Mind-brain identity is also challenged by nondualists who think that the development of computers
reveals the hypothesis that minds are brains to be much too narrow. The possibility of artificial
intelligence, which is the construction of computers capable of reasoning and learning, suggests that
we should identify mental processes more generally with computational processes that can occur, not
just in brains, but also in machines made out of silicon chips or other kinds of hardware. This view is
called functionalism, because it says that mental states are inherently functional, providing causal
connections between inputs and outputs in ways that produce intelligent behaviors. Computers and
other machines, or maybe even extraterrestrial organisms, can have such functional states without
having brains, so identification of mind and brain is a mistake. It is mental software that makes minds
work, and the particular hardware on which it runs is not very important. I found this computational
view appealing when I first got interested in cognitive science in 1978, but came to doubt it in the late
1980s when I began to work on neural network models, and even more in the 1990s when I started
research on emotion.