Sunday, August 10, 2008
Comparing a brain to a computer
It's funny how some people make a comparison of an organic brain to a computer when making comparisons of processing power. But in a lot of cases, the over-simplification gets it entirely wrong.
The wrong way: Saying a transistor is comparable to a neuron. No! No! No! A transistor pretty much has two states I/O. And then relays back or holds one of those two states depending on what its function is. A neuron however has many many states, and relays back much more complicated information.
So what would be a better way?
A neuron in itself, is a CPU + Memory + a very nice routing management system. So a neuron isn't the equivalent of a transistor, but rather an entire computer. The neuron gets its information in an analog waveform which contains more and much subtler information than I/O. (It also gets other data via chemical means, with the chemical route being slower - it probably provides bias as to how the electronic data is handled.) Upon recieving the analog waveform, the neuron can route the information, evaluate it, store it, command a physical action, or any combination of those. Seeing how it operates in a massively parallel fashion and using weighting for evaluation, it gives up precision and accuracy for the speed and raw power of a somewhat noisy yet noise-tolerant processing cluster.
Instead of comparing a brain to a computer, instead it should be compared to a large network of computers. A mouse's brain might be a pretty good sized server farm, as where a human brain would be a nice chunk of the internet. Using this analogy would be much closer towards how the architecture of an organic brain actually works.
The wrong way: Saying a transistor is comparable to a neuron. No! No! No! A transistor pretty much has two states I/O. And then relays back or holds one of those two states depending on what its function is. A neuron however has many many states, and relays back much more complicated information.
So what would be a better way?
A neuron in itself, is a CPU + Memory + a very nice routing management system. So a neuron isn't the equivalent of a transistor, but rather an entire computer. The neuron gets its information in an analog waveform which contains more and much subtler information than I/O. (It also gets other data via chemical means, with the chemical route being slower - it probably provides bias as to how the electronic data is handled.) Upon recieving the analog waveform, the neuron can route the information, evaluate it, store it, command a physical action, or any combination of those. Seeing how it operates in a massively parallel fashion and using weighting for evaluation, it gives up precision and accuracy for the speed and raw power of a somewhat noisy yet noise-tolerant processing cluster.
Instead of comparing a brain to a computer, instead it should be compared to a large network of computers. A mouse's brain might be a pretty good sized server farm, as where a human brain would be a nice chunk of the internet. Using this analogy would be much closer towards how the architecture of an organic brain actually works.
Labels: brain, computer, computing, processing
# posted by pauljs75 @ 8:30 AM 
