A neuron is (conceptually - if that's the right word) a fairly straightforward device. Signals come in through its dendrites and are added up. If the sum exceeds a value, the neuron sends a signal out along its axon, which is connected to the dendrites of lots of other neurons. Otherwise it stays schtum.
The number of neurons you have (1011) is quite impressive, but nowhere near as impressive as the number of links each neuron makes with others (typically thousands). Considered as a whole, it is that interconnectivity that makes the human mind the most fabulous object in the Universe.
But it is also annoyingly fragile. After one has got past the grief when someone dies, a remaining irritation is the realisation that all that they knew has died too. Who is that person in the photograph? you want to be able to ask them. What do you think of Jim and Clara's buying that house?
An abiding and very old science fiction response to this mortal fragility is to propose some sort of brain scan that allows the recording of the entire structure and then the use of that data to create its emulation in a large computer.
Unlikely.
But let's return to the inherent simplicity of a single neuron. In a few years time we ought to be able to do the same in silicon. (We could now, but for the very high connectivity.) And we'll soon - apparently - have hordes of nanobots swimming round our insides fixing up molecular breakages. If they also replaced about four million neurons (a tiny fraction) every day with their silicon equivalents, then, by the time someone was eighty, their brain would be entirely electronic. And the life-long process would be so gradual and smooth that it would become part of the normal acquisition of wisdom with age.
But when someone's body died, you would be able to crack their scull, take out their silicon brain, plug in a five-volt power supply, a webcam, a microphone, one or two robot avatars and a network connection, and they would live forever.
And no one would ever get Alzheimer's disease again...
No comments:
Post a Comment