Modularity and Constraints in Language and Cognition

Frank C. Keil Cornell University

In the heat of battle, nativists sometimes forget that your everyday empiricist would be perfectly happy with some modules and constraints on learning that are specific to certain kinds of inputs and that are innately predetermined. After all, no one doubts that the sense organs such as the eye and the ear have unique specializations for the kinds of inputs they receive nor that their structural properties exert constraints that bias how input information is initially picked up and how it is ultimately interpreted. Of course the remarkable demonstrations of perception across modules and the difficulties of decomposing perception into discrete stages of the sort so near and dear to traditional information- processing theories temper these observations somewhat, but they cannot do away with them altogether.

Much of the emerging interest in modules and constraints, and certainly most of the controversy, revolves not around their existence in any instance, and not only about domain specificity, but also about just how cognitive and belief-laden these constraints might be. Our everyday empiricist would not be nearly so complacent, for example, with claims that there is an innately predetermined module for constructing beliefs about the mechanics of physical objects or about the dynamics of a belief- desire psychology. He or she would want to keep any constraints and module-like notions as close to the periphery as possible. This peripheral/central issue may be why many empiricists and neo-empiricists found Fodor ( 1983) book The Modularity of Mind to be much tamer than they feared. His proposed modules were quasi-perceptual in nature and not at all embodiments of the most belief-laden aspects of cognition.

-103-