For the past few decades, Raymond Boudon has revelled in uncovering the counterintuitive aspects of social thought and action, most recently by examining irregularities within the edifice of rational choice. This book is a collection of Boudon's essays revolving around his "cognitivist" or "judicatory" model of rationality, a generalization of the conventional rational choice model. The book builds on ideas presented in his earlier work, The Art of Self-Persuasion, extending its analysis of positive beliefs to that of values, which are defined here as explicitly held normative beliefs.

Before presenting his model in detail, Boudon examines existing models of values in the social sciences, dividing them between non-rationalist (fideist, skeptical) and rationalist types. While largely dismissing the former, he is unsatisfied with the conventional rational choice model, which portrays values solely as the product of instrumental reasoning applied to self- interest. His cognitivist model essentially expands the conventional one to encompass both non-instrumental and instrumental reasoning.

Much of the book is devoted to describing cases or general phenomena in which values are generated through reasoning processeses with non-instrumental components. As is typical of his work, Boudon delves into a impressively wide range of sources, including many that are not typically considered in relation to either rational choice or a theory of values. Of particular note is his extraction of ideas about values from the writings of well-known social theorists, including Toqueville on why the Cult of Reason prevailed in 18th century France but not Britain, Marx on why workers accept exploitation (not false consciousness), Adam Smith on the why miners are paid more than soldiers, and C. Wright Mills on the reasons for strict workplace egalitarianism among white collar workers. Besides this, Boudon also engages in sustained analysis of the relationship of his cognitivist model to the concept of axiological rationality in Weber's work, as well as to relativist and naturalistic theories of value, particularly the work of James Q. Wilson.

Overall, the book is convincing in arguing that a model containing a broader view of values has both greater verisimilitude and fewer anomalies than the conventional rational choice model. It is also convincing in showing that the cognitivist model, unlike irrationalist models, accomodates the fact that individuals can typically construct a set of logical arguments for the values that they hold. More broadly, the book's examples aid the reader in grasping for a verstehen-type understanding of a variety of diverse social phenomena.

Nonetheless, three interrelated issues could be addressed in order to clarify the applicability of the model to rational choice theory:

First, Boudon could be clearer in stating the role of self- interest in his conception of instrumentalism. He equates instrumentalism with cost-benefit calculation, and his examples of instrumentalism all seem to involve individuals seeking to maximize their own welfare. Does Boudon accept that instrumental reasoning can be applied to wider notions of good than self- interest? If so, could at least some of the phenomena he examines be explained as a result of altruistic or more complex other-regarding maximands rather than as result of non-instrumental reasoning?

Second, Boudon differentiates between values and mere "tastes", unreasoned preferences that have no moral standing and do not generate conviction. However, the rational choice model makes no theoretical differentiation between preferences based upon their moral standing, since any well-ordered ranking of outcomes can determine the ends of rational action. It is not clear what Boudon's position is on the role of tastes in a positive rational choice model. Does he believe they are insignificant in causing behavior, and hence unnecessary to explain? If they are significant, what does he attribute them to? Does he believe that the origin of tastes has any relationship with the origin of values?

Finally, Boudon notes that reasoning processes depend upon normative assumptions that are not themselves the product of reason, but may instead have other causes such as social background. For instance, values based upon fairness depend on each individual's particular definition of fairness and the role it plays in his or her conception of the good. Boudon seems to resist viewing such assumptions as a kind of taste, or as consistent with instrumental reasoning. However, he could be more specific about their ontological status, and describe whether he believes they can be accounted for in a determinate, predictive fashion. If not, does he believe this is a problem for theory-building?

Perhaps such questions can be addressed in his next book!