The functional architecture can be viewed as the set of basic information processing capabilities available to an information processing system. "Specifying the functional architecture of a system is like providing a manual that defines some programming language. Indeed, defining a programming language is equivalent to specifying the functional architecture of a virtual machine" (Pylyshyn, 1984, p. 92). In other words, if it is assumed that cognition is the result of the brain's "running of a program", then the functional architecture is the language in which that program has been written.
The functional architecture is of interest to cognitive science because if offers an escape from Ryle's Regress (a.k.a. the homunculus problem). The functional architecture is comprised of a set of primitive operations or functions. This means that these basic functions cannot be explained by being further decomposed into less complex ("smaller") subfunctions. Instead, they must be explained by appealing to implementational properties (e.g., for human cognition, properties of the human brain). As a result, the functional architecture represents the point at which the decomposition of mental state terms into other mental state terms via functional analysis can stop. By specifying the functional architecture, one converts the black box descriptions that cognitivists create into explanations.
Pylyshyn, Z.W. (1984). Computation and cognition. Cambridge, MA: MIT Press.
Functional Analysis | Ryle's Regress | Primitive
Contributed by M.R.W. Dawson, October 12, 1995