Unfortunately, that's only for a number of cells equivalent to 1% of a human brain. A full brain's worth of simulation would take 2.5 days....although that much is probably not necessary. It will fall soon enough, though.
The thought that we are maybe two decades away from having the computing power to simulate a human brain in realtime (and much more than that subsequently) is astounding, even in this age of technological wonder.
Assuming (1) it's parallelisable (2) this is an accurate enough similation, and (3) a whole human brain is 100 times bigger, we'd need 406083k*100 processors which is about 20 billion.
We could do it now (hardware-wise), if we really wanted to.
Admittedly, I was just spitballing from the parent poster's numbers. But the idea that brain simulation is not that far off still seems plausible given those extra considerations.
If each processor draws down 50 Watts, 20 billion of them would draw 1 TW. The US uses on average ~3 TW across every power source (including e.g. combustion engines).
So it will take about 26.6 years before we can simulate a brain in real time, only considering Moore's Law. (Calculated this without pen and paper using my iPhone calculator, so somebody should check this ;) )
Only if you assume that there won't be any other advances in technology that will speed up simulating a brain (such as more specialized processors). Look at genetic sequencing. It was really hard to get from 0% to 1% of sequenced human DNA. But sequencing the remaining percent happened much faster, and nowadays the cost for is exponentially decreasing (which means that soon you will be able to buy your sequencing kit at your local supermarket for a few dollars).
I would be very surprised if it took us another 26.6 years to simulate a brain in real-time. Maybe half of that time. And a few years later you will be able to cheaply backup your own brain in the cloud in case you have an accident.
I'm not certain if that extrapolation you made of the time it takes to simulate a brain at the same abstraction level is valid. You're assuming the computation cycles needed will scale linearly, which might not be true considering that the brain is a network of neurons, where each neuron has a high number of in/out links to many other neurons. Consider the case where you take n seconds to simulate n neurons. Now add another neuron with connections to perhaps n/2 existing neurons. You'll have to iterate over each in/out edge of that neuron. Since adding just one neuron increases the time by a non-constant factor, the resources needed would increase more than just linearly. All this of course depends on the abstraction model, and I admit I know nothing about it.
At what point does this become unethical (if ever)?
Before you answer, please consider how certain you are that you are not a simulation.
Edit: I don't know the answer, and there probably is a large gray area. But I do know this starts to make me uncomfortable the more accurate it gets. Time to go reread Egan's Axiomatic again, I guess.
Even if you are a simulation, would you prefer not to exist? What, really, is the difference between reality and a perfectly-executed simulation? It's hard to condemn one without condemning the other.
I've yet to see a convincing definition of "free will" that makes any sense as anything other than an abstraction on top of determinism or randomness. To propose that we have any ability to make decisions beyond state-machine-dynamics possibly with some randomness sprinkled on top, is totally nonsensical to me.
Precisely. It's almost certain that we don't have free will. I would define free will as the mere appearance that thought processes are governed by self in a non-local time frame. I would gander that a simulated brain might have less of this illusory 'free will.'
I think the GP is talking about current simulations, not future ones. Current simulations don't have sensory organs, and I doubt the simulated neurons accurately reflect a real brain at this stage. Most likely, early simulated minds would be insane and in great pain, with no way to communicate.
Considering we experiment on animals to reduce human suffering (heck, we kill and eat them just because they're tasty), running a tiny fraction of a simulated brain for a few subjective seconds doesn't worry me much. Of course, as computers get faster and cheaper this could become a huge issue.
It's only tangentially-related, but Robin Hanson has some good talks and papers about emulated minds. The shortest summary of his views is probably this talk: http://www.youtube.com/watch?v=9qcIsjrHENU. A more in-depth version of that talk is at http://vimeo.com/9508131.
If we are in a simulation, then ipso facto, we have no idea what 'reality' outside our simulation is actually like. We can only surmise that it probably resembles our simulation, in as much as the creators probably used their own reality as a template.
Even more reason why there is no functional difference, subjectively speaking, whether we're living in a simulation or in actual reality.
Only when given the choice to enter or leave a simulation does the distinction becomes meaningful.
I think it becomes unethical once we see strong evidence that the system could be self-aware. Before that, it's just a simulation. Simulating storm activity doesn't make it rain.
It isn't a big leap to imagine a debate between aliens about whether the hairless apes on the third rock from Sol should be considered "sentient", or merely clever animals.
I'd rather err far on the side of assuming sentience and personhood if there is a shadow of a doubt, simulated or otherwise.
Why would they debate? It's very easy to test for [sentience or fake sentience programmed by a sentient]. Unless the aliens think we're robots they can do a mirror test or something.
But such a test does not work in a possibly-tainted environment. A neural net that seems to communicate might just have a hidden encoding of ELIZA.
You can err on the side of assuming sentience if you want, but don't think it's out of self-preservation-empathy-logic.
Isn't that only true if you assume that our definition of sentience is universal or that there is no higher development of consciousness? It seems to me that it doesn't even require a very skilled science fiction/fantasy writer to develop such a concept or ability.
As long as you assume whatever you're testing moves around or communicates instead of constantly naval-gazing, I'd bet even crazy-alien intelligence can be seen to be intelligent.
Sure there might be higher levels, but we're not talking about higher levels?
You were explicitly talking about testing sentience (which does not require "moves around" or "communicates"). Sentience is simply a concept we have come up with to differentiate organisms. It wouldn't seem surprising to me if something more evolved used a different concept more appropriate to their physiology or philosophy. Hence I think you need to be careful if we are about to apply our concepts to classify something.
They might have a different standard of important but that has nothing to do with checking for sentience. If they're so advanced beyond that the original analogy breaks and while I might have that discussion at some point I don't think it fits here.
On the topic of sentience, I just think it's exceedingly unlikely for there to be an intelligent organism that does literally nothing with its intelligence. Some kind of self-aware rock. And on top of that the mechanism of the intelligence would have to be almost impossible to study or plug wires into to try to force it to communicate.
We choose the definition of sentience so that it includes us. There's no reason to believe that an alien species might not do the same, and that we might not make the cut. Hell, some of the worst racist rhetoric on the planet involves seeing other humans as non-sentient.
The whole point is that we're blind to our own biases, as fish are blind to water. That includes our bias towards physical matter over simulated matter. (It's funny how the notion of a simulated universe unsettles people and destabilizes their narratives, despite the fact that discovering it would change precisely nothing about the subjective reality of our universe.)
Without having an abortion debate, what's actually happening in the simulation is just a random interconnection of very primitive nodes at this point. The network has no refined architecture and, at best, some kind of spontaneous activity. In the number of nodes, it could be (generously) compared to flipping on a cat for one second. But a cat's brain obviously has extremely fine architecture, so the better analogy would be that they're flipping on a completely scrambled cat for one second.
Does anyone know what it means to simulate brain activity? Is the brain being emulated in any meaningful way? Or is it just modeling the physical behavior of neurons?
To put this claim in perspective, we are still unable to simulate a single cell by modeling everything we know from biochemistry, even if we ignore molecular dynamics. Mostly this is because we don't know what many genes and the proteins they encode do.
This matters, for example, if we want to simulate neuronal plasticity or the effect of antidepressants. We're an incredibly long way from a full simulation, which would also require integration with the rest of the nervous system.
> To put this claim in perspective, we are still unable to simulate a single cell by modeling everything we know from biochemistry, even if we ignore molecular dynamics. Mostly this is because we don't know what many genes and the proteins they encode do.
To me, this mostly shows that naive beliefs about difficulty aren't really accurate ('but it should it harder to simulate a brain than a cell, because it's bigger right!').
Sure, but that's also for a bacterial cell. We have 50 times as many genes and you can't easily culture humans on a Petri dish for experimentation. A single gene can take years of work to understand.
Yes, you can simulate anything if you abstract away enough. But if the goal is a convincing replication, you'll need to include lots of intracellular effects.
In the end, I believe that practical simulation of a whole brain that you can have a conversation about its existence with is more difficult than the development of quantum computing powerful enough to break all existing crypto.
The work is still interesting, if only because it pushes the limits of what we can do. But I might want to start with an ant brain.
Why does it show that the belief isn't accurate? This isn't simulating a brain, it's simulating the connections in a brain and pretending the neurons are about as complex as wires. That is a model that is possibly correct, but there's certainly no evidence it's correct.
It shows, at a minimum, that 'simulating' is not a binary either-or proposition, and that the higher-level a simulation, the easier/more efficient the simulation can be, and so it is perfectly possible to have a high-level simulation of a 'big' thing which is easier to run than a low-level simulation of a 'small' thing - both of these points, as obvious as they may seem to you or me, are badly underappreciated by a lot of people.
When people say 'simulate a brain' they mean a very specific thing: simulating it in enough detail that it can have thoughts in fundamentally the same manner as a fleshy brain.
>it is perfectly possible to have a high-level simulation of a 'big' thing which is easier to run than a low-level simulation of a 'small' thing
While this is true, it's not being shown here until we know that this particular high-level simulation actually works. If it doesn't work then it's more of a failed attempt at creating a simulation.
I happen to work in the synthetic biology lab at the venter institute, although not on this project - two of our researchers are coauthors on this paper - and my personal opinion is that this paper is a joke. Note that the guys who made the simulation didn't even pick the correct organism. The cell that we have synthetic control over is M. mycoides, not M. genitalium. Lest this seem like a nit-pickey detail, until they rewrite their software for mycoides, in broad strokes the model is essentially untestable, since we don't have broad-scale control over the genitalium genome like we do mycoides.
Also note that: "The model accounts for previously observed
gene essentiality with 79% accuracy." This is REALLY bad. Considering they are 'seeding' the model using a biased structure that automatically "knows" that certain genes are essential (for example, they have a 'module' for replication that probably zeroes out if you are missing an important replication component, as opposed to simulating how those genes work at a low level) this is probably somewhere close to as good as flipping a coin.
A common theme where I work is me making a vociferous criticism, being ignored, and the things I say coming true later. (this is not to say that all of the objections I make wind up being correct)
To correctly model a cell you would need to simulate every atom and there are about 10^11 to 10^14 (!) of them in a single cell.
I think an interesting number would be how many atoms does it take to simulate one atom reasonably well. Then you could get correct results for some ( really ) small biological functions.
Atom is still an abstraction of sub particles ( that may be made of even lower ones, we don't know ) and their physical interactions so i think it is a model. And a reasonably good one.
In cases where qualitative meaning is nontrivial, you don't know 100% what is going on, it is nevertheless possible to implement parametric studies for some target parameter.
Your point is valid, but many people are trying to "correctly model a cell" without making a total atomistic simulation (which is clearly not going to happen any time soon). It's all about what questions you're looking to ask of a model.
Usually people modeling cells don't consider molecular dynamics, they consider concentrations of proteins, ions, and metabolites. A full MD simulation would obviously be superior, but useful predictions could be made without it.
It's hard to say which is the better approach. On the one hand you have MD from which you could technically build an accurate model "from scratch", folding all of your own proteins as part of the simulation, but it is exceedingly expensive to do that. On the other hand, you have a model built at the level of biochemical reactions and molecular biology, which is computationally feasible, but then you don't have enough information from bench work to flesh it out.
What would you get from simulating both hydrogen atoms and the oxygen atom that you wouldn't get from simulating plain old water? The vast majority of a cell is just water, seems like you could cut your complexity in half with that abstraction.
I'm no molecular biologist, so i'll happily agree if you say there are some fraction of water molecules broken down, or formed by various cellular processes, but that must be a tiny fraction of the total, and could just be special cased.
It just seems like you're willfully ignoring everything physics has to say about how the universe works so you can get a big number for your calculation.
How would you plan on simulating all the variety of states, reactions and solutions involving plain old water without worrying about hydrogen and oxygen atoms?
Again, i would assert that a cell is 70% water. some tiny fraction of water that's actually used in a reaction would just be special cased, and modeled independently.
My understanding of "solution" is stuff floating in water. If there's some value in breaking out each atom, i'm all for it, but i kind of suspect the spherical cow is good enough in that case.
Even if i'm wrong there, each atom would introduce at least 6 degrees of freedom. IIRC, water is slightly polarized, and that's likely relevant to at least some reactions. Replacing the 18 degrees of freedom with 6 to represent a rigid water, or perhaps 10 or so to represent a flexible version, modeling as the molecule will dramatically lower complexity.
"My understanding of "solution" is stuff floating in water."
In that case, don't even bother with trying to work out what it would take to mathematically model biochemistry yet. Go study some basic physics and chemistry for a few years instead.
Ah yes, i called out water specifically and therefore i am an idiot. Mia Culpa. Best of luck with your simulation, i'm sure that individual atom model will work out great.
I don't know if you are an idiot, however you appear to have not even the slightest bit of technical understanding of the subject you are offering opinions on.
I am not attempting to be insulting, I am attempting to point out to you that you need to go and learn a hell of a lot more about the sciences than you seem to know at the moment if you are going to have any sort of productive discussion about the technical aspects of modelling biochemistry.
Note that cells are in fact completely packed- everything is grinding against everything else. THere's water there, but it's not a bunch of free solutes.
check out "macrophage and bacterium", by the same guy. All of the shapes of the proteins are informed by their crystal structures. The concentrations of them in the cell/plasma are also informed by known data about those environments.
I don't understand why it's easier to simulate each individual atom than it is to simulate things with structure. I would assume a simulation would regard the long yellow strands of dna as dna rather than the component parts.
I'm used to a world where having structure implies constraints. With constraints, simulation can avoid a bunch of special cases and is therefore faster. I find foobarbazqux's approach of simulating each atom upsetting. But that's ok. You guys seem to think it's the way to go, and you all get results. shrug
-edit-
I don't think simulating water is easy, but i believe it's likely easier to simulate water than it is to simulate all possible states of the constituant atoms.
yeah, i don't think simulating each atom will work either. I'm not sure physics-based constraints at all is correct, and that maybe we need to try to generate heuristic constraints and then try to justify those heuristics using physics later.
At an atomic level, water plays critical roles in very small quantities and understanding these role requires detailed simulations. See this recent paper [1] for an example of how water molecules affect potassium channel inactivation, a key process in cardiac electrophysiology and other areas.
That's not to say that you can't get useful results by treating solvent as a continuum (or even ignoring it entirely), they'll just have a different kind of explanatory power.
Yes we are very far away from the human brain on many levels. However, I don't like this kind of article because it doesn't make sense on the software level.
Mathematically, we could figure out how much computing power we would need to match the human brain (24 bytes per synapse * number of synapse, etc).
I think I would be more interested in what was the specific of their experiment.
These attempts at brain simulation perplex me to no end.
Simulating a lump of tissue is definitely useful, it's when they start talking about simulating a complete brain that I get lost.
A brain doesn't happen overnight. It is the result of a long developmental process encompassing embryology, learning and experience acquisition.
How do they plan to wire the whole thing?
Micro- and macroscopic connections are well understood, it is the so-called meso-scale (in between) that is troublesome.
There will never be a non-destructive way to do extract the information of a live brain, and I'm not sure there will ever be a way to extract it at all. The current methods[0] allow to extract either the wiring or the genes expression.
They use it to understand brain. So they wire it like the brain is wired.
If they simulate cortex, they wire arrange it into cortical columns (size of each column is between 50,000 and 100,000 neurons and there is over two million of them). Those columns have many different layers and neuron types. Understanding how cortical columns and minicolumns work is very important.
Each neuron is growing, making decisions about it's environment, reacting to a multitude of neurotransmitters with a mulititude of complex reactions. Deciding where are when to make a new connection. If each node is a little computer system how do we even begin to emulate its probably extensive and convoluted 'programming.'
You made my point regarding small scale simulation, but completely missed the rest.
> So they wire it like the brain is wired.
As if anyone knew how to do that. Even assuming that you can do both accurate micro-tractography and extract cell-by cell epigenetic information out of a single column, you'd still be missing other cricial information like individual synaptic strength.
I'm not talking about cortico-thalamic connections (required to model epilepsy) inter-columnar connections, and whatever regulation happens in the white matter.
Relying on the probability distribution of connections as a crude approximation is useful, but far from the real deal.
Yup...people thought the Wright brothers should stick to building bicycles too. I mean how can a bicycle repairman replicate a mechanism that nature took a million years to develop in a bird? It was ridiculous. Couple years later they were watching them fly around in the air.
The title statement is almost meaningless. It would be better just to say "simulating reality takes a LONG time".
My research is in molecular dynamics. I simulate systems up to 1 billion atoms on Kraken & Titan. HUGE approximations are made in simulating these systems, but depending on what exactly it is you're studying, these simulations still provide useful results. That is the key to all these studies: how well does your approximation reproduce whatever it is that you're attempting to model? In some cases, very well. For instance, I'm not going to get exact energy levels of a large system, but the system will qualitatively evolve in the same fashion that the experimental system does (which then guides the experimental counterpart to the research). I don't know the details of this brain simulation, but there is certainly some aspect of it that is not being reproduced anywhere close to real-life, and hopefully this isn't what they're interested in (and I'm sure they know that, but I don't think the article author does).
The very best simulations of reality that we can perform handle at most just a few H/He/C atoms. And an issue called the fermion-sign problem means that the computational power necessary to simulate larger systems scales exponentially with the number of particles. Unfortunately what that means is -- short of developing quantum computers (which are polynomial order for the sign problem) -- we aren't ever going to simulate more than a few atoms with near-perfect accuracy, and certainly nothing like a human brain.
EDIT: Didn't mean for my comment to sound so negative. Obviously, the researchers know exactly what they're doing. I was just trying to dispel the impression that we're close to simulating the human brain.
If we have a reliable model for how synapses fire, we can simulate the brain. Whatever problems quantum mechanics impose on molecular simulation is not relevant, the same way that you don't need to simulate the nucleons on your molecules even if you need precise energy measurements.
Generally, the people doing brain simulations like to claim that they can use a massively reduced representation (little more than a graph model that pushes signals around with some constants) and yet, somehow, the simulation is inherently capable of reproducing something complex about thought. I'm not sure how they get to this conclusion, but it's a massive reductionism. It would be nice if it were true- it would mean that for a reasonable expenditure, we could build a machine that, basically, "thought" and had a "mind" and we could convince ourselves that it was "living" much like we think of other humans.
To date nobody has come up with a really compelling disproof that they are more or less correct- thought and mind seem to be best explained as "emergent behavior of a complex system".
Now, whether can we bootstrap a thinking mind by running complex simulations on computers remains to be seen. I'd like to believe there is some interesting physics going on in brains we can't simulate with straightforward physical models, but there really isn't any good evidence for that.
Can you elaborate on this class of 'something'? There are intelligent behaviours that can be reproduced this way, so without more specifics I feel like you're making the same mistake you're condemning.
For example, the glial cells are now believed to be vital to the brain's processing, but most "brain simulations" are still modeling neurons only, not glia.
Not helpful. My question demands an answer from the person making the claim, unless you happen to know what they're thinking. Besides, this sounds nothing like what they're talking about.
Worse, I don't think you're prepared to prove that a mostly pure-neuron simulation is incapable of giving rise to that certain complex quality apparently far removed from the level of the simulation.
I was literally asking someone what they were talking about. Instead I got an answer from someone else about something else. It's just some asshole who has one little nugget of information they want to spout at someone.
I've noticed this too. I blame a combination of Zen and the Art of Motorcycle Maintenance and not knowing what the word actually means.
From my observation, it appears people assume "reductionist" means "simplistic" as opposed to "representing a complex system as no more than the sum of its components". To be honest, I think it's just an etymological problem: "reduction" sounds bad, so "reductionism" sounds bad.
The grandparent here follows this pattern. Using the actual meaning, it would be paradoxical for an approximation to be reductionist. In fact, it's quite the opposite. A reductionist would hold that the most accurate high-level model of a brain would be the most low-level model of a brain. If we could model each atom precisely then we would, necessarily, model the whole brain precisely.
The whole thing is pretty amusing when you take into account that anyone who manipulates software or systems in any serious way needs to engage in reductionism to be able to work. It's not as if "first, we write module A, then module B, then a system magically appears from the ether" is a viable architecture. At least, not since we stopped taking neural nets seriously.
Reductionism is fundamental to science. There are places where it shows it limits, but I've always equated being reductionist with being clever, with taking a complex question and re-casting it in terms that can be measured/studied using the current state of the art.
Some parts of the mind are emergent, but I'm sure some parts or features of the mind are a consequence of the particular organization of the neurons itself, and they won't 'emerge' without careful tuning.
Otherwise Sperm Whales would simply be much smarter than us.
About conscience itself, my guess is as good as yours.
That's true. However, unless I'm mistaken (my field is more chemistry-simulation than physics-simulation), the internal nature of the proton is not necessary for achieving chemical accuracy level results from something like FCIQMC, namely because the chemical properties of interest are the result of only the charge interactions of the protons/electrons as potential terms in the Schrodinger equation. I believe applying QMC to solve that equation for all charge-charge interactions (including electron correlation) is enough to reach essentially exact energy levels (in the limit of simulation time -> infinity) ... am I incorrect? I mean, there's probably some difference in the eleventh decimal place due to the lack of relativistic and radiative corrections.
Sure, just wanted to point out how complicated truly "simulating reality" is. :)
I mean, my doctorate involved calculating tiny, tiny corrections (8 or 9 decimal places) to hydrogen-like energy levels, but even there I could still treat the proton as point like!
For all the reasons you listed, if we’re going to crack the problem of strong AI in the next century, we’re not going to be doing it by simulating the brain at the atomic or molecular level. However the lower level simulations are important because we still don’t have deep understanding of how neurons work.
An overwhelming amount of neuroscience research data points to the neocortex reusing the same functional unit composed of 100's-1000's of neurons called a cortical column. For example one patient was able to regain her balance by rerouting the signal from her vestibular system to her tongue. So the cortical column is the layer of abstraction we should be shooting for.
Hopefully we'll move away from the 1950’s style perceptron, whose mathematical simplicity has attracted researchers but has been a major distraction to AI as they share almost nothing in common with their biological counterparts. If we don’t want to see a 3rd AI winter, we need to move to more biologically inspired approaches.
I would imagine that the analogue nature of the brain is going to always entail a lot of approximation on simulation. I am not sure you could even accurately simulate even 1 neuron.
Disclaimer: this is completely out of my field - just a layman speculating.
How many Moore law doublings before it is real time? Seems like ~23 generations or ~35 years. Of course, if you leave out the bits constantly thinking about sex, we might get there a lot faster :)
Recent research has shown glial cells, the uncelebrated insulator of the axon, in fact provide significant chemical signaling and modulation to the process of neural activity. Only phenotypical changes have been observed, it's not understood yet. Suffice to say that computational limits aside, we still couldn't simulate a brain because we wouldn't know what to make.
If Moore's law were to continue to hold, this would mean the same simulation could operate in real-time on an 83K processor cluster in approximately 25 years or on one processor in about 50 years. Of course, this simulation is 1% of the neurons that a human brain has, and I don't know if scaling up to a full-brain would be linear (This probably depends on the average length of axons in the brain.). This completely unrealistic estimate also assumes no improvements in algorithms, etc..
Perhaps a more interesting question is whether or not the brain can really be simulated by a classical computer at all. If quantum mechanics plays a role in the function of the brain we will need a quantum computer to do the job. A classical simulation of a quantum brain would be a very odd beast indeed! It might exhibit behavior that seems like intelligence but somehow falls short.
Roger Penrose has theorized the brain relies on quantum effects, but it seems to have been as discredited as something currently unfalsifiable can be discredited.
It's not like there is a conspiracy against Penrose, there is just a prevailing feeling (which I for one subscribe to) that: "computation via quantum mechanical processes is irrelevant to explaining how brains produce thought, contrary to the ongoing speculations of many theorists. First, quantum effects do not have the temporal properties required for neural information processing. Second, there are substantial physical obstacles to any organic instantiation of quantum computation. Third, there is no psychological evidence that such mental phenomena as consciousness and mathematical thinking require explanation via quantum theory. We conclude that understanding brain function is unlikely to require quantum computation or similar mechanisms."
Oh, yeah. Totally agree, I must have worded it poorly. I'm just saying all theories are technically "speculative" until we can replicate cognition, but his quantum theories sound unnecessary and hand-wavey, akin to "The Soul".
Stanford and other groups have been working on "brain asics" for a few years now. The eventually success of these chip-sets will probably dramatically increase our ability to simulate brain processes more efficiently in the near future.
Maybe the simulation algorithm is the thing that needs improvement. Some day one finds a way to reduce the complexity an order of magnitude, then the era of human brain is over.
~10 billion nerves * 10,000 connections per nerve * 100HZ * ~ 100 neurotransmitters = a complex beast to simulate even if you could get away with 1 calculation per connection per transmitter and ignore things like propagation delays and fatigue.
This has been done before with a full human brain scale model in 2005. It took 50 days on 27 processor for that simulation. It's always good to have more research being produced, but this linked article is a bit short on information.
123 comments
[ 2.9 ms ] story [ 180 ms ] threadWe could do it now (hardware-wise), if we really wanted to.
I would be very surprised if it took us another 26.6 years to simulate a brain in real-time. Maybe half of that time. And a few years later you will be able to cheaply backup your own brain in the cloud in case you have an accident.
Before you answer, please consider how certain you are that you are not a simulation.
Edit: I don't know the answer, and there probably is a large gray area. But I do know this starts to make me uncomfortable the more accurate it gets. Time to go reread Egan's Axiomatic again, I guess.
Considering we experiment on animals to reduce human suffering (heck, we kill and eat them just because they're tasty), running a tiny fraction of a simulated brain for a few subjective seconds doesn't worry me much. Of course, as computers get faster and cheaper this could become a huge issue.
It's only tangentially-related, but Robin Hanson has some good talks and papers about emulated minds. The shortest summary of his views is probably this talk: http://www.youtube.com/watch?v=9qcIsjrHENU. A more in-depth version of that talk is at http://vimeo.com/9508131.
Insane maybe, but why in pain? Why would you add pain signals without also communication?
Even currently existing lower animals (lobsters and down) don't feel pain - why would you expect a simulation to do so?
If we are in a simulation, then ipso facto, we have no idea what 'reality' outside our simulation is actually like. We can only surmise that it probably resembles our simulation, in as much as the creators probably used their own reality as a template.
Even more reason why there is no functional difference, subjectively speaking, whether we're living in a simulation or in actual reality.
Only when given the choice to enter or leave a simulation does the distinction becomes meaningful.
I'd rather err far on the side of assuming sentience and personhood if there is a shadow of a doubt, simulated or otherwise.
But such a test does not work in a possibly-tainted environment. A neural net that seems to communicate might just have a hidden encoding of ELIZA.
You can err on the side of assuming sentience if you want, but don't think it's out of self-preservation-empathy-logic.
Isn't that only true if you assume that our definition of sentience is universal or that there is no higher development of consciousness? It seems to me that it doesn't even require a very skilled science fiction/fantasy writer to develop such a concept or ability.
Sure there might be higher levels, but we're not talking about higher levels?
On the topic of sentience, I just think it's exceedingly unlikely for there to be an intelligent organism that does literally nothing with its intelligence. Some kind of self-aware rock. And on top of that the mechanism of the intelligence would have to be almost impossible to study or plug wires into to try to force it to communicate.
See also: http://terrybison.com/page6/page6.html
It's not like this is some crazy threshold designed specifically to apply to humans, it applies to a bunch of animals too.
At a certain level it's kind of like checking for turing completeness.
I find it difficult to see how this could ever be unethical (given the status quo).
This matters, for example, if we want to simulate neuronal plasticity or the effect of antidepressants. We're an incredibly long way from a full simulation, which would also require integration with the rest of the nervous system.
To me, this mostly shows that naive beliefs about difficulty aren't really accurate ('but it should it harder to simulate a brain than a cell, because it's bigger right!').
But of course, progress is being made on simulating cells: http://lesswrong.com/lw/drk/paper_simulation_of_a_complete_c... It's a high-level enough model that you can do it with only 1 core per cell.
Yes, you can simulate anything if you abstract away enough. But if the goal is a convincing replication, you'll need to include lots of intracellular effects.
In the end, I believe that practical simulation of a whole brain that you can have a conversation about its existence with is more difficult than the development of quantum computing powerful enough to break all existing crypto.
The work is still interesting, if only because it pushes the limits of what we can do. But I might want to start with an ant brain.
>it is perfectly possible to have a high-level simulation of a 'big' thing which is easier to run than a low-level simulation of a 'small' thing
While this is true, it's not being shown here until we know that this particular high-level simulation actually works. If it doesn't work then it's more of a failed attempt at creating a simulation.
Also note that: "The model accounts for previously observed gene essentiality with 79% accuracy." This is REALLY bad. Considering they are 'seeding' the model using a biased structure that automatically "knows" that certain genes are essential (for example, they have a 'module' for replication that probably zeroes out if you are missing an important replication component, as opposed to simulating how those genes work at a low level) this is probably somewhere close to as good as flipping a coin.
I think an interesting number would be how many atoms does it take to simulate one atom reasonably well. Then you could get correct results for some ( really ) small biological functions.
It's hard to say which is the better approach. On the one hand you have MD from which you could technically build an accurate model "from scratch", folding all of your own proteins as part of the simulation, but it is exceedingly expensive to do that. On the other hand, you have a model built at the level of biochemical reactions and molecular biology, which is computationally feasible, but then you don't have enough information from bench work to flesh it out.
Massive waste of cycles. Never gonna make a useful prediction.
Reduced representations that contain the minimum of complexity to express the phenomena of interest are far more useful.
I'm no molecular biologist, so i'll happily agree if you say there are some fraction of water molecules broken down, or formed by various cellular processes, but that must be a tiny fraction of the total, and could just be special cased.
It just seems like you're willfully ignoring everything physics has to say about how the universe works so you can get a big number for your calculation.
My understanding of "solution" is stuff floating in water. If there's some value in breaking out each atom, i'm all for it, but i kind of suspect the spherical cow is good enough in that case.
Even if i'm wrong there, each atom would introduce at least 6 degrees of freedom. IIRC, water is slightly polarized, and that's likely relevant to at least some reactions. Replacing the 18 degrees of freedom with 6 to represent a rigid water, or perhaps 10 or so to represent a flexible version, modeling as the molecule will dramatically lower complexity.
In that case, don't even bother with trying to work out what it would take to mathematically model biochemistry yet. Go study some basic physics and chemistry for a few years instead.
I am not attempting to be insulting, I am attempting to point out to you that you need to go and learn a hell of a lot more about the sciences than you seem to know at the moment if you are going to have any sort of productive discussion about the technical aspects of modelling biochemistry.
http://en.wikipedia.org/wiki/Properties_of_water
http://mgl.scripps.edu/people/goodsell/illustration/mycoplas...
I don't understand why it's easier to simulate each individual atom than it is to simulate things with structure. I would assume a simulation would regard the long yellow strands of dna as dna rather than the component parts.
I'm used to a world where having structure implies constraints. With constraints, simulation can avoid a bunch of special cases and is therefore faster. I find foobarbazqux's approach of simulating each atom upsetting. But that's ok. You guys seem to think it's the way to go, and you all get results. shrug
-edit-
I don't think simulating water is easy, but i believe it's likely easier to simulate water than it is to simulate all possible states of the constituant atoms.
That's not to say that you can't get useful results by treating solvent as a continuum (or even ignoring it entirely), they'll just have a different kind of explanatory power.
[1] http://www.sciencedaily.com/releases/2013/07/130728134055.ht...
Mathematically, we could figure out how much computing power we would need to match the human brain (24 bytes per synapse * number of synapse, etc).
I think I would be more interested in what was the specific of their experiment.
- What kind of software were they running?
- Was the software bug-free(yeah right)?
- Was the software optimized?
- Caching?
Simulating a lump of tissue is definitely useful, it's when they start talking about simulating a complete brain that I get lost.
A brain doesn't happen overnight. It is the result of a long developmental process encompassing embryology, learning and experience acquisition.
How do they plan to wire the whole thing?
Micro- and macroscopic connections are well understood, it is the so-called meso-scale (in between) that is troublesome.
There will never be a non-destructive way to do extract the information of a live brain, and I'm not sure there will ever be a way to extract it at all. The current methods[0] allow to extract either the wiring or the genes expression.
[0] http://www.brain-map.org/
If they simulate cortex, they wire arrange it into cortical columns (size of each column is between 50,000 and 100,000 neurons and there is over two million of them). Those columns have many different layers and neuron types. Understanding how cortical columns and minicolumns work is very important.
> So they wire it like the brain is wired.
As if anyone knew how to do that. Even assuming that you can do both accurate micro-tractography and extract cell-by cell epigenetic information out of a single column, you'd still be missing other cricial information like individual synaptic strength.
I'm not talking about cortico-thalamic connections (required to model epilepsy) inter-columnar connections, and whatever regulation happens in the white matter.
Relying on the probability distribution of connections as a crude approximation is useful, but far from the real deal.
Your real deal has nothing to do with current brain science and these simulations.
With current brain science, call me perplexed. Could detail were you think I'm off base?
From the press release (emphasis mine):
"The nerve cells were randomly connected and the simulation itself was not supposed to provide new insight into the brain"
My research is in molecular dynamics. I simulate systems up to 1 billion atoms on Kraken & Titan. HUGE approximations are made in simulating these systems, but depending on what exactly it is you're studying, these simulations still provide useful results. That is the key to all these studies: how well does your approximation reproduce whatever it is that you're attempting to model? In some cases, very well. For instance, I'm not going to get exact energy levels of a large system, but the system will qualitatively evolve in the same fashion that the experimental system does (which then guides the experimental counterpart to the research). I don't know the details of this brain simulation, but there is certainly some aspect of it that is not being reproduced anywhere close to real-life, and hopefully this isn't what they're interested in (and I'm sure they know that, but I don't think the article author does).
The very best simulations of reality that we can perform handle at most just a few H/He/C atoms. And an issue called the fermion-sign problem means that the computational power necessary to simulate larger systems scales exponentially with the number of particles. Unfortunately what that means is -- short of developing quantum computers (which are polynomial order for the sign problem) -- we aren't ever going to simulate more than a few atoms with near-perfect accuracy, and certainly nothing like a human brain.
EDIT: Didn't mean for my comment to sound so negative. Obviously, the researchers know exactly what they're doing. I was just trying to dispel the impression that we're close to simulating the human brain.
To date nobody has come up with a really compelling disproof that they are more or less correct- thought and mind seem to be best explained as "emergent behavior of a complex system".
Now, whether can we bootstrap a thinking mind by running complex simulations on computers remains to be seen. I'd like to believe there is some interesting physics going on in brains we can't simulate with straightforward physical models, but there really isn't any good evidence for that.
Can you elaborate on this class of 'something'? There are intelligent behaviours that can be reproduced this way, so without more specifics I feel like you're making the same mistake you're condemning.
http://shahamlab.rockefeller.edu/pdf/CTDB_69_C_69003.pdf
Worse, I don't think you're prepared to prove that a mostly pure-neuron simulation is incapable of giving rise to that certain complex quality apparently far removed from the level of the simulation.
No, questions can be answered by anyone, if they provide a relevant answer.
What the person making the claim thought is irrelavent. The parent gave you a valid reply.
So, basically, you're the only one being an 'asshole' here.
From my observation, it appears people assume "reductionist" means "simplistic" as opposed to "representing a complex system as no more than the sum of its components". To be honest, I think it's just an etymological problem: "reduction" sounds bad, so "reductionism" sounds bad.
The grandparent here follows this pattern. Using the actual meaning, it would be paradoxical for an approximation to be reductionist. In fact, it's quite the opposite. A reductionist would hold that the most accurate high-level model of a brain would be the most low-level model of a brain. If we could model each atom precisely then we would, necessarily, model the whole brain precisely.
The whole thing is pretty amusing when you take into account that anyone who manipulates software or systems in any serious way needs to engage in reductionism to be able to work. It's not as if "first, we write module A, then module B, then a system magically appears from the ether" is a viable architecture. At least, not since we stopped taking neural nets seriously.
Otherwise Sperm Whales would simply be much smarter than us.
About conscience itself, my guess is as good as yours.
On a deeper level, we still have trouble simulating even a single proton.
http://en.wikipedia.org/wiki/Lattice_gauge_theory
Hmm. I need to research this some more.
I mean, my doctorate involved calculating tiny, tiny corrections (8 or 9 decimal places) to hydrogen-like energy levels, but even there I could still treat the proton as point like!
An overwhelming amount of neuroscience research data points to the neocortex reusing the same functional unit composed of 100's-1000's of neurons called a cortical column. For example one patient was able to regain her balance by rerouting the signal from her vestibular system to her tongue. So the cortical column is the layer of abstraction we should be shooting for.
Hopefully we'll move away from the 1950’s style perceptron, whose mathematical simplicity has attracted researchers but has been a major distraction to AI as they share almost nothing in common with their biological counterparts. If we don’t want to see a 3rd AI winter, we need to move to more biologically inspired approaches.
Disclaimer: this is completely out of my field - just a layman speculating.
Is there any research on what other chemical/electrical interactions might be necessary to simulate the brain?
How many gears from a mechanical computer does it take to simulate a micro processor?
If we were to one day to closely simulate the brain, I am sure the method will not involve computers as we know them today.
Imo brains themselves are like mechanical gears compared to the potential of processor.
Perhaps a more interesting question is whether or not the brain can really be simulated by a classical computer at all. If quantum mechanics plays a role in the function of the brain we will need a quantum computer to do the job. A classical simulation of a quantum brain would be a very odd beast indeed! It might exhibit behavior that seems like intelligence but somehow falls short.
http://en.wikipedia.org/wiki/Quantum_mind
(http://watarts.uwaterloo.ca/~pthagard/Articles/quantum.pdf)
Massive array at 8nm and there's your positronic brain.
See http://www.stanford.edu/group/brainsinsilicon/about.html
Maybe the simulation algorithm is the thing that needs improvement. Some day one finds a way to reduce the complexity an order of magnitude, then the era of human brain is over.
Not at all. A classical computer can accurately simulate a quantum computer, albeit slowly.
http://izhikevich.org/human_brain_simulation/Blue_Brain.htm