A Moorean paradox of desire

Abstract

Moore's paradox is a paradox in which certain kinds of belief or assertion, such as a belief that ‘it is raining and I do not believe that it is raining’, are irrational despite involving no obvious contradiction in what is believed. But is there a parallel paradox involving other kinds of attitude, in particular desire? I argue that certain kinds of desire would be irrational to have for similar, distinctive reasons that having Moorean beliefs would be irrational to have. Hence, I argue that such desires, a desire that ‘one have a particular desire that was frustrated’ or a desire that ‘some state of affairs obtain about which one was indifferent’, are a parallel Moorean paradox of desire. I further argue that this analogous paradox has implications for practical reasoning, in particular by presenting a problem for instrumentalism about the objects of desire.

Keywords:

• Moore's paradox
• desire
• belief
• instrumentalism
• philosophy of mind
• philosophy of action

Acknowledgements

Versions of this paper were presented at a meeting of the Philosophical Society at the Australian National University in April 2006, and at the Australasian Association of Philosophy Conference, also at the ANU in July 2006. I thank members of those audiences for helpful comments. I am especially grateful to Michael Brady, David Chalmers, Daniel Friedrich, Alan Hájek, Fiona Macpherson, Martin Smith, Jean Stace, and Daniel Stoljar for discussion of these ideas and comments on earlier drafts.

Notes

What Sorenson calls an ‘adequate solution’ to the paradox should both diagnose what is wrong with such beliefs and assertions and allow us to predict what other propositions will be similarly faulty if believed or asserted (1988, 16).

This is assuming that sincere requests or demands are speech acts through which we express desires (see e.g. Searle 1969), analogous to the way that sincere assertions are speech acts through which we express beliefs. However, as I will focus on the psychological version of the paradox rather than the linguistic version, here this detail is not essential to the argument.

I will assume throughout that the objects of desires are propositional attitudes, such as beliefs; so the objects of desires are propositions. It is true that we sometimes describe desires by saying things such as ‘I desire an ice cream’, or ‘I want to kick the football’, where the object of the desire seems to be an object or performing an action. However, I take it that these can be paraphrased without change of meaning into desires for propositions involving those objects or performance of actions, for example, ‘I desire that I have an ice cream’ and ‘I want that I kick the football’, respectively. I take it that this is the standard approach in the literature on desire (see e.g. Fernandez 2007, Stalnaker 1984).

Heal claims that it is a condition on an adequate solution to Moore's paradox that it must ‘identify a contradiction, or something contradiction-like, in the Moorean claims (i.e. the assertion of, or belief in, Moorean propositions)’ (1994, 6).

Although I have focused here on the norms of avoiding ignorance and falsity, this is merely because these appear the most relevant to Moorean beliefs. It should not be taken to imply that these are the only, or the most important, doxastic norms. Indeed, there is an extensive literature on this kind of subject, both by contemporary writers on virtue epistemology such as Zagzebski, Greco, and Sosa, and people writing through history about intellectual virtues more generally. In her Virtues of them mind,, Zagzebski (1996) describes some of this tradition, in addition to presenting her own account of what constitutes an intellectual virtue, citing people such as Hobbes, Emerson, Descartes, and Dewey. What is especially interesting in this discussion is how much consensus there has been about what particular intellectual attitudes and habits people are expected to cultivate and that regulate our intellectual assessment of people, despite the different traditional backgrounds from which people have written. There is general agreement that people should be motivated to form true beliefs, and in such a way that they constitute knowledge, and that to reliably achieve this, they should aim to avoid prejudices, have enthusiasm for inquiry, resist distractions, take responsibility for, and follow through, the commitments that come from their particular beliefs, and so on. Hence, it is uncontroversial that there are standards regulating the way we should form beliefs, or doxastic norms, and that we do take these into consideration, although often implicitly and in complicated ways, when assessing someone intellectually.

See, for example, Harman (1984, 1986) for discussion of such issues. Harman argues for an empirical and naturalistic approach to belief revision and denies that logical implication does or should have a privileged role in real-life belief formation. For example, he suggests that it is guided by, among other things, a second-order principle of ‘clutter avoidance’ (1984, 47); that we should avoid learning and storing beliefs unnecessarily, say by not trying to infer all of the logical consequences of our beliefs, by trying to store in memory only those key matters one will later need to recall, by not trying to retain trivial information, and so on. See also the previous note referring to other norms that, in contrast with Harman's greater focus on empirical considerations, are perhaps more purely epistemological.

This might also suggest that there can be versions of Moore's paradox involving other norms for forming beliefs (see Hájek 2007 for discussion and examples).

See, for example, McGinn (1982) and Shoemaker (1996). Both take this requirement to regulate one's beliefs to show that someone who is able to have propositional attitudes must be able to become aware of those propositional attitudes. That is, that being a rational believer entails the capacity for self-consciousness. However, this further claim is not required to support the intuition appealed to here, that if one is aware of certain kinds of fault in one's beliefs, then one will typically regulate one's beliefs accordingly.

Perhaps we could supply a context in which we might not expect her to regulate her other beliefs if she has a Moorean belief. For example, if someone were offered a large sum of money if they could truly believe that she had a particular belief that was false, we might think that she should try to maintain both beliefs. But if we consider a Moorean belief in isolation from such a context then intuitively if the agent is a proper epistemic agent she should respond by recognising the mistake and regulating her beliefs. Moreover, even in such a context what is over-ridding that demand is a pragmatic norm and not an epistemic norm. An analogous point holds for the omissive case.

Shoemaker, for example, claims that someone who grasps the concept of belief will not assert a Moorean sentence (unless they were being insincere perhaps) even if she did not have first-person access to her own beliefs (1996, 35). He claims that someone who grasps this concept should recognise the paradoxical nature of such a sentence nonetheless.

Shoemaker also briefly discusses this possibility (1996, 91–2).

Though it might imply that she should regulate her belief system in a different way. One way of understanding this case is as a recognition that her belief system is fragmented, and that she has conflicting, but compartmentalised, beliefs. If we think that a competent believer should respond to such a recognition by repairing such fragmentation, other things being equal (see, e.g. Lewis 1982), yet she does not do so, then we might doubt whether she is a competent believer at all. This case would then become a variant of the third kind of doubt we might have about the Moorean belief.

And analogous doubts are raised about the omissive version, mutatis mutandis.

There is considerable disagreement and debate about what is involved in possessing, or grasping, a concept. Much of this is related to disagreement about what concepts are, their ontological status: for example, according to people such as Locke (1975), Fodor (1987), and Harman (1987), concepts are psychological entities that are tokened in mental states that employ those particular concepts, such that grasping a concept is a matter of having the appropriate psychological structure in one's brain available to be used in beliefs and other kinds of mental states; in contrast, according to people such as Dummett (1993), concepts are abilities such that possessing a particular concept is a matter of being able to consistently employ the concept correctly in thought and language to communicate with others and refer to what it represents; others such as Peacocke (1992) claim that concepts are abstract objects, sometimes identified with Fregean Senses, that are ways of thinking about and referring to, or modes of presentation of, potential referents. According to this latter view, grasping a concept would be a matter of thinking or talking about what the concept purported to refer to under the appropriate mode of presentation. There is not scope to engage substantially in this debate in this paper; however, it is sufficient for my purposes that all of these views about what it is to grasp a concept agree that it involves being able to reliably employ the concept correctly in thought and language, whether this is a matter of tokening the right psychological entity, communicating with others, of thinking about something under the right mode of presentation. And if someone cannot do this, then she does not possess, or fails to grasp, the corresponding concept.

And conversely, there might be things that it would be odd to hope for, or wish for, that it might not be odd to desire. For example, in his discussion Shoemaker also suggests examples of counterparts of Moore-paradoxical utterances for hoping, like “Would that he would come, but I hope that he doesn't”, and intention, like “I'll be there, but I intend not to be” (Shoemaker 1996, 46).

An analogous claim is true with belief and other kinds of cognitive state such as supposition or imagining. It does not seem odd to suppose that it is raining and I do not believe that it is raining. But this is no objection to the claim that it is odd to believe this and that there can be Moorean beliefs.

I take this term from Millgram (1997), though my use of it is different from his. He uses ‘orectic state’ to ‘denote psychological states that seem to involve an attraction to their objects’ although he allows that there are orectic states other than desire (Millgram 1997, 13). I will use ‘orectic norms’ to talk about those norms for the formation of desire alone, and not to include norms for the formation of other kinds of pro-attitude.

Or rather, that we might expect to be violated by paradigm cases of Moorean desires. Just as there might be variations of Moorean beliefs that violate doxastic norms other than avoiding falsity and ignorance, so there might be variations of Moorean desires that violate the orectic norms that are analogous to these (again see Hájek 2007).

Prima facie this is likely to seem especially surprising, that it is an orectic norm that one should have desires about more aspects of the world, and that one might be going wrong in an orectic sense if one fails to have a desire about something. However, on closer consideration, this complements our ordinary understanding of agents doing well and flourishing. For instance, people who have very narrow interests are often criticised as being obsessive and encouraged to ‘broaden their horizons’, and take an interest in the world. This is one of the reasons why people are sometimes suspicious of academics, for example, as their focus on a particular subject matter to the exclusion of interest in other issues is seen as unusual. Indeed, people who are particularly concerned with only few aspects of the world are sometimes referred to derogatively as ‘geeks’ or ‘nerds’. So it is part of our everyday understanding of agents that they are healthier or flourishing to the extent that they are interested in, and have desires about, more aspects of the world than fewer. That is, agents flourish more to the extent that they are less indifferent.

See, for example, Velleman (1992), Fehige (2001), and Platts (1979). Note that this sense of satisfaction is independent of the desiring agent's awareness of the satisfaction of her desire. This allows for the possibility of desires persisting though their satisfaction, for example, if the object of the desire in fact obtains while the desirer remains ignorant of this.

Similar cases of varying liability for violating the doxastic norm of avoiding ignorance can also be easily imagined but for reasons of space I will not describe any here.

While we are perhaps more familiar with cases of ambivalence involving desires for different, incompatible things, such as when someone desires that she goes to the beach this afternoon, and also desires that she stays home to watch the television this afternoon, this is because we assume that the desire for one entails having a desire not to have the other that is incompatible. However, just as we might fail to believe everything that is entailed by one of our beliefs, we might similarly fail to desire everything that is entailed by one of our desires; so it is possible that someone could have desires for incompatible things without actually desiring not to have one or other. So these cases might not actually be cases of ambivalence that, strictly speaking, are cases of having conflicting attitudes towards a particular state of affairs.

Of course, the sense of error or fault that is attributed to the agent here is a very attenuated, purely orectic sense. She is going wrong when judged against some or other orectic norms alone. This may not be a sense of going wrong that we take very seriously in everyday assessment of agents, but it is a fault nonetheless, as it is a failure to attain certain standards that apply to the agent.

See, for example, Hume (1975), Ryle (1949), Lewis (1983), Stalnaker (1984), Smith (1994), and Braddon-Mitchell and Jackson (1996) for arguments in support of the view that having a desire entails being motivated. See, for example, Strawson (1994) and Schroeder (2004) for arguments against this view.

Indeed, according to Anscombe, ‘the primitive sign of desiring is trying to get’ (1957)

Consider an elderly man's desire that he retire and spends more time with his wife. We can imagine that if his wife were to die suddenly, or if they were to separate, then he might not want to retire, despite maintaining his original desire that he retire and spends more time with his wife. So his desire for the conjunction does not entail an independent desire for at least one of the conjuncts simpliciter. This failure to distribute across conjunction is one respect in which desire differs from belief.

For example, I might want my department to hire philosopher A and want the department to hire philosopher B yet not want the department to hire both philosophers A and B because I am also aware that A and B do not get on and so would disrupt the harmony of the department. (Thanks to Alan Hájek for this example.) But the possibility of the kind of fragmentation of belief described by Lewis (1982) suggests that in this respect of failing to agglomerate over conjunction desire is similar to belief.

Thanks to an anonymous reviewer for this example and for pressing me on this point.

Indeed, Frankfurt agrees on this point, saying that ‘Someone who wants only in this truncated way to want to X stands at the margin of precocity’ (1971, 15). Thanks to an anonymous reviewer for bringing this point of agreement to my attention.

Furthermore, in this respect, these desires are again similar to Moorean beliefs. It seems possible to have Moorean beliefs (and Alan Hájek has recently argued that certain people actually do) and in such cases close inspection reveals the belief to be odd. One of the examples discussed by Hájek is the preface paradox in which ‘[y]ou preface a long book that you have written with the modest words: “Despite all my efforts, I am sure that there is at least one mistake somewhere in the book”. But the book itself can be regarded as a very long assertion – the conjunction of many individual assertions’ (Hájek 2007, 222). He argues that ‘[y]ou should be as committed to this unwieldy sentence as you were to the original book […]. Conjoining the unwieldy sentence to you preface, we have an assertion of the form:

[MY BOOK] (the unwieldy sentence), and I believe that (corner quote)MY BOOK(corner quote) is false (since I believe that at least some conjunct in it is false).

Schematically, we have the dreaded

p, and I believe that (corner)p(corner) is false. (Hájek 2007, 222)

Is this kind of case so odd? In fact, there seem to be actual cases in which someone might reasonably have such a combination of attitudes, for example, if she is self-deceived about wanting to break her diet or about liking low-brow television. In such cases, someone might intelligibly both want the cake and believe that she does not want the cake, or might both want to watch Days of our Lives and believe that she does not want to watch Days of our Lives. But perhaps such cases can be explained away by claiming that the belief is not a belief that the agent does not have that desire at all, but rather that it is not her strongest relevant desire. That is, that she both desire that she has some cake and believes that she does not really want some cake, because she also believes that she wants to stick to her diet. Nonetheless, these cases do not strike us as odd to a similar extent as Moorean beliefs.

However, as I understand him, Millgram does not claim that this case is Moorean himself. His concern is to show that desiring involves having certain backward-directed commitments (about one's reasons for desiring something) and certain forward-directed commitments (about making certain practical and other inferences), with the aim of showing that instrumentalism is false and that the rational objects of desire are not unconstrained (Millgram 1997, 29).

Note that as Millgram describes the case he excludes the agent from believing that she has her desire for means-ends reasons, or merely because she feels like having it where this might be satisfying or pleasant.

One traditional analysis of desire claims that it is necessary for having a desire that p that someone is motivated to act to bring about p: that is, if someone desires that p then she is disposed to act in ways she believes will bring about p. This is a view expressed in another of Hume's slogans, that desires are ‘motivating passions’. However, an objection to this Humean view appeals to the possibility of desiring something that one takes to be impossible. If someone desires something that she takes to be impossible, perhaps merely nomologically impossible, such as desiring that she travel faster than the speed of light, then she cannot reasonably also believe that there is anything she can do to bring about what she desires. But then, according to the objection, she cannot be disposed to act in any way in virtue of having this desire. So desires for things that are believed to be impossible are presented as counter examples to the Humean view that desiring something entails being motivated to try to get it.

Yet, if what I have argued here is correct, that a desire to have a particular desire that is frustrated is Moorean, and that this implies that one cannot rationally desire something that one believes one cannot obtain, then this objection can be avoided to a certain extent. One plausible way of understanding the project of giving a theory of desire is as aiming at specifying the conditions on a rational agent having a desire. If the Humean view is understood in this context, then desires for something that you take to be impossible are not counter examples at all: they are not desires that a rational agent could appropriately have so the fact that someone might not be motivated to act in virtue of having such a desire is not a problem for a theory about the desires of rational agents.

As mentioned earlier, my aim here was to suggest a counterpart of Moore's paradox for desire alone. Yet desire is only one of a number of pro-attitudes or conative states that we are familiar with from everyday psychology. Others include wishing, hoping, intending, and so on. Intuitively these seem to be genuinely different (if related) kinds of mental state but it is not clear exactly what it is that distinguishes between them. One proposal for distinguishing between them is in terms of the constraints on the objects of the different kinds of pro-attitude: it is plausible that what it is rational or reasonable to desire is different from what it is rational or reasonable to hope for or wish for. For example, one might think that there is something wrong with desiring that something in the past had been different but think that there is nothing wrong with wishing that something in the past had been different. Moorean desires suggest constraints on what can be the object of desire. If we could identify counterparts of Moore's paradox for the other pro-attitudes as well, and in doing so identify the constraints on the objects of those pro-attitudes, we would then have a principled way of distinguishing between kinds of pro-attitude.