Peter Bentley

Digital Biology

The Creation of Life inside Computers and How it Will Affect us

Headline Book Publishing 2001


KEYWORDS : brain - sense of self - consciousness - decision-making - learning - memory

pg 65

Brains

Our brains are the most impressive achievement of evolution we know of. They are several million times more powerful than any computer we have created. Every one of our brains is so complicated that it would take thousands of our best experts in many lifetimes to design and construct anything a fraction as powerful using our current technology.

Our brains also create our sense of self, our consciousness. This very personal part of each of us is created by cells in our skull - the concept so hard to accept that the large percentage of the population believe instead that our self is generated by a supernatural soul. But, whatever you choose to believe, we are long way from having all the answers about the brain. Psychologists tell us what brains do, neurologists tell us how they are organised and neuroscientists tell us what they are made of, but nobody seems to know exactly how it all works.

Nevertheless, despite the ongoing arguments, we know enough to understand the rudiments of decision-making, learning, memory and even consciousness. We certainly know enough to be able to build digital brains within our computers that can make decisions, learn and remember.

pg 100

Understanding brains is a ticklish subject. But science is beginning to shed light on the mysteries within our heads. We now know that our brains think and learn using neurons. These neurons are cleverly via together into networks, each of which can then learn a little like a company with many levels of employees and bosses. The brain is also made from many different regions, and there's regions were slowly added by evolution like the extensions to a building. Finally, consciousness is made from many processes - some unconscious, some not - badly managed like the crowd extremity and the football game.

Digital brains are much less complicated, and we know how they work. Digital brains are usually found within the digital universes of our computers. They are also made up from neurons, accepting signals that they weight, sum, adjust with a threshold and transformed with an activation function. They are also wired into networks, some feedforward, some recurrent.

And they also learn. For example, a neural network might be used for quality control, learning by adjusting the input weights of every neuron to make the output match training examples. Thus subsumption architecture has resulted in robot brains with anatomies - specialised regions or process that perform parallel independent tasks. So that's brains for you: neurons, networks and regions leading to thought.










 

HOME      BOE     SAL     TEXTE