“What about the computer? Does it think, and if so does that make it conscious? For the present at least, the answer is that it does not think in the sense that human beings and other mammals think and so is not conscious in the same sense. Furthermore, the mammalian brain is enormously more complex than any present computer, not only in the number of functional elements but also in its connections, the individual neuron frequently having synaptic connection with upwards of a thousand others. What I propose is that we have here a parallel with the physicist’s concept of a critical mass. Consciousness, that is, depends on a critical degree of complexity of neural action. Probably also it requires the kind or pattern of complexity characteristic of the structure of mammalian cortex. The individual neuron then is not conscious, nor any small group of interacting neurons. Consciousness appeared in evolution when thought became possible, and there is no evidence of thought in lower animals, even those with quite extensive nervous systems. It probably exists in birds such as the crow, but it has not actually been demonstrated except in mammals. In them the cortex is well developed, but vestigial or absent in other animal forms. The argument then is that a computer built on the plan of the mammalian brain, and of a complexity at least equal to that of the brain of the laboratory rat, might be conscious—given the same capacity to learn and a suitable early experience. This is unlikely, but conceivable.”