Why and When is Pure Moral Motivation Defective

Abstract

Agents sometimes have a final, de dicto desire to do what is right. They desire to do what is right for its own sake and under this description. These agents have pure moral motivation (PMM). It is often surmised that PMM is in some sense defective. Most famously, it has been suggested that PMM manifests a kind of moral fetishism. However, it also seems defective if an agent shows no concern whatsoever for moral rightness in their motivations. In this paper, I attempt to resolve this puzzling tension. I argue, first, that PMM is defective insofar as it manifests a failure to respond to reasons. I argue, second, that not every instance of PMM manifests a responsiveness failure. In particular, we need to distinguish between the de dicto final desire to do what is right and the de dicto final desire to do what is right for the right reasons.

1 Pure Moral Motivation

Agents sometimes have what I will call—following Lerner (2018)—pure moral motivation (PMM). An agent has PMM if they have a final, de dicto desire¹ to do the right thing. They have a final de dicto desire to do the right thing if they desire to do the right thing not merely as a means to satisfying some other desire, such as the desire to fit in or the desire to do what pleases God. Instead, to have a final desire to do the right thing is to desire to do the right thing for its own sake. They have a final de dicto desire to do the right thing if they desire to do the right thing as such. An agent who desires to do the right thing de dicto desires to do what is right under this description—the right thing to do. Other agents may desire to save the drowning child or feed the hungry man. They thus arguably desire to do the right thing, but not under the description of it being the right thing. They desire to do the right thing de re.

The possibility of PMM has given rise to various forms of philosophical puzzlement (Carbonell, 2013; Lerner, 2018). Susan Wolf (1982) was famously puzzled in part by the fact that while having PMM seems to be an essential ingredient in the description of the very best moral agents—the moral saints—it would at the same time make these agents appear abnormally and even defectively focussed on rightness/wrongness itself. There is prima facie something off about agents who care about doing what is right, but don’t care at all about why it is right. These agents seem to be guilty of making morality into a fetish—as Smith (1994, p. 71–76) has called this defect. However, in the wake of Smith’s criticism, many have argued that there are many cases in which PMM not only seems perfectly fine and non-defective, it even seems necessary for having a good kind of moral motivation, especially in cases of uncertainty about what the right thing to do is (Lillehammer, 1997; Olson, 2002; Sliwa, 2012, 2015; Sobel, 2001; Svavarsadottir, 1999). These responses lend support to the first part of Wolf’s puzzlement: the assumption that PMM is often indispensable for the make-up of a good moral agent.

There is then a noticeable tension in our thinking about PMM. One way to respond to this tension would be to disregard one of the intuitions that lead to it—that is, to either deny that there is anything wrong with PMM or to deny that PMM is at least sometimes important for good moral motivation.² In this paper, I shall explore a more conciliatory strategy. My suggestion will be that the two attitudes towards PMM are not in fact in tension once we have cleared up two imprecisions in our thinking about PMM: First, we need to explore more thoroughly what exactly is defective about moral fetishism, what kind of defect it is, or, in other words, why PMM is defective. This will afford us a clearer view about what exactly the tension consists in. Second, we have to be more precise about whether all instances of PMM are defective in this way, or, in other words, when PMM is defective.

My answer to the first question is that PMM is defective insofar as it manifests a failure to respond to reasons—which makes it a rational failure. PMM is unresponsive to reasons because it is a response to moral rightness, and moral rightness is not itself a reasons-giving property, it is the property of having such properties. Ideas like this have been implicit in some parts of the literature, but it is worth spelling them out. The account of the defect will rest on controversial assumptions, but it will prove illuminating even to those who don’t share these assumptions. My consequent answer to the second question will be that we need to distinguish the final de dicto desire to do the right thing from the final de dicto desire to do the right thing for the right reasons. While the former manifests a failure to respond to reasons, the latter does not. Thus, what I propose is that while we should rightfully find unresponsive instances of PMM defective, we can still maintain that responsive instances of PMM are a perfectly fine or even indispensable part of a good kind of moral motivation.

The paper proceeds as follows: In Sect. 2, I argue that moral fetishism is a rational defect because it is a failure to respond to reasons. In Sect. 3, I review example cases that strongly favour the conclusion that PMM is at least sometimes rational. In Sect. 4, I attempt to distinguish two types of PMM, only one of which counts as unresponsive to reasons. Section 5 develops my solution further by addressing worries about it.

2 Why PMM is Irrational

Despite the fact that many share the intuition that PMM is somehow defective, a curious lacuna exists in the literature when it comes to spelling out this defect. What is it that we find defective about an agent motivated by PMM and—let us suppose for the moment—no additional motivation?³ One way of expressing the intuition, I take it, is that an agent motivated by PMM does not show concern ‘at the right level of morality’, as it were. They should be concerned for the right-making features of an action, yet what they are concerned for is only whether the action is morally right or not.

But what does it mean that an agent with PMM does not show concern at the right level? In order to unpack this, it is helpful to look at what Susan Wolf had to say about her version of an agent motivated by PMM. She expresses the intuition about the defect as follows:

[T]here is something odd about the idea of morality itself, or moral goodness, serving as the object of a dominant passion in the way that a more concrete and specific vision of a goal (even a concrete moral goal) might be imagined to serve. Morality itself does not seem to be a suitable object of passion. (Wolf, 1982, p. 424, my italics)

I want to focus on the last sentence of the quote. What Wolf seems to be saying here is that motivational states are subject to a standard of correctness such that some of their actual objects are suitable and some are unsuitable. Her explanation for why PMM is defective is that it is a motivational state with an unsuitable object. I think this is the correct explanation of why PMM is defective and a correct explication of our intuitions about it. But it leaves two crucial questions open: What gives us the relevant standard of correctness for motivational states and why does PMM violate this standard? I will construct my explanation of why PMM is defective by answering these questions in turn. I should note at the outset that my answer—that PMM fails to respond to reasons—is implicit in some of the literature on the topic (for example in Stratton-Lake, 2000, ch.1; Toppinen, 2004; Shoemaker, 2007, 88; Markovits, 2010, 207; Johnson-King, 2019; Hicks, 2019).⁴ So it is not entirely new. My effort is to understand these tendencies better by making them explicit.

What is the standard of correctness for moral motivation then? In recent literature, reasons have played an increasingly important role as fundamental normative concepts, so they offer a good common ground for spelling out Wolf’s thought. I will follow orthodoxy in assuming that reasons are considerations that speak in favour of actions and attitudes. “Considerations” are roughly facts.⁵

To say that motivational states have a standard of correctness, speaking in terms of reasons, is to say that they are attitudes that are sensitive to a range of reasons that apply to them. An attitude is correct in this sense then, if it is the correct response to the relevant reasons.⁶ What exactly qualifies as a correct response to reasons cannot be answered exhaustively here. As a preliminary characterisation, it should suffice to say that some attitude will count as a response to reasons only if those reasons figure in an explanation⁷ of why the agent is holding that attitude. If my action of saving a drowning child is motivated by the thought that the child needed help, then the fact that the child needed help should figure in an explanation of why I have this thought. Otherwise, the connection between my thought that the child needs help and the fact that it needs help may turn out to be merely accidental. And whatever we will say about my motivation then, we will surely not say that it was a response to the fact that the child needed help. I will say more about how desires can be subject to the relevant type of explanation in Sect. 5.

However, if we want to say that a motivational state is correct qua moral motivation, it does not suffice to require it to be a response to reasons. It does not suffice because this fails to distinguish responses to non-moral reasons from responses to moral reasons. A moral motivation to φ is correct only if it is a response to moral reasons to φ.⁸ An agent motivated by prudential reasons to save a drowning child might do the right thing, but not because it is the right thing. This manifests a failure to be responsive to the right kind of reasons for the motivation in question (Markovits, 2010). The right kind of reasons in the moral case are the reasons that favour performing the action morally, that is, the kind of reasons that make the action morally right to perform.

What I have said so far amounts to the following claim: A moral motivation to φ is correct only if it is a (correct) response to the moral reasons to φ—those features of a situation that make it morally right to φ. It is also possible to say more about what kind of standard of correctness is at issue now. To be responsive to reasons, traditionally [but not uncontroversially: Broome (2013) has famously denied this position] has been viewed as a way of being rational. Another way to explain the relation between reasons and rationality according to reasons-responsiveness views is to say that reasons are those features of a situation that together give a definite answer to the question what the agent ought to do overall. So, if the agent performs the action supported by the balance of reasons, the agent will do what they overall ought to do, that is, what they rationally ought to do. This explanation points to an important caveat in my requirement about responding to the right reasons. To respond to the right reasons to φ morally is to respond to those reasons that determine what the agent ought to do morally. Insofar as rationality concerns what the agent ought to do overall, there will be a rational constraint of acting for the right (that is, in our case, moral) reasons in only and all of those cases where the moral reasons decisively determine what the agent ought to do overall. I do not take this caveat to detract from the argument of this paper, as I think it is a precondition of the debate about de dicto moral motivation that it is interesting to think about especially in cases where moral reasons are clearly decisive.

This being said, what I have spelled out so far is a rational and markedly rationalist⁹ constraint on moral motivation:

(R) A moral motivation to φ is rational only if it is a response to the moral reasons to φ.

This seems to me not only a plausible explication of what Wolf means by “suitable” in the above quote, it is also an extremely plausible constraint on moral motivation in general, provided you think there ought to be any connection between moral motivation and rationality.

(R) is a plausible explication of the standard of correctness that Wolf seems to have in mind for motivational states. This answers my first question about her quote. But it remains unclear why PMM would not satisfy (R). That is to say, we have yet to see why PMM is not a response to moral reasons to φ.

The answer to this second question comes from a view that naturally fits with the picture about reasons for action that I started with. If reasons are the ground-level normative features of a situation, then it is plausible to assume that moral rightness and wrongness themselves are not reasons-giving features of a situation. They are, as it is sometimes put, second order properties: to say that an action is right is to say that it has the property of having (sufficient) right-making features. A rough and ready argument for this view is to point out that when all (or a sufficient amount) of the right-making features of an action are listed, the fact that the action is also morally right will not do any additional justificatory work. It does not therefore present an additional moral reason to perform the action. This view of rightness as a sort of “summary concept” (Wallace, 2006, p. 332) has come to be known as deontic buck-passing¹⁰:

(D) Moral rightness is not a reasons-giving property; it is the property of having such properties.

However, I need to clear up a short complication before I can use (D). The literature on buck-passing has distinguished between a negative and a positive buck-passing thesis (see for example Schroeder, 2009, 341; Väyrynen, 2006, 296). The positive thesis claims that rightness (in our case) is just a higher order property (or fact). The negative thesis claims that this property does not provide a reason. The rationale for separating these two theses is that the negative thesis does not follow from the positive thesis, as the latter does not exclude the idea that rightness provides a derivative or non-additional reason. We might say, very roughly, that a fact F is a non-additional reason to φ if there is a distinct fact F’ such that F is a reason to φ because F’ is a reason to φ. If (D) is true, rightness may not provide an additional reason, but it may very well provide a non-additional reason.

While this complication is an important result for the debate about buck-passing, it is not important for my argument. The argument in this paper is that PMM is irrational because it is not a response to reasons. In this picture, attitudes that fail to be responses to reasons are irrational because they fail to latch onto determiners of what the agent overall ought to do/desire (Lord, 2018, ch.8). But non-additional or derivative reasons are, by definition, not determiners of what the agent overall ought to do. They have no “independent normative force” (Parfit, 2011, p. 39). This is because they are explained by facts about what the agent ought to do, or by the facts that determine what the agent ought to do. And it is impossible for p to both explain why the agent ought to ϕ and be explained by the fact that the agent ought to ϕ. There might be a weaker notion of reasons-responsiveness as rationality which is not connected to what the agent overall ought to do. If there is, it is not the notion at issue.

Hence, we should, for my purposes, read (R) above as saying that moral motivation to φ is rational only if it is a response to non-derivative or additional reasons to φ. Accordingly we should also read (D) as saying that rightness does not give us non-derivative or additional reasons, which is an uncontroversial claim even if we reject the negative claim about deontic buck-passing.

With this out of the way, the conjunction of (D) and (R) provides us with an interpretation of Wolf’s claim that morality itself, that is, moral rightness, is not a suitable object for moral motivation. Any moral motivation to do what is right de dicto will be a response to the property of moral rightness. However, as (D) tells us, the fact that φ-ing is right does not provide an (additional) reason to φ. And as (R) tells us, any moral motivation that is not a response to an (additional) moral reason to φ is not rational. Hence, moral rightness is an unsuitable object of moral motivation. (R) and (D) therefore together imply that PMM is not rational. Since it is a necessary feature of PMM that it is a response to the moral rightness of a situation de dicto, PMM necessarily is not a response to a reasons-giving property. So, as it turns out, PMM is necessarily irrational.¹¹ For purposes of clarity, here is the argument in premise/conclusion form with the modification “additional” omitted for brevity’s sake:

Rational Motivation

• (1) Moral motivation to φ is rational only if it is a response to moral reasons to φ. (R)

• (2) The fact that φ-ing is right is not a moral reason to φ. (D)

• (3) PMM is a response to the fact that φ-ing is right. (Definition of PMM)

• (4) Therefore PMM is not a response to moral reasons to φ. (from 2,3)

• (5) Therefore, PMM is not rational. (from 1,4)

I started out this paper by wondering about what sort of defect moral fetishism could be. Rational Motivation provides us with a straightforward explanation of the defect. As I said, I think the driving intuition behind the fetishism charge is that agents with PMM seem to aim at the wrong level of morality, as it were. With Rational Motivation, we can spell out this charge clearly. To say that a motivational state “aims at the wrong level of morality” just is to say that it is a response to a property that is not itself a moral reason to do anything. To be preoccupied with bringing about actions that bear this property therefore is to miss the point about moral action. To be correctly morally motivated is for our motivation to be a response to moral reasons—the ground-level normative features of a situation. The discomfort we feel with agents motivated by PMM is then that they are not properly responsive to reasons, which many think counts as a rational failure. PMM is defective because it is irrational.

I should point out two things about Rational Motivation:

First, you may ask yourself what is gained by calling PMM irrational instead of tracing its defect to another feature such as a lack of moral worth (Lord, 2017; Markovits, 2010). Notice that in the rationalist framework of Rational Motivation, if PMM is irrational, then it is also not morally worthy.¹² So the two proposals are not incompatible. However, I think that it is helpful to focus on PMM as a rational defect because it allows us to see that moral fetishism is only a particular instance of a more general mistake—we might call it normative fetishism. It seems to me that we can find cases structurally identical to moral fetishism cases in many or all normative domains. Imagine for example someone who knows a lot about what makes art beautiful, but who is not motivated to collect art for those reasons. Instead, they collect art solely because it is beautiful. We can legitimately call such a person an aesthetic fetishist. They are defective in the very same way the moral fetishist is, but their defect clearly can’t be about moral worth. If there is normative fetishism, then we should expect Rational Motivation to generalise. And it does. Just replace all instances of “moral” with a variable for normative domains and all instances of “right” with a variable for a normative property in that domain. If that property is a second order property (as we might think beauty is), then the generalised Rational Motivation will support and explain our intuitions about fetishism in that domain.¹³ Only a focus on fetishism as a rational failure will reveal this significant explanatory potential of Rational Motivation.

Second, relatedly, I should emphasise that one central point of Rational Motivation is to situate the problem with moral fetishism dialectically. So it should provide an interesting framework for thinking about PMM even for philosophers who reject one (or both) of Rational Motivation’s controversial premises. That is, even if you are not convinced of the moral rationalism inherent in (R) and even if you are not convinced that a buck-passing account of moral rightness is correct, it should be informative that if these views are correct, then they together imply that PMM is irrational. In addition to helping us situate which kind of natural combination of views will most easily come with a theoretical (and intuitive) commitment to PMM being defective, this conditional information should also help represent the dialectic about de dicto moral motivation better. For example, it clearly identifies the views that will have to be given up in order to avoid the conclusion that PMM is irrational or—if you prefer the tollens formulation—pinpoints the views falsified by the assumption that PMM is clearly rational.

However, if you, like me, have at least sympathies towards both substantial premises in Rational Motivation, then the assumption that PMM seems at least sometimes perfectly rational should instil a sense of puzzlement, a puzzlement we are now in a position to better understand: PMM cannot both be necessarily irrational and sometimes rational. It is this tension that will occupy the next sections. In order to see whether PMM really does seem rational in some cases, we have to briefly rehearse these purported examples. I will do this in the next section. After having rehearsed the examples and having found no flaw in them, I will attempt to give a sketch of a solution to the puzzle based on the thought that PMM might have different types of de dicto content.

3 Why PMM is Sometimes Rational

The argument for thinking that PMM is sometimes rational is based on examples. Some of these cases are less suitable than others to establish the conclusion that PMM is sometimes rational however. Let me therefore briefly discuss some cases which I find less helpful before I present my preferred class of examples.

One prominent idea in the literature is that de dicto motivation is unproblematic, and even performs crucial tasks (such as, but not limited to, functioning as a motivational backup should the relevant de re belief fail to produce the corresponding motivation) when it occurs in conjunction with de re motivation (Carbonell, 2013, pp. 470 ff.; Copp, 1997, pp. 49–50; Lillehammer, 1997, p. 192; Olson, 2002, p. 91; Svavarsdottir, 1999, p. 199). This idea grasps something important about why and when PMM is defective, but, I believe, imperfectly so. For the central puzzle with which this paper is concerned arises with respect to cases where the agent is motivated solely by PMM. If there are cases in which such an agent is rational in having PMM, then they are more pertinent to figuring out why and when PMM is defective. For they show that PMM may lack the relevant rational defect even though it does not occur in conjunction with a suitable de re motivational state. Thus, they show that whether or not PMM occurs in conjunction with de re states is not essential to our puzzlement about it. Methodologically speaking, we should prefer those pure cases because they allow us to isolate factors more clearly. And indeed I think it is evident that there are such cases.

One easy formula for devising cases in which an agent solely motivated by PMM still seems rational in being so motivated is to construct circumstances which render the agent unable to have the relevant de re states. Lerner (2018) presents the case of Painfree, an individual who cannot feel and has never felt pain before, and so does not know what it is like to feel pain. This individual needs to rely on de dicto testimony that, say, pain is a reason not to torture. And if they do, they may do so because they want to do the right thing de dicto. In such situations, an agent who has PMM still seems rational.

However, I worry about how much such cases can really be taken to show. The intuition that underlies them is some version of ‘Ought Implies Can’ (OIC). How can an agent be faulted for acting from PMM, the intuition goes, when this is the only motivation she can act upon. But it is notoriously unclear how we should interpret OIC. Many take it to establish that agents can be subjects to deontic requirements without being blameworthy for violating them (Zimmerman, 2004). But if we choose this interpretation of OIC, we are not establishing that some agents might be rational in having PMM, we are establishing that although they are irrational in having PMM, we cannot rationally blame them for it. But there is no puzzle in thinking that we are sometimes not rationally blameworthy for failing to respond to reasons. So this way of reading the cases establishes too little.

We might then instead adopt an interpretation according to which OIC is a general constraint on what kinds of actions and attitudes can be subject to the demands of rationality. Under this interpretation, agents who have PMM because they cannot have any de re states are non-defective on the deontic level, because the demands of rationality do not even apply to them. But all we can establish with OIC-reliant cases is that PMM is sometimes subject to a very general constraint on rationality—that PMM is “protected”, as it were, from the charge of irrationality for general reasons that say nothing about the nature of PMM. PMM might then still be irrational qua the motivational state that it is, while this judgment is overridden by a general judgement triggered by OIC. Because we cannot plausibly rule out that such overriding takes place, intuitions cannot be neatly separated, and OIC-reliant cases like Painfree lack dialectical teeth. They move us no closer to deciphering the puzzle of whether PMM, qua the motivational state that it is, is irrational.

I will therefore focus on a type of case discussed by Lerner that most clearly, and without interfering intuitions, presents an instance of PMM that is clearly rational. In this type of case, the agent has not yet gained moral knowledge about a certain domain, but is motivated to gain knowledge because they want to do the right thing. Because they don’t know what the right thing to do is de re before having completed their inquiry, they can only have a desire to do the right thing de dicto. So they engage in moral inquiry because they have PMM. This seems rational.¹⁴ Here is Lerner’s version of the case.

Factory

Delilah owns a factory farm. She knows everything there is to know about what goes on inside a factory farm. She also knows everything there is to know about how factory farms impact the environment, the economy, and people’s gustatory experiences. But Delilah has never thought about whether it might be wrong to run a factory farm. Or maybe she has given it some thought, but she hasn't come to a settled view. She now faces a decision: should she sink time and effort into trying to figure out whether running a factory farm is right?

Suppose Delilah chooses to deliberate. She seems perfectly rational in making this choice. Why would Delilah make this choice? Clearly, she has a desire to know what’s morally right in this case. But presumably she doesn't want this bit of moral knowledge for its own sake. She wants it because she wants to do what’s right.” (Lerner, 2018, 125).

In Factory, Delilah has PMM. She wants to do the right thing, which is why she chooses to investigate. Moreover, Delilah seems to be clearly rational¹⁵ in her pursuit. And if her moral inquiry is rational, then the motivational state that motivates it, PMM, must also be rational. Delilah has no accompanying de re motivational states, nor is she in principle unable to form them—but she does need to decide to investigate in order to form them. So this case exhibits none of the flaws of other case groups discussed above. It seems then that Factory provides us with a quite solid case for thinking that PMM is rational.

We can now take a clear view of what I take to be one of the central philosophical puzzles about PMM. While Rational Motivation provides us with a precise and plausible account of what makes PMM defective that implies that PMM is irrational in every instance, Factory presents us with a clear instance of PMM which seems perfectly rational. Thus, we have both good reason to believe that PMM is always irrational and good reason to believe that PMM is sometimes rational. This is puzzling, to say the least. Absent any third option, we either have to bite the bullet and claim that despite appearances, Delilah is really not rational in Factory, or give up one of the premises of Rational Motivation. As I said in Sect. 2, if you think either deontic buck-passing or rationalism should be abandoned anyway, then Rational Motivation + Factory will have given you a further argument for this position. But those convinced both by the premises of Rational Motivation and Factory will have to respond to the tension between the two. I think that attempting to solve that puzzle can produce an interesting and more precise answer to the question of when, that is, in which cases, PMM is irrational. I will attempt to sketch a solution in the next section. The solution is based on the idea that there are at least two ways to have PMM, only one of which will trigger Rational Motivation.

4 A Solution

Puzzles like the one created by Rational Motivation and Factory often point to a lack of conceptual precision in our thinking about the context in which the puzzle arises. More specifically, they point to a failure to make an important conceptual distinction somewhere in the train of thought that leads to the puzzle. I think this is true of the puzzle at hand as well. Indeed, I think we should pay closer attention to two different ways in which Delilah might instantiate PMM in Factory.

In order to see what I mean, consider two versions of Delilah–Delilah* and Delilah#. Delilah* and Delilah# exhibit a dispositional difference in what they consider an acceptable answer to their moral inquiry. This difference manifests in the following way: Imagine that there are two books, written by a moral expert on factory farming. One book simply contains a list of all the right and wrong actions in factory farming, and might end with a simple answer to whether factory farming, as whole, is right or wrong. What it does not contain, importantly, are any explanations as to why any given action appears in the right (wrong)-column of the list. It contains, in other words, no information about any of the right-making features of these actions. The other book however does contain precisely this kind of information. It is about what makes factory farming right or wrong, not merely about whether it is right or wrong. The difference between Delilah* and Delilah# is that Delilah# is disposed to consider her moral inquiry successfully completed after reading the list-book, while Delilah* is not so disposed. Delilah* will only consider her moral inquiry successfully completed once she has read the more comprehensive book. So, for example, Delilah# will be happy to accept a list that contains “putting chicken in small boxes is wrong” and “cooping up cattle in small spaces is wrong” without further explanation, while Delilah* will only be satisfied by a list that contains “putting chicken in small boxes is wrong because it causes them pain”.

Now, going back to the original Factory case, we can see that it fails to provide us with enough information to ascertain whether its protagonist is like Delilah* or Delilah#. However, this information is crucial for our intuitions about the case. For consider that Delilah in Factory is like Delilah#, happy to just accept a list of rights and wrongs about factory farming. This agent is the paradigm of a moral fetishist, an agent solely concerned with the right, and not motivated at all by the reasons that make the action right. So, if Delilah in Factory is like Delilah#, then there is no puzzle, for Delilah is clearly not rational in that case. It seems then that in judging Factory to be a case of rationality, we implicitly assume that Delilah is like Delilah*: Interested in how the de re contents, which she has under non-moral descriptions (Lerner, 2018, 125), make factory farming right or wrong—and therefore, it seems, concerned for the level of right-makers.

How does this distinction help us resolve the puzzle? It helps us because the difference in what answers to their guiding question Delilah* and Delilah# accept points to a difference in their de dicto motivational states. Recall that Delilah in the original example does not collect moral knowledge for its own sake, but because she wants to do the right thing (has PMM). This is simply a stipulation of the case. So insofar as she gathers moral knowledge instrumentally because she wants to do the right thing, constraints on what type of moral knowledge will be acceptable can only be generated by what exactly it is she ultimately wants to do. Therefore, a difference in the type of knowledge she accepts as an answer to her inquiry implies a difference in the motivational state that ultimately guides this inquiry. The motivational state that guides Delilah’s inquiry is PMM. So the difference between Delilah* and Delilah# implies a difference in their respective PMM. Both Delilah* and Delilah# still want to do the right thing for its own sake as such, so their motivations both still satisfy the sufficient condition I gave for PMM. But Delilah*’s PMM is different because it comes with stricter constraints on what type of moral knowledge will count as instrumentally sufficient for the goal spelled out in PMM.

These considerations are sufficient for showing that there are at least two types of PMM, but they do not yet tell us how these versions of PMM differ in content, if they differ in content at all. It is here where all I can do is sketch an answer that prima facie appears to solve the puzzle. The answer I want to sketch can be derived from my reasoning about the instrumental desire to gain moral knowledge that is guiding Delilah’s moral inquiry. As I said, the fact that what satisfies this desire differs between Delilah* and Delilah# can only be explained by a difference in what the object of this desire is instrumental to, i.e. to doing the right thing as such for its own sake. This type of reasoning also provides us with a guide to finding the right kind of content for Delilah*’s PMM, let’s call it PMM*. In order to find out what exactly it is that Delilah* wants to do, we have to ask what sort of description of doing the right thing will be such that only gaining knowledge of the right-making features of a situation will enable the agent to do the right thing under this description. As we saw, the object of PMM* cannot simply be the action that is the right thing to do. If Delilah* only wanted to perform the action that is the right thing to do, Delilah* would be satisfied with being given a list of right actions. She would then be indistinguishable from Delilah#. Instead, the desire to do the right thing that is only satisfied if the agent performs the right actions with de re knowledge of what makes them right is the desire to do the right thing for the right reasons. It is this de dicto desire—the desire to do the right thing for the right reasons—that distinguishes Delilah* from Delilah#. To have PMM* then, instead of PMM# is to desire to do the right thing for the right reasons instead of merely desiring to do the right thing.¹⁶

The distinction between PMM* and PMM# solves the puzzle created by Rational Motivation and Factory. This is because while PMM* is still a de dicto desire, it manifests a concern for the right level of morality. In fact, it manifests concern for both levels of morality. It is the desire to do the right thing because of those features that make it the right thing. Because the content of PMM* is this complex relational property, we can also still classify it as a response to reasons—albeit a de dicto response to reasons. Recall that PMM# turns out to be irrational because it is not a response to a reasons-giving property. Some attitude is a response to a feature, I assumed throughout this paper, only if the feature that figures in the content of the attitude is involved in explaining why the agent has an attitude with this content. We can visualize this by bracketing the object of the relevant desire. For any given instance of PMM#, this would come out as:

The desire to [do the right thing].

Given that rightness is not a reasons-giving property, and given that the fact that φ-ing is the right thing to do explains why the agent has PMM#, no reasons figure in the explanation of why the agent has PMM# and therefore PMM# is not a response to reasons. Now compare my version of having PMM*, the desire to do the right thing for the right reasons. This is a desire with a more complex content and explanation. It can be represented as:

The desire to [do the right thing for the right reasons].

The object of this desire is not the property of moral rightness, but the relational property instantiated when actions with the rightness property are performed on the basis of the right-making features. Unlike in the case of PMM#, the right-making features are therefore an indispensable part of why the agent comes to have PMM*. And so, PMM* does not violate the crucial necessary condition on a state being a response to reasons that PMM# violates. Given that I have throughout this paper assumed that moral motivation will satisfy all other conditions for responsiveness to reasons, we can thus hold that PMM* is a response to reasons. It is a de dicto response to reasons. I will elaborate on this notion more below. First, I want to show that it can solve the puzzle created by Rational Motivation and Factory. Recall:

Rational Motivation

• (1) Moral motivation to φ is rational only if it is a response to moral reasons to φ. (R)

• (2) The fact that φ-ing is right is not a moral reason to φ. (D)

• (3) PMM is a response to the fact that φ-ing is right. (Definition of PMM)

• (4) Therefore, PMM is not a response to moral reasons to φ. (from 2,3)

• (5) Therefore, PMM is not rational. (from 1,4)

In terms of Rational Motivation, if we insert PMM* into the argument, this will make premise (3) false, thereby blocking the inference from (2) and (3) to (4). Hence, Rational Motivation fails to apply to PMM*. So PMM* is rational, while PMM# is not.

These considerations give us the correct results for the Rational Motivation and Factory: Delilah in Factory is either like Delilah* or like Delilah#, so she either has PMM* or PMM#. If Delilah has PMM*, then we have the intuition that she is rational in Factory, but this is no longer contradicted by Rational Motivation, which fails to apply to PMM*. If Delilah has PMM#, then Rational Motivation applies, but the intuition that she is rational in Factory vanishes, so again no tension arises. Thus, the distinction between PMM* and PMM# resolves the puzzle. And it gives us a more precise answer to when PMM is irrational: It is irrational in all instances of PMM#.

However, so far, my proposed solution is just a sketch. It leaves some crucial questions about the nature of PMM* unanswered. In the next section, I will attempt to answer these questions and thereby hopefully stave off some objections.

5 The Nature of PMM* and the Revenge of Fetishism

I am grateful to two anonymous reviewers for making me substantially revise and extend the following section.

I start this section with three further characterizations (I), (II), (III) of PMM*.

(I) The agent who has PMM* desires to do the right thing for the right reasons de dicto, for its own sake. PMM* is the desire to, in part, have a specific kind of moral motivation—the de re desire that is appropriate for the relevant action.¹⁸ Whatever she does in the end—close the factory farm or run it even more efficiently—Delilah* wants to perform this action based on the right motivation, that is, the motivation with the right de re contents. If Delilah is otherwise rational, she will therefore update her desire set in accordance with new information. So if she acquires new information about how the fact that her farm emits CO₂ bears on the rightness/wrongness of running the farm, for example, this will transform her PMM* into a motivational state that is about both the right-making features and the fact the relevant action is right. It will be the desire to do the right thing for the reason that it is the thing that ends animal suffering or avoids harmful CO₂ emissions. This seems to me to be in general the right kind of motivation to have for a moral agent.

(II) Since PMM is the desire to do the right thing for the right reasons, it is also the desire for a reasons-explanation in terms of the right reasons to be true of the agent. This is not the desire for an explanation of the type “the agent ϕ-ed because there was a reason for them to ϕ” to be true. Delilah* wants it to be true of her eventual ϕ-ing, her ending to factory farm, say, that she did it because it would end animal suffering or that she did it because it would avoid harmful CO₂ emissions. Again, this is why she is instrumentally motivated to find out about the right-makers involving factory farming.

(III) Somewhat controversially,¹⁹ desiring to do the right thing for the right reasons is desiring to do the right thing non-accidentally. This is because, plausibly, when an agent saves the drowning child because they needed help, then it will not be an accident that they did what morality required of them. The desire to do the right thing for the right reasons is then inter alia the desire for this non-accidental connection between the agent’s action and the fact that it was the right thing to do to be realized. It should not surprise us that my proposed solution has this implication. The desire to do the right thing non-accidentally should always have struck us as part of the makeup of a good moral agent. Good moral agents don’t just do the right non-accidentally—they also want to do the right thing non-accidentally. They don’t just want to end up doing the right thing by sheer luck.

Let me now address the most pressing objection to my solution. The objection is a kind of revenge problem: If the desire to do the right thing is defective in the way I described in this paper, then so is the desire to do the right thing for the right reasons. This basic idea comes in three main guises, which I shall address in ascending order of threat-level to my solution.

First, we might think that PMM* turns the good moral agent into a kind of narcissist—overly focused on their own standing, on how saving the drowning child or feeding the hungry man will reflect on them and their actions. If this is a problem, then it is a different problem from how I have tried to understand the charge of fetishism in this paper. I have argued that PMM# is a failure to respond to reasons, a local rational defect on the agent’s part. It is not, according to this reading, a vice. And a discussion about how fetishism might be understood as a vice is beyond the scope of this paper. Briefly: Desiring to be the kind of agent who is in touch with moral reality, it seems to me, is not at all a vice. It is simply the expression of a concern for morality put in agentive terms. It is not the instrumental—and rightly criticizable—motivation to be praiseworthy in order to be admired, nor the desire to enjoy the privileges of a moral hero. It is the desire to do the right thing for the right reasons for its own sake.

Second, we might still think that there is something off with the relation between de re motivation and de dicto motivation in my picture. Assume for example, that I give my friend Daniele some medicine that will alleviate his pain. The notion that I can be motivated to do this by the desire to do the right thing for the right reasons might make it seem like I desire to alleviate my friend’s pain in a defectively derivative sense. I care about alleviating Daniele’s pain, it might be thought, only because it is a right-making feature, or because caring for it will result in doing the right thing for the right reasons.

However, PMM* is meant to be a dynamic state that will express an agent’s non-defective orientation towards the right level of morality as long as they do not have de re thoughts about it or do not have information about how those contents bear on the moral status of an action (like Delilah). It is the state that instrumentally motivates inquiry into these contents and their bearing. Thus, PMM* is the state that agents like Delilah* have before they acquire the relevant contents. Importantly, the idea is not that PMM* will persist once the relevant de re motivations are acquired, thereby making it true that the agent has these motivations “only because” she has PMM*. The relation between the desire to do the right thing for the right reasons and the desire to help Daniele because it would spare him pain is not one of instrumental explanation. Rather, the two should be seen as temporal stages of the same dynamic state, so their relation is one of parthood. It will still be true, of course, that the agent has the de re stage because they had the de dicto state first. It is the sense in which the temporarily prior stage in a dynamic state is required for the later stage to set in. But this is is not the problematic instrumental sense denoted by the locution “only because”.

It is a more general feature of our orientation towards things—our cravings, inclinations, and desires—that they sometimes unfold or evolve in this way. I might desire to watch a particular movie without being able to put my finger on which movie it is. Until I gain new insight, I can only say that it is the movie which has such and such features and instils this and that feeling in me. Once it turns out that that movie is 2015 horror classic “It Follows”, I desire to watch “It Follows”. But it would be misleading to claim that I desire to watch “It Follows” ‘only because’ I desire to watch the movie that has such and such features. For my mental situation in this case is markedly different from another type of situation I sometimes find myself in, where I just want to watch something scary, and “It Follows” will do. In the original situation, my desire to watch “It Follows” isn’t technically distinct from my earlier desire. It is a “filled in” or “evolved” later stage of it.

Third, and most pressingly, we might worry that PMM* is defective by my own lights. The thought goes like this: My argument for why PMM# is defective is that it is a response to moral rightness as such, and moral rightness as such is not a reasons-giving property (it is the property of having such properties). Thus, PMM# is not a response to reasons, which makes it a rational defect. But if moral rightness as such isn’t a reasons-giving property, surely the complex relational property of “being the right thing done for the right reasons” isn’t a reason-giving property either. PMM* is a response to this property. Thus, PMM* is also not a response to reasons.

This worry goes to the core of the general philosophical problem in the background of the issues discussed in this paper. The problem, most basically, is how it is possible, if it is possible, to have thoughts that are about something without having access to that thing de re. The problem as it pertains to the literature on moral motivation is how your desires can be about morality when you have no access to the level of right-makers. Desiring to do the right thing as such, of course, will be about morality. But the problem is precisely that it is ‘about’ morality in a sense that is off target. Now, if you think it is impossible to have de dicto thoughts that are about something in the right sense, then you must either think that de dicto motivations are always non-defective in the moral domain or that they are never non-defective. What I have proposed in this paper, however, is that it is possible for certain complex forms of de dicto thoughts to be about something in the right sense. Thus, some forms of de dicto desires really do latch onto the right level of morality without being off target in the way PMM# is. The third worry is really the challenge to flesh out this proposal in more detail and explain just how the desire to do the right thing for the right reasons latches onto—is ‘about’—the right-making level of morality. I will now attempt to answer this challenge, but insofar as the challenge connects to the larger issue just sketched, my answer will remain incomplete.

The way I chose to engage with the general puzzle of aboutness in terms of desires being responses to reasons. I offered a preliminary explanationist understanding of the notion of a response. S’s desire to ϕ is a response to the reason that p, according to this characterisation, only if the reason that p plays a role in explaining why S has the desire to ϕ. Thus, the current challenge is to say more about how reasons play a role in explaining PMM* but do not play a role in explaining PMM#.

Here is a less simplified view on desires as responses to reasons: Under what conditions a mental state is a response to a reason is in part determined by the nature of that mental state. Desires—at least those with propositional content—are prospective in an important sense. They represent unactualized states of affairs as “to-be-actualised”. If reasons are to explain this type of teleological content, they must provide information as to why a certain set of conditions is represented as to-be-realised in the desire. And they do this, presumably, in virtue of creating a normative demand that is satisfied under these conditions. Thus, a desire is a response to reasons only if the conditions the desire represents as to-be-realised are such that if they were true, they would satisfy the demand(s) of the relevant reasons. For example, if I desire to help Daniele because—or for the reason that—he needs help, then my desire should be explainable via the correct relationship between the fact that he needs help and my action of helping him. After all, desiring to help him is rational because it specifies a way of discharging the normative demand created by the fact that he needs help. Thus, what I really desire when I desire to help Daniele as a response to his needing help, is to help him as a way of dissolving his need for help. This might sound trivial in the current toy example, but it gives us a rough test for which sorts of desires for action are responses to the reason that p. Only desires for actions that can conceivably discharge the normative demands of p can count as responses to the reason that p. More generally: desires that are responses to the normative property N specify in their content the conditions that would have to be true in order for N to be satisfied. This is the way in which N explains the desire to ϕ: by imposing constraints on just what counts as satisfiers for the desire. So, for example my desire to help Daniele is a response to the reason that he needs help, on this explanationist picture, because it spells out the conditions that would have to be true in order for Daniele to be helped. This is the way in which desires—as prospective states—can be responses to reasons.

Now we can see better, I hope, why the desire to do the right thing is not a response to reasons de re in the sense that they play no role in the explanation of the desire. All that is specified in the content of this desire is that the agent perform the action that counts as the right thing. Of course, in order for the action to be the right thing, it needs to be true that there are reasons that play the right-making role. But no particular fact beyond this existentially generalised fact needs to hold. Even agents who do the right thing accidentally will count as doing the right thing, because all that is required for the satisfaction of the desire is that the agent do the right thing, irrespective of why they do it. This is why the desire to do the right thing is compatible with a lack of care for the right-making features. No particular reason de re is among the conditions that would have to be true for the desire to do the right thing to be satisfied. Hence, this desire is not a response to reasons.

Crucially, the same is not true of the desire to do the right thing for the right reasons. For this desire specifies in its content that a reasons-explanation be true of the agent. And this reasons-explanation cannot be true of the agent unless they have access to the particular reasons-facts in question de re, to the right-makers that is. For the relevant reasons-explanation works in terms of reasons, not facts about those reasons. In order for it to be true that the agent acts for reasons, in other words, it is not enough that the agent acts on the basis of the fact that there are reasons to act (the existentially generalised fact). An agent who acts on the basis of the fact that there are reasons to act does not act for those reasons. Perhaps they act on the basis of the additional reason provided by that fact, if it does provide such a reason. Or perhaps they act for no reason at all. In any case, acting for a reason requires being acquainted with that reason, since to act for that reason is, among other things, for the content of the reason to make one’s action intelligible. Since acting for reasons requires de re acquaintance, the desire to do the right thing for the right reasons has reasons de re as its satisfaction conditions. That is, it is satisfied only if the agent acts for reasons, not just when the agent acts on the basis of the fact that there are such reasons. Therefore, the desire to do the right thing for the right reasons has particular de re right-makers among its satisfaction conditions (albeit under a de dicto description). And these reasons therefore play an indispensable role in explaining the desire according to the model just introduced. Thus, the desire to do the right thing for the right reasons is a response to reasons.

With this, I conclude my investigation into PMM. The question this paper overall asked was why and when PMM is defective. I take myself to have provided an answer to this question that proves illuminating²⁰ even if you do not accept it: PMM is defective because and insofar as it is not a response to reasons. I hope that this answer has made the possibility of PMM—or at least the notion that it is somehow defective—less puzzling overall.