This insightful book covers an impressive range of topics in moral psychology and moral philosophy. Arpaly and Schroeder (‘A&S’) initially bill their book as defending ‘the side of Appetite in the long-standing philosophical dispute between Reason and Appetite’ (2). But it soon becomes clear that their project is something more: to present an overarching view in which the most central phenomena studied by moral philosophers are to be understood in terms of having the right desires. Along the way, they offer accounts of good and ill will, virtue and vice, praiseworthiness and blameworthiness, acting for reasons, love, prejudice, addiction and more. At times, one would have liked A&S to go further in fleshing out their views. Nonetheless, they consistently engage with the literature in interesting ways, and there is likely to be something of interest here for everyone working in moral philosophy. Moreover, A&S are well-versed in the psychological and neuroscience literature and consequently are admirably attuned to the nuances of our inner experience. In this review, I sketch A&S's most important views and then raise questions about their theories of desire and blameworthiness.

A&S name their package of views ‘Spare Conativism’. At its core lies their theory of good and ill will. The basic thought is that ‘having good will is desiring the right or good (via the right concepts) while having ill will is desiring the wrong or bad (via the right concepts)’ (160). Thus, to have a perfectly good will is to have a desire for the right or good with not only the right reference, but also the right sense (14):

A ‘person has more good will insofar as her intrinsic desire for the right or good is stronger, and less good will insofar as this intrinsic desire is weaker’ (162).

Of course, few will desire the right or good via the correct conceptualization because few fully grasp the true normative theory. Therefore, to avoid the implication that nobody has good will, A&S distinguish perfect good will from partial good will. ‘Partial good will is an intrinsic desire for some part of the right or good, correctly conceptualized’ (162). But what is a part of the right or good? A&S claim that ‘pro tanto moral reasons suggest a principled way of dividing up the whole of the right or good into parts’, such that ‘partial good will can be thought of as an intrinsic desire for something that there are pro tanto moral reasons to bring about’ (166). Accordingly, partial good will is reasonably attainable. After all, it's easy enough to have desires that ‘approximate complete good will by having referential contents that are pro tanto moral reasons, presented by the concepts that the correct normative theory would pick out for such reasons’ (290). (Analogous points hold for ill will.)

Still, good and ill will do not exhaust the field of moral evaluation. In addition, we also think it morally significant to display a conspicuous lack of good or ill will. Thus, A&S say that a ‘person is more morally indifferent the less good will she has’, and ‘more reverse morally indifferent the less ill will she has’ (163).

A&S put this basic account to an impressive array of uses. For instance, they think being ‘virtuous is to have significant good will and lack ill will’, and being ‘vicious is to have significant ill will or significant moral indifference’ (202). Moreover, they claim that a person is praiseworthy for action A ‘to the extent that A manifests good will (or reverse moral indifference) through being rationalized by it’ and blameworthy for A ‘to the extent that A manifests ill will (or moral indifference) through being rationalized by it’ (170). A&S also use their theory of good and ill will to explain other moral phenomena like the cognitions characteristic of virtue or vice (e.g. open-mindedness or prejudice) and specific virtues or vices like modesty or immodesty.

Given the central role assigned to desires, it is crucial to explain A&S's novel theory of intrinsic desire. On their view, neither dispositions to act nor dispositions to feel, say, pleasure, are essential to having a desire (110–25). Rather, they think ‘intrinsic desires cause, but are not constituted by, their familiar effects on motivation or pleasure’ (126). Their positive view – the ‘reward theory’ – centres on the brain's reward-based learning system. Specifically, they claim that having ‘an intrinsic appetitive desire that P is [for one's reward system] to constitute P as a reward’, and having ‘an intrinsic aversion to P is [for this system] to constitute P as a punishment’ (128).

The reward system implements a kind of algorithmic learning wherein ‘one mental state [causes] another, and then that causal sequence is followed by the unconscious release of a signal in the brain’ (129). The signal associated with moving from the first mental state, M1 (e.g. a perception, belief or intention) to the second, M2, can take two main forms. A positive learning signal strengthens the connection between M1 and M2 (i.e. makes M2 more likely to follow M1), while a negative learning signal weakens the connection between M1 and M2. Rewards, then, are understood as follows: ‘[t]he representation of certain states of affairs . . . directly increases the chance that a positive learning signal will be released, and this is what makes these represented states of affairs rewards’ (130). The converse characterizes punishments (131). Thus, what counts as rewards or punishments has ‘nothing to do with feelings of pleasure or pain, self-conscious stances toward states of affairs’, etc. (131). Instead, it ‘come[s] down to variations on the strengthening and weakening of causal connections, nothing more’ (131). Hence, whether P counts as a reward or punishment for a person is determined ‘by the contingent power of that person's representation that P to cause . . . a reward or punishment signal’ (132).Footnote ¹ For example, when you constitute the regard of your friends as a reward, representing that your friends regard you highly causes the release of a positive learning signal – and this is what the desire for the regard of your friends consists in, on A&S's view (136). The result of reward-based learning ‘is that the mental processes – perceptual, intellectual, agentive, and so on – that lead to rewards (so defined) and the avoidance of punishments (so defined) become more dominant over time in the agent's perceptions, thoughts and actions, while those leading to punishments and to missing rewards play a smaller role’ (135).

To make the case for their view, A&S argue that the evidence from neuroscience shows that the reward system causes what desires cause – primarily, actions and pleasure. First consider action. A&S observe that ‘[i]nformation about what is the case and what is intrinsically desired combine in the dorsal striatum with information about what basic actions are available[.] The output of the dorsal striatum controls the . . . motor and premotor cortex’, which in turn controls bodily movements (137–138). While other ‘parts of the brain can override the output of the dorsal striatum, and thus the output of the reward . . . system’, ‘these movements are experienced by their makers as not their own’ (138). Concerning pleasure, A&S claim ‘[t]he evidence that increases in the activity of the reward system causes increases in feelings of pleasure is widespread and powerful[.] Expected but not actualized rewards cause drops in the reward signal, [which] cause[s] feelings of displeasure’ (139). Accordingly, the reward system plausibly causes what we intuitively think desires cause.

This, then, is the notion of intrinsic desire to be plugged into A&S's theories of good and ill will, virtue and vice, etc. However, various aspects of their views might be questioned.

Begin with A&S's theory of desire. It seems to depart from our intuitive concept of desire. This in itself isn't a problem for A&S, since they think desire is a natural kind, the investigation of which can lead to surprising discoveries about the phenomenon being studied (as studying H₂O might yield surprises about water) (143–146). Still, enough differences between the reward theory and the intuitive concept of desire might raise doubts about whether the theory tells the complete story about what desires are.

One difference is that desires intuitively motivate action directly (together with instrumental beliefs), while this is not so on the reward theory. Rather, the theory says that desires impact action only indirectly by affecting mechanisms that make certain behaviours, thoughts or feelings more likely to follow from others. Desires, on this theory, are second-order operators on mental states, while the first-order operators are learning signals. Thus, while it might be that ‘the reward system causes what desires cause’ (as one section is tellingly labelled; 137), it seems the mental states that A&S's theory identifies with desires do not themselves cause what we intuitively thought desires cause.

A second worry is that while desires are thought to figure in our rationalizing reasons for acting (as A&S argue in Chapter 3), it's not clear the reward theory captures all the ways in which desires rationalize action. When we act, we do so to obtain things we desire. But it may be that what we seek in acting is not the same as what our reward systems constitute as rewards. Suppose John's desire for world peace frequently motivates him to act in pursuit of that goal. It does not obviously follow, though, that John's brain constitutes world peace as a reward, such that when he represents that the world is peaceful (or more so) a positive learning signal is released. Empirical investigation is needed to see whether his brain responds this way. Were we to discover that positive learning signals are not released as predicted when he represents that the world is peaceful, would we be forced to say he does not desire world peace? That seems implausible, since this goal frequently motivates John to action. Accordingly, one wonders if some desires (especially complex abstract goals) might not be reducible to states of the brain's reward system. Perhaps A&S can tell a story about how various desires combine to make up complex values of this sort, but more work seems needed.

Another area where more detail would have been helpful concerns blameworthiness. Plugging A&S's theory of ill will into their basic account of blameworthiness yields this:

However, A&S do not say precisely what's involved in manifesting ill will or moral indifference.

First, a few clarifications about rationalization. It is familiar that belief-and-desire pairs can rationalize action – i.e. if the agent believes the act would satisfy a desire of hers. But A&S note that not ‘every action is rationalized by exactly one belief and exactly one desire’ (63). Next, A&S suppose ‘that something akin to maximizing expected satisfaction of intrinsic desires is what rationalizes action’ (63). Thus, an action is fully rationalized if it is the alternative that maximizes expected satisfaction of the agent's intrinsic desires. But an action can also be partially rationalized if it apparently provides ‘some expected desire satisfaction’ (74). Thus, desire D helps rationalize action A iff D is included among those desires of the agent's whose satisfaction she believes would be promoted (e.g. made more likely) by doing A.

Now consider manifesting desires (or their absence). Contrast (i) cases where one happens to have desires and beliefs that would rationalize A with (ii) cases where certain beliefs and desires formed the agent's rationalization for doing A – i.e. were the agent's ‘rationalizing reason’ (54) for it.Footnote ² It is only in (ii) that A plausibly manifests the desires in question, so what matters for blameworthiness is (ii), not (i). In evaluating how blameworthy I am for, say, not picking John up from the airport as promised, it's irrelevant that my desire for revenge against John would rationalize my action if this desire did not actually figure into the rationalizing reason(s) that caused me to act as I did. (Perhaps I didn't pick him up only because I was mugged.) This suggests the following first-pass account:

But this is only half the story. What about moral indifference? A&S aren't explicit about how actions manifest the lack of a desire. But one clue is that they think the absence of a rationalization, say, not to turn left can play a causal role in bringing about the agent's actual left turn – especially if this rationalization's absence seems explanatorily salient (83). Thus, we might think an action manifests the lack of a desire if we’d have expected it to provide a rationalization for not doing the action, but this counter-rationalization was conspicuously absent from the factors that actually combined to produce the action. This suggests:

However, if we use this first-pass account – (a)+(b) – to flesh out A&S's basic account of blameworthiness, then the resulting view is inadequate, as A&S note (188). Suppose Jack and Jill each knowingly give a lost motorist misleading directions. Jack justifiably feels aggrieved by the motorist and misleads her to satisfy his desire for revenge. His action is performed from a desire for part of the bad, and seems somewhat blameworthy. But Jill is truly evil and has bottomless ill will. She misleads the motorist just to inflict gratuitous harm. Since Jill acts on a much worse desire than Jack, she might seem to manifest ill will to a much greater extent. However, it is counterintuitive that Jill is vastly more blameworthy for this particular action than Jack.

In response, A&S claim that the amount of ‘ill will that an action manifests is not the same as the amount of . . . ill will that exists and is being acted on’ (188). ‘An opportunity to [inconvenience] a lost motorist is not typically an occasion for a full display of a powerful [antipathy] to morality. Hence, the strength of desire for the [bad] that is actually manifested in the two cases we imagined is the same. It is on occasions when only a person of extraordinary [hostility to morality] has a rationalization to act in a certain way that the action is one that might manifest extraordinary [ill will]’ (189).

Nonetheless, A&S's proposed restriction – that an action can manifest extreme ill will only if only someone with extreme ill will would have a rationalization to do it – is problematic. Consider breaking up your brother's happy marriage. One possible rationalization here is ill will: you might simply desire to see your brother suffer – clearly part of the bad. Alternatively, the action might be rationalized by a less bad desire, like wanting to protect him from (what you see as) a corrupting influence. Thus, not only a person of extreme ill will would have a rationalization for breaking up this happy marriage. Nonetheless, this action does plausibly provide an ‘occasion’ for manifesting extreme ill will. In particular, whether the action manifests it or not depends on whether the action was actually performed because of the first rationalization (a strong desire for some bad) or a more benign rationalization like the second. Thus, it seems false that actions can provide occasion to manifest extreme ill will only if rationalized solely by a strong desire for (part) of the wrong or bad.

One might object that breaking up your brother's marriage from a desire to harm him is not the same action as doing so from a desire to protect him, and it is only the former that provides occasion to manifest extreme ill will. However, this response fails. If one adopts it, A&S's proposed constraint no longer solves the problem we started with. This response entails that inconveniencing the lost motorist out of extreme ill will is a different action from doing so out of mild ill will, and while the latter couldn't be the occasion to manifest extreme ill will, the former still is. After all, only someone with extreme ill will would have a rationalization for the fine-grained act of inconveniencing-the-motorist-from-a-strong-desire-for-the-bad. So the initial problem resurfaces. We again face the implausible result that a minor act of inconveniencing a lost motorist can be enormously blameworthy.

A second problem is that the extent to which an action manifests a desire may depend not just on its strength, as A&S suggest (162, 189), but also on how much causal impact it had on the action. Suppose Jerry and George each cut off another driver because of two desires: (i) to get home in time for a daughter's birthday party and (ii) to avenge having been cursed at. Suppose the desires have the same strength for both men. But while (i) was more causally efficacious in producing Jerry's bad behaviour than (ii), (ii) played a larger role than (i) for George. Jerry was intently focused on getting home in time, and was only slightly moved by his desire for revenge. George was more focused on getting revenge than on getting home. Thus, George's action seems to manifest ill will to a greater degree than Jerry's action – even though the desire for revenge was supposedly equally strong for both. Accordingly, more clarification seems needed of what's involved in an action's manifesting ill will.

Although A&S's views thus prompt questions, that is the mark of any thought-provoking contribution in philosophy. Their book warrants a careful read.