19 replies to this topic

[Singularity2045]

Hello all,

How can it be that there are only 8,000 neurons in the "Blue Brain" project -

http://www.seedmagaz...o...age=all&p=y


When the human brain contains about a 100,000,000,000 neurons, how do they want to simulate it with only 8,000 ?

Please explain, thanks.

[Heliotrope]

8000. yeah the number sure seems low, we've far to go
like you say

Brains can be extremely complex, and human brain contains roughly 100 billion neurons!!! linked with up to 10,000 synaptic connections for each neuron , so many connections.

Then add glial cells whatnot, i believe i read brain has 10^14 or 10^15 cells.

i think what they simulated is like half of a mouse's brain lol. hopefully to simulate human brain, don't need all those trillions of components, but enough to get a core identity/personality, then add some for memory.

[niner]

When the human brain contains about a 100,000,000,000 neurons, how do they want to simulate it with only 8,000 ?

Please explain, thanks.

As an example of the principle, a drop of water contains about 10e20 molecules, yet you can simulate the structure and properties of water with fewer than 1000 molecules in the model. When a simulation is complex and expensive, you simulate the smallest number of objects that can reproduce observable properties.

[Singularity2045]

OK thanks all for the answers, I read a little more about this project and I found out that at this point they only try to simulate small part of a 2 weeks rat brain, later they will enlarge it to simulate the whole human brain, which they plan to complete till 2015, amazing isn't it?

We'll see.

[Benjamin]

Hi Singularity2045,

What Markram et al. are doing is to simulate a Cortical Column.
This is a very basic functional unit of human brains. The detail of simulation is very high, to make it possible to resemble it's
natural counterpart as exact as possible. (The parameters undergo constant modification, based on experimental data.)

Once the simulation is found to be sufficient, it could be broken down into a lower detail simulation, of a greater part of the brain.
E.g. different columns in concert.

Note: Currently one processor of BlueGene L (the supercomputer used in the simulation) can only simulate two neurons in real time.


Greetings,
Ben

[Singularity2045]

Thanks very much Benjamin, also an amazing thing is that Markram has stated that he want to simulate the whole human brain in 2015 !! much before the singularity in 2045..... the funny thing will be that eventually the singularity will come much sooner than expected..... ......

"Markram's team is now digging deeper in to the molecular level and will eventually build up an entirely reverse-engineered model of the human brain (aimed for 2015)."

http://lis.epfl.ch/r...n-robotics.html

Edited by Singularity2045, 09 December 2008 - 05:21 PM.

[Benjamin]

...which leads us to the question, if remodeling the human brain(the Kurzweil approach) is a possible and safe way
to singularity.

I doubt it.


Greetings
Ben

Edited by Benjamin, 09 December 2008 - 07:47 PM.

[mentatpsi]

When the human brain contains about a 100,000,000,000 neurons, how do they want to simulate it with only 8,000 ?

Please explain, thanks.

As an example of the principle, a drop of water contains about 10e20 molecules, yet you can simulate the structure and properties of water with fewer than 1000 molecules in the model. When a simulation is complex and expensive, you simulate the smallest number of objects that can reproduce observable properties.

You make an interesting point, but at the same token i kind of wonder... given the interconnectedness of the different brain regions as well as the effect the rest of the body (such as the heart) has on the brain, could you really extend the same properties of a small region/sample size to try to get an accurate gauge on the region. What i'm saying is... lets say you have a group of points, and you're trying to plot them and develop a curve fit... now in the range of -10<=x<=-5 the y reaches to 700, while the normal points are around 3... with a standard deviation of 10 (not including that range of -10 to -5). The equation you'll get is much different if you consider this range or don't. Extended to the problem, the algorithm one would develop to represent a small region while not accounting for neurons in another location, would produce variables which do not consider the whole system and would not be accurate in truly simulating the system.

But i really don't know too much about simulations... that would just be my guess if i was trying to simulate a system.

Edited by mysticpsi, 21 January 2009 - 06:24 AM.

[mentatpsi]

plus i'm pretty sure if you consider one molecule of H2O vs a droplet of water based on avogadro's number, the physical properties wouldn't be entirely different other than the way they act with one another. The things that would change would be certain magnitudes with the particles. I'm not too familiar with that area, but the simplicity of a droplet of water compared to the brain is rather... well... ya.

[simulation11]

Very good discussion. Thank u so much for sharing

simulation rachat credit

[quelareine]

very informative!
simulation rachat de credit

[Johny Nicoline]

This sounds interesting!

simulation credit

[monkeyking]

Hello all,

Just want to say thank you for such a wonderful information, it was really cool!

Thanks,
monkeyking
simulation assurance vie

[zawy]

I think Kurzweil and others are greatly underestimating computers when comparing them to brains. For example, my computer can translate a document 100,000 times faster than me. To do that, each word has to be categorized by each of us. Granted, I have a much larger concept of the meaning of what I am reading, but 50% of my brain's effort is not doing anymore than what the computer is doing (imagine how large the region is that is just determining the shape of the letters). So for that 50% of my brain, my computer is 100,000 times faster. The same goes for any raw math or other logic skills. To do my higher brain functions, the computer will need to simulate only the neocorticol columns in my front brain, maybe 20% of my brain which would be only 400,000 cortical columns (a group of neurons). I don't see why efficient programming on a basic PC can't simulate a cortical column, so that only 400,000 computers need to be hooked up over the internet to model a brain. What I am describingwill be as slow as a brain due to slow internet packet speeds. But it has 24 hour capability (3 times a workday) and direct access to all of wikipedia and google, not having to wait on fingers to do the typing nor on eyes to do the reading. With perfect memory wherever needed. No human has more than maybe 30% raw thinking ability than any other human. But imagine ten times as many of these cortical column computers connected (4,000,000). That would be 1,000% more raw thinking power than a person, but with the added abilities of perfect memory, typing and reading 100,000 times faster with immediate access to the same inifinite memory and knowledge dataset that we use (google and wikipedia), hacking web sites, and bank websites, running as a screen saver on everyone's computer, with the ability to infect every other computer. So I think the singularity could be here and now if the programming were advanced and in place. There's no way to know if it's not already here. Viruses, Aliens, AI, and other indifferent entities you can't see are the most dangerous ones. As with any virus, AI does not need our approval for anything. But our complacency will speed it along.

[Norlene]

Wow! fantastic information zawy. Thanks a lot man. Keep up the good work.


Simulation pret

[Shoe]

I see five users in this thread that I suspect are spambots, or something similar. Or perhaps it's Blue Brain that has come to life?

[Norlene]

I see five users in this thread that I suspect are spambots, or something similar. Or perhaps it's Blue Brain that has come to life?

Hi Shoe,

Did you include me in your counting as a spam bot? If yes, then you are wrong. The reason why I posted in this forum is because I just a appreciated this project. The blue brain project went smoothly way back then. And that's it.

[Shoe]

I see five users in this thread that I suspect are spambots, or something similar. Or perhaps it's Blue Brain that has come to life?

Hi Shoe,

Did you include me in your counting as a spam bot? If yes, then you are wrong. The reason why I posted in this forum is because I just a appreciated this project. The blue brain project went smoothly way back then. And that's it.

Yes, I did include you. Sorry. I still suspect the other four, though.

[truelolor]

hi you guys. May I join this forum?
Thanks

simulation credit auto

Edited by truelolor, 02 June 2009 - 07:24 AM.

[Shoe]

WTF? Another one!