Counting on the Brain
Evolution has made a lot of inventions, as it evolved the nervous system…
The nervous system is not accompanied by a manual explaining the principles of operation.
The blueprints and the early prototypes were thrown away a long time ago.
Now we are stuck with an artifact, so we must try to reverse engineer it.
It is often said that the human brain in
“ The most complex object in the known universe”
It is easy to list impressive statistics about the brain, like “A piece of brain tissue the size of a grain of sand contains 100,000 neurons“
So it would seem almost impossible to duplicate this in a mechanical or digital system. And yet, over the last two centuries this is precisely what has happened. Step by step we have managed to understand and copy more and more bits and pieces of the brain. For practical purposes, it is not so much relevant what the brain is, as a biological part of the human body that uses chemical signals, but what the brain does. We started to view the brain as an information processing system: it takes some input, uses and algorithm, and produces an output. We then also looked at how the brain does it, from a very abstract point of view: each module, even each neuron itself could be understood as an information processor. Individual neurons receive input, both positive and negative, then basically use a threshold to compute whether or not to fire, to send the signal onwards.
A system of neurons, a neural network, then becomes an incredibly efficient computer. .
The brain apparently only uses 20 Watts to do its thing.
A supercomputer trying to emulate the brain requires orders of magnitude more than that!
But we are getting increasingly good at making hardware that is organized more and more like the brain (Link)
Not by using chemical signals, but simply by sending a 1 or 0 or by sending integers in a digital electronic format. Neurons, natural or digital, produce a spike, a signal, and send it onwards to the rest of the network. The overall computation, like finding a face in a photo, is not done by individual neurons, but by the network as a whole. And yet we still don’t fully grasp how the brain does it.