The computer metaphor of mind has been developed in an era when the serial
digital computer is in ascendancy, and those classical cognitivists who
support a notion of strong equivalence between mental and computational
processes have had a von Neumann architecture in mind. Analog computers
have offered an alternative picture of computation, and von Neumann's The
Computer and the Brain, published in 1958, brought this sense
of an alternative into the philosophy of mind by suggesting that human cognition
is a function of processes some of which are analog not digital.
I have examined the analog/digital distinction in the light of this suggestion,
looking first at the engineering community's uses of the contrast, and then
at several sorts of philosophic construal of the terms. I conclude that,
seen at the hardware level, there is no philosophically important difference
between analog and digital computation, and that the contrast has its primary
use in a dispute among language communities - those who offer explanations
in formal/linguistic terms, and those who offer explanations in physical/mathematical
terms.
Analog or connectionist systems are not easily interpreted as symbol-using
systems, because they lack disjoint, finitely differentiated elements. Their
intransigance in code terms, combined as it is with computational efficacy,
suggests that we do not have to think of computation in terms of symbols.
But those who offer a logical systems explanation have tended to think of
the brain as code-using as well as code-describable. Those who say that
some if not all intelligent processes do not use code, have tended to avoid
logical systems explanation in favor of explanation in dynamical systems
terms. I argue that this separation of vocabularies is not necessary if
we do not assume symbol-describable cognition is symbol-using cognition,
and that any sort of formal modeling, whether logical or mathematical, implies
symbol-describability at some level. The larger importance of connectionist
processing does not lie in its resistance to description in symbol terms,
but in the suggestions it offers about how cognitive states may have intrinsic
content |