Consciousness in meme machines
Abstract of lecture for Banbury Center conference on machine consciousness, May 2001
What kind of machine could have something like human consciousness? I shall argue that humans are best thought of as meme machines, and that human consciousness is the way it is because of the memes that we store and propagate. If this is so then the only machines that could have human-like consciousness would be machines capable of imitation.
Memes are ideas, habits, skills or any kind of information that is copied from person to person, mainly by imitation. I argue that since imitation first evolved and the first memes appeared (perhaps 2 million years ago), human evolution has been shaped by the interaction between two replicators - memes and genes. In particular, the expansion of the human brain, and its redesign for language, were a result of memetic drive. The human brain is best thought of as a selective imitation device.
If this is so, the implication is that only machines capable of imitation could be conscious in a way comparable with humans. I discuss two kinds of machine imitation. If machines could imitate humans they should acquire a sense of self and consciousness in the same way that humans do. If machines could imitate each other, the same processes should occur as occurred during human evolution. For example, if robots could imitate sounds made by other robots, language should spontaneously appear. This could be tested.
In conclusion, the way to create conscious machines is by making them capable of imitation, and thus setting off a new memetic evolutionary process among them. But if this worked we should not expect to understand the language they create or to know what it is like to be them.