Are these researchers really making progress? The full text of Markram's letter was too long to print here, so I'll just summarize the science behind his vitriol. In short, Blue Brain is composed of model neurons that are highly sophisticated in their handling of electrical and chemical signals. They are more faithful to real neurons than are the model neurons of Modha's simulation, which in turn are more realistic than the weighted voting model discussed in this book.
There is plenty of empirical evidence that the weighted voting model approximates many neurons well. But we also know that the model is not perfect, and can even fail badly for some neurons. Markram is correct that real neurons have many complexities that are not captured by simple models. A single neuron is an entire world in itself. Like any cell, it's a highly complex assembly of many molecules, a machine built from molecular parts. And each of these molecules in turn is a minuscule machine made of atoms.
As I mentioned earlier, ion channels are an important class of molecule, because they are responsible for the electrical signals in neurons. Axons, dendrites, and synapses contain different types of ion channels, or at least have them in differing numbers, which is why these parts of neurons have distinct electrical properties. In principle, every neuron is unique in its behavior, owing to the unique configuration of its ion channels. This is a far cry from the weighted voting model, according to which all neurons are essentially the same. But it sounds like bad news for brain simulation. If neurons were infinitely diverse, how could we ever succeed at modeling them? Measuring the properties of one neuron would tell you nothing about another.
There is one hope for escaping the morass of infinite variation: neuron types. You may recall that Cajal [neuroscientist pioneer Ramón y Cajal] classified neurons into types based on location and shape. You can think of these properties as being like an animals’ habitat and appearance. When a neuroscientist speaks of the double bouquet cell of the neocortex, it reminds me of the way that a naturalist speaks of the polar bear of the Arctic. The naturalist might also point out that polar bears, unlike brown bears, hunt for seals. Likewise, neurons of the same type generally exhibit the same electrical behaviors. This is presumably because their ion channels are distributed in the same way.
If this is the case, then neural diversity is actually finite. We should compile a catalog of all the neuron types, a “parts list” for the brain, and then construct a model for each type. We’ll assume that each model is valid for all neurons of that type in all normal brain, much as we assume that all resistors behave the same way in any electronic device. Once all neuron types have been modeled, we’ll be ready to simulate brains.
Markram's laboratory has characterized the electrical properties of many neocortical neuron types through experiments in vitro, and constructed a model for each. Each neuron type is modeled as hundreds of interacting electrical "compartments;” which is an approximation to simulating the millions of ion channels in a neuron. Markram deserves credit for the realism of the multicompartmental model neurons used in Blue Brain.
But Blue Brain [which has now morphed into the Human Brain Project] is severely lacking in one respect. Since no cortical connectome is known yet, it's not clear how to connect the model neurons with each other. Markram follows Peters' Rule, a theoretical principle stating that connectivity is random. The accidental collisions of axons and dendrites in the tangled "spaghetti" of the brain lead to contact points. At everyone of these, a synapse occurs with some probability, as if it were the outcome of tossing a biased coin.
Peters' Rule is conceptually related to an idea introduced earlier, the random synapse creation of neural Darwinism. The ideas are not equivalent, however. Neural Darwinism includes activity-dependent synapse elimination, which makes the surviving connections end up nonrandom. Violations of Peters' Rule have already been discovered. I suspect that many more will be found, and that the rule has managed to survive only because of our ignorance of connectomes.
See what we're tweeting about
10 Comments
Add CommentFirst of all, I would like to say that I learned a lot from the article about the current problems of neuronal simulation, thanks a lot!
Reply | Report Abuse | Link to thisHowever, I think it's neccessary to say, that claiming to "simulate a cat's/mouse's brain" is not a philosophical question of what a brain is, but simply not possible, as was pointed out 3 pages later in this same article. Therefore, I would support Markram in his opposition, although I would choose different words.
Apart from that, the article gives the impression, that the only purpose of the "Human Brain Project" is to provide a complete simulation of the human brain. Certainly, that was never the intention but rather to provide a reduced, yet realistic model or am I wrong?
Greetings!
I get the impression from the article that a lot of people believe that, “You are your brain”, and/or “You are your connectome”. Let me explain why I think this is an incorrect assumption.
Reply | Report Abuse | Link to thisIn information systems there is a concept called software, which means that the system has a memory that contains information (like a computer or a cortex). If you don’t know that your system is running a software system, it will seem to behave in utterly inexplicable ways to the hardware experts (biologists) who will attribute the complex behavior to some hardware element that they don’t quite understand (like the connectome). The connectome is a hardware detail, not software.
The brain and mind are separate entities. The human brain is just a computer made of biological hardware. The mind system is a software system being processed by the cortex. That software is stored as weights on the neurons, not somehow embedded in the brain, or in the connectome designs. So neither the brain nor the connectome is “you”. You are the sum of information stored in memory (the learned software system). The Home sapiens version of the software system that develops over someone’s lifespan is what psychologists study. When you are born you have a brain, but no mind, because only a trivial amount of learned information has been accumulated, and it takes years to make a person.
If what I’m saying is true, then uploading a mind is a much easier exercise, because you just need to read out the memory contents, and process the mind software on a more robust computing platform.
The Openworm project is working on the C. Elegans connectome. This document contains the excitatory and inhibitory neurons. The Google docs file is attached. Search for Openworm if the link does not work. https://docs.google.com/spreadsheet/ccc?key=0Avt3mQaA-HaMdHZuZnFuZmI5Q1VRU0VMekZ5d1QyZVE#gid=0
Reply | Report Abuse | Link to thisNot everyone's neurons are going to be identical, methinks. First consider people who have brain damage, and second consider that brains do not all develop identically. How much of development is neurons and how much is connectomes (or something else, if connectomes don't include the 'weights' jsweck was referring to), I do not know.
Reply | Report Abuse | Link to thisSuppose a simulated brain passes the Turing test, how then do we know we've managed to simulate a human brain? I admit, this may be more of a philosophical question than a computational one, but still...
Reply | Report Abuse | Link to thisBut philosophy aside, I've always thought that to be able to properly simulate something, you'd first have to know exactly just what you're trying to simulate. After all, how else are you going to know whether you've succeeded? And these scientists are telling us that they do know how the human brain is wired together and how it functions? They can come up with objective metrics that tell us about succes or failure? I am very impressed.
The problem with seeing it as a "Software" vs. "Hardware" is this: At some level, all software is stored by a physical change in hardware.
Reply | Report Abuse | Link to thisWhether that's transistors set to a particular position or the specific pattern of grooves in a DVD, "software" is not physically, intrinsically independent of hardware like you say it is. Software must change the hardware in order to store and process information.
In this case, the argument is being made that the connectome may be the intrinsic "hardware" marker of "software" on the brain. Like the grooves on a DVD, changing the connectome changes the information stored. By using your analogy, you are born with a brain but without a mind because the connectome is random, and the "software" is random nonsense. As it is refined and becomes non-random you "learn" and your neurons are programmed to interact in certain, specific ways.
You can see the connectome as "software" but at some level the software/hardware dichotomy becomes arbitrary as software is itself encoded on and processed by minute physical changes in hardware.
Yes, software must ultimately be stored in some physical way in a memory system, but the software system is almost completely independent in operation. It is not some high level hardware abstraction; it’s a separate and independent system in memory.
Reply | Report Abuse | Link to thisTo see this, think about battling monsters in a computer game. What exactly is the nature of the monsters? It’s not the billions of randomly flipping logic gates in the computer; it’s a piece of software in independent operation, and there’s nothing random about its operation. All you need to have software is a memory system – the stuff you store in that system is always software. The cortex is primarily a memory system, so it must have a huge software system. When we learn, we are adding to those software data structures.
The reason all computers exist is to solve problems. But you can only solve the simplest computational problems with logic gates or neurons – there is no way you can solve the big stuff, unless you create a memory system with those components. Then you simply move informational versions of the problem elements into memory, compute a bit, and the problem has the potential to be solved. So that having a memory system is the key to solve all big problems. In the computer world the only reason the hardware exists is to serve as the basis for the software system, and that software system is the one that solves all of the significant problems of the system. When the software is active the hardware people see random behavior, because they are not in control of the system.
I don't see how to avoid the idea that every neuron is actually unique, even if they also fall into subtypes. A single neuron fires when a person sees a picture of a particular person, and not when the person sees other people. Those neurons must be unique just by the memories stored in/by them.
Reply | Report Abuse | Link to thisSuppose you try to simulate a computer that has no information stored in it? What would such a computer do? Maybe you can have fun watching the various connections light up randomly, but you aren't actually simulating a computer. You aren't simulating any actual thought process or transmission of information.
It seems a bit like making a map of every road, highway, driveway and building in a country, and then simulating traffic patterns on them. Does that tell you anything about what is happening inside the houses, what the people think, or what the decisions of the government will be? Suppose you add all the phones, internet, cable networks, etc., and simulating electrical signals on them. Does that tell you what the people think or talk about? No, you have to model the components that remember and process memories and make decisions and then send that information out through the various forms of communication. Those components would be the people.
What I am saying is that until you know how neurons store, process and transmit information, and actually include the information itself in the simulation, you can't simulate a nervous system.
RE: Beware of pseudoscientific pretensions in Brain-reductionist cognitions! (Continued from above)
Reply | Report Abuse | Link to thisFrom a practical science and philosophy perspective today: Reductionism -- or specifically Physico-reductionism -- has its roots in our innate analytical thinking and inductive reasoning of physical facts and observations on Nature -- the Physical World -- since the primeval epistemological claims of the several pre and post-Socratic philosophers: Among whom, Thales (624-546 BCE) being the first who proclaimed that all entities come from water; and then, Anaximenes (circa 6th century BCE) who assumed that all things are composed of air; and centuries later, Epicurius (341-270 BCE) who intuited that all entities are made of atoms; and of course, of no lesser significant in his imagination, Aristarchus (circa 3rd century BCE) who was the first to pronounce that the sun is the center of the universe, and that the earth moves around the sun -- almost 2 millennia before Astronomy was first established as a Science (Physical Science) by Galileo Galilei (1564-1642) during the Renaissance in Europe and beyond!
Fast forward to the 21st century: The latest antithetical (or anti-theism) reductionist-cosmologist pretension, is one that has been recently expressed by Stephen Hawking: in which he boldly declares that our Cosmos (or Universe) is created by Time and that Time is created by the Big Bang; and therefore, no theological imaginations or inspirations shall be necessary in his field of Physico-reductionist cosmology* and in which, Hawking has in fact (unknowingly and utterly) negated his own fundamental epistemological inquiry and reductionist speculation: What has had caused the Big Bang -- if there was indeed a Big Bang (nor God) at all!? Thus, Hawking should have had not irrelevantly or immaterially raised nor hackled Theology in his field of universally Physico-epistemological inquiry at all! -- [*See my comment on Physico-reductionism in cosmology today here: http://www.scientificamerican.com/article.cfm?id=the-consolation-of-philos&posted=1&posted=1&posted=1#comment-110 -- "The Consolation of Philosophy -- RE: Physico-reductionism at its Best: When in a miraculous shebang Nothing became Everything in our Dynamic Universe! -- Or, could physico-reductionists (including evolutionary sophists and neo-Darwinists alike) be able to self-examine and heal themselves of their socio-reductionist-bias (or hubris) syndrome!?" (ScientificAmericanUSA; May 5)] (To be continued below)
I think these people simply don't get it how tremendously complex nature is. The naive belief that funding a super supercomputer feeding it with several complicate rules won't work. It will not meet the expectations, as it was the case with the human genome project. It would be much better to use that money for several smaller researches than throwing it into another disappointing mammouth project.
Reply | Report Abuse | Link to this