3 ms·

Will this project "emulate" a human brain down to the subatomic level? Down to even the atomic level? Both are probably far too ambitious for this project. I

by gnosis 15y ago

Will this project "emulate" a human brain down to the subatomic level? Down to even the atomic level? Both are probably far too ambitious for this project. I'd even be quite surprised if they managed to accurately simulate a brain down to a molecular level. What this sounds like is yet another attempt at a grossly simplified simulation of a subset of what is actually going on in the brain. It remains to be seen how useful this is, nevermind whether this will bring us any closer to AI (how would you even measure the "intelligence" of a simulated brain without any sensory organs or an ability to communicate?)

a-priori 15y ago
Will this project "emulate" a human brain down to the subatomic level? Down to even the atomic level? The short answer is that each part of the simulation, from ion channels and synapses to cell bodies to network topologies and cell distributions, is done at the lowest level that people in the laboratories can effectively isolate and study. From that they can derive mathematical models of each part, and integrate those models to simulate them interacting over time and space. Is it a simplified model of the brain? Yes, but its the most detailed model that science can produce (or at least it aims to be). how would you even measure the "intelligence" of a simulated brain without any sensory organs or an ability to communicate? What makes you think that's difficult to simulate? It's been shown that primate brains are capable of learning to interpret novel inputs, such as a grid of electrodes 'painting' an image onto cortex. There was also a study where they implanted electrodes in a monkey's brain where the rate of activation of the cells surrounding the electrodes moved a cursor on a computer screen. The monkey learned to move the cursor. This shows that simulated inputs and outputs wouldn't have to perfectly correspond with reality in order for the brain to be able to process it.
roel_v 15y ago
Does it matter? As far as I know, we don't know yet what level we need to simulate at to simulate consciousness. If it's at the molecular level and we achieve that, who cares about the atomic level.
- possibilistic 15y ago
  It likely relies on subatomic phenomena. The stochasticity of microtubules, which certainly store state information (as do most components in the entire cell) derives from quantum mechanical phenomena. In any case, unraveling the molecular level details of the cell is the most astoundingly complex problem humanity has yet faced. We can't even model bacterial state (regulation, etc.) right now. I strongly feel that we won't be solving this problem within our lifetimes. Probably not even in a hundred years.
  - roel_v 15y ago
    Maybe, but do we know if this stochasticity is fundamental to our consciousness? It may work just as well if we use random values from a lava lamp. (I don't know, I'm just speculating). I think we will solve it in the next 20 years or so, certainly while I'm alive (I'm 30). Like I mentioned in another post, it's a great time to be alive.

by gnosis 15y ago

a-priori 15y ago

Will this project "emulate" a human brain down to the subatomic level? Down to even the atomic level? The short answer is that each part of the simulation, from ion channels and synapses to cell bodies to network topologies and cell distributions, is done at the lowest level that people in the laboratories can effectively isolate and study. From that they can derive mathematical models of each part, and integrate those models to simulate them interacting over time and space. Is it a simplified model of the brain? Yes, but its the most detailed model that science can produce (or at least it aims to be). how would you even measure the "intelligence" of a simulated brain without any sensory organs or an ability to communicate? What makes you think that's difficult to simulate? It's been shown that primate brains are capable of learning to interpret novel inputs, such as a grid of electrodes 'painting' an image onto cortex. There was also a study where they implanted electrodes in a monkey's brain where the rate of activation of the cells surrounding the electrodes moved a cursor on a computer screen. The monkey learned to move the cursor. This shows that simulated inputs and outputs wouldn't have to perfectly correspond with reality in order for the brain to be able to process it.

roel_v 15y ago

Does it matter? As far as I know, we don't know yet what level we need to simulate at to simulate consciousness. If it's at the molecular level and we achieve that, who cares about the atomic level.

possibilistic 15y ago
It likely relies on subatomic phenomena. The stochasticity of microtubules, which certainly store state information (as do most components in the entire cell) derives from quantum mechanical phenomena. In any case, unraveling the molecular level details of the cell is the most astoundingly complex problem humanity has yet faced. We can't even model bacterial state (regulation, etc.) right now. I strongly feel that we won't be solving this problem within our lifetimes. Probably not even in a hundred years.
- roel_v 15y ago
  Maybe, but do we know if this stochasticity is fundamental to our consciousness? It may work just as well if we use random values from a lava lamp. (I don't know, I'm just speculating). I think we will solve it in the next 20 years or so, certainly while I'm alive (I'm 30). Like I mentioned in another post, it's a great time to be alive.