Toward a cognitively responsible theory of inference: Or, what can synapses tell us about ambiguity?

The recent fascination of literary and linguistic theorists with the brain sciences seems predicated on an expectation that the latter will ultimately reveal the existence of basic mechanisms of interpretation which are binary, hard-wired, and generally free from the ever-present ambiguity of human language interchange. The information already available to us about communication between cells of the nervous system at the synapse reveals this to be a misguided hope. The neurophysiology of the synapse reveals the existence, at the cellular level, of the biological equivalent of inference. The nervous system is apparently no different at the level of its basic components than at the level of higher level human language behavior. It is built to cope with insufficient or fragmentary data, and to make decisions when necessary in spite of the insufficiency of the input. The implications for theory building are important. A system of well-formedness rules will not be sufficiently flexible to describe language interpretation. A preference system which can describe not only well-formedness conditions, but also necessary conditions that can be satisfied along a gradient, and conditions which are not necessary but are only typical (both gradient and non-gradient) would seem to be more responsive to biological reality.