0:00

Let's go to the 5th frontier. What I call the computer simulation of

neuronal circuits or computer, computer or theoretical modeling of the brain.

As I said in the beginning, my claim, the claim of many scientists, as it will be

able to integrate data, all levels of data.

And also, the anatomical and physiological data and set it on the

standard system, we will need some theoretical tool, mathematical tool.

And I'd like to site a, a very prominent scientist, a very, very prominent

scientist, William Thomson or Lord Kelvin, a great physicist and

mathematician. He said the following, I'm never content

until I have constructed a mathematical model of what I'm studying.

If I succeeded in making one, I understand.

Otherwise, I do not. that's a very powerful claim, that

understanding of complex systems, physical system or others, a complex

system, biological system, requires mathematical formulation of the system.

Not everybody agrees to that. Not always you need the mathematics to

say I understand. But in terms of complex systems, I

totally, absolutely, of course, agree with Lord Kelvin.

And this is what brought us to the blue brain project.

which is a project that I will discuss later on.

But here, the idea was the following. As the idea was that we have so much

data, we have a lot anatomical data, we have a lot of physiological data at the

circuit level. How do we integrate it all together?

What do we do in order to put this data, that we already have not complete?

We need still to work very hard to get the whole data.

We need to connect the whole brain. But what do we do?

So, the idea was to take a large computer, here is the Blue-Gene Computer.

It's blue, by the way, because IBM used to have blue colors as it's own color,

today it's black. But still blue, so the term blue comes

from the IBM machine that we are using. and blue brain has its all very good

connotations, but it's also blue because of the IBM.

And we're now using a very powerful IBM machine.

It's now the EPFL Lausanne, in order to model, to simulate cells.

But, what does it mean to model a cell? So, suppose somebody gives me a cell.

This is a nerve cell. We'll talk a lot about the nerve cells

because they are the unit. You give me a cell and you tell me that

when you record from this cell, it has a particular pattern of firing.

Let's say this is the case. So, this is the real recordings from this

particular cell in isolation. So, this cell likes to do tu, tu, tu, tu,

tu ,tu stop tu, tu, tu, tu, tu stop. We call it a burster.

So, the type of cell that is called the burster because it bursts with spikes.

I need to write mathematical models, and I'm not going to do it here because there

would be a particular lesson to do the mathematics of spikes.

But I want eventually, to write an equation or a set of equations, to

describe this electrical activity. So, after writing these equations and you

learn about these equations, you will see that I can replicate mathematically.

I can replicate the activity or closely replicate the activity of this particular

cell. This means that I have a mathematical

model of this cell. And if I have another cell that fire

differently, I need to write another equation.

And so, I build different equations to describe different set types.

This is what I mean, to model mathematically the cell.

Then, I can start to assemble, to put the cells together and this is what we do.

We take the supercomputer, each processor is now solving mathematical equation for

this cell, and another processor for another cell.

And then, we put them together, and also connects them together anatomically.

And so, we need to model the connection too, not only the activity, but also the

connection. And eventually, I have in the computer, a

model, a simulation of the system that I want to simulate.

It looks something like this. So, this is a real, a real graphical

demonstration from the blue brain project of about 10,000 cells in a piece of about

two cubic millimeter of cortex. In this case, of a rat.

You see the cells modeled in the computer.

You see all the wires. The axon and dendrites that we'll talk

about. And so, this is a realistic replica.

It's a realistic copy of the complexity of the anatomy of the cells in the

computer. But this is just the anatomy.

I need also to show the activity. That means that each cell has to really

replicate its electrical activity that I just showed you before.

This is just to show you the jungle in your brain, how complex a piece of a

brain looks like. This is the blue brain project.

So, let me show you the activity. So, you see the same piece of brain that

you saw before. We call it cortical column.

The same piece of brain, but now this piece of brain in the model is now

active. So, in this case, you don't see the

spikes, but each time it's red. It's color coded for spikes.

So, red cell means that it fires a spike. Blue cell means that it did not fire a

spike here. So, you can see that the network starts

to, to act. This is a simulation of 10,000 cells

which is a very small number relative to the top number.

But it is a beginning. It is a beginning of a simulation of a

whole brain. But why do I want to simulate a whole

brain? Why do I want to do this modeling?

As I said, I think that we will be able to understand the network.

And when I say understand, I mean that when I see a particular activity, which

is a result of many, many, many interacting elements, each one with its

own music, so to speak. So, this is doing ta, ta, ta, this is

doing ta,ta,ta, ta. Yes, everyone is doing another ta, ta,

ta, ta. But eventually, the network as a whole is

generating a whole behavior, a wave of activity, maybe a sleep activity, maybe

activity related to Parkinson's. Maybe activity related to another

disease. Maybe activity related to happiness, to

emotions. How do these activity merges and what

goes wrong when an activity becomes Parkinsonian activity?

What goes wrong? I believe that this type of modeling

technique, where you have a handle on each parameter, because you built it.

And you can manipulate the parameter, so your computer will become sick with

Parkinson. Your computer model will become sick with

Alzheimer's. Then, I will be able to repair it in the

computer. And then, come with ideas, sophisticated

ideas, well-established ideas, based on what we did to develop medication or to

manipulate this particular region or this particular cell.

Maybe with optogenetics, maybe with other tools.

So, this is what we call today, simulation based medicine and simulation

based research. The research is based on a model that we

built, on a computer that we built, the brain built computer, the brain built

models, the brain built mathematics. In order to integrate an anatomical and

physiological data to build a replica of a piece of a brain, in order to be able

to eventually fix the replica and extend the replica.

And then say, through this mathematical exercise to understand the brain.

So, that's computer based simulation of the brain.

And that, I think, we are going towards this, in our information age.

So, thank you for the first lesson. This is the end of it.

And in the next one, we'll go to study the ingredients.

The next lesson will be about nerve cells, about the synapses, about the

signals. You will go into the brain, into the

brain and start to learn more and more about the ingredients of this system.

Thank you very much.