In Normative Externalism, Brian Weatherson argues that living up to one's principles is overrated: ‘If one's own principles are good, then one should conform to them. But that's because they are good, not because they are one's own’ (224). Weatherson argues that there is no reason to avoid being a hypocrite, or having incoherent beliefs, because the first-order question of what you ought to do (or believe) is independent of the second-order question of what you ought to believe about what you ought to do (or believe).

Part 1 argues for normative externalism in ethics: ‘moral norms are independent of both what one thinks the moral norms are, and what one should think the moral norms are’ (115). Weatherson argues that all arguments against this view generate the implausible result that hypocrisy is the only moral vice. This would make moral truth ‘irrelevant to our judgment of actions and persons’ (84), and turn misguided fanatics into moral exemplars. For Weatherson, Robespierre is the perfect example of how an iron resolve to do what is right, combined with false moral beliefs about what is, in fact, right can have disastrous consequences (such as the Reign of Terror) that we should not evaluate positively.

One such argument to the contrary begins from the thought that uncertainty matters to moral evaluation. Doing the right thing when one is uncertain that it is the right thing is reckless. If moral uncertainty is to be treated in the same way as factual uncertainty, then our moral beliefs matter morally in a way that normative externalism rules out. Weatherson argues that treating factual and moral uncertainty in the same way would require agents to be intrinsically motivated by moral rightness itself. It would require that they attempt to do what is morally right, while remaining neutral on what that is. Weatherson argues that this cannot be what is required (except when we are in the process of changing our moral views) because it is always ‘permissible to not be intrinsically motivated by maximally thin moral properties de dicto’ (50). Following Nomy Arpaly, Michael Smith and Bernard Williams, Weatherson thinks that that good moral agents are motivated by what is, in fact, morally right – fairness, justice, kindness (although the details depend on the first-order moral theory). People who are motivated by these things can act well without also being motivated by moral rightness itself. Weatherson also endorses the stronger claim that “in most circumstances, it is impermissible to be at all intrinsically motivated” by moral rightness itself (51) – it can be dangerous, as Robespierre illustrates. However, the central argument does not rest on this.

Another respect in which we might have thought our moral beliefs have implications for moral evaluation is in their capacity to excuse wrongdoing. For example, we might have thought that, ‘Wrong actions are blameless, and hence do not reflect badly on the actor who performs them, if that actor believes the action is right, and has good reason for that belief’. However, Weatherson argues that this is not true – moral ignorance cannot mitigate blameworthiness. Weatherson's argument for this depends on accepting a desire-based view of blameworthiness. If blameworthiness is a matter of ‘having the right desires’ (88), moral ignorance does not excuse because moral beliefs never affect our desires in a way that can mitigate blameworthiness. Of course, it sometimes seems that our moral beliefs do affect our desires, perhaps in ways that mitigate blameworthiness. For any such case Weatherson must say either that the agent does not have defective desires, or that the agent did not do anything wrong. He suggests that many such cases are cases of practical irrationality, in which agents have the right desires, but fail to realise them. In such cases, the agent is not blameworthy (89). It is worth noting how strong this claim is. The desire-based view of blame implies that that there is no possible situation in which one's beliefs or evidence could cause one to have defective desires and yet be blameless.

Normative externalism is essentially a meta-normative thesis, and its applications to ethics and epistemology are sold as different flavours of the same view. However, the arguments of each half sometimes depends on first-order claims that are difficult to translate between domains: that moral blameworthiness depends on desires (88), and that Change Evidentialism is true (‘a person with a rational attitude towards p is under no rational obligation to change that attitude unless their evidence for or against p changes’ (120)), to mention just two. One interesting difference between the treatment of the two domains is that in the ethical domain the moral norms held to be external are both those governing moral rightness, and those governing praise- and blameworthiness. One surprising consequence of this is that there is no blameless wrongdoing (108). The application to epistemology focuses for the most part on only one kind of evaluation – which beliefs it would be rational to have. It would be interesting to consider what normative externalists might say about epistemic praise- and blameworthiness.

Part 2 transposes normative externalism into epistemology. The core idea is that epistemic norms are independent of both what one thinks they are, and what one should think they are. In other words, there are no ‘level-crossing principles’, or principles of coherence describing ‘necessary connections between first-order and second-order claims’ (120).

Weatherson argues that level-crossing principles are under-motivated, and unnecessary for explaining the phenomena that they are usually brought in to explain (higher-order evidence (Ch. 8), epistemic akrasia (Ch. 10), peer disagreement (Ch. 12)). Higher-order evidence, epistemic akrasia, and peer disagreement are often thought to support level-crossing principles, because in such cases agents would, supposedly, be irrational if they did not revise their attitudes – even if those attitudes are otherwise rational. However, Weatherson points out that these cases are often under-described. When we flesh them out, and if we assume Weatherson's preferred first-order epistemic theory, Change Evidentialism, these phenomena provide no independent normative pressure to revise attitudes. The first-order theory requires the usual attitude changes all by itself. On some ways of fleshing out the cases, the phenomena also change the evidence, thus also changing what is rationally required. For example, learning that you are too tired to make reliable diagnoses will mean you should reduce credence in some cases (tricky cases in which you know the evidence might support distinct conclusions), but not in others (easy cases where you can see that your patient's hair is on fire) (130-1). However, this is not because some level-crossing principle holds, but because your evidence has changed.

Weatherson also dismisses some of the underlying motivations for coherence requirements in epistemology. For example, it has been thought impermissibly circular to rely on one's own original reasoning about the evidence to dismiss apparent pressure to change beliefs in response to higher-order evidence (Ch. 9). Weatherson responds to this by arguing that not all circular reasoning is epistemically bad, and there is nothing particularly bad about normative externalism's recommendation that you should trust (only) your evidence in all situations – even though this reasoning sometimes has a ‘whiff’ of circularity (147). Weatherson also dismisses arguments from ‘weirdness’ (175) that have been used to argue that akratic belief combinations (‘p, but I ought not believe that p’) are always irrational. Normative externalism is not afraid of weirdness, when what the evidence supports is weird. Non-ideal agents may sometimes be rational to adopt akratic belief combinations, because their being non-ideal means that they may sometimes have evidence for p and also evidence that their belief p is not rational (for example, when they have good testimonial evidence for p, and also compelling but misleading arguments for testimonial scepticism (170)). Although, such agents do have reason to reconsider their reasoning (178). Weatherson also suggests that level-crossing is motivated by an implicit commitment to the principle ‘judgments screen evidence’ (188). This says that one's rational judgments about the evidence exert independent normative pressure, in addition to and sometimes superseding the evidence itself. Weatherson argues that such principles lead to regress. If agents always adjusted their credences to accommodate their own rational judgments about the evidence, they would end up with absurd credences (193-7).

The central claim that epistemology is not about living up to your epistemic principles is strictly independent of whether Change Evidentialism is true. However, most of the arguments of Part 2 rely heavily on the truth of Change Evidentialism, and one reason for this is that normative externalism is less plausible if we assume first-order epistemic theories that include level-crossing principles. Although the arguments are Part 2 provide reasons to resist such first-order theories, the interdependency is worth noting.

Normative Externalism is a rare example of an ambitious and serious project that makes for an engaging and enjoyable read. The writing is clear, the examples well chosen, and the relevance of the many interesting sub-fields dabbled in is never lost. The helpful road map for future objectors is particularly thoughtful (114-6)). The book also deserves special mention as a model of responsible philosophical scholarship. Weatherson is mindful of his intellectual debts, and scrupulous in acknowledging them. This is an important contribution that deserves to be engaged with by anyone serious about normativity, particularly in ethics and epistemology.