VALIDATING COMPUTATIONAL MODELS - CRITIQUE OF ANDERSON INDETERMINACY OF REPRESENTATION CLAIM

It has occasionally been claimed that computational models are underdetermined by behavioral data in a way that is qualitatively different from the usual inductive underdetermination of all scientific theories. The present article takes issue with such claims, which have been gaining some currency, by challenging a particularly explicit version presented by J. R. Anderson (see record 1979-22767-001). It is argued that Anderson's version of the indeterminacy thesis, which is directed at deciding the form of mental representations, rests on 4 assumptions, each of which can be shown to be untenable (and some of which involve mathematical errors). (15 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved).