On Perceptual Information and Fitting Models to Data: A Reply to Massaro and Cohen (1993)

Cutting, Bruno, Brady, and Moore (1992) investigated the properties of Massaro's (1987) fuzzy-logical model of perception (FLMP) and other models. Among their results, they found that FLMP and an additive model fit the data of individuals' judgments of depth about equally well, that the additive model fit the group data better, and that FLMP fit random data better. Massaro and Cohen (1993) questioned the pertinence of the latter finding, questioned the fiting of the additive model (but not that of FLMP) to group data, and also compared our general approaches to perception. Here we show two new ways in which FLMP absorbs random error, raise a question about fits of FLMP to group data, and provide new support for Cutting et al's claims about the importance of fitting models to random data. We then make further comparisons between the approaches and explicate the concept of information specification which they queried.