The Logic of Decision [ペーパーバック]

The outlines of modern decision theory were laid down by Ramsay, di Finetti, von Neumann, and Savage in the mid-Twentieth century. Their treatments tended to be inspired either by mathematical or behavioral concerns. Jeffery's contribution is that it takes a philosopher's viewpoint, and while by no means ignoring mathematical arguments, is accessible to readers with a limited tolerance for or proficiency in mathematical analysis.

The introductory student will find this book hospitable, but for an appreciation of the classics of the field and the varied intentions of its architects, the student will have to refer to the original sources. Jeffrey does not provide such an overview. This book was written in 1965 and updated in 1983, so it is not surprising that it does not deal with some of the salient issues in decision theory today. In particular, it does not deal with the now huge literature in experimental psychology and behavioral game theory that sheds much light on how people actually make decisions. The reader interested in such issues will be forced to go elsewhere for such an analysis. Similarly, it does not deal with the bounded rationality literature, such as Gigerenzer's contributions to "fast and frugal" decision-making.

Several of the behavioral sciences do not include the rational actor described by decision theory in its general toolbox, and most practitioners in these fields are swift to condemn this model, usually on clearly spurious grounds. For instance, many sociologists believe that Bayesian rationality presupposes that beliefs (a.k.a. subjective priors) must be "rationally justifiable,'' and reject the model because beliefs in the real world cannot be accurately described through the criterion of rational justifiability. In fact, as Jeffrey's argument makes clear, subjective priors are just that---priors. They must be updated using Bayes' Rule, but that is all. Many psychologists reject decision theory because experimental results indicates that there are important biases in human decision-making that are not accounted for by the Savage-von-Neumann axioms. This is correct and important, but it does not follow that decision theory should be rejected as a contribution to understanding human behavior. These and related contemporary arguments are not treated in Jeffrey, but he does give the reader a solid set of insights into the meaning and operation of modern decision theory.