Printer Friendly

The abductive case for humeanism over quasi-perceptual theories of desire.

THE THESIS THAT WHATEVER WE PURSUE we pursue sub specie boni seems incompatible with the obvious facts that desires persist after we have judged their objects worthless, and that akrasia is real. But in recent decades philosophers attracted to the scholastic formula have proposed a new way to accommodate temptation: Desires should be understood as analogous to perceptual states rather than full-fledged judgments (cf. Stampe 1987; Scanlon 1998; Wallace 1999a; Johnston 2001; Helm 2002; Oddie 2005 and 2010; Tenenbaum 2007 and 2008; and Schafer 2013). Perceptual states have limited rational corrigibility. Optical illusions frequently persist even after we have judged them illusory. Similarly, a desire, as a quasi-perceiving, will often present its object as good even after one comes to the conclusion that it is worthless. Perceptual states seem phenomenologically richer than judgments, and desires often come with a vivid phenomenology of their own. Perception is largely a capacity to be affected, a capacity with respect to which we as agents are passive; it does not seem accidental that the older name for desire is "passion," something with respect to which one is patient rather than agent.

Percptualism, as this paper will call it, has been consciously offered as an alternative to the Humean orthodoxy about desires, which holds that desires do not have a representational aspect (Smith 1987; Sinhababu 2009). The arguments in its favor are largely abductive. Perceptualists claim their theory is able to explain phenomena that Humeanism cannot. This paper will provide a defense of orthodoxy, on the same abductive grounds. When we consider the phenomena that supposedly motivate perceptualism, we discover that Humeanism can explain most equally well; in one case it is able to provide an obviously superior explanation. Perceptualism, the paper argues, is for the most part unmotivated.

The first section will make clearer the theoretical commitments of Humeanism and perceptualism, and briefly consider the advantages and costs these theories face in explaining the possibility of practical reason.

1. Psychological Humeanism, the Guise of the Good, and the Problem of Practical Reason

The Humean theory of motivation divides the psychological states responsible for action into two fundamental classes: states that represent the way the world is, also called cognitive states, and states that motivate, also called conative states. These states are distinct existences; they can vary independently and there are no necessary connections between them. One can want something while cognizing it as good, bad or having no opinion on its value at all. And the cognition of something as good cannot cause action on its own: Desire impels action, sometimes for the sake of the perceived good, and sometimes for the sake of the perceived worthless.

This thesis, that cognitions and motivations can vary independently, is psychological Humeanism in the broad sense. Note that it is neutral on whether there is heterogeneity within these two basic classes of attitude--that is, whether motives include intentions, whims and so on, along with desires, or whether cognitions include perceivings and supposings along with belief. This broad Humeanism can be contrasted with a more narrow psychological Humeanism, which amounts to endorsement of the belief-desire model, and hence a rejection of heterogeneity. This paper is neutral on the issue of narrow versus broad Humeanism. (1)

Broad Humeanism about motivation should be contrasted with another thesis, normative Humeanism. Normative Humeanism is often understood to be some sort of instrumentalism about practical reason, but it is probably better to think of it as a coherentist theory of practical reason (Dreier 1996)--with instrumentalism as a specific account of coherence. There is a natural argument from psychological Humeanism to normative Humeanism. Cognitive attitudes can clearly be mistaken. They represent the way the world is, and so they can misrepresent the way the world is. They can be false, or at least non-veridical. However, conative attitudes are not representations, nor are they necessarily connected with any representational states. It follows from this that an agent could want any sort of object--whether painful, boring, or destructive--without any mistakes in her judgments, inferences, or assessments of evidence. The agent could desire this object without any failure in what is ordinarily called "reasoning." (2)

At the same time, desires have the job in one's psychology of leading to action. Desires that conflict or are otherwise inconsistent would plausibly undermine each other's ability to direct action. So, the thought is, desires cannot be criticized on the basis of having mistaken objects (the way beliefs can), but they can be criticized because they fail to cohere with the agent's other motives. This failure of coherence can then be given an instrumentalist gloss or some other gloss. (3)

But philosophers can accept the normative Humean's basic methodology and still reject her conclusions. One might propose, for example, that fully informed agents with perfectly coherent motivational sets will actually show a high degree of convergence in their desires (cf. Smith 1994; 1995). Alternately, one might hold that certain motives are necessary to full-fledged agency (cf. Velleman 2000); (4) or that we will all have reason to perform certain actions regardless of our contingent ends (cf. Schroeder 2007); or that certain motives are necessary to maintain one's status as a full-fledged agent over time and in potentially compromising situations (cf. Smith 2013). Finally, a Humean could adopt some sort of normative primitivism: She could accept that reasons for action are primitive, for example (as in Scanlon 1998; Parfit 2011); or she could take another normative category, such as practical virtue, as primitive (cf. Setiya 2007). These pictures will result in some substantive and universal requirements of practical reason--requirements that will apply to all agents, regardless of their actual contingent desires.

Notice, though, that even on these more robust conceptions of practical reason, psychological Humeanism can be understood as involving a split between practical and theoretical reasoning corresponding to the split in the two types of attitudes (Tenenbaum 2007). According to the pictures considered, theoretical reasoning ultimately has the aim of bringing one's theoretical attitudes (beliefs) into accord with an independent standard--the truth. On non-primitivist accounts, practical reasoning does not aim at an independent standard; rather, it brings one's practical attitudes (motives) into accord with each other, though on the more extensive pictures of practical reason this can be augmented by standards of accuracy in beliefs and the fact that certain motives may be in some sense necessary. Primitivists, on the other hand, are committed to practical normative requirements that cannot be explained by reference to an aim, whereas theoretical normativity can be explained in terms of an aim of truth or accuracy. (5)

In contrast to psychological Humeanism, we can accept the guise of the good, which effectively is the denial of the Humean thesis that motivating and representing states are distinct. Motivating attitudes, on this view, are judgments or "seemings" about value, or at least necessarily depend on them. (6) The guise of the good thereby avoids the split within reason (or so its advocates claim): Our motives are directed at the good, it holds, in the same way that our judgments are directed at the truth. As Joseph Raz explains:
   The division between substantive and procedural rationality (and
   between substantive and instrumental rationality) took hold among
   philosophers who doubted that reason is directly involved in the
   choice of ends, but believed that reason has a role to play in
   practical thought, which the notion of procedural rationality
   captures (1999: 360-61).

He then goes on to say:
   But since we desire only what we think of as worth desiring, our
   desires are among our responses to perceived reasons. This is true
   of a desire to drink when thirsty as much as of the desire to
   become a good teacher. In conceiving desires and in adopting and
   maintaining goals we deploy all the capacities which are involved
   in so-called procedural rationality, and there seems to be no other
   capacity involved, at least none which can relate to rationality.
   The reasoning ability and other capacities which make people
   rational in forming beliefs about scientific matters, or about the
   weather, or anything else which can be said not to be in itself
   normative, are the same abilities which make people rational in the
   way they adopt and maintain goals. Therefore, there is only one
   kind of rationality (361).

Sergio Tenenbaum (2007: 5-6) likewise notes that the guise of the good has an uncomplicated explanation of how intrinsic ends can be mistaken: They can be inaccurate, because their objects are not in fact good (or they are not things that there is reason to pursue, etc.). He goes on to write that the guise of the good "conceives of our rational faculties as a unified whole. They are the same rational faculties employed in two different endeavors: theoretical inquiry and practical inquiry" (6; emphasis mine).

This is the first explanatory advantage of the guise of the good. It offers a straightforward explanation of how an agent's ultimate ends can be mistaken, and it provides an account of the nature of reason that makes reason a unified normative system and a unified set of capacities for meeting those norms. Now, perceptualism is a species of the guise of the good--one that takes the more specific position that desires are quasi-perceivings of value (typically it is intentions that correspond to judgments about value). An advocate of the guise of the good could accept a more intellectualist picture of motivation, according to which we only desire what we judge to be good or worthwhile. (In the quote above, Raz seems to endorse such a view.) But as we noted, this seems unable to explain the experience of akrasia, or persistent temptation against one's better judgment. Perceptualism was introduced partly to explain this. In any case, as a species of the guise of the good, perceptualism will also possess an explanatory advantage over psychological Humeanism with respect to practical reason, or so the argument goes.

Nonetheless, there are also substantial costs to the guise of the good. There are two possible ways of spelling out the relation between motivating attitudes and goodness, the sense in which motives present their objects as good. On the one hand, a motivating attitude with the object p may be identical to (or perhaps necessarily depend on) a cognitive attitude that represents that p is good. Call this the content view, because good (or some other normative or evaluative notion) must be part of the content of every motivating state.

The other option is explicitly proposed in Tenenbaum (2008), and developed further in Schafer (2013). On this view, motivating states aim at the good in the same way that beliefs aim at truth. Tenenbaum emphasizes that this does not merely mean that goodness sets a standard of correctness for desires. As Schroeder (2008) points out, this would not distinguish the guise of the good from psychological Humeanism, as any psychological Humean who rejected normative Humeanism could accept something like this. As an illustration, Smith (2013) argues for psychological Humeanism, but still agrees that there is some sense in which desire aims at the good--namely, desires are open to rational criticism when their objects are not in fact good. (7) But any theorist who holds that desires are subject to rational criticism can hold that the objects of desires are good when the desires are rational, and not good otherwise.

If there is to be a distinct claim about the nature of (rather than the normative standards on) desire, the connection between desires and the good must be stronger than a simple standard of correctness. Fortunately, as Tenenbaum points out, the tie between beliefs and truth is much tighter than that. Consider a variation on Moore's paradox: "I believe it is raining, but it is false that it is raining." The person who utters this is not simply believing incorrectly or irrationally. It is hard to make sense of what state of mind the agent might be in, because to believe that p is in some sense to hold p true, even if true is not part of the content of every belief (2008). Call this the formal object view, following the terminology used by Tenenbaum.

We can grant that both the content view and the formal object view offer a more straightforward explanation of substantive requirements of practical reason. The Humean must either accept a revisionary, minimalist picture of practical reason, or else offer a very complicated explanation in terms of convergence among extraordinarily idealized agents or in terms of motives constitutive of agency. But if we accept the guise of the good, the person who desires worthless ends is mistaken because she is simply wrong about whether those ends are good, just as one who believes that the sun orbits the earth is wrong about that proposition's truth. The problem is that the guise of the good purchases explanatory straightforwardness here by sacrificing it elsewhere.

The content view seems to rule out desires or intentions in small children and animals, since possession of the concept good is necessary to forming an attitude with the appropriate content (Schroeder 2008). (8) The formal end view faces an objection of obscurity. It provides a more straightforward explanation of the possibility of practical reason by insisting that motives stand in a special relation to the good, a relation that is poorly understood, and of which there is only one uncontroversial example--the relation between belief and truth. Both views are committed to some necessary connection between motivation and some normative property. This is a necessary truth that Humeanism avoids. (9)

Finally, there is the fact that sentences such as "Smoking is not at all good, but I desire to smoke" may sound weird--it may sound like the speaker desires incorrectly and irrationally--but they do not sound unintelligible, as Moore-paradoxical sentences do. But both the content view and the formal-end view seem to predict Moore-paradoxicality. (10) And while the perceptualist may be able to account for why "Smoking is not at all good, but I desire to smoke" sounds acceptable--because the desire is more analogous to a perceptual state than a judgment--note that sentences such as "Smoking is not at all good, but I intend to smoke" or "Smoking is not at all good, but I choose to smoke" also fail to sound like pragmatic contradictions. (11) This appears to be true of conative attitudes generally (Woods 2014).

None of these objections is meant to be decisive. Tenenbaum (2007) for example, tries to show that his formal-end view has the resources to explain the absence of unintelligibility in utterances self-attributing a desire for an object one believes worthless (73-75, and chs. 6-8). Perhaps holders of the content view could offer an explanation of how children and small animals could possess motivating attitudes. But these explanations will, necessarily, be more complicated, or involve more substantial theoretical commitments, than the Humean's explanation of the same. Straightforwardness cuts both ways.

In light of these considerations, whether Humeanism or the guise of the good provides a better explanation of the possibility of practical reason is simply inconclusive. (12) The guise of the good offers a more straightforward explanation, but it does so by assuming more, and because of these assumptions it risks explaining too much.

However one chooses to weigh these considerations, the remainder of this paper will argue that there are no other points in favor of perceptualism. All of the other considerations offered on behalf of the theory are, at best, neutral between perceptualism and Humeanism, and one of these considerations actually tells in favor of Humeanism. In the course of the discussion, it will also emerge that there is serious reason to doubt the claims of Raz and Tenenbaum that practical and theoretical reasoning deploy the same capacities, only toward different subject matters or ends. If this is correct, compatibility with the unity of reason is less certain as a desideratum on our theories (because we have independent reason to doubt its unity), and so the fact that perceptualism offers a simple explanation of this unity becomes a much weaker point in its favor. There may, after all, be nothing to explain.

We will start by considering phenomenological evidence.

2. Phenomenological Evidence

When we want something we typically feel an attraction; we feel happy or content at the prospect of getting it, anxiety at the possibility of loss. Part of the appeal of perceptualism can be that it acknowledges the phenomenology of desires, rather than treating them as pure behavioral dispositions.

Of course, that desires have a phenomenology does not tell against Hume. As Donald Hubin points out, Humeanism and behaviorism are distinct theses (2001: 451-52; 460-61). Contemporary Humeans like Hubin, Peter Railton (2004: 193-95), and Neil Sinhababu (2009: 469) have all claimed that phenomenology is an important aspect of desire, and, of course, the famous Scottish Humean, David Hume, believed desires to have a phenomenology. (13)

So the question is, do perceptualists get the phenomenology right? My argument will be negative: Perceptualists have failed to give a compelling argument that the phenomenology of desire is best characterized as a representation of goodness or an attitude that takes goodness as a formal end, even after some fairly generous assumptions on behalf of the perceptualist. Perceptualist and Humean interpretations of the phenomenology should be regarded as equally plausible.

But let us start with the arguments of the perceptualists. Graham Oddie gives an explicitly introspective argument: "When I desire that P, P has a certain magnetic appeal for me. It presents itself to me as something needing to be pursued, or promoted, or preserved, or embraced. Now the good just is that which needs to be pursued, or promoted, or preserved, or embraced" (2005: 55).

Oddie's argument amounts to a pun. He appeals to the fact that we are inclined to accept certain expressions as ways of evoking what it feels like to desire something, but then moves to a very theoretically loaded interpretation of that same expression. We do, for example, sometimes express our desires using the language of need. After a hard day of work I might say, "I need a beer." The good, though, needs to be pursued (when it needs to be pursued) in the sense that its pursuit is obligatory or required. "Need" here indicates practical necessity. But it would only be in the interests of humor that one would interpret the sentence "I need a beer" as describing an obligation or requirement of practical reason. (14)

The sentence "P has a certain magnetic appeal for me" is even more clearly without normative implication--it is straightforwardly metaphorical. The simple fact that we might use sentences like these to communicate what it feels like to desire does not by itself give us reason to think that perceptualism is true. A Humean could agree with Oddie's claims; she simply gives them a different gloss. In order for Oddie's claim to support his conclusion, the phenomenology he points to must represent P as normatively significant. P must not simply have "magnetic appeal": It must feel like it should excite interest.

Similar points can be made about Scanlon's famous directed-attention account of desire. Scanlon initially writes that to have a directed-attention desire for something is for it to keep "occurring to him or her in a favorable light" (1998: 39). But Scanlon goes on to say, a few sentences later, that one desires something in the directed-attention sense "if the person's attention is directed insistently toward considerations that present themselves as counting in favor of [it]" (ibid.). As Sinhababu (2009) points out, the first formulation of directed attention is not obviously equivalent to the second. (This should be unsurprising, given that the first formulation is a metaphor. (15)) The Humean can accept the first formulation, only she must interpret it to mean that people who desire P will tend to focus their attention or imagination on things they associate with P, and they will tend to experience the possibility of P's realization as pleasant and the possibility of ~P as distressing (469; 489-90). None of this is equivalent to Scanlon's second formulation: judging or otherwise representing that there are reasons that would justify pursuing P.

These arguments for perceptualism start by pointing to an evocative expression ("magnetic appeal," "needing to be pursued," "in a favorable light") that communicates what our experience of wanting is often like--an expression everyone familiar with how to express themselves in English would accept. But then they immediately move, without supporting considerations, to a very precise and contentious interpretation of that expression.

Nevertheless, it may still seem that direct reflection on the phenomenology supports perceptualism. Think about the experience of wanting a cigarette. One is vividly aware of how much pleasure smoking would bring. One is nagged by the thought of the pleasure. What else could this experience be but a representation of the pleasure of smoking as good?

We should be extremely skeptical that introspection will tell us the answer. (16) When I have a vivid experience of desire, it does not feel to me like a proposition (whatever that would be like); nor is it a sentence whispering in my ear "eating the cake would be good"; nor is it an image of God pointing at the cake and giving me the thumbs up. It is also not experienced as any of the things that would favor a Humean interpretation (nothing feels like a functional disposition). What I find is feelings such as the nebulous nagging sensation described above, which does not really feel like anything else except a nagging sensation. Attributing content to this nebulous feeling requires an act of interpretation, a judgment call. But which interpretation seems most fitting will likely depend on one's background theory. This means that using phenomenological introspection to adjudicate disputes between such theories will produce, at best, very tenuous evidence.

We should conclude that the phenomenology of desire is neutral between Humeanism and perceptualism, unless some independent theoretical reason for interpreting that phenomenology as representational can be offered. But there may be an independent theoretical motive: intentionalism about phenomenology.

a. Intentionalism and the possibility of nonrepresentational experiences

Oddie and Scanlon move from the fact that there is a phenomenology to desire to the conclusion that desires represent value too fast, without considering other possible interpretations of that phenomenology or giving us reason to prefer their own. This might be because they are assuming that intentionalism is true, that all phenomenological experiences have intentional content. So, the feeling of wanting would have to have propositional content, and that content must be such, moreover, that it allows us to distinguish the experience of wanting P from other experiences, such as visually perceiving P. Since wanting is closely tied to action, it is natural to assume that it must present P as normatively significant, as providing a reason for the actions the want motivates.

Some evidence for this interpretation of the perceptualist assumptions is found in Helm (2002), who makes the connection between perceptualism and intentionalism explicit. He first offers an intentionalist thesis regarding pain: Pain represents disvalue. (17) Conversely, desires represent certain possibilities as worthy of realization.

But the Humean is already committed to denying the claim that all intentional states are representational states. Broad Humeanism, remember, is committed to the thesis that motives are not inherently representational. Nonetheless, a Humean must agree that desires are intentional states: They are, after all, desires for something. Humeans avoid conflict in these two commitments by invoking the idea of direction of fit (Anscombe 2000/1957; Smith 1987 Humberstone 1992).

Consider two sentences: "The door is closed" and "Close the door!" Both sentences have the same propositional content, the door is closed, but the first sentence presents that proposition as true, whereas the second presents it as to be made true. "The door is closed" could be falsified by the discovery that the door is open, whereas that discovery would not falsify the command. (Responding to "Close the door!" with "No it is not" would be unintelligible.) Likewise, responding to the discovery that the door is open by closing it would show understanding in the case of the command, but would simply be bizarre as a response to the report. The content of the report is meant to fit onto the way the world is. The world is meant to be altered to fit the content of the command. (18)

Beliefs and desires both have propositional content, and can even take the same proposition for their content; but, the psychological Humean will insist, beliefs present their content as true, and desires present their content as to be made true. If we discover that the propositional content of a belief is false, the belief should be given up; if the propositional content of a desire is false, this will, ceteris paribus, motivate the agent to take steps to realize that content in the world. It makes sense to call a belief false; desires cannot be false. So, our propositional attitudes seem to display the same directions of fit as the sentences provided above, or so the Humean will argue.

Colin Klein (2007; 2012) and Richard Hall (2008) have argued that a phenomenological state can have nonrepresentational intentionality. For example, Klein argues that pain presents its content in the imperative mood, effectively issuing us commands such as "Do not touch that!" or "Do not put weight on that foot!" Hall offers a similar interpretation of other sensations like itches, tickles and noxious smells. (19)

The appeal to the Humean of the imperatival interpretation of phenomenology should be obvious. It allows her to respond to the insistence on intentionalism by noting that she always considered desires intentional; she simply denied that they were representational. (20) So the fact that desires have a phenomenology does not force the Humean to attribute any additional (i.e., representational) intentional properties to them, beyond what was already posited. The phenomenological properties, on her view, correspond to a nonrepresentational direction of fit.

b. Does the imperatival aspect of desire entail the guise of the good?

It is worth noting here Karl Schafer's (2013) argument that desires should be understood as quasi-perceivings with a normative property (ought-to-be-done, in this case) as their formal object, precisely because they present their content in an imperatival mood. Such an argument may look like it could be used to show that the Humean position is unstable: If desires have imperatival intentionality, they must present their objects as good. I will show that this is not the case. Schafer's objection to the abductive case for Humeanism lies elsewhere.

Schafer introduces a notion of attitudinal force, which is analogous to the illocutionary force of utterances. For the purposes of this discussion, we can say that force is like direction of fit, but finer grained. Following Velleman (2000: 99-122, 244-82), Schafer points out that attitudes such as supposing or imagining seem to have the same direction of fit as belief--they seem representational in some broad sense--but they clearly lack the strict connection to truth that belief possesses. (21) But we see the same fact in language: Suppositional claims and assertions both present propositions representationally; both can figure as premises in an argument, for example, unlike commands or questions. But only assertions commit the speaker to the truth of their content.

Beliefs, Schafer thus proposes, present their contents with assertoric force. (Imaginings and supposings presumably present their contents with something like suppositional force) Desires, for their part, present their content with imperatival force. (22) So far, this should all be acceptable to a Humean. But Schafer then proposes the following parallelism: To present p with assertoric force is to present p as true; similarly, to present p with imperatival force is to present p as something that ought to be done (276-78). Note that Schafer here is taking up a formal-end view, rather than a content view.

So what does Schafer's argument accomplish? It gives reason to think that, even if desires are imperatival, this does not show the guise of the good to be false. This is significant because direction-of-fit considerations have been used as arguments against the guise (cf. Smith 1987). Schafer's argument helps us see that, regardless of direction of fit, or attitudinal force, attitudes could still have formal ends.

What Schafer's argument does not do is establish, by itself, any sort of explanatory superiority for the guise of the good (to be fair, this does not appear to be Schafer's intention). One might reply to this that only the guise of the good gives a story of what the formal end of desires could be. But Humeanism denies that desires have a formal end, at least in the way beliefs have formal ends. So this argument for perceptualism proposes an explanans to a non-neutral explanandum. An independent reason for thinking that desires have a formal end would be needed before this argument against Humeanism could be pursued.

One such reason might be this: In the same way that assertion conversationally commits the speaker to the truth of what is asserted, issuing an imperative conversationally commits the speaker to the requisiteness (or oughtness) of the action the imperative instructs. (23) If I assert that it is raining, and you ask me whether that is true, and I respond, "No," my communicative intentions become opaque. If I say "Close the door!" and you ask, "Should I close the door?" and I say, "No," my communicative intentions are similarly opaque.

The problem is that there are two possible explanations of the conversational commitment. One is that utterances in the imperatival mood present their content as requisite in the same way that assertions present their content as true. The other explanation is that imperatival utterances are made with the intention of bringing about actions: That is, speakers aim to make it the case that you will X by telling you to X. In the latter case, we can offer the following explanation of the communicative breakdown: The speaker, by saying that you should not obey her command, makes it less likely the command will be obeyed, and the whole point of issuing the command is to get the interlocutor to do something. What she is trying to bring about, then, with this collection of sentences, becomes mysterious.

It is worth briefly noting that this account of the relation between imperatives and ought-claims parallels a Humean account of the relation between desires and normative facts. A Humean will agree that an agent is criticizable for acting on a desire that p when it is also the case that she ought not to bring about p. Acting on the desire may be irrational (at least if the agent knew what was required of her); the desire itself may be irrational. So there is a sense in which the facts about what there was decisive reason to do defeat the desire, just as there is a sense in which what one ought to do can defeat an imperative. The special commitment of the Humean is that she must not explain why the desire is defeated, why the agent may be guilty of irrationality, by treating the desire as a state presenting its object as of normative significance, a state that must be mistaken or presenting the world other than how it is, given the facts about what ought to be done. She must offer some other explanation of the defeat and the irrationality (or treat them as primitive). Providing an explanation of the irrationality is hard; but providing an explanation of the defeat is straightforward. Desires aim at their own satisfaction, but an agent who knows that she ought not to satisfy a desire will not satisfy it so long as she is rational.

The Humean can agree that imperatives take goodness, or ought-ness, or some other normatively relevant property as their standard of correctness, just as she can agree that the same properties serve as the standard of correctness for motivating attitudes. What she must deny is that these are the formal end of either imperatives or motives. But the fact that ought-claims pragmatically "defeat" imperatives, in some sense, is explained well enough with the standard-of-correctness relation. The formal-end relation is stronger, and unnecessary.

One might try to construct another type of argument. We explain illocutionary or attitudinal force by appeal to the property that the content is presented by the force as having. We explain assertion as the speech act and belief as the attitude presenting their contents as true. So how else would we account for imperatival force, except by appeal to normative properties?

But it is fair at this point to deny that all forms of illocutionary force present their contents as having a certain property. Assertoric force may present content as true, but it is not obvious that suppositional force or interrogative force (the illocutionary force of questions) present content as having some property. Likewise, it is unclear that attitudes such as supposing, imagining, or wondering present their content as having some property. And again, there seems to be a disanalogy in the relation that beliefs bear to truth compared with the relation that motives bear to goodness or ought-ness, as the discussion of Moore-paradoxicality in section 1 illustrated. So it is unlikely that force in general should be explained in terms of properties to which the content is related, and the Humean has reasons for denying that the explanation works in this particular case.

Schafer, as mentioned above, does not seem to assume that simply committing to the imperatival nature of desire forces one to accept perceptualism (or any other form of the guise of the good). He argues that we should accept that desires present their contents as normatively favored, because desires can then serve as a source of normative evidence (277-81). This consideration in favor of perceptualism will be dealt with later.

3. The Possibility of Deliberative Choice

R. Jay Wallace (1999a) offers a different reason for accepting perceptualism: Psychological Humeanism is incompatible with genuine agency. On the Humean model: "Action is traced to the operation of forces within us, with respect to which we are ultimately passive, and in a picture of this kind real agency seems to drop out of view" (633). Genuine agency requires rational control over our actions:
   [Agents'] actions are not merely the causal products of
   psychological states that they happen to find themselves in at the
   time, rather they have the power to determine for themselves what
   they are going to do. By exercising this power, such persons can
   bring about a kind of rational action that is not merely due to the
   fortuitous coincidence of rational judgment and a given desire, but
   that is a manifestation of the very capacities that make them,
   distinctively, agents (637-38).

This deliberative picture seems attractive, but it leaves us with an apparent problem. Desires figure in commonsense action-explanations, but how do they do this if they are not motivational dispositions?
   Action is an expression of choice or decision, not the result of
   psychological forces operating on us. This raises an important
   question, however. If ordinary, given desires are not needed to
   account for the capacity of persons to comply with their
   deliberated verdicts about action, what role is left for them to
   play in the etiology of reflective agency (638)?

Wallace answers that desires must explain action by influencing our assessment of reasons. Treating them as quasi-perceptual states, with the attendant incorrigibility and vividness, gives the best picture of how they influence that assessment:
   Let us suppose that the anticipated pleasure and release from
   discomfort is of little normative significance in comparison to the
   clear prudential and moral disadvantages of continued heroin
   consumption ... Adding such a [phenomenologically intense,
   quasi-perceptual] desire into the mix, it seems, would make it much
   harder for the agent to reach his conclusion and to keep it firmly
   in view (646).

So, to summarize, genuine agency requires deliberative control of choice. This requirement is met, however, only so long as psychological states explain action solely by means of influencing our assessment of reasons. Desires figure in action explanations without undermining freedom, and so they must influence our assessment of reasons. The most plausible picture of how they do so is provided by perceptualism. So the theoretical advantage over Humeanism is that it explains how we are free, or genuinely agents, even though subject to desires.

I have attributed to Wallace a commitment to the following desideratum on psychological theories: Psychological states must explain action solely by means of influencing our assessment of reasons (or else we are unfree). As we will see, Wallace himself has commitments that seem inconsistent with this desideratum, so I would like to note that, unless influencing one's assessment of reasons is supposed to be the sole means by which a psychological state can explain free action, it is unclear why nonrepresentational desires would be incompatible with genuine agency. A Humean can allow a contingent connection between normative judgments and motivation, and such a connection need not be, in Wallace's words "a fortuitous coincidence of rational judgment and a given desire." Contingent connections can be stable and non-coincidental. Most causal connections are like this. The Humean can claim, then, that normative judgments will (for most agents at least) reliably change the agent's motives to a greater or lesser extent. Perhaps Wallace's point is only that free agency requires that the agent's deliberative verdict about her reasons must be part of the mix of explaining an action, but that psychological states that motivate action independently of a deliberative verdict can also be present, and do not necessarily undermine freedom. But this is a requirement that broad Humeanism can meet. The stronger thesis is necessary if Wallace is to succeed at objecting to Humeanism.

If any of this sounds surprising, the reader should remember that psychological Humeans are not necessarily committed to instrumentalism about practical reason, which is a normative thesis. The Humean might accept the meta-normative positions of Smith, Velleman, Schroeder--which all imply the existence of noncontingent reasons for action--or they might embrace some sort of normative primitivism. In any of these cases, they will hold that agents can deliberate about reasons, and insofar as they are rational, their judgments will lead to shifts in their motivation; so the agent's assessment of reasons is part of the mix determining her action. (How rational an agent is may be contingent, but that will turn out to be true on Wallace's perceptualist account as well.)

The problem with Wallace's argument is that akratic action is, by definition, action that must be explained by something other than the agent's assessment of reasons. If I am acting akratically, then, according to the perceptualist, it must be a case where a desire presents its object as worthy of pursuit, but I judge that it is not worthy of pursuit. But then something other than my deliberative verdict must be at work in causing me to act akratically--that simply follows from the nature of akrasia.

Of course, I may be distracted, rushed, distressed, or intoxicated, and so fail to call to attention what I believe. I fall for the "optical illusion of value." But this is a case where I temporarily judge myself to have most reason to perform the action--i.e, I temporarily cease to be aware that it is a temptation. It is not full-blooded akrasia, in which I am aware of myself as acting contrary to reason. But, if I genuinely judge myself to be acting contrary to reason, desire must explain my action via something other than its impact on my deliberative verdict. (24)

What is truly puzzling is that Wallace seems to agree that this happens. He writes:
   To be in the grip of such an impulse [i.e., an incessant and vivid
   desire] is to be made vividly aware of a concrete alternative for
   action, presented in terms that appear highly attractive. Even if
   one succeeds, in the face of such a desire, in reasoning correctly
   to the conclusion that it should not be acted on, its continued
   presence and urgency will make it comparatively difficult to choose
   to comply with the deliberated verdict one has arrived at (648).

Wallace plainly states here that the desire influences choice, in ways that exceed its influence on the agent's assessment of reasons. But this is just to say that his perceptualist account does not meet his own desideratum on explanations of action. (And, it seems, no account that allowed for genuine akrasia could.)

This may lead to the thought that it must be a misinterpretation of Wallace to attribute commitment to this desideratum to him. This is possible: His argument is complicated. But, as noted, if this is not the desideratum he has insisted on, it is obscure what his objection to Humean psychology is supposed to be. To illustrate this, we should step back and look at his initial complaint that Humean psychology left us passive with respect to our own motivations.

If we are active with respect to our motives, it must be because we have capacities for controlling them. Let us call this collection of capacities the will. Now, the will could be identified with capacities that are narrowly deliberative: the capacities for reaching a verdict about what I have most reason to do. On this view, the will just is the intellect, or that portion of the intellect that reasons about practical matters. The alternative to this is that the will, while it does make use of the powers of the intellect, also has additional, non-intellectual powers for doing things like beating down unruly desires or generating extra motivational force. So the capacities of the will are either solely capacities of reasoning, or else they include additional capacities for controlling action.

If the former is correct, it is not an objection to Humeanism that it fails to make us active with respect to motives that persist in motivating us after we have judged against acting on them, because we are not active with respect to such motives. If the will is just a part of the intellect, then coming to judgments about what one ought to do is the sole means of controlling one's motives. But sometimes an agent judges that she ought to X, and the motivation to Y remains unaffected. She must then be passive with respect to this motive, because she has no further ways of controlling it. (25)

Wallace may be thinking that Humeanism leaves us passive with respect to all of our motives, or at least all intrinsic motives, and that that is simply too much passivity. But as noted above, this is simply a mistake. On the assumption that what makes us free is our ability to determine our action through deliberative verdicts, any Humean that allows that there are non-instrumental reasons for action predicts that most normal agents are active with respect to a great many of their intrinsic desires. These accounts may leave us passive with respect to desires that persist in motivating us in the face of our deliberative verdicts, but on the purely intellectualist account of the will we are considering, that must be true for the perceptualist as well.

Wallace's worry may be that, on the Humean account, our deliberative verdict might only determine our actions because of some contingent causal connection. But on his own account, whether a desire is such that it will persist in motivating even after we have judged it illusory must itself be a contingent fact. It is hard to see how any account will avoid the conclusion that how active an agent is with respect to her own psychology is a contingent matter, given the contingency of how weak or strong willed an agent is.

So if we reduce the will to the intellect, it is unclear why embracing perceptualism would lead to agents being more active with respect to their motives, at least once we allow that people do act akratically. If we instead take the latter option, there is no reason at all why Humean psychology would leave us passive with respect to our motivations, even our akratic motives. We have a distinctive power of will that allows us to resist motives. Since this power is not identical to any of our powers of judgment, of weighing evidence, or of reasoning (to any of our intellectual powers, that is), there is absolutely no reason why our motives would need a representational component before we could resist them.

Again, perplexingly, Wallace is clear that he does not reduce the will to the intellect. He writes that temptations and addictions "represent potential impairments not only of our capacities for rational deliberation, but also of our volitional power to comply with the verdicts of such deliberation" (649). But all of this suggests that Wallace has misplaced his complaint against Humean psychology. Perhaps no extant broadly Humean theory makes room for sufficiently robust powers of will. This would be surprising, given the number of distinctive accounts of robust self-control argued for by Humeans (cf. Smith and Kennett 1996 and 1997; Bratman 1999, chs. 3-4, and 2000; Smith 2003; Holton 2009; Mele 2012). But in any case, potential problems with Humean self-control only favor perceptualism if there is some additional capacity for self-control that perceptualism predicts. Yet if the powers of self-control are not intellectual powers, this problem is not in any way solved by making desires into a class of cognition--or at least it is completely obscure why attributing a representational component to our desires would give us an additional "volitional power to comply with the verdicts of deliberation." Given Wallace's particular concerns, perceptualism is a non sequitur. (26) The fact that we have robust powers of self-control does not favor either position.

A final point is worth making. Wallace insists that self-controlled agency depends on both deliberative and volitional capacities, conceived of as distinct. But notice that this is inconsistent with Raz's claim that "The reasoning ability and other capacities which make people rational in forming beliefs about scientific matters ... are the same abilities which make people rational in the way they adopt and maintain goals." Philosophers, of course, are entitled to come to similar conclusions but for different reasons, reasons that may even be inconsistent. Nonetheless, there is an important lesson to be 26 drawn from Wallace's commitments.

Wallace holds that there are cases of temptation that persist in motivating us even after reason has concluded that such temptation is not to be acted upon (and this is a very intuitive position). He also holds that we are still active, not passive, with respect to such motivation (another intuitively attractive position). But then one must accept some means of controlling the motive other than simply judging it contrary to reason--because by stipulation one has already done that. So Wallace must posit that there is a "volitional power to comply with the verdicts of" reason. But this is inconsistent with the initial motive offered in favor of the guise of the good. Practical agency depends on specific capacities, "volitional powers," that do not have any counterpart in the case of theoretical reason. Agents can come to the correct conclusions given their evidence--or according to whatever other standard one takes as relevant to theoretical reason--but still fail to comply with practical reason's demands. This suggests a major discontinuity between practical and theoretical reason, and casts doubt on the idea that they should be given a unified theoretical treatment.

So, while the existence of robust powers of self-control does not directly favor Humeanism, or directly disfavor perceptualism, it does cast doubt on a seemingly powerful consideration in favor of the latter thesis.

4. Intelligibility

Mark Johnston (2001) argues for perceptualism on grounds that, if desires did not represent their objects as worthwhile, the agent performing the action would not experience her own action or his desire as intelligible. He writes:
   Perceptual experience makes certain immediate perceptual beliefs
   based on them seem apt or fitting ... In the same way, affective
   disclosure of sensuous goods makes desire readily intelligible from
   the inside. Were a desire to come upon one without one's being in
   some way taken with the desired object or outcome then there would
   be a lack of intelligibility in one's having the desire. The
   question of justification--"Why am I doing or wanting this?" would
   naturally arise.... I could find myself with a very strong urge to
   turn knobs of doors counterclockwise, even though I have no
   interest in opening doors and no interest in feeling the solidity
   and texture of the knobs in my hand. The urge could be very
   effective, but it would not render intelligible what I am doing,
   least of all to myself (189-90).

So, according to Johnston, the action would not be experienced as rationally intelligible unless the desire motivating it were a perceiving of value. Note that he must have in mind rational intelligibility, and not merely explanatory intelligibility. A mere urge would suffice for the latter.

The obvious response is that sometimes our own actions are not rationally intelligible to us. They sometimes feel, and we sometimes judge them to be, foolish, shameful or wrong. The question "Why am I doing this?" arises quite frequently, I find.

It is also unclear why a quasi-perceiving of value would contribute intelligibility to the action. Consider Johnston's example, again. I have a mere brute disposition to turn the knob, but no belief that doing so is good. There is a sense in which my action will strike me as unintelligible. Why do I keep turning this knob, even though I do not think there is any reason to do so? Now let us add that I have an unreflective feeling that there is a reason to turn the knob. Remember, though, that my normative beliefs are still in place, so I know that this feeling is inaccurate. I do not see how my action would become more intelligible to me at all. The question simply changes to, why do I keep acting on this delusional state even though I am well aware that it is delusional?

Perhaps Johnston could agree that akratic actions, along with perverse actions, are to a high degree unintelligible. Nonetheless, they must be experienced as intelligible to some minimal degree. If they were not, we would not experience them as actions at all, or at least not free actions. At best, they would seem like compulsive actions over which we had no control; they might even seem like the mere action-like moving of our limbs.

This claim would need support, however; in the context of this debate it is simply question begging to assume that actions would not be experienced as free unless the desires that prompted them were quasi-perceptual states. One possible defense might be the position articulated by Wallace: A theory that presents our actions as the result of motivational dispositions operating within us leaves us overly passive with respect to those motives. It is inconsistent with our understanding of ourselves as rational deliberators; such a picture has no room for agents. But we already saw that, once we allow for akrasia, this argument against Humeanism is a non sequitur.

Alternatively, the thought may be that actions must be minimally reasons-responsive in order to be actions at all. This is reasonable, but again it is hard to see how an argument for perceptualism over Humeanism can be constructed out of it. For the perceptualist, remember, akrasia is a case of the agent acting on a persistent "optical illusion" of value, which she believes to be illusory (otherwise it is not akrasia, but simply being mistaken about what one ought to do). But it is unclear why a quasi-perceptual state presenting an action as good or reasonable would make choosing that action more reasons-responsive than a brute motivational state, if it is stipulated that one judges the quasi-perceiving to be inaccurate and misleading. In both cases, the agent acts on a motive that she is convinced she has no reason to act on. The question "Why am I doing this?" would arise equally for both acts, and be equally without answer (at least of a justifying, rather than explanatory, sort).

We should also note that, while almost everyone agrees that actions must be minimally reasons-responsive, how minimal is itself up for debate. The strict Humean could point out that even an akratic action caused by a brute motive for an object absolutely devoid of value could still be reasons-responsive in the following respects: (1) the action still shows means--end coherence; (2) even if the agent did not respond to the particular reasons she had in this situation, had other reasons been present she could have responded to those--she would have resisted temptation if someone had pointed a gun at her head, or offered her a million dollars to do so; or (3) she could have responded to her reasons had she been less stressed, less distracted, or otherwise in a more favorable situation for maintaining rationality. (27) If a perceptualist is to motivate her theory by appeal to the idea that minimal reasons-responsiveness is a necessary condition on action, she must demonstrate that each of (1)--(3) falls short of the minimum, otherwise the argument is question begging.

A final thought might be that Johnston intends a weaker claim then we have interpreted him to be making, that in fact he does not mean that a quasi-perceiving of value invariably brings intelligibility, he simply means to account for the default, unreflective sense of intelligibility that accompanies actions that we do not explicitly recognize as akratic or otherwise defective, and that we have without forming explicit judgments about the appropriateness of our actions. I doubt that this is Johnston's position, (28) but someone attempting to develop similar ideas may prefer a position like this. There are unfortunately two difficulties with pointing to this--a default sense of intelligibility in ordinary cases--as evidence for perceptualism. First, it is unclear what the target explanandum is actually supposed to be. While my actions typically do not feel unintelligible, I have to confess that it is unclear to me that they typically feel intelligible either. If there is a feeling of intelligibility accompanying most of my actions, it is not particularly vivid or intense. If the argument for perceptualism is that it can account for this datum, it is a weak argument, because one could reasonably deny the datum (in the absence of further argument I am inclined to deny it). As was noted in the beginning of section 2, using introspective phenomenology to adjudicate disputes about the intentional properties of attitudes will produce tenuous evidence at best, because which interpretation of the phenomenology seems correct will often depend on the particular theorist's background commitments.

Even if we accept the datum, it is unclear why the Humean cannot explain it. Since agents will typically (perhaps necessarily) meet some minimal threshold of rationality, most of their choices will conform to their background beliefs about what they ought to do. Alternately, since the Humean can hold that desires have a (nonrepresentational) phenomenology, they can explain any default intelligibility of our actions in terms of feelings of excitement, pleasure, and distress that the potential satisfaction and distress of our desires brings. (29)

In summary, if it were the case that perceptualism predicted that we would experience acting on our desires as rationally intelligible, that would be a point in favor of Humeanism, because we are oftentimes consciously at odds with those desires. It is unclear that perceptualism even predicts that, however. So it appears that both theories are equally well placed to account for our experience of akrasia. One might argue that perceptualism is better-placed: Akratic actions are experienced as free, and free actions are minimally reasons-responsive. But there is no obvious reason to think that perceptualism predicts a greater degree of reasons-responsiveness in our akratic actions, and the Humean has a story about minimal reasons-responsiveness that she can tell. Finally, if we are interested in explaining the sense of intelligibility that ordinarily attaches to our actions, then on this issue as well, perceptualism and Humeanism have explanatory parity.

5. Normative Evidence

A number of philosophers have argued that if desires are quasi-perceptual states presenting their objects as valuable, this would explain how we have evidence about the normative. Oddie, for example, writes:
   If the desire that P just is the experience of P's seeming good to
   me, then it can serve as a reason, albeit in some sense defeasible
   or overridable, for thinking that P is good (2005: 56).

Schafer writes that if imperatival states have a formal end:
   [Then] our desires provide us with a basic form of normative
   experience. For, given this, we need not have any normative or
   evaluative beliefs in order to experience the world as, in
   Sellars's phrase, "fraught with ought." Rather, in order to
   experience the world in these terms, we simply need to feel a
   desire. In this way, prior to moral or normative reflection, simply
   in feeling desire, the world is already presented to us in
   normatively laden terms (2013: 277).

So perceptualism may provide a superior story about how we come to have evidence about which actions are justified and unjustified. There is, unfortunately, a serious defect with the perceptualist argument. It is completely unclear what problem in normative epistemology perceptualists take themselves to be solving by identifying the sources of our normative evidence with desires.

Consider the view that our normative intuitions are the source of our normative evidence. Now, there are a number of reasons for dissatisfaction with this thesis. (Note, none of these are presented as reasons to reject intuitionism, as will become clear; they are to set the dialectical stage.) We might be worried that our normative intuitions are not caused by their objects, and so we have no explanation of their reliability. We might worry about postulating a mysterious faculty of insight into some objective realm of facts. We might be worried about biases in our intuitions, or ways that they can be manipulated by context or mood. We might worry that our "intuitions" are really just expressions of cultural indoctrination or biologically predetermined heuristics.

But notice that none of these problems would be solved by the following conjecture: Normative intuitions are identical with motivating attitudes or imperatival attitudes. This would simply compound our problems. We would still want to know how normative intuitions can serve as evidence. And we would also want to know why the intuition is identical to a motivating or imperatival attitude, given that the functional role of those mental states seems significantly different. Notice, moreover, that these two problems--the problem of vindicating the evidential status of the normative intuitions and the problem of explaining why we should believe one state plays both roles--do not look, at first glance, like they will have a common answer.

So here is one Humean account of how we have normative evidence: via our normative intuitions. What would make the perceptualist account explanatorily superior? Is the complaint that values cannot cause our intuitions? But then it would seem that they cannot cause our quasi-perceivings as well. Is the claim that values can cause quasi-perceivings (cf. Oddie 2005: 198203)? But if evaluative properties are part of the causal order, why can they not cause our normative intuitions? Perhaps the thought is that quasiperceptual states can still be evidence even if they are not caused by their object (cf. Schafer 2013: 280-81). But then we are owed an explanation of why intuitions cannot still serve as good evidence.

Presumably, what the perceptualist needs is a solution to the standard problems with intuitionism, which works because the relevant state of seeming is identical with a motivating or imperatival attitude. But it is unclear how the fact that a state motivates, or presents with imperatival force, would improve that state's epistemic value. But without that, perceptualist epistemology just looks like a form of intuitionism, combined with some very bold and epistemically irrelevant conjectures about the other functional or rationalizing roles the intuitions (or intuition-like states) play within our psychologies.

Perhaps the motivational role of desire makes it epistemically relevant because of the following platitude about the good: The good is that which agents with well-functioning psychologies are motivated to pursue. If we combine this platitude with the very reasonable assumption that most people's psychologies are reasonably healthy, we have an explanation of why our actual motivating states could serve as reliable indicators of the good.

The problem is, once we accept this as a platitude, Humeans become equally well positioned to explain how desires could serve as normative evidence.

The Humean believes desires are motivating states. The platitude about the good combined with the assumption that people mostly have wellfunctioning psychologies predicts that actual motives will generally be reliable indicators of the good. So Humean-style desires will be reliable indicators of the good. Attributing representational properties to our motives becomes explanatorily otiose. (30)

When it comes to explaining normative evidence, it is simply unclear what problem in normative epistemology perceptualism is supposed to be addressing.

6. Linguistic Considerations

Oddie (2005; 2010) sees theoretical advantage in perceptualism in its apparent ability to explain why the sentence "P is good, but I do not desire that P" is strange. Oddie starts by noting that the Moore-paradoxical sentence "It is raining, but I do not believe that it is raining" sounds incoherent, even though both conjuncts could be true at the same time. The general explanation for this seeming incoherence is that asserting "P" conversationally implies that the speaker believes that P.

Oddie goes on to identify a type of sentence that we could call weakly Moore-paradoxical:
   Q is true, although Q doesn't seem true to me.

   Now this utterance isn't as obviously paradoxical as the Moorean
   assertion. But in typical circumstances there is something a little
   bit odd about it(2005: 44).

Oddie's explanation of the oddness is that normally by asserting "P" we also conversationally imply that we have evidence for P, and the fact that P seems true is often a form of defeasible evidence for P (ibid.).

Oddie then points out that, by assuming perceptualism, we can show the following sentence to be weakly Moore-paradoxical:
   P is good, but I don't desire that P.

   Whatever oddness attaches to this can be explained in terms of the
   oddness of the proposition above which ... is necessarily
   equivalent to it: namely, P is good, but it doesn't seem good to me

First, we should note that, pace Oddie, absolutely no oddness attaches to "P is good, but I don't desire that P." "Exercise is good, but I don't desire to exercise," sounds perfectly fine.

More significantly, on Oddie's analysis, "I desire that P" and "P seems good to me" are necessarily equivalent, and implicitly recognizing this is a matter of linguistic competence. But then the sentences "I desire to smoke, even though smoking does not seem good to me," and "I don't desire to exercise, even though exercise seems good to me," should sound unintelligible, because we should, as a matter of linguistic competence, implicitly recognize that they express contradictions--and not even pragmatic contradictions but literal ones. But they do not sound unintelligible (even Oddie acknowledges that at most they sound "odd"). (31)

So on further inspection, the linguistic evidence cited by Oddie fails to support his thesis. Does it support the Humean thesis? It is fairly strong evidence that "I desire that P" and "P seems good to me" are not necessarily equivalent, but not all perceptualists will be committed to a necessary equivalence; they may hold instead that there are multiple ways of seeming good, only one of which is desiring. (32) This response does little to dispel the worries, however; if desiring is a way of seeming good, then they are still committed to holding that the truth of "I desire that P" entails the truth of "P seems good to me," even if the two claims are not equivalent. Presumably this entailment is a priori, unless the guise of the good is defended as an empirical hypothesis. But then "I desire to smoke, even though smoking doesn't seem good to me" is still a contradiction, and one that persons competent with the relevant concepts should implicitly recognize.

The perceptualist is of course welcome to argue that we are able to make sense of the sentence because we are generally forgiving in our interpretations of other people's utterances. Our interpretations are often forgiving. Nonetheless, that a theory has to appeal to the forgivingness of interpreters to explain away countervailing linguistic data should be regarded as a cost to the theory, if not a decisive one. It is worth noting that we are not forgiving with the following sentences, which sound awful:

(S1) Smoking seems good to me in a quasi-perceptual way, but it doesn't seem good to me.

(S2) Smoking seems good to me in a motivating way, but it doesn't seem good to me.

(S3) I see smoking as good, but it doesn't seem good to me.

(S4) I feel like smoking is good, but it doesn't seem good to me.

It is very hard for the perceptualist to deny that at least one of these, or a very similar sentence, is necessarily equivalent to "I desire to smoke, but smoking doesn't seem good to me." We are owed, then, an explanation of why we are more forgiving in one case than another.

7. Conclusion

The only consideration that genuinely speaks in favor of perceptualism is the promise that, as a species of the guise of the good, it will allow for a very straightforward explanation of the possibility of practical reason. But this consideration is inconclusive. The Humean has explanations of the possibility of practical reason too. They must either give up on straightforwardness or be revisionary, but they can start from weaker assumptions, and they do not need to explain away the fact that certain sentences fail to sound Moore-paradoxical. So considerations speak both ways.

The remaining considerations simply fail to support perceptualism at all. As we saw, in one case, the "weak" Moore-paradoxicality averted to by Oddie, the data supports Humeanism if it supports anything. The remaining considerations appear neutral.

It should be acknowledged that none of this amounts to a decisive refutation of the position. Abductive arguments generally do not provide those, and in any case there is, as just noted, a genuine motive for the view. But, at present, it is the only proposed motive for perceptualism. Otherwise, we have a list of admittedly mysterious phenomena that look like they might be explained by the quasi-perceptual thesis. The precise way that explanation is supposed to work has never been made clear, however; nor has any attempt been made to show that the Humean explanation of the same would be overly cumbersome, let alone impossible. In several cases, it is not even particularly clear what the target phenomena are supposed to be.

Finally, it has emerged from the discussion that the possibility of self-control and its lack in the face of temptation creates significant pressure to accept a capacity for willpower not reducible to the capacities of the intellect. We need, as Wallace puts it, both "capacities for rational deliberation" and a distinctive "volitional power to comply with the verdicts of such deliberation" (1999a: 649). This suggests that the original motivation for the guise of the good was misguided: Acting correctly is not reducible to reasoning correctly about the world. If the will and intellect are distinct capacities, and both are rational capacities, a disjunctive theory of reason may be unavoidable. It is less clear, then, that providing a straightforward explanation of the unity of reason really is a point in favor of the guise, since that might mean only that the theory is again guilty of explaining too much. (33) So while the case is hardly decisive, the evidence at present seems to favor Humeanism. (34)


Anscombe, E. (2000/1957) Intention, Cambridge, Mass.: Harvard University Press.

Baker, D. (forthcoming) "Akrasia and the Problem of the Unity of Reason," Ratio.

Bratman, M. (2000) "Reflection, Planning, and Temporally Extended Agency," The Philosophical Review 109(1): 35-61.

--(1999) Faces of Intention, Cambridge: Cambridge University Press.

Byrne, A. (2001) "Intentionalism Defended," The Philosophical Review 90(2): 199-239.

Dreier, J. (1996) "Rational Preference: Decision Theory as a Theory of Practical Rationality," Theory and Decision 40(3): 249-76.

Hall, R. J. (2008) "If It Itches, Scratch!" Australasian Journal of Philosophy 86(4): 525-35.

Helm, B. W. (2002) "Felt Evaluations: A Theory of Pleasure and Pain," American Philosophical Quarterly 39: 13-30.

Holton, R. (2009) Willing, Wanting, Waiting, Oxford: Oxford University Press.

Hubin, D. (2001) "The Groundless Normativity of Instrumental Rationality," The Journal of Philosophy 48(9): 445-68.

Humberstone, I. L. (1992) "Direction of Fit," Mind 101(401): 59-83.

--(1990) "Wanting, Getting, Having," Philosophical Papers 19(2): 99-118.

Jackson, F. (1985) "Internal Conflicts in Desires and Morals," American Philosophical Quarterly 22(2): 105-14.

Johnston, M. (2001) "The Authority of Affect," Philosophy and Phenomenological Research 63(1): 181-214.

Klein, C. (2012) "Imperatives, Phantom Pains, and Hallucination by Presupposition," Philosophical Psychology 25(6): 917-28.

--(2007) "An Imperative Theory of Pain," The Journal of Philosophy 104(10): 517-32.

Korsgaard, C. (2008) "The Normativity of Instrumental Reason," in The Constitution of Agency: Essays on Practical Reason and Moral Psychology, Oxford: Oxford University Press: 27-68.

Mele, A. (2012) Backsliding: Understanding Weakness of Will, Oxford: Oxford University Press.

Oddie, G. (2010) "Experiences of Value," in C. Pigden, ed., Hume on Motivation and Virtue, Basingstoke: Palgrave-Macmillan, pp. 121-41.

--(2005) Value, Reality, and Desire, Oxford: Oxford University Press.

Parfit, D. (2011) On What Matters, Volume 1, Oxford: Oxford University Press.

Railton, P. (2004) "How to Engage Regress," in R. J. Wallace, et al., eds., Reason and Value: Themes from the Moral Philosophy of Joseph Ray, Oxford: Oxford University Press.

Raz, J. (1999) "Explaining Rationality: On Rationality and the Justification of Reason," Ratio 12(4): 354-79.

Scanlon, T. M. (1998) What We Owe To Each Other, Cambridge, Mass.: Belknap Press of Harvard University Press.

Schafer, K. (2013) "Perception and the Rational Force of Desire," The Journal of Philosophy 60(5): 258-81.

Schroeder, M. (2012) "Reply to Shafer-Landau, McPherson, and Dancy," Philosophical Studies 157(3): 463-74.

--(2008) "How Does the Good Appear to Us?" Social Theory and Practice 34(1): 119-30.

--(2007) Slaves of the Passions, Oxford: Oxford University Press.

Setiya, K. (2007) Reasons without Rationalism, Oxford: Oxford University Press.

Sinhababu, N. (2013) "The Belief-Desire Model Explains Everything," Nous 47(4): 680-96.

--(2011) "The Humean Theory of Practical Irrationality," The Journal of Ethics and Social Philosophy 6(1): 1-13.

--(2009) "The Humean Theory of Motivation Reformulated and Defended," The Philosophical Review 118(4): 465-500.

Smith, M. (2013) "A Constitutivist Theory of Reasons: Its Promise and Parts," Law, Ethics, and Philosophy 1.

--(2003) "Rational Capacities (or How to Distinguish Recklessness, Weakness, and Compulsion)," in C. Tappolet and S. Stroud, eds., Weakness of Will and Varieties of Practical Irrationality, Cambridge: Cambridge University Press, pp. 17-38.

--(1995) "Internal Reasons," Philosophy and Phenomenological Research 55(1): 105-31.

--(1994) The Moral Problem, Malden, Mass.: Blackwell Publishing.

--(1987) "The Humean Theory of Motivation," Mind 96(381): 36-61.

Smith, M. and Kennett, J. (1997) "Synchronic Self-Control Is Always Non-Actional," Analysis 57(2): 123-31.

--(1996) "Frog and Toad Lose Control," Analysis 56(2): 63-73.

Stampe, D. W. (1987) "The Authority of Desire," The Philosophical Review 96(3): 335-81.

Tenenbaum, S. (2008) "Appearing Good: A Reply to Schroeder," Social Theory and Practice 34(1): 131-8.

--(2007) Appearances of the Good: An Essay on Practical Reason, Cambridge: Cambridge University Press.

Tye, M. (1995) "A Representational Theory of Pains and Their Phenomenal Character," Philosophical Perspectives 9: 223-39.

Velleman, J. D. (2000) The Possibility of Practical Reason, The Scholarly Publishing Office: The University of Michigan.

Wallace, R. J. (1999a) "Addiction as a Defect of the Will: Some Philosophical Remarks," Law and Philosophy 18(6): 621-54.

--(1999b) "Three Conceptions of Rational Agency," Ethical Theory and Moral Practice 2: 217-42.

Watson, G. (1977) "Skepticism about Weakness of Will," The Philosophical Review 86(3): 31639.

Woods, J. (2014) "Expressivism and Moore's Paradox," Philosophers' Imprint 14(5): 1-12.

Derek Baker

Lingnan University

Department of Philosophy

(1) For example, Holton (2009: 112-36) rejects this narrow Humeanism, holding that intentions are not reducible to beliefs and desires, while Sinhababu (2013) defends it.

(2) Note that I am assuming here that a theory of practical normativity is primarily a theory of practical reason. Those unhappy with this assumption can treat my use of "normative Humeanism" as shorthand for "Humeanism about practical reason." It should also be noted that some have taken the implication of Humean psychology to be error theory about practical reason. According to Korsgaard (2008), this was the view of the historical Hume; but see Baker (forthcoming) for the point that this might more naturally be understood as a reductive theory about practical reason, on which practical reason is just a special case of theoretical reason.

(3) See Smith (1994, ch. 1; 2013) for a more detailed explanation of the relation between psychological and normative Humeanism.

(4) Also see (Schroeder 2012, [section][section] 6-7).

(5) We should note, however, that both Velleman and Smith (especially Smith 2013) go to great lengths to establish that theoretical and practical reason are in fact continuous. I will not go into their arguments here, except to note that, within the Humean model, substantially more philosophical work is required to show the unity of reason. Also see Smith (2013) and Baker (forthcoming) for arguments that normative primitivists are especially vulnerable to the charge that their accounts of reason are disjunctive.

(6) It is worth noting, though, that the key perceptualists all seem to embrace the identity of desires with quasi-perceptual states (see Stampe 1987; Wallace 1999a; Johnston 2001; Oddie 2005 and 2010; Tenenbaum 2007; and Schafer 2013).

(7) This position is achieved by explaining goodness in terms of the desires of a fully rational and perfectly informed agent.

(8) One may object that if desires are quasi-perceptual states representing goodness, then animals and small children could have such states even if they lacked the concept of good, in the same way that they can have visual experiences of greenness despite lacking the concept of green. There are several complications here: disagreements about the nature of concepts or the conditions of their possession, for example.

These complications can be sidestepped, however, if we keep in mind that the concern is really about whether all intuitively desiring beings have the psychological sophistication necessary to form intentional states in which good is part of the content. If we assume that the content of an intentional state is individuated by the functional role of the state, then agents must be capable of very sophisticated behavioral responses and inferences before we attribute intentional states about good or bad to them: They must not only pursue certain ends and avoid others, but also show that they regard those ends as appropriate or inappropriate, as worthy or unworthy of pursuit. The difficulty for the content view is that such behavior and inferences seem much more sophisticated than the behavior and inferences involved in pursuit and avoidance; so it seems there could be and are creatures that can do the latter, and hence have desires, without doing the former. The perceptualist may insist on some other way of individuating content, of course. But the burden is on the perceptualist, then, to offer and motivate an alternative account of individuation.

(9) A referee points out that the necessary truth looks like less of a theoretical burden if it is interpreted as a necessary truth about rational motivation. But notice that this leaves the advocate of the guise with no more resources than the Humean to explain why irrational motives are irrational (they cannot in any sense be inaccurate, or failing to meet their formal end, since we are assuming that they do not inaccurately present their objects as good). Philosophers attracted to the guise seem to want the stronger position, for this very reason. Wallace (1999a), Oddie (2005), and Tenenbaum (2007) all attempt to explain the irrationality of certain desires--akratic and perverse desires -- by citing their role of presenting their objects as good.

(10) Oddie (2005; 2010) takes the view that these sentences have something paradoxical about them. But by something paradoxical, Oddie seems to have in mind that they sound strange. The significance of Moore's paradox, however, is that the speaker sounds like she is contradicting herself even when she is not. Hence, the mere oddness of a statement does not show it to be "somewhat" Moore-paradoxical (also see n. 31). Oddie's argument will be addressed in more detail in section 6.

(11) Intentions are identified with normative judgments explicitly in Tenenbaum (2007) and Schafer (2013).

(12) But see Baker (forthcoming) for an argument that further considerations show that the guise of the good actually fails to provide a straightforward explanation of practical reason at all. The argument is too involved to present here, and so for the sake of this argument we will simply assume that the guise is successful on this score. But if the argument of that paper is correct, the guise of the good is on extremely weak footing.

(13) Humberstone (1990) does not advocate a Humean thesis, but also makes a convincing case that feelings of positive or negative effect are important components of our typical understanding of desire.

(14) Also notice that if Oddie is not relying on a pun or some other kind of equivocation, the premise of his argument must be that when I desire P, P presents itself to me as practically necessary. But this is just to assert the very point under debate: It is not to offer an argument for perceptualism but to restate (a version of) the thesis. Perhaps Oddie simply takes the thesis to be self-evident, because of the phenomenology of desire. But given the controversy surrounding the thesis, even among people who have desires and so presumably phenomenology of their own, it cannot be taken as self-evident.

(15) Or if it is not a metaphor, it is simply asserting that the phenomenology has a certain content, which the Humean is entitled to deny in the absence of further argument; see n. 14.

(16) Thanks to Colin Klein and Jack Woods for leading the author (who previously argued that the phenomenology obviously favored Humeanism) to see the light on this.

(17) Helm's thesis is an alternative to the more standard intentionalist interpretation of pain as a representation of tissue damage (cf. Tye 1995; Byrne 2001).

(18) A referee wonders whether critics of Humeanism could insist that commands such as "Close the door!" are really elliptical ought-claims (i.e., "You ought to close the door"). This is not an unreasonable position--there are accounts of the semantics of imperatives that do interpret them as modals. At the same time, this is hardly the orthodox or generally accepted position, and at present a popular semantic account treats imperatives as nonrepresentational (a brief and informative overview of this issue is found in Charlow 2014, [section] 4). The Humean is thus entitled to take as her starting point the assumption that there are nonrepresentational intentional states. If, of course, representational accounts of imperatives are ultimately successful, this would be a serious blow to Humeanism. But at present the assumptions are theoretically respectable.

(19) Hall argues that some experiences, noxious smells for example, have both representational and imperatival content.

(20) This may strike some as a misuse of "intentionalism," which one might hold is by definition the thesis that phenomenological states are representational states. For what it is worth, Klein (2007) describes his position as intentionalist, whereas Hall (2008) clearly regards his theory as continuing in the spirit of the intentionalist program, though it is less clear whether he thinks the theory itself is intentionalist. In any case, we can refer to the position as broad or expansive intentionalism, to be distinguished from the classical form.

(21) Also see Bratman (1999: 15-34).

(22) Schafer also argues that perceivings are assertoric attitudes and that intentions are imperatival attitudes as well. This supports a suggestive parallelism: Desires are to intentions as perceptions are to beliefs.

(23) I would like to thank an anonymous referee for pressing me to say more about this kind of argument.

(24) Can the desire explain the action through a quasi-perceptual assessment regarding the reasons? Perhaps it can, but since such a quasi-perceptual assessment does not--in the akratic cases-- reflect the conclusion of the deliberating agent, Wallace's concern that the agent is passive with respect to the desire (and hence with respect to the quasi-perceptual verdict) will remain. Thanks to an anonymous referee for asking me to consider this point.

(25) This kind of a position tells us that what we typically call "akratic" or "weak-willed" action should actually be thought of as compulsive or otherwise pathological--that these are cases in which the agent's action is not completely free. The position is counterintuitive, but there are powerful points in its favor (see Watson 1977).

(26) See Wallace (1999b) for a more explicit rejection of extant Humean accounts on the grounds that they do not allow robust enough powers for resisting temptation. Wallace's reasons for insisting that more robust powers of self-control lead to something in the neighborhood of a guise of the good thesis (239-41) remain, unfortunately, obscure.

(27) See (Smith 2003; Sinhababu 2011).

(28) Johnston explicitly claims that a desire can grant intelligibility to an action "from the inside" even when acting on the desire conflicts with one's normative judgments, and that the absence of a phenomenologically rich desire can make an action seem unintelligible, even when one judges that there is reason to do it (190-91). Also see n. 29.

(29) Johnston dismisses nonrepresentational phenomenology and normative beliefs as sources of the sense of intelligibility. But his reason for dismissing nonrepresentational phenomenology is that we could have such experiences but still find our behavior unintelligible (191). The preceding argument established that we could have a representational response but still find our behavior unintelligible as well. So nonrepresentational phenomenology would not invariably grant intelligibility, but neither would representational phenomenology. He dismisses normative beliefs as explanans on the grounds that sometimes we feel alienated from our actions despite normative beliefs being in place and supporting them (190-91). But that is only to say that normative beliefs do not invariably grant intelligibility, and quasi-perceivings of value could not invariably grant intelligibility either. (It may well be that no mental state invariably grants a sense of intelligibility.) This is another reason to assume that Johnston endorses the stronger claim initially attributed to him. If he were embracing the weaker claim, these arguments would be instances of special pleading.

(30) This response is inspired by the following point from Schroeder (2008: 128):
   Take, for example, the very simple subjectivist view that the good
   is what anyone would desire in reflective equilibrium in the
   presence of full information. On such a theory, it is both possible
   to desire something that is not good and to fail to desire
   something that is good.... But there is still a match between the
   structure of desire and the structure of the good. If something is
   good, then if you are in "ideal circumstances"--you are fully
   informed and are in reflective equilibrium--then you will desire
   it. And if you desire something, then if you are in ideal
   circumstances, it is good.

Note that if we combine the position just described by Schroeder with the assumption that most of our beliefs are true and our attitudes are generally coherent (basic principles of charity), we have an argument for why we should take our desires to be reliable prima facie indicators of the good. As rational agents, most of us will be close enough to the ideal that our desires will more often be for good than bad things. It should also be noted, however, that the argument in the main text does not depend on subjectivism; it simply follows from the proposed platitude about the connection between motivation and the good.

(31) A referee argues that "I desire to smoke, but smoking doesn't seem good to me in any way at all" sounds odder. This may be (the speaker is expressing a more seriously irrational state of mind), but the sentence does not sound like a contradiction, or otherwise unintelligible, which is what Oddie's analysis requires. For some worthwhile points on the gap between mere oddness and Moore-paradoxicality in sentences, see Woods (2014).

(32) I would like to thank an anonymous referee for raising this point.

(33) But see Baker (forthcoming) for a discussion of the theoretical drawbacks of a disjunctive account of reason.

(34) Thanks to Colin Klein, Tristram McPherson, Darrell Rowbottom, Neil Sinhababu, Jack Woods, and two anonymous referees for critical feedback. The research in this paper was substantially funded by a grant from the Research Grants Council of Hong Kong Special Administrative Region, China (Project No. LU342612).
COPYRIGHT 2014 Journal of Ethics & Social Philosophy
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Baker, Derek
Publication:Journal of Ethics & Social Philosophy
Date:Sep 1, 2014
Previous Article:Money pumps, diachronic and synchronic.
Next Article:Moral disagreement and epistemic advantages: a challenge to McGrath.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters