Tuesday, August 18, 2009

The liar paradox and desire

The standard desire version of the liar paradox is to consider a person whose only desire is to have no satisfied desires. But that's a weird enough desire that one might wonder if it's possible to have it. Here is a version of the liar paradox using desires that are more imaginable.

Malefa has only one desire: That none of Bonnie's desires be satisfied. Bonnie has only one desire: That all of Malefa's desires be satisfied. Whose desire is satisfied? If Malefa's is, then Bonnie's isn't, and Malefa's isn't. If Malefa's isn't, then Bonnie's is, and Malefa's is, too.

What assumptions does this paradox depend on?

  1. It is possible that Malefa and Bonnie both have the above desires.
  2. The following disquotational schema for desire satisfaction is correct: A desire that p is satisfied iff I(p) (where I(p) is p rewritten with the subjunctive mood replaced by the indicative; thus, I("he eat ice cream") is "he eats ice cream"; to be more precise in the schema, I need to put in quotation marks of the right sort, but I'm not going to bother);
  3. Classical logic.

In regard to (1), one might worry that it's not possible to have only one desire. But that's easily handled by modifying the cases. Maybe Malefa's strongest (or most intense or latest acquired) desire is that Bonnie's strongest (or most intense or latest acquired) desire not be satisfied, and Bonnie's strongest (or ...) desire is that Malefa's strongest (or ...) desire be satisfied.

Moreover, there is nothing absurd about having desires that someone else's desires be or not be satisfied.

One could do what I did in my Monday post and argue that whether Malefa actually manages to have the indicated desires depends on what Bonnie desires (or vice versa or both). Somehow, I find this less plausible in the case of desire—I guess I feel a pull of a certain internalism about desire.

A different move would be based on the Gorgias. In the Gorgias, Socrates argues at length that the tyrant, though he is able to put enemies to death and all that, gets less and less of what he wants the more powerful he is. The reason for that is that he does not really desire to put enemies to death and all that—what he really wants is happiness. There are a couple of ways of taking Socrates' point. One way is to say that there are no instrumental desires. If the tyrant had a desire to have enemies put to death, that would be merely instrumental. Another way (I think Heda Segvic took this view) is to make desire have a normative dimension, such that to desire is to desire appropriately, so that the tyrant does not desire the deaths of his enemies.

Both of these two readings undercut the view that everything that can be put in a (subjunctified) "that clause" can be an object of desire. Moreover, they in particular make questionable the possibility of one person desiring that another's desires not be satisfied: that desire seems too much akin to the tyrant's desire that so-and-so die.

The paradox gives support for the thesis of the Gorgias. But there is something uncomfortable in using a paradox to give support to a substantive philosophical position.

Moreover, one might think that the solution in the case of the desire-satisfaction form of the paradox should be the same as in the case of the truth form. I am not completely sure. (Here is a consideration to back up my uncertainty: the complements of desires are subjunctified that-clauses, while the complements of beliefs and assertions are indicative that-clauses. This observation weakens—ever so slightly—the standard view that the object of a desire is a proposition. But the object of belief and assertion is a proposition. (I say this without committing to a realism about propositions.))

12 comments:

  1. Here's an idea. Suppose someone played a prank on you by telling you in a very serious voice, "Toves are slithy." Because you respect them, you come to "believe" that toves are slithy, in that you are prepared to say these words in an assertoric tone. But because they are nonsense, there is no content to your belief. (I want to say you do have a belief, but one with no content; I'm not sure whether this is right.)

    By parity of reasoning, you could come to want it to be true that toves are slithy. But equally there would be no content to this desire (and again, I want to say you have a desire, but one with no content; again I am not sure this is right.)

    Perhaps the best way to understand liar-paradoxical desires is in something like this way: they are desires all right, but without content. So they are neither satisfied nor unsatisfied, strictly speaking. (And mutatis mutandis for beliefs and sentences.)

    ReplyDelete
  2. A readiness to utter noises does not suffice for a belief. Suppose I know no French and an authority says: "Le neige est blanc." I accept her utterance as true, and am prepared to repeat it. What beliefs do I have? Obviously, both here and in the toves case, I have the second order belief that a certain utterance expresses a true proposition. This is unproblematic. On your suggestion, in the toves case I additionally have a content-free belief. By analogy, I should have some second belief in the French case. A content-free belief, again? That doesn't sound right, as then how do we distinguish this from the toves case? But the only content that I can see is that snow is white, and this surely I do not believe (supposing for simplicity that I didn't earlier believe it).
    For another case, suppose a little girl tells me: "Je suis dix ans". I trust her and repeat her words. Surely that's not a belief. For if it's a belief, it would be a belief that *I* am 10 years old.
    I wonder, actually, if it isn't the case fairly often that we do not actually believe that p, but only that "p" is true, even in the case of fairly ordinary sentences of one's native tongue, say ones involving scientific terminology like "electron". I wonder if belief isn't more of an accomplishment than is usually thought. That would also help explain why faith is a gift. That "Christ is Lord" is true needs no grace to believe; that Christ is Lord needs grace to believe.

    ReplyDelete
  3. The reply you outlined is, I find, pretty common when I push this hypothesis. Here’s my problem with it. If it were right, then there would be a sharp distinction between first-order beliefs, say that snow is white, and second-order beliefs that some sentence(-type) is true. There would be a correspondingly sharp distinction between understanding propositions, as in the first case, and having no idea what they mean, as in the second.

    But in fact, the distinction is not sharp. There are many things we believe (in any ordinary sense of that word) but do not fully understand. We would need some account of what ‘understanding/grasping a proposition’ amounted to, but it might be true that we do not fully understand hardly anything. We can competently use words in sentences and in thought and inference, having only a partial definition of them.

    For example, suppose on my first day of work my boss tells me, “You need a stapler. You can get one from the secretary.” Antecedently, suppose, I have never heard the word ‘stapler.’ But it is wrong to say that now I only believe “I need a stapler” is true and “I can get a stapler from the secretary” is true; I believe that I need a stapler, whatever that is, and that I can get one from the secretary. I can also infer that staplers are physical objects, probably office supplies, are relatively small and lightweight, that automobiles are not staplers, and so on.

    So I think the most viable position is approximately this. Understanding is a matter of knowing how to use a particular word, in e.g. observation and inference. Very often, understanding is partial, and it comes in degrees. A non-French-speaker’s relations to “snow is white” and to “la neige est blanc” are simply at opposite poles of the understanding spectrum. In principle we should say that her attitude toward them may be the same (i.e. belief) but in practice we often use ‘belief’ for more-paradigmatic cases of near-full understanding. (I think this view has close connections with the medieval slogans of “faith seeking understanding” and “believing in order to understand.”)

    What’s going on in Liar cases is that there is literally nothing there to be understood. But whether or not something is there to be understood is not a matter of analytic truth or introspection or phenomenology. And since belief can come before understanding, it is possible to believe when there is nothing to understand.

    (Also, I don’t see why I can’t trust the little girl and come to believe something ridiculous. The worst that could happen would be that I come to have contradictory beliefs. But so what, since there is a clear story to be told about why I wouldn’t realize this?)

    ReplyDelete
  4. 1. I actually find the general view you express much more plausible than the application to the paradox cases. In the latter case, I have a hard time getting over the observation that to believe is to believe something to be the case, and to desire is to desire that something be the case. If believings and desirings are entities, then an exact duplicate of a person will presumably have entities corresponding to the believings and desirings of the original, and it will be plausible to say that these entities are of the same kind as those in the original, and hence that they are also believings and desirings. But believings and desirings, likely, aren't entities (or aren't all entities--but nevermind that), and to have a believing is simply to believe something. Exact duplicates of a person, then, might differ in their believings. And whether one is desiring or believing is not something one can always introspect.

    2. The stapler case does not seem very difficult for the opponent of your view to accommodate. I come to believe that I need something called a "stapler", and I know some properties of the thing called a "stapler". I may later come to know the essential properties of the thing, and then I know that I need a stapler.

    3. Language has to be anchored to the world, and that, I think, will require someone's use of the word to involve more than just this kind of verbal transmission from one person to another. (An overall best-fit approach will fall to indeterminacy of translation issues.) So there will have to be some cases where one has a concept in a qualitatively more substantial way. Let's call that "internally having" the concept. (There are two ways of internally having a concept. For explicitly stipulated concepts, like that of the meter (=def 1/299792458 of the distance light travels in a vacuum in a second) or the mile (=def 1609.344 meters), one internally has the concept by knowing the definition. It is possible in a case like that to internally have the concept in the definiendum without internally having all the concepts in the definiens. Non-stipulative concepts will require something else.)

    4. The claim in (2) is compatible, interestingly, with the possibility that, excepting God, nobody has any beliefs where they internally have all the concepts in the belief.

    5. Here is a sketch of an argument for your view. If one tries to come up with an account of the case where one really has a first order belief, when one tries to build in all the understanding that is necessary, then it is likely that the resulting account of belief will either make false belief impossible or at least will make it impossible to have necessarily false beliefs (e.g., false mathematical beliefs). But this Spinozistic conclusion is unacceptable to many. (I am less scared of it than most.)

    6. There is an interesting metaphysicalized variant of your approach, which I find appealing. When you tell me that I need a stapler, and I don't know what staplers are, parallel to the physical communication process there is a non-natural causal process where the concept of a stapler is acquired by me (e.g., a non-natural state of your mind causes a non-natural state of my mind of the same type). I now have the concept of a stapler, and am the different for it. If, on the other hand, you tell me that I need a tove, and in fact there is no concept of a tove (I'm speaking a lot of nonsense here by using the word "tove", but let's not worry about that too much--ladders can be pushed away), then the non-natural aspect of the communication process does not occur. My mind does not come to have a new non-natural state. On this metaphysicalized approach, having a concept of C is an intrinsic property of a mind, exact duplicates of persons have the same concepts (but maybe not the same beliefs, because compositionality may not hold), and in fact all concepts are internally had. But there is a variation in the mastery of the inferential connections.

    ReplyDelete
  5. Re 1: So the choice is (a) whether one believes is introspectible, but some believings can be contentless, or (b) all believings have content but whether one believes is not introspectible. I think identifying that dichotomy is a considerable achievement right there, since it rules out the standard view on which all believing have content and one can introspect whether one believes. In neither case are intrinsic duplicates guaranteed to have the same beliefs. If (a) is true, the contents of the belief (or whether it has content) is an extrinsic matter. If (b) is true, whether one believes at all is an extrinsic matter.

    I am trying to figure out how these two possibilities apply to Liar cases and I don’t yet know.

    Re 2: Your solution implies that everything has essential properties. That would be a lot for some people to swallow. It is also a pretty heavy metaphysical conclusion from premises in the philosophy of mind and language…maybe that’s okay. But the problem can also be raised for any kind of words, not just common nouns. My boss might tell me that I need to “perambulate” or be “obsequious.” Are there parallels for verbs and adjectives of essential properties?

    Re 3: I want to hear more about this.

    Re 5: I think your Spinozistic conclusion is a reductio. There is also the possibility that no one has any false (or necessarily false) beliefs if they are not omniscient, which would be even worse. This would follow if belief required complete understanding of everything connected with it, which required complete understanding of everything connected with that….

    ReplyDelete
  6. Yes, the dichotomy of solutions is helpful. I hadn't thought of the possibility of a contentless belief.

    I wonder whether the distinction between the two solutions isn't merely verbal. Suppose my solution is correct. Then there are beliefs and pseudo-beliefs. Nonetheless, there is a fairly natural kind, call it quasi-belief, which subsumes the two, since the two must have something significant in common if one can switch between them by moving between duplicates.

    Suppose your solution is correct. Then there are beliefs and contentless beliefs. But, of course, there also will be a class of contentful beliefs, and it will, surely, be a fairly natural class.

    So the only question is whether to identify the ordinary-language "beliefs" with contentful beliefs or with quasi-beliefs. But I am not sure anything rides on this question.

    Maybe the way to think about this is as follows. The ordinary concept of a "belief" has two sides to it--a semantic side and a (broadly) explanatory side. On the semantic side, the belief is a belief in a proposition, which proposition may be about certain things that are referred to; the belief may be true or false. On the explanatory side, the belief helps to causally explain certain actions of mine (averrals, jumpings, etc.--it all depends on the belief) and certain phenomenological states of mine, and its presence in me is also causally explained by other beliefs, sensory inputs, whatever.

    The analogous thing is true for desires.

    (I have omitted normativity here. That is because there is probably normativity both on the semantic and the explanatory side.)

    What the contingent liar shows is that we can have states that have the explanatory aspects of belief without the semantic aspects (unless we're prepared to ditch classical logic). And analogously for desire.

    So, we have these states that have the explanatory aspects of belief without the semantic ones. We can make two moves. We can stress the (broadly) explanatory aspect in the ordinary notion of belief (including the phenomenology--feelings of conviction, etc.). If we do that, then your solution will appear less revisionary--we simply extend the concept of "belief" to include some contentless states. Or we can stress either the semantic aspect, or the need for a conjunction of the semantic and the explanatory, and we get my solution.


    On both solutions, assuming that phenomenal properties do not differ between exact duplicates (I am actually not sure of this!), it follows that whether some mental state is a contentful belief is not introspectible. I do not know how revisionary that is, given that a lot of people are willing to say things like: "Reflecting on what I felt, I realized that I never really believed p."

    I probably shouldn't have used the word "essential". I certainly didn't mean "essential" in the contemporary sense. I meant it in something closer to the medieval sense. The essential properties are the central ones, the ones that define the thing, or action, or whatever. They are the ones that answer the Aristotelian "What is it?" questions. :-)

    I don't have anything deep to say about my point 3. I hope that some version of the Thomistic account of mental indwelling of forms can help, but it's not something I've worked on.

    ReplyDelete
  7. It would be nice if one could find an argument for the same, or similar, conclusion not based on a paradox. People are generally not receptive to arguments that are centered on a paradox. Maybe they worry that there is another way out, or that in the vicinity of a paradox there is so much contradiction that one might simply be using Explosion.

    Here would be a non-paradoxical way to construct a quasi-belief that is not a contentful belief. Take a word w that (like all English words) can only occur in certain sentential positions in contentful sentences. (We'll have to have to worry a bit about the contentfulness of ungrammatical sentences. Let's bracket that.) Suppose that x is a user of the language whose interaction with the other language users has been sufficient to ensure that x can generate many grammatically correct sentences using w, but not sufficient to ensure x's being apprised of those rules of the use of w that ensure that w cannot be contentfully used in a quasi-sentence s(w). (A quasi-sentence is a sequence of marks that is treated as if it were a sentence by the speaker.) Then, we construct a world W* with a duplicate x* of x, but where in W* the linguistic rules that x* has not come in contact with do allow the quasi-sentence s(w) as a meaningful sentence. (If we're holistic about word-meaning, in W* the word w has a different meaning from the one it has in the actual world. That's fine.)

    Now, suppose that in the actual world, x forms a mental state that he would naturally identify as "the belief that s(w)". (Maybe he sees the inscription "s(w)" in a book that he takes to be authoritative.) Then, x* forms a mental state that he would naturally identify as "the belief that s(w)". Now, x*'s mental state is in fact a belief, a belief that is correctly identified in x*'s language as "the belief that s(w)", and hence a contentful belief. However, x's mental state is not a contentful belief; for the only contentful belief it could be would be one described as "the belief that s(w)", a description that is nonsense. Thus, x has a mental state that is not a contentful belief, and x* has a mental state that is a contentful belief, even though the two states are exact duplicates.

    (David Alexander and I have a way out of this argument, assuming essentiality of origins.)

    The word "sake" would perhaps lend itself to such an argument.

    ReplyDelete
  8. I think there is a lot to be said for your “merely verbal dispute” diagnosis. I was trying to figure out if there were some important way the solutions differed but the problem is hard and I didn’t come up with anything yet.

    I also think there are important arguments to be made along these lines that do not depend on paradoxes. In fact, what interested me was that perhaps light could be shed on the Liar by connecting it with these lines of argument. Here are three, related:

    1. (Quasi-)beliefs by testimony. People will generally trust authoritative-seeming testimony. They may understand it only partially. There doesn’t seem to be any problem with believing testimony you mostly, but not fully understand. Since understanding is a continuum, however, you can believe what you only barely understand. And then it is only a short step to the claim that you may believe something you do not understand at all. (One might draw the line at being able to parse the sentence grammatically.) But then there seems to be no barrier to (quasi-) believing parseable nonsense which cannot be understood.

    2. (Quasi-)beliefs by Burgean Variation. Burge has an argument, persuasive to me, that the contents of my beliefs involving arthritis are not determined by what’s in my head but by the social use of the concept. Now simply vary the cases so that my understanding of arthritis gets farther and farther from everyone else’s. At the limit, I have some totally bogus nonsensical (quasi-)belief. I’m not sure this argument works because one might reply that at some point Davidsonian considerations of charity kick in and we re-interpret my use of ‘arthritis’ to mean some other concept. I think this is most like your x* argument.

    3. Skeptical scenario (Quasi-)beliefs. Putnam argues that we are not BIVs because if we were BIVs our thoughts would have no content and our words would not refer; but our thoughts do have content and our words do refer. If the first premise is right, then people trapped in skeptical scenarios have lots of quasi-beliefs. [And there is another continuum between us and them; we could be trapped in a “pretty skeptical” scenario, where an evil demon systematically deceives us about the chemical makeup of our water. We would then be systematically deceived not only about water but about the contents of our beliefs about water, at least once we had read Putnam on meaning and reference. Such “pretty skeptical” scenarios can be as skeptical as you like.]

    ReplyDelete
  9. Heath's solution (first comment above) seems to me to be that the paradoxical (as I find them) desires cannot be satisfied or unsatisfied. But I don't see how that solves the paradox. Bonnie has a desire, on Heath's approach, and it is not satisfied. I don't see how, if Bonnie's desire is not satisfied, Malefa's desire would fail to be satisfied. But even if it does so fail, if a desire is neither satisfied nor not satisfied, then it seems to me that that desire does not exist (via reductio ad absurdum). And it seems to me to be obvious that such desires could conceivably exist (insofar as desires exist). As Heath says, I might desire that toves be slithy.

    ReplyDelete
  10. enigMan:

    The argument I give shows that at least one of the two desires is not contentful. Paradox would not ensue if we supposed Malefa's not to be really a desire that Bonnie have no satisfied desires, but a quasi-desire that Malefa describes as "a desire that Malefa have no satisfied desires", but supposed Bonnie to have a contentful (and unsatisfied) desire. If we have the Socratic normative ideas, we might think that Bonnie has a desire but Malefa does not, since Malefa is malevolent, or something like that. Or we might think: They're both on par, so if we say one is a mere quasi-desire, we should say that both are.

    Heath:

    Maybe we should write something together on this eventually?

    I don't think standard Davidsonian charity will escape the Burgean variation argument, because if we vary enough, there will be no truths in the vicinity. Though there is an extension of Davidsonian charity that goes like this (I've never heard it, but it seems sensible): When choosing an interpretation:
    1. First, maximize the number of contentful claims.
    2. Second, maximize truth.

    ReplyDelete
  11. I see what you mean, but are those desires really not contentful? The argument you give shows that the two desire are paradoxical in the sense that at least one of them can be neither satisfied nor unsatisfied. It does not show that there cannot be such paradoxical -- or rather, quasi-paradoxical -- desires -- I don't see why such desires should be impossible, why their possibility is beyond belief; whereas that they are desires seems self-evident.

    Surely one's strongest desire might be that some other's strongest desire is not satisfied (as you say). A desire is something I have, something I feel. Why should it have to be satisfiable or not in order to exist as a desire? I would even go so far as to suggest that people do have such desires, e.g. when a couple is splitting up. Why should the content of a desire disappear just because substitution makes it inconsistent? If I see a mirage and desire to drink what I take to be its waters, I do not have the desire to drink its waters, as there are no such waters, but I do have a desire. That desire will quite simply not be satisfiable, but I think that your argument (which I love) shows that some desires are quasi-paradoxical.

    ReplyDelete
  12. Hmm... upon reflection my last comment was confused, sorry; but I think that you have the best argument I've seen for dialetheism. Why should the desires not be, each of them, both satisfied and not satisfied? I normally reject dialetheism on the grounds of the obvious meaning of 'not'. But desires can be complex things. Even the simplest desire might be only sort of satisfied in many ways... What I think you have shown is that such complexity lies between simple satisfiability and dialetheism inclusively.

    ReplyDelete