's better to fund many small exploratory parties to map out the terrain, rather than launching a huge invasion with charts that are made out of speculation. We know a computer simulation is going to fail, because we don't know what it's going to simulate. So why are they doing this? Maybe it's a question of who "they" are.
Alexandre Pouget, a signatory of the letter at Geneva University, said that while simulations were valuable, they would not be enough to explain how the brain works. "There is a danger that Europe thinks it is investing in a big neuroscience project here, but it's not. It's an IT project," he said. "They need to widen the scope and take advantage of the expertise we have in neuroscience. It's not too late. We can fix it. It's up to Europe to make the right decision."
I've noticed this, that a lot of gung-ho futurists and computer scientist types have this very naive vision of how the brain works -- it's just another computer. We can build those. Build a big enough computer, and it'll be just like the brain. Nope. That's operating on ignorance. And handing ignorant people billions of dollars to implement a glorious model of their ignorance is an exercise in futility.
Two things have to be considered here.
The first one is that Markram never claimed (from what I know) that this is going to be anything more than a computational model build from bottom-up. Start from single neurons, form cortical columns, etc. So basically, what these scientists are protesting is what has already been said in the description of Human Brain Project.
The second is that, from what I've read in the news, the signatories are from cognitive neuroscience field, and they have not been included in the scope of the project from the start.
So, while goals of HBP are... ambitious, while there are many risks involved (including a potential backlash), and even while much of what this petition claims is correct, by considering the above two things I read this initiative as "Markram got all the money and we didn't, and we're upset". Well, tough luck. What HBP is trying to do is still going to produce some great results, whether it fails in its ultimate goal or not.
More precisely, the petition says:
"In the case that the review is not able to secure these objectives, we call for the European Commission and Member States to reallocate the funding currently allocated to the HBP core and partnering projects to broad neuroscience-directed funding to meet the original goals of the HBP—understanding brain function and its effect on society. "
After reading the petition in more detail, I realised I missed a crucial part ("even further narrowing of goals and funding allocation, including the removal of an entire neuroscience subproject and the consequent deletion of 18 additional laboratories, as well as further withdrawals and the resignation of one member of the internal scientific advisory board"), and would thus like to withdraw my previous comment. Unfortunately I do not know how to delete comments on this website.
It is so crazy to hear that the EU and US are going to be spending huge sums of money to try and build a computer to emulate the human brain. The “what are you going to simulate?” question is spot on. Are they going to simulate some device that responds to stimuli, like a character appearing on the screen when a key is typed? Newsflash, go to your local computer store and buy one for less than $1,000. Do they want something that’s more akin to stimuli and responses of humans? Newsflash #2, the Japanese (and others) are working on incredible robotics to do just that. Do they think they’re going to design some super-powerful Turing test? Why not just contract the whole thing out to Apple’s Sirius or Google’s Robin project. Gosh, I hope that’s not what they have in mind. Do they think they’re going to map the actual neural structure and function of the brain? This immediately opens up a can of worms, exposing many questions that we’re not even close to having answers to, and building some super-computer is going to get us no closer. For instance, vision. Sure, we’ve got a pretty good understanding of how our eyes work and how they send signals to our occipital lobes. However, when it comes to understanding how we do pattern recognition and then apply names (words) to the things we see, we’re fairly clueless. Furthermore, sensory cognition doesn’t even get at the most fundamental aspect of what it is to be alive, much less human. I would argue that there are two somewhat intertwined words that best describe what it means to have a functioning human brain: awareness and choice. Awareness would include self-awareness, and it would also suggest a fundamental “understanding” of what we are doing when we do things. The second word, “choice”, can lead to discussions of free will and even get into spirituality if we allow. However, there is an easier way to define it. An entity that has choice is “fundamentally” unpredictable from any cause-and-effect deterministic perspective. In other words, it has fundamental stochastic properties to outside observers. Newsflash, digital computers (regardless of how powerful) do not have stochastic properties (unless they are broken). One simply needs to read a bit of John Searle or Roger Penrose to understand these core concepts of what it means to have a functioning brain. I feel compelled to say a bit more here regarding my own beliefs. I believe that our brain, our mind, our consciousness, and our unconsciousness arise from the periodic table of material from which we are made. I believe this as strongly as I believe that we evolved from simpler life forms, that the Earth is approximately round, and and that it’s revolving around the Sun. However, digital computers are fundamentally the wrong medium from which to attempt an accurate functional model of the human brain. Their programmers may pretend they are, they may fool us at times, and they undoubtedly provide incredible and fascinating tasks for us, but in the end, they are machines with no sense of awareness or choice.
PZ... Knowing about your atheist background (which I share) I find this statement of yours quite interesting:
"...and computer scientist types have this very naive vision of how the brain works — it’s just another computer. We can build those..."
If I didn't know better, I would think you are dualist. If the brain isn't "just another computer", what is it then? I agree that we don't really know. But so far, medicine has made pathetically little progress on this subject, especially considering the great progress that has been made in artificial intelligence. Maybe it IS time to give the computer scientist types a bit of time with the subject.
Who says that simulating the interaction of a great many neurons (with admittedly simply input/output and unknown initial states) will not lead to insights into the causes of the macro-effects we observe in the human brain. Examples of where this might be useful is in Migraine, epilepsy, mental disorders - which are mostly described with very vague analogies by neuroscientists.
Small scale simulation of neural networks has already lead to progress in the realms of image recognition, logistics planning, computer gaming, self driving cars and adaptive, artificial limbs.
If nothing else comes out of this, building a machine of this magnitude will create a great pressure on chip makers to come up with more powerful CPU architecture. This is something that everyone will benefit from. With that perspective, 3B USD is a bargain - many big companies spend more than that on IT consultants who do very little good in this world. Axing a bit of funding to neuroscience seems like the lesser evil in that context.
Disclaimer: I am a computer scientist type!
The brain is indeed just another computer. Different substrate and mechanisms, but it's a physical thing and it can most certainly be simulated in a computer, because anything can.
The point everyone misses; nobody is claiming this will be a perfectly accurate real-time simulation. The higher the resolution of your simulation, the slower it runs.
A home PC could 'simulate a human brain'. Only it would take 1000 years to model even a fraction of a second of it. Like mapping the human genome, what is unrealistic and expensive at first eventually becomes routine and cheap - but not if you never start the research at all.
It's a Moon shot, for sure. But last time humanity did that, we did actually hit the Moon.
PZ, here's what's up with that:
They've bought into The Singularity, a pseudoscience Computer-God religion, the goal of which is eternal life via "uploading" your mind into a machine.
See also _The Book of Immortality_ by Gollner, who apparently does a pretty good job dissecting this stuff. Written for laypeople but shows how many Silicon Valley types (who really should know better) have bought into this nonsense.
See also anything written by Ray Kurzweil. I frankly have compassion for the guy because he's stuck between the rock of atheism and the hard place of being terrified of death. He really needs to learn to meditate, and spend enough time contemplating nothingness that he loses his fear of it.
Thomas & Mark @ 5 and 6:
Simulation is not duplication. Much of human decision making is based on emotions. Emotions are the subjective sensation of the action of neurochemicals on neurons. You can't duplicate that in silicon, and you can't even simulate it, because as of yet we don't understand the mechanism by which chemistry translates to qualia.
Nor does even a hardware random number generator substitute for the kind of stochasticity found in human behavior.
Face this: you're going to die some day. Either there's an afterlife or there isn't. But computerized reincarnation is nothing more than a new religion wrapped up in high-tech to make it seem scientific. If you can reincarnate into a computer, you can also reincarnate into a cat. Given the choice, I'll take the cat.
Elroy @ 4:
Nice compact definition: "An entity that has choice is “fundamentally” unpredictable from any cause-and-effect deterministic perspective. In other words, it has fundamental stochastic properties to outside observers." It's behavioral, so it can be operationalized easily enough, without reliance on measuring brain activity or getting trained observers to report their subjective states.
BTW, if Penrose & Hameroff are right, the complexity of the human brain isn't 10 ^ 16, it's closer to 10 ^ 27, which also means bye-bye to the Singularity. For which reason Singularitarians despise both of them.
There is an entire segment of the high tech world that is devoted to the holy grail of predicting human behavior right down to the level of the moment-to-moment activity of any given individual. This, for obvious commercial reasons, though with some truly unpleasant implications for "freedom and dignity." Aside from the Singularitarians, this is where you'll also find much opposition to the idea of free will.
Mark, I hate to disagree, but the functioning human brain isn't just another computer. And Thomas, I applaud you for at least admitting that you don't know what the brain is. I will readily admit that many aspects of the brain work like a computer, particularly those rather uninteresting aspects of the entire brain-stem. For instance, we have "feedback loops" in our spinal cord that allow us to flinch when we put our fingers on something hot. We know this because the response times are faster than neurons are capable of getting signals all the way from our fingers, to our neocortex, and then back again. The signal gets as far as our spinal column, and some "computer like" decision making is made there, sending back a quicker signal. In other words, the order of events is: 1) touch something hot, 2) flinch by moving fingers away, 3) being aware that we touched something hot.
And sure, there may even be "software like programming" in parts of our actual brain. For instance, habits have very much the feel of a program running. Turn brain off and blithely go through our day, being only mildly aware of what we're doing.
However, the word I've used twice in the above is "aware". That awareness is the aspect of the brain that we most want to "get at", and that is precisely the aspect that digital computers have NO CHANCE of getting at.
Thomas, among other things, I also have a strong background in computer science, and am quite a good programmer when I need to be. People sometimes forget this, but digital computers, in the end, do nothing but push ones and zeros around, ANDing and ORing them to make it appear that they're doing fancy things. Even the alphabet is a completely foreign concept in the bowels of a computer. The moment you hit a key, a stream of ones and zeros is created, with the letter only being a bit of paint on the keyboard. The letters you see on the screen are just RGB pixels being turned on and off with those same ones and zeros, and it is your wonderful brain that actually “sees” the letters. Sure, we have high level programming languages for computers, but at every level, they are just ones and zeros in the memory (or other storage devices) of the computer. Computers may get VERY fast at ANDing and ORing these ones and zeros, and they may have the capacity to store and manipulate HUGE numbers of them, but they will never be anything more than machines that AND and OR ones and zeros.
John Searle has developed some wonderful and bullet proof metaphors to make this point (which no other philosophers have even dented), but it should be obvious that a machine that simply ANDs and ORs ones and zeros is completely incapable of ever developing anything even remotely resembling awareness. Yes, the programmers may set up programs that "fool" us for periods of time, but even then, they are not aware. They're just running a program (of ones and zeros). This is the classic Turing test. However, passing the Turing test still gets us no closer to modelling an actual human brain, with all of its most interesting aspects, particularly awareness.
No, I haven't said what functioning brains are, but I have stated what they're not (or at least not in their entirety). They are not "just" digital computers. They are something more. And using digital computers to attempt a complete simulation of the human brain is at best a fool's errand, and at worst an awful distraction from more fruitful endeavors and a sinkhole for wasted funds.
How would a brain not be a computer? We have no evidence that it is doing anything other than receiving sensory signals, processing them, forming cognitive network models, and producing output.
I'm a former mathematician, not a compsi guy directly, but I've modeled systems that self-build similar networks. They're incredible, and they happen to behave very similarly to the developing brain in a living species like a human.
Lastly, to say that the brain isn't a computer is a suspicious claim when we are able to disable those "awareness/consciousness" traits by snipping the wiring.
"Brain is not a computer"
There is the Church-Turing thesis that says that any process could be expressed using a Turing machine or lambda calculus.
If the brain can not be expressed with lambda calculus or Turing machine then we have in front of us something almost magical or divine.
There is the Church-Turing thesis that says that any process could be expressed using a Turing machine or lambda calculus.
The Church-Turing thesis states that any algorithmic process can be expressed using a Turing machine or lambda calculus. There's no reason to assume that brain processes need to be algorithmic.