ARCHITECTURE OF COGNITIVE NUMERICAL PROCESSING MECHANISMS - CONTRASTING PERSPECTIVES ON THEORY DEVELOPMENT AND EVALUATION

This article discusses two alternative approaches to theory developmen t and evaluation in the domain of numerical processing. Taking as an e xample a recent study of a brain-damaged patient with deficits in nume rical processing (Macaruso, McCloskey, & Aliminosa, 1993), we contrast the approach taken by our colleagues and ourselves (e.g., McCloskey e t al., 1985; McCloskey, 1992) with that adopted by Campbell and Clark (1988, 1992; Campbell, this issue; Clark & Campbell, 1991). We suggest that whereas the modular model proposed by our colleagues and ourselv es represents an attempt to develop a model sufficiently constrained a nd explicit to generate specific testable predictions, Campbell and Cl ark's encoding complex approach is unconstrained and inexplicit, and d oes not constitute a specific theory that is subject to empirical test .