Slashdot Mirror
Scientists Begin Mapping the Brain
Raindance writes "A team at the University of Utah has unveiled a system to map and digitize brain tissue — thus fulfilling one of the long-standing holy grails of neuroscience and enabling for the first time in-depth analysis of how mammalian neural networks function. So far, maps for the entire retina and related neural networks have been released; no ETA on a full-brain digital reconstruction yet. (One of the lead authors hangs out here on Slashdot.)"
A neural network running a simulation of a human brain would be a Turing-complete strong AI
I understand what you're saying here but there are a lot of non-trivial hurdles to get over, even assuming you can accurately scan and simulate the brain.
First, the brain also includes a lot of chemical transmitters which we really don't understand the function of yet. You would have to include them in your model as well, including the ones that don't originate in the brain.
Second, you have to interpret the simulated neuron firing into something that you can actually understand. It's pretty pointless for your brain-on-a-hard-drive to be saying 'hello' if you can't understand what it's saying. An accurate simulation of my mind would have all the neurons that control my breathing, lips, tongue, and vocal cords firing like crazy, but good luck figuring out what I'm saying.
Third, you have to be able to supply meaningful input into the brain. You're essentially talking about submitting a consciousness (assuming your simulation is 100% accurate) to the most horrible sensory deprivation imaginable. In order for your research to be useful, you would have to supply it realistic input (including feedback based on it's output) otherwise the brain would change drastically just from that.