Michael A Bishop and J. D Trout New York, New York: Oxford University Press. 2005. Pp. xii, 205 . This book is highly original and engaging. Michael Bishop and J. D. Trout propose a radical reorientation of epistemology. They propose to shift its focus from justification and knowledge to the findings of cognitive science about good reasoning, Ameliorative Psychology (AP). In the book they defend the importance of AP, criticize existing epistemology, and propound a theory of epistemic excellence in reasoning. Here is one proposal by Bishop and Trout about the aim of the new epistemology. On our view, the goal of epistemology is to articulate the epistemic generalizations that guide the prescriptions of Ameliorative Psychology. (15) This suggests the following taxonomy. AP offers prescriptions about reasoning. The prescriptions are guided by epistemic principles. Epistemology states those principles. There is more to Bishop and Trout’s new epistemology. They propose an epistemology that offers “real-world recommendations for living”(6). The recommendations are to be derived from AP. Here is an expression of this view of the work of epistemologists. Ameliorative Psychology is the science of applied epistemology, and theoretical epistemology is theoretical Ameliorative Psychology (i.e., a theoretical science). (18) The division of labor here seems to be as follows. Epistemological theories are scientific theories of reasoning. The practical work of epistemology, applied epistemology, explains how to apply the scientific theories to actual reasoning problems. These proposals for epistemology assign it empirical work, which is in keeping with Bishop and Trout’s avowed naturalism (22). But the projects seem disparate. The first aim gives epistemology certain epistemic principles to articulate. The second proposal seems to abandon anything recognizably epistemic. It seems to identify theoretical epistemology with the part of cognitive science that theorizes about reasoning. Applied epistemology applies those theories to actual examples. Neither endeavor in this proposal seems to involve any epistemic evaluation. Seeing how Bishop and Trout proceed in the book suggests how the two project descriptions are to be reconciled. Their contribution to the new epistemology is Strategic Reliabilism, a view about the conditions that make for epistemic excellence in reasoning (101). This topic is where the new epistemology comes in. The following reconciling picture emerges. The rest of AP investigates reasoning procedures and recommends some of them. Theoretical epistemology takes these recommendations as data and theorizes about the principles that determine what is thereby counted as excellence in reasoning. This is empirical theorizing about the tacit epistemic principles from which the recommendations derive. Thus, theoretical epistemology does indeed aim to state epistemic principles—the principles concerning excellence in reasoning that guide the recommendations of AP. And theoretical epistemology is indeed empirical science within AP—it aims to identify the psychologically realized epistemic principles that guide the AP recommendations. Applied epistemology applies the resulting theory of reasoning excellence to real-world problems. This is an objectionable project for epistemology. The science about reasoning bears no direct rational connection to the conditions that make for excellent reasoning or the conditions that justify recommendations about how to reason. In particular, the science gives no evidence for any theory that might guide recommendations about reasoning. Bishop and Trout report results of AP in the chapter, “The Amazing Success of Statistical Predictive Rules” (SPRs). An SPR is a formula that predicts the membership of some socially determined classification, such as the recidivism rate of a type of inmate or the future price of a type of wine, using the values of correlated variables. Numerous SPRs have been found that predict at a higher rate of accuracy than do human experts who have the same data or more (26–32). Clearly these findings about relative accuracy issue no recommendations. They are silent about the relevance of relative accuracy to epistemic excellence in reasoning. Whatever the prevalence in the AP literature of such evaluations prescriptions and evaluations, any epistemic principles behind them do not derive from the science. Again, the project assigned to the new theoretical epistemology is to state tacit principles of excellence in reasoning that guide AP recommendations. That project is of doubtful value to an epistemology, old or new, that aims to state the conditions that actually determine any given epistemic status. We have no good grounds to think that the epistemic principles that are tacitly accepted by AP researchers are correct. The recommending researchers need not even be guided by epistemic principles. Bishop and Trout describe their theory of epistemic excellence in reasoning as “the normative framework that supports the recommendations of Ameliorative Psychology” (58). But the supporting evaluative framework need not be epistemic. Sensible ethical and prudential concerns support recommending the more accurate SPR procedures for making important judgments. Bishop and Trout object to analytic epistemology as it is typically practiced, which they call Standard Analytic Epistemology (SAE). They have two main objections. One is methodological. They regard the method of SAE, in theorizing about justification or knowledge, as that of testing theories by the considered judgments about examples of practitioners of SAE. Bishop and Trout cite empirical findings of diversity in example evaluations. They challenge practitioners of SAE to defend the claim that their considered judgments are most trustworthy (109–115). There is a reasonable response to this challenge. The most trusted SAE case judgments survive intense critical reflection. At their best, the considerations that inform them include appreciation of the historical and contemporary theories and arguments about justification and knowledge. The relevant empirical evidence, including evidence about diversity, also informs them. The resulting judgments are fallible. Their trustworthiness is open to philosophical or empirical discrediting. Still, no other visible basis for evaluating examples has comparably good credentials. Bishop and Trout’s other objection to SAE is practical. Again, they seek an epistemology that gives practical advice, going as far as to say at one point: The primary aim of epistemology, from our perspective, is to provide useful general guidance about reasoning. (94) They find only unhelpful guidance from SAE. It appears to them that the “normative force” of SAE theories of justification is exhausted by the advice to adopt some subset or other of the beliefs that are justified for the advisee. They say that in the absence of useable guidance about how to reason, this advice is no better than the investment recommendation to “buy low, sell high” (79, 144). SAE theories give no guidance at all. A theory of the conditions that determine justified belief gives no advice about what to believe, not even the advice to believe only what is justified. Justification is something positive about a belief, but not decisive on its own. The advisability of holding a belief is independent of its justification. Even when only epistemic value is at stake, having a particular justified belief can be inadvisable. The belief might be distracting; it might obstruct or destroy other justification or knowledge. For corresponding positive reasons, having some unjustified belief might be epistemically constructive. More fundamentally, the advisability of believing is determined by more than its epistemic merits. Believing can help or hinder realizing various non-epistemic sorts of value, such as health and happiness. Optimal epistemic value need not make believing best for the person or best by any other non-epistemic standard. Bishop and Trout seem to be committed to disagreeing with the last claim. Their Strategic Reliabilism evaluates the epistemic excellence of a person’s reasoning by a cost/benefit analysis. The costs and benefits are determined by the reasoning’s impact on “epistemically significant” problems for the person. In their view this sort of significance “is a function of the weight of the objective reasons that [the person] has for devoting resources to solving the problem” (95). So their view makes epistemic excellence depend on the balance of all sorts of objective reasons. Here is an example. An oncoming vehicle threatens your life. You are able to reason productively about how to avoid this threat. Instead you ignore it. You spend the time reasoning with impeccable cogency about the mind/body problem. Bishop and Trout use this very example to illustrate that relative epistemic significance can be obvious (94). They would deny that the mind/body reasoning is as epistemically excellent as reasoning effectively about avoiding the threat. But in one respect your philosophical reasoning is as good as reasoning gets. It is perfectly cogent. That is excellence of the specifically epistemic kind. The example also shows that epistemic considerations are too limited to justify advice about how to reason. The advisability of any conduct, including reasoning, depends partly on non-epistemic evaluations, including the prudential and ethical merits of the alternatives. It also depends on empirical matters beyond the results of AP. It depends on the factors that contribute to all sorts of well-being. Nevertheless, the findings of AP are valuable evidence about our reasoning. They deserve our attention, as do the topic of epistemic excellence in reasoning and Bishop and Trout’s views about it.