Some interesting things to ponder - all interrelated

[Explodicle]

My point is basically that the devil is in the details. The brain is probably an order of magnitude more complex than people currently think. Like every synapse is a laptop complex.

Isn't Markram allocating about the equivalent of one laptop per neuron? Granted, that's a neuron, not a synapse, but still. Also, isn't Hameroff basically saying that the computing power inside microtubules is huge?

Assuming the computer only needs to simulate the 100 trillion synapses, and Moore's law continues, that means that in 70 years you can buy a "laptop" capable of simulating the human brain at 100% speed.

How did you arrive at this?

Current processing power for one synapse = 1 laptop
Synapses in a human brain = 100 trillion
Moore's law means a "laptop" (of fixed price) doubles in speed every 1.5 years
So after about 46.7 doubling periods of 1.5 years, a laptop will be 100 trillion times faster.

[FirstAscent]

Regarding Henry Markram and brain simulation for those who don't know: http://bluebrain.epfl.ch/cms/lang/en/pid/56882

[bb113]

Current processing power for one synapse = 1 laptop

Ah, I see. Maybe I am thinking about this analogy wrong.

Assume each synapse to do 2 billion calculations per second (we can probably get a better number than this... but for now)  and then output binary (it is probably not binary... but let us assume this as a lower bound) information that is readable by the neuron.

Further assume each neuron has 10,000 synapses.

So each second the neuron could be in any one of 2^10,000 states. Would you take that to mean that each neuron does 2^10,000 calculations per second? Or would it be 10,000 calculations per second?

I may have confused myself.

[FirstAscent]

Notice how the conversation as of late has been: Singularity -> ( The Easy Problem + Quantum Physics ). We'll see where it goes from here.

[bb113]

Or is it 10,000² different states? I think I am just confused.

[FirstAscent]

Current processing power for one synapse = 1 laptop

Ah, I see. Maybe I am thinking about this analogy wrong.

Assume each synapse to do 2 billion calculations per second (we can probably get a better number than this... but for now)  and then output binary (it is probably not binary... but let us assume this as a lower bound) information that is readable by the neuron.

Further assume each neuron has 10,000 synapses.

So each second the neuron could be in any one of 2^10,000 states. Would you take that to mean that each neuron does 2^10,000 calculations per second? Or would it be 10,000 calculations per second?

I may have confused myself.

I think you're getting ahead of yourself with regard to the computer model used to model synapses and neurons. Neurons have action potentials, and accumulate charge/energy/potential (not sure what to call it) from the synaptic connections upstream until a threshold is reached, at which point they fire and propagate downstream.

[bb113]

Each synapse integrates information from many different receptors plus the local internal state, etc which determines if there is a local depolarization or not. The simplest possible model would view each synapse as either depolarizing or not. Then each neuron integrates all the information from each synapse and translates this into an action potential (either fires or not). Then the brain integrates the information from each neuron to determine behavioral output. At this point I'm just wondering how this is translated into processing power.

[FirstAscent]

Then the brain integrates the information from each neuron to determine behavioral output. At this point I'm just wondering how this is translated into processing power.

At any given time, a certain distributed pattern of activity (the subset of neurons firing) exists within the brain. Any pattern could represent a timeslice of a memory, thought, etc. These patterns in turn stimulate neurons and create new patterns, and thus the mind operates. Some of those neurons which get activated ultimately are responsible for sending their output to the body for motor control.

If the proper pattern of neurons activates, the signals are sent outbound to create behavior.

[bb113]

Ok, but what does that mean in "processing power".

[FirstAscent]

Ok, but what does that mean in "processing power".

It doesn't mean anything to me. That's not a term I would use.

[Explodicle]

Ok, but what does that mean in "processing power".

It doesn't mean anything to me. That's not a term I would use.

I just used it as a shorthand while assuming 1 laptop = 1 synapse. It doesn't really mean anything, and I have no idea how hard it actually is to simulate a synapse.

[FirstAscent]

Ok, but what does that mean in "processing power".

It doesn't mean anything to me. That's not a term I would use.

I just used it as a shorthand while assuming 1 laptop = 1 synapse. It doesn't really mean anything, and I have no idea how hard it actually is to simulate a synapse.

As hard as you want it to be. It can be a relatively simple equation that operates on the fact that the upstream neuron fired, in which case the math is calculated to accumulate to the downstream neuron. Or, it can model ion flow, molecular reactions, etc. taking into account the physical geometry of this particular synapse, as opposed to the 100 trillion other synapses, etc.

[bb113]

Ok, but what does that mean in "processing power".

It doesn't mean anything to me. That's not a term I would use.

I just used it as a shorthand while assuming 1 laptop = 1 synapse. It doesn't really mean anything, and I have no idea how hard it actually is to simulate a synapse.

I understand, but what would be the combined processing power of 10,000 laptops working in parallel on a problem at 2 ghz? Do you just add them up?

[Explodicle]

Ok, but what does that mean in "processing power".

It doesn't mean anything to me. That's not a term I would use.

I just used it as a shorthand while assuming 1 laptop = 1 synapse. It doesn't really mean anything, and I have no idea how hard it actually is to simulate a synapse.

I understand, but what would be the combined processing power of 10,000 laptops working in parallel on a problem at 2 ghz? Do you just add them up?

Yes. So a 100% speed upload would be possible sooner if you're willing to buy more computers.

[FirstAscent]

Need I point out that we're only one hundredth of one percent the way through pondering that which should be pondered in this thread?