On the analysis of irrational data selection: A critique of Oaksford and Chater (1994)

A critique is offered of the theory by M. Oaksford and N. Chater (1994) of rational analysis in P. C. Wason's (1966) selection task. That theory turns on the idea that participants choose among different cards according to the amount of information they expect to gain; however, Oaksford and Chater have chosen a measure of information (the Shannon-Wiener measure) that is irrelevant to the question actually put to the participants. Consequent to that choice, certain assumptions and parameter values are dictated by the rank order of selection frequencies that the theory seeks to match. Those relationships are picked out, and the mathematical assumptions are identified and transposed into their psychological equivalents. Those psychological equivalents are (a) more in number than the results to be explained and (b) psychologically implausible. Finally, a correct Bayesian analysis of the selection task is shown to confirm the prescription of propositional logic.