Редактирование: Whole brain emulation (раздел)

= Misconceptions =

'''A synaptic spike is like a FLOP'''

Most calculations of the computing power of the brain are based on synaptic spikes. These are assumed to be the basic, atomistic interactions over which everything else is built. Thus, if one knows the number of synapses and the mean number of spikes, one could measure the computational capacity of the brain in "Synapse operations per second" or "Spikes per second". See, for example, Moravec<ref>''When will computer hardware match the human brain?'', Hans Moravec, December 1997. [http://www.transhumanist.com/volume1/moravec.htm Available here].</ref>.

The problem with these models is that they only take spikes into account. Measuring computational capacity in "spikes per second" only measures the computational capacity of the brain ''if its neurons were to be shaped into biological, computer circuitry''. It does ''not'' measure all the underlying computation that takes place to determine whether or not there'll be an spike.

Consider gene expression: It is fairly low level, it has nothing to do with spikes, and it is required for neuroplasticity; the basis of learning. Neural structures can't be modified or created without neuroplasticity. An upload ''wouldn't be able to learn'' without neuroplasticity, which means that a spike-only model of the brain would need special considerations for gene expression -- And that's a toughie, FLOPS-wise. Other things, like the chemical environment of the brain, would also need to be included in the model. In short, while spiking models are accurate for instantaneous activity, long-term changes and neuroplasticity require the lower-level events to be taken into account. Moreover, purely spiking models are still very computationally complex, and the computations underlying a single spike in such models are far, far more than a single FLOP.

There exist, however, computationally cheap models that are very accurate. See the [[Izhikevich model of spiking neurons]]. However, this model consists of two differential equations. These are the underlying computations leading up to the spike, and they are sufficiently complex to destroy the idea that a synaptic spike is anything like a floating-point operation.

'''IBM has emulated the brain of a cat'''

No, no, no, no, no. Absolutely not.

[http://www.acceleratingfuture.com/michael/blog/2011/11/more-nonsense-reporting-overblowing-ibms-accomplishments/ Let Michael explain everything]

'''We can send nanobots to scan the brain'''

'''The Blue Brain Project is trying to emulate the mind'''

The Blue Brain Project is a pharmacological simulation to run drug trials and obtain a better understanding of the brain. It is not an uploading effort and none of its participants make any guarantee that it will produce a mind.