In Making Morality Work, Holly M. Smith investigates the requirement that moral theories should be action guiding. Smith outlines how our cognitive limitations prevent moral theories from being action guiding, and thereby give rise to what she calls the problems of error and uncertainty. She then considers various responses to these problems. In this review of her book, I first provide a brief summary of its contents, and then go on to make some critical remarks.

The first chapters set the stage for the rest of the book. In chapter 1, Smith notes that moral theories are commonly required to play two roles. First, they should play the theoretical role of providing ‘a correct account of the features that make actions right or wrong’ (p. 2). Second, they should play the practical role of helping people choose which actions to perform and not to perform (p. 2). The second role gives rise to the requirement that moral theories should be action guiding. In chapter 2, Smith distinguishes between two ways in which a principle can be action guiding. A principle is usable in the core sense if and only if an agent would derive a prescription for action from it if she wanted to (pp. 16, 31). In contrast, a principle is usable in the extended sense if and only if an agent would derive a prescription for action from it if she wanted to, and where this action is such that it can actually be performed by her and is in fact prescribed by the principle (pp. 21, 31–2).

In chapter 3, Smith moves on to discuss the various cognitive limitations which can stand in the way of moral principles being action guiding. She focuses on two problems. First, a principle may fail to be action guiding because we make errors about non-moral facts, such as whether a health care policy will lower medicine prices – Smith calls this the problem of error. Second, it may fail to be action guiding because we are uncertain about non-moral facts – Smith calls this the problem of uncertainty (Smith treats ignorance about non-moral facts under the same label of uncertainty).

Non-moral error is a problem for the extended usability of a principle, since we may fail to derive a prescription that we can perform or that the principle prescribes. Non-moral uncertainty, however, is a problem not only for the extended but also for the core usability of a principle, since when we are uncertain, we may fail to derive any prescription whatsoever.

In the rest of the book, Smith systematically applies three different responses to the problems of error and uncertainty. She calls these the pragmatic, austere and hybrid responses.

Pragmatic responses say that ‘the theoretical function of morality cannot be isolated from its practical or regulative function’ (p. 47) and that a moral principle ‘must … be capable of guiding us’ (pp. 47–8). According to pragmatic responses, ‘one crucial test of a moral principle's theoretical correctness just is its practical usability’ (p. 47).

In sharp contrast to pragmatic responses, austere responses say that ‘a moral principle's practical usability, or lack thereof, is no test of its adequacy or inadequacy as a theoretical account of right and wrong’ (p. 49). If we cannot use a moral principle to guide our actions, ‘that is a defect in us, not a defect in the theory’ (p. 50).

Finally, hybrid responses agree with austere responses in that whether a moral principle is true is not determined by whether it is action guiding. However, according to the hybrid response, a principle which is not action guiding must be supplemented by appropriate second-level decision rules which are (pp. 51–3). According to hybrid responses, the complete moral theory consists of both (i) the moral principle giving the correct account of right and wrong, and (ii) the appropriate second-level decision rules.

In chapters 4–7, Smith discusses pragmatic responses to the problem of error. Ideal pragmatic responses say that a moral theory should be fully usable, while non-ideal responses only say that a moral theory is better than another theory in so far as it has higher so-called usability value. Smith shows that both ideal and non-ideal responses leave us with new epistemic difficulties – it is difficult to identify which theory is fully usable, or which theory has the highest usability value.

In chapter 8, Smith instead considers austere and hybrid responses to the problem of error. She rejects hybrid responses to the problem, because they run into epistemic difficulties – it is epistemically challenging to identify the second-level rules which we need to apply to our first-level principles. She concludes that austere responses to the problem of error are superior to both pragmatic and hybrid ones. Austere responses do not have the epistemic problems of pragmatic and hybrid responses, and they still respect part of the rationale for why theories should be action guiding. For example, even if we are in error about what a principle prescribes, we can have what Smith calls a degree of autonomy, which is ‘a way to make decisions that accord with [our] values’ (p. 206). This is because we can still use the principle in the core sense – we can still use it to derive a prescription for action, even if it is not an action that we can perform, or one that the principle actually prescribes.

In chapter 9, Smith goes from discussing the problem of error to discussing the problem of uncertainty. As I mentioned earlier, uncertainty threatens the core usability, and not just the extended usability, of moral theories. Because of this, Smith argues that an austere response to the problem of uncertainty cannot guarantee us autonomy.

At this point in the book, Smith proposes a combination of responses: the austere response to the problem of error, and the hybrid response to the problem of uncertainty. She calls combinations of austere and hybrid responses hybrid systems. In chapters 10–13, Smith then tries to find an acceptable hybrid system. Very briefly, this is what happens next: Smith distinguishes between what is objectively and subjectively right and wrong. She then proposes that we accept a hybrid system which consists, among other things, of (a) a principle giving us the correct account of what is objectively right and wrong, and (b) multiple hierarchically ordered decision-guides stating what is prima facie subjectively right and wrong, and which are tailored to different agents’ individual epistemic situations. Working out the details of this view takes up the rest of the book.

Having sketched the main themes of Making Morality Work, let me now point out a few passages where I was less than fully satisfied with Smith's discussion.

First, I would have liked more discussion of which responses are available to the problems of error and uncertainty. In particular, there seems to exist a mixed austere and pragmatic response which Smith does not discuss. According to this response, a moral theory should play both the theoretical and practical role, but failing to play the practical role does not matter for whether the theory is true. Instead, it matters for whether it is normatively interesting, useful or important. On the mixed response, both truth and these other qualities then matter when we evaluate how good a moral theory is. This is similar to how Newton's law of universal gravitation is evaluated: it is interesting, useful and important, and thereby a good theory to some extent, even if it is also false. I do not know what we should say of these mixed responses to the problem of error and uncertainty, but they strike me as at least as plausible as the pragmatic responses, and as worthy of discussion.

Second, I would have liked more clarification of the concept or property of autonomy. According to Smith, a person lacks autonomy if she cannot ‘make decisions that accord with [her] values’ (p. 206) or ‘translate her moral values into a choice of what to do’ (p. 219, cf. pp. 194–5). Smith then claims that if a moral theory is not usable for guidance in the core sense, it prevents agents from having autonomy. But few of us have our values aligned with any moral theory whatsoever. For example, few of us have our values aligned with utilitarianism, including utilitarians themselves. Therefore, a utilitarian austere response which lacks core usability will have no issue with respecting our autonomy, because if we do not have our values aligned with utilitarianism, then utilitarianism does not prevent us from making decisions that accord with our values.

To sum up my impressions of Making Morality Work, I found it clearly written, carefully researched and filled with original contributions. I am happy to recommend it to any researcher interested in the topic of action guidance.