In this book, Dancy extends his theory of normative reasons to give a neo-Aristotelian account of practical reasoning by way of a novel account of reasoning in general. According to neo-Aristotelianism, ‘action is as direct, as unmediated a response to reasoning as belief can be’ (p. 42). Just as theoretical reasoning starts from one's beliefs and leads, directly, to the formation or abandonment of a belief, practical reasoning starts from one's beliefs and leads, directly, to the formation or abandonment of an action. This stands in opposition to accounts according to which practical reasoning starts and ends with normative beliefs or intentions or only ends with action indirectly (via some motivational state). Dancy argues for this in three major moves. First, he dismantles arguments against neo-Aristotelianism. Second, he gives an account of reasoning in general and uses this account to give more specific accounts of formal and non-formal theoretical reasoning, moral reasoning, and non-instrumental and instrumental practical reasoning. Third, he argues that these specific accounts are better than prevailing accounts – the major complaint throughout being that the prevailing accounts cannot give an adequate explanation of how the premises ‘make a case’ for the conclusion.

The strongest parts of the book are where Dancy dismantles the arguments against neo-Aristotelianism. (This happens primarily in chapters 1, 8, and 9.) One exits the book thinking that neo-Aristotelianism really should be taken more seriously. But I won't rehearse those arguments here. Instead, I want to lay out Dancy's accounts. I will gloss over a lot of important details and qualifications for the sake of brevity.

For Dancy, reasoning is a process. In general, this process involves six major elements: beliefs (Bn), states of affairs (Sn), a response-type (R), a response-token (r), a dependence relation (D), and a favouring relation (F). The account is this:

The broad distinction between the theoretical and the practical is simply a difference in response (R, and correspondingly r). In theoretical reasoning, the response is believing. In practical reasoning, the response is acting. So, we get the following two accounts:

Acting is as much a direct response to reasoning as believing because (a) acting is as much a kind of response as believing, even if distinct, (b) acting can be directly favoured by states of affairs, and (c) the response-token in 2 depends solely on 1.

Notice that where the orthodoxy would claim that reasoning operates on propositions, Dancy claims that reasoning operates on states of affairs. For example, consider the following argument:

The orthodoxy would claim that the sentences in this argument represent, primarily, propositions. The reasoner believes 1, believes 2, and (if rational) forms a belief in c. On Dancy's account, the sentences in this argument represent, primarily, states of affairs. The reasoner believes that all men are mortal, and this belief represents 1. The reasoner believes that Socrates is a man, and this belief represents 2. These states of affairs (not the propositions) favour believing that Socrates is mortal; c is the state of affairs represented by this response. Where the orthodoxy would claim that the reasoning involves somehow extracting Proposition c from Propositions 1 and 2, Dancy claims that States of Affairs 1 and 2 favour responding in a certain way, and the reasoning tracks this favouring relation, leading to responding in that way. As you can see, favouring plays a central role in Dancy's account.

Now consider an example of practical reasoning:

The orthodoxy would claim that the sentence next to ‘1.’ represents a proposition, but it is not clear what to claim about the sentence next to ‘c’. If we claim that it represents a proposition, all sorts of complications arise. If we claim that it represents a state of affairs (which seems most accurate), it no longer seems a proper part of the reasoning. And it is not at all clear what we should say about the move from 1 to c: how is Proposition or State of Affairs c extracted from Proposition 1? And is this in any way similar to the way the conclusion is extracted from the premises in the theoretical case? For Dancy, the account is straightforward. The sentence next to ‘1.’ represents a state of affairs, and that state of affairs favours acting in a certain way. The sentence next to ‘c.’ represents the state of affairs of responding that way. (Notice that this is different from the theoretical example. In that example, the sentence next to ‘c.’ represents the state of affairs represented by the favoured response, not the instantiating of the response itself. This may signal a problem, which I'll raise below.)

At this point, one might think that surely there must be something more important about the difference between the theoretical and the practical than merely the response involved. And indeed there is. For Dancy, the difference in response makes all the difference in the world, since the difference in response calls for a difference in the explanation of why that kind of response is favoured. When believing is favoured, this explanation involves truth. When acting is favoured, this explanation involves value. This explanation of the favouring relation plays a crucial role in Dancy's accounts of formal and non-formal theoretical reasoning and moral reasoning.

Formal theoretical reasoning is theoretical reasoning where the explanation of the favouring relation involves the structural relations between the propositions representing the relevant state of affairs and the proposition in the relevant response. In other words, this favouring-explanation will have the following form: Sn favours R because the propositions representing Sn stand in a certain truth-preserving, structural relation to the proposition in R. For example, all men are mortal and Socrates is a man favour believing that Socrates is mortal because the propositions that all men are mortal and that Socrates is a man stand in a certain truth-preserving, structural relation to that Socrates is mortal.

Non-formal theoretical reasoning is theoretical reasoning where the favouring-explanation has the following form: Sn favours R because Sn makes it more likely (than otherwise) that the state of affairs represented by R obtains. For example, he had the motive, he had the means, and he had the opportunity favour believing that he did it because those states of affairs make it more likely (than otherwise) that he did it obtains.

Dancy's account of moral reasoning is surprising but compelling. There are two forms: moral theoretical reasoning and moral practical reasoning. The latter is simply practical reasoning where the favouring-explanation is in terms of specifically moral value. Not so surprising. But moral theoretical reasoning is not simply an instance of formal or non-formal theoretical reasoning. Moral theoretical reasoning stands alongside formal and non-formal. In the formal case, the believing is favoured because of structural relations between the relevant propositions. In the non-formal case, it is because the state of affairs to be believed is probabilified. In the moral case, it is because the state of affairs to be believed is made to obtain by the value involved. In other words, the explanation for why the believing is favoured involves both truth and value. Moral theoretical reasoning is still strictly theoretical reasoning because the response favoured is believing, not acting, but it is fundamentally distinct from formal and non-formal reasoning precisely because the state of affairs to be believed is not guaranteed or probabilified but morally valuable.

The distinction between non-instrumental and instrumental practical reasoning is not based on the favouring-explanation. Instead, it is based on the ‘practical purpose’ (Dancy's words) of the reasoning. The practical purpose of instrumental reasoning is to inquire into how to achieve some ‘material’ aim. A material aim is one that is not simply entailed by reasoning practically itself. The practical purpose of non-instrumental reasoning is otherwise. Thus, practical reasoning with the purpose of inquiring into whether to achieve some material aim would be non-instrumental. Notably, the distinction is not based on the propositional attitudes involved. For instance, instrumental reasoning does not, characteristically, involve reasoning from intentions or to intentions. The reasoning starts from beliefs, as all reasoning does, and ends with acting, as all practical reasoning does. The practical purpose frames the reasoning but is not itself an intention or desire from which the reasoning starts, on par with the beliefs that represent the premises.

Dancy argues, for each of these accounts in turn, that they have a significant advantage over prevailing views. In different ways, the prevailing views struggle, where Dancy's do not, to explain how the premises make a case for the conclusion. For instance, prevailing views of instrumental reasoning can't explain how ‘the fact that a necessary means is also sufficient strengthens the case for taking that means’ (pp. 113–14).

Anyone who is suspicious of the favouring relation will be suspicious of Dancy's accounts. The favouring relation is in the general account, so all forms of reasoning involve it. Thus any problem with it infects all the specific accounts.

More seriously, Dancy has not said enough to convincingly argue that his view is neo-Aristotelian. To do this, he must convincingly argue for Part 2 of his account of practical reasoning. It seems to me that Dancy ends up denying Part 2 of this account with his introduction of a practical purpose (pp. 122–4). Further, where Part 2 of the theoretical account is plausible, Part 2 of the practical account is not. Part 2 of the theoretical account states that, in a non-defective case of the reasoning, one comes to believe in a certain way because the states of affairs represented by one's beliefs favour believing in that way. This is plausible because one believes in those states of affairs and solely thereby is also in the business of believing. Therefore, recognition of what belief-type is favoured is enough to lead to the belief-token. Part 2 of the practical account states in parallel that, in a non-defective case of the reasoning, one comes to act in a certain way because the states of affairs represented by one's beliefs favour acting in that way. This is implausible because simply believing in those states of affairs does not entail that one is in the business of acting at all. Therefore, recognition of what act-type is favoured is not enough to lead to the act-token. In other words, believing in the theoretical case is more direct than acting in the practical case.

Consequently, even if Dancy's accounts are better than the prevailing ones, he has failed to give a positive defence of neo-Aristotelianism. But also, the above problem signals an advantage for prevailing views of practical reasoning: they seem to have a better explanation of the practicality of practical reasoning. On John Broome's account, for instance, practical reasoning starts with intending, so the reasoner is, automatically, in the business of taking action. It is no surprise, then, that practical reasoning characteristically leads to an act-token. And Dancy can't simply modify his account to state that practical reasoning starts with intending because the states of affairs represented by future-directed intentions do not obtain and thus can't favour anything. Therefore, reasoning in general would not simply amount to tracking the favouring relation, and Dancy would lose much of what makes his view powerful.

Ultimately, this book is an excellent attempt to upset our orthodox ways of thinking about reasoning.