Organizational Behavior and Human Decision Processes

Lopes (1996) clearly has limited patience with the Try to work through with undergraduates the equivalence concept of ‘‘rationality’’ as a basis for, or descriptor of, human decision making. I, in contrast, find some of the principles of rationality reasonably compel(x, E; (y, D; z)) B ((x, E; y), D; (x, E; z)), (1) ling—as do subjects when the situation allows them to see the structure of the decision they face—and where the symbol (u, E; v) means one receives u if event E occurs and v otherwise and the underlying chance so it strikes me as relevant to understand in some detail just which principles of rationality are and are experiment is independently run twice to realize the events D and E. I believe that the empirical evidence not being violated. This, I believe, is the thrust of a good deal of the contemporary work, some of which is heavily against the behavioral accuracy of most structural rationality properties, but it is much less so she mentions, relating data and theory. The first question is just what does the behavioral against preference rationality. The rank-dependent utility (RDU) models of the evidence against subjective expected utility (SEU), which surely is overwhelming, tell us is wrong with 1980s (for a summary see Quiggin, 1993) have exactly the feature of maintaining preference rationality and that theory. I have argued (Luce, 1990, 1992; Luce & von Winterfeldt, 1994) that it is very important to abandoning most of structural rationality. Their most complex structural property is what has been called distinguish between two types of rationality embodied in SEU. The one, called preference rationality, event commutativity: includes such principles as transitivity and consequence monotonicity (i.e., replacing a consequence in ((x, D; 0), E; 0) Ç ((x, E; 0), D; 0). (2) a gamble by something better can only make the gamble better). Violations of these properties typically Here the bottom line is reasonably clear: x if D and E both occur and 0 otherwise; they differ only in the are viewed as errors when they are pointed out to subjects. The other type, called structural rationality, order in which D and E are realized. This has been tested and although the results are somewhat mixed, says that one’s preferences should not be altered by changing the way a gamble is formulated so long as the most careful laboratory study (Chung, von Winterfeldt, & Luce, 1994) supports Eq. (2) for a large the bottom line remains unchanged. It may well be that people agree to this in principle, but most unproportion of subjects. Adding further, more complex conditions of structural rationality, such as Eq. (1), aided people simply fail to implement it successfully.