The Brain as a Computer
In 1943, neurophysiologist-cybernetician Warren Sturgis McCulloch (1898-1969) and logician Walter Pitts (1923-1969) wrote a paper titled “A Logical Calculus of the Ideas Immanent in Nervous Activity” (McCulloch and Pitts 1943), in which they provided a simplified model of brain activity with neurons as on-off switches in a computational account of the brain. They illustrated how logic circuits could be implemented in networks of artificial neurons. In the following diagram, we see how two Boolean circuits are implemented. To make an AND circuit, two neurons must fire together on a third neuron. In an OR circuit, it is sufficient that either one of two neurons fire, because they each fire over stronger connections.
If the brain is a neural network machine along the lines of McCulloch’s and Pitts’s suggestion, it must neurobiologically support Boolean processing. Many thought that as Shannon had shown there could be thinking digital machines, McCulloch and Pitts had shown that the brain was such a machine.
The behaviorists had objected to the study of the mind because, as Watson explained, “mind” is just another word for “soul.” This was erroneous, said emerging cognitive scientists. The mind is not a soul—it is a digital processing machine. But questions remained for McCulloch and Pitts. How, for example, did neural information processing relate to knowledge, understanding, and values? There seemed to be a gap between our understanding of neurons as networked on-off switches and mental life. But McCulloch and Pitts maintained that there was no gap. Their solution to the mind-body problem was that the brain is a computer and the mind is a program. Along with this realization came several consequences. All mental life, thought, understanding, willing, value, purpose, and so on were to be understood computationally. If we understood how the nervous net of the brain did “mental” computations, there would be nothing more to find out in psychology:
To psychology, however defined, specification of the net would contribute all that could be achieved in that field . . . in psychology, introspective, behavioristic or physiological, the fundamental relations are those of two-valued logic. (McCulloch and Pitts 1943, p. 129)
In the view of McCulloch and Pitts, psychology must be thought of as a computational science, given that mental life is computational.
McCulloch and Pitts also argue that there are no metaphysical problems left, because of their computational insights into the mind. McCulloch adopted a neoKantian perspective, which involved viewing at least some of Kant’s a priori conditions as neurobiologically hard-wired (Kay 2001, p. 594). At one point in the article, they write about Kant’s “thing in itself” and suggest that there are no metaphysical barriers and that we can know things in themselves by appeal to the neurocomputational transparency of the brain. Their argument, albeit unclear, reveals their enthusiasm for the project. Old philosophical problems, including the mind-body problem, were solved, and the mind had, in its basic principles, been understood computationally.