Model Selection 3

Model Selection 2 Quantitative methods used to compare the performance of mathematical models of cognition measure the ability of models to redescribe experimental data. Some also weight various properties of the models. Two theoretical approaches to model selection were compared in the first part of the paper. One emphasizes flexibility, the other generalizability. In the second part, simulations were carried out to gain a better understanding of the results of Massaro, Cohen, Campbell, and Rodriguez (2001). Findings provide further insight into why measures of generalizability, such as BMS, are preferable to measures of flexibility, such as RMSD, and also show that the results from model recovery tests can be misleading if not interpreted relative to the data on which they were evaluated