Can we make a fake brain?

15 October 2013



How far away are we from being able to simulate a whole brain in software?


We posed this question to our brain panel. Bill - Well, we've been simulating neurons for a long time. the question is, how good are we at simulating a particular neuron and its fullness. It depends on which neurons you're looking at. All the neurons are different from each other and people have been trying to simulate different neurons and we can do it pretty well, but never to completion. The same thing with the brain - people simulate pathways in the brain, work out models for how information goes from one place to another and it's getting more and more sophisticated, but simulate a brain, which brain are we talking about - a frog brain, a human brain, you're brain? That may be more and more challenging.

Mike - There is a lot of interest now in trying to model whole brains but using super computers. So, there's a lot of funding just been awarded in Switzerland which is called a Blue Brain Project where people are trying to use very powerful super computers to model the whole brain. We're really at the beginning of that and I think it's going to be many years before useful data arise out of that project. Bill - When you say useful data, what do you mean by that? Because I think every model produces some data that's at least kind of hypothesis for what it might be doing so it's testable. So, models are valuable as far as they go and they throw up problems with our current understanding. So, I might argue that the Blue Brain project, even in its early phases will give you answers that may ring true or may not, and that lead to further sophistication of the model.

Mike - I think that's a good point. To be honest, I'm slightly sceptical about whether this is a good use of money and whether I would fund this or not. I think if it was up to me, I would probably fund far more slightly less ambitious projects. So, I'm actually slightly sceptical about how useful this will be.

Hannah - I mean, do we know enough about the brain in order to have the knowledge to simulate it in the first place?

Katie - So, there's a number of large multimillion pound projects going on at the moment and they're trying to map what's been called the human connectome, so using some of the newer imaging techniques that we've got to look at the brain to look at both the structural and the functional connectivity between different areas. We know quite a lot about this in some pathways and in some sort of networks within the brain, but we don't know in its entirety exactly how all areas of the brain work together, how the patterns of activity in one set of pathways might influence a pattern of activities in another set of pathways completely in all its complexity. Obviously, if we could, it would be a very useful way to perhaps be able to model what we might expect to happen in the use of certain drugs or certain medications in development.

Bill - Let's say you look at a circuit in the spinal cord that's involved in walking. People have been studying something like that for quite a long time and understand to a medium degree of understanding I guess, the neurons that are involved. But it's not by any means complete how we walk faster or slower, how we change gait from walking to running, how we jump and hop and skip, what's the circuitry involved, what are the neurons involved? We don't even know those in a very small part of the brain. Similarly, when light goes into our retina and all the cells are connected up in the retina. We don't know what all the cells are doing. We know what some of them are doing, many of them are doing, but completion of this task, how far away from completion of the task is kind of hard to say.

Katie - I think that just because we don't have like fairly complex and in-depth models of a fully functioning brain in terms of software doesn't mean that people should think that we don't have useful models. So obviously, there's entire fields of neuroscience that are largely devoted to using software and using it to model the way that the brain works and that both helps us in developing theories that we can test and also, taking the things that we have learned and putting them together in trying to work out exactly what is going on by running different iterations of them.

Hannah - Along that line, there's a paper that was published actually last month. So, published in Nature, researchers at the Institute of Molecular Biology in Vienna, Austria managed to develop a 3D organ which resembled a 9-week embryonic human brain. They developed this from a human stem cells and it grew from 3 to 4 mm cubed. They think that they kind of seem to mirror some of the activity of a functioning kind of brain-oid they called it and may be useful in the future for trying to uncover more about Schizophrenia and autism for example.

Bill - I'm not sure what this little mini brain in culture has to do with modelling, simulating a whole brain. I'd take it has to do with getting a mathematical model that predicts the interactions and the activity of different things. This little mini brain made out of embryonic stem cells certainly has some wiring to it as bunch of neurons in a culture dish would if you put the dish together. And it may be a little bit more alike a real brain than a bunch of neurons in a dish that are randomly connected to each other. But that's such a far away from a real animal's brain or a human brain.

Hannah - But do you think some of the information that could come from this brain-oid could help to input into the computer simulations?

Bill - No. I would study real brain rather than little mini brains. Each one of which is going to be different from each other.


Add a comment