Thursday, May 22, 2008

Constitutive Features of Desire

I want to lay down a few constitutive features that, it seems to me, desires must have in order to be desires. I welcome criticisms of the features I mention, and would love to see other suggestions added. I suggest three features (I will address the last two in future posts): conditions of satisfaction, affect, and representational content.

The important distinction, I think, is between occurrent desires (for which the word is normally reserved in ordinary language) and dispositional desires, which are a trickier breed of animal. This is all further complicated by the fact that what we ordinarily mean by “desire” is only one of the many different kinds of pro attitudes. Here I will be using “desire” confusingly to stand for both the narrower and the broader class; that is, I will be taking anything falling under the pro attitude heading as a desire, but will also appeal to the narrower sense. This creates particular complications in light of the argument many have raised against the BDI (belief-desire-intention) theory of action, i.e., the argument that for plenty of actions no desire is needed—the making of a promise, for example, is taken as providing sufficient incentive for keeping it. To say that that incentive itself must be a desire on formal grounds (i.e., because all actions involve desires) would be question-begging. The key, then, is precisely to try to give the constitutive features of desire, such that some such entity can be posited in the etiology of any action without begging the question.

1. Aiming at Satisfaction. Desires necessarily aim at their own satisfaction. I think this is fairly uncontroversial: it is hard to imagine what it would mean to have a desire that does not contain a push toward its fulfillment. Before moving on to this issue, let me note something I think is important: it is in a sense the desire, not the agent, that aims at satisfaction. The agent can, of course, aim at the satisfaction of the desire, but she does so by adopting or incorporating the desire. This is clear from the fact that we have (or can have) plenty of desires that we do not seek to fulfill—violent, fantastic, or incompatible with our other goals. These desires do have their own conditions of satisfaction; but those are not the agent’s conditions unless the agent makes them her own. I will develop this point in a future post, but I think it is important to the literature on the place of agents in event-causal theories of action. Another brief hint at the issue: akrasia clearly involves satisfying some desire; what makes it problematic is the issue of how agents can be taken to commit themselves to actions that aim to satisfy desires the agent has not (or not wholeheartedly) endorsed.

The conditions of satisfaction of occurrent desires are also conditions of their termination: if you have an occurrent desire, and you fulfill it, the desire goes away. If it fails to go away, that means you have not fulfilled the conditions of satisfaction; perhaps you haven’t quite got what those conditions are, or perhaps you’ve only partially satisfied them.

Dispositional desires are more complicated. First off, most of them are not what we would—in everyday language—call desires. Imagine, for example, that whenever I see a chocolate cake, given certain conditions (the cake is not too expensive, I am not already full, etc) I will eat a piece. What this indicates is a reliable disposition toward chocolate cake, but it would be awkward—in ordinary language—to call this a desire. Rather, we might be drawn to another alternative: the disposition is really a disposition toward desiring chocolate cake whenever one is present, namely, a disposition toward developing occurrent desires of a certain kind under certain conditions. But this alternative has serious problems. For one, it is not at all clear that an occurrent desire needs to be present every time I encounter a cake—the disposition alone is sufficient to explain the action, both rationally and causally. Second, this gets even more complicated for non-appetitive things, such as showering regularly (some people really enjoy showering and want to do so as much as possible; I am thinking more of the people who do it just because it is socially required). A disposition to shower regularly provides a good predictor of the behavior of many people, but to say that all people who shower regularly have occurrent desires to shower whenever they do so seems excessive. Similar points for brushing teeth, dressing appropriately for work, and so on.

This latter point is sometimes raised as criticism of the BDI model of action. The idea of the criticisms is something like this: if you assume that occurrent desires are needed to rationalize every action, then you are merely introducing a theoretical element. There are certainly plenty of actions that we undertake without the presence of occurrent desires, and inserting these as premises in a practical syllogism that rationalizes every action simply distorts the nature of action in favor of a theory (externalists about practical reason are particularly prone to such arguments). But the criticism doesn’t work. We can be motivated dispositionally without any occurrent desires taking place. The question is whether these dispositions deserve to be called desires at all; can’t belief alone be enough?

Brandom, for example, argues that we really do not need desire as a premise in the practical syllogism because such syllogisms can involve material, rather than formal, inferences. That is, reasoning from “I am going out today” to “therefore I shall take a shower” is perfectly legitimate; we do not need to insert the extra premise that “I want to be clean around other people”; the premise is already implied, insofar as another premise (e.g., “I want to be dirty around other people”) is lacking. But I don’t think this sort of argument against the BDI theory is right. That the inference goes through in this case shows something about my motivational states (broadly construed), e.g., that I belong to a culture where regular showering is expected. Were I from a culture where it is not, to be sure, my regular showering would require a further explanation. But this does not mean that, in the first case, the postulate of a “desire” is mistaken. It means only that crude BDI theories, ones that take the beliefs and desires that enter into an action as necessarily occurrent states, are mistaken. Dispositional desires may well play a role; it is only that, against a social background in which the disposition in question is assumed, the particular disposition in question is not informative or particularly worth mentioning. But the disposition is still needed as part of the explanation of my action, even though we can easily leave it off when talking to other members of our society. It cannot be eliminated in favor simply of the background existence of social convention, or my recognition of the convention. For one thing, not everyone accepts social conventions; for another, there is no reason to think that just because something is a social convention, it cannot be a desire—many, if not most, of our desires are conventional in nature.

Positing a state—occurrent or dispositional—that is both goal directed and irreducible to either beliefs or social conventions or simply commitments (a disposition to keep my promises, for example, cannot be explained simply by reference to my beliefs that I have made commitments) seems necessary to the explanation of actions.

Continue Reading...

Thursday, May 15, 2008

Subjective Time and the A-Series

I’ve been lightly dabbling in philosophy of time (not very successfully, if last week’s letter from the APA is any indication) and wanted to throw out a couple of thoughts. The present line is largely in response to Alexander Pruss’s recent post on the famous “Thank goodness that’s over” argument, first formulated by Arthur Prior (no way is that name a coincidence, though perhaps “Paster” would have been more appropriate). Pruss argues, based on time-travel scenarios, that subjective time differs from objective time and then suggests that subjective time is logically primitive. I agree with these points. But then something a bit odd happens: Pruss claims that both subjective and objective time are B-theoretical, which seems to be the opposite of what his conclusions imply.

For those unfamiliar with the distinction, it derives from McTaggart’s “The Unreality of Time” and breaks up theories of time into two possibilities. According to A-theorists, the past, present, and future are real. According to B-theorists, we can arrange all temporal events according to earlier/later relations. B-theorists generally argue that earlier/later relations serve as truth-makers for all superficially A-theoretic claims; A-theorists respond that the B-theory cannot account for temporal indexicals or for the subjective experience of time.

Pruss’s argument, basically, is this: Let’s say that, while suffering from a headache, I go back in time. My headache magically vanishes in the very instant I find myself in the past. In this case, I think that my headache was in the past, but in fact it is clearly in the future. So it is in the future in objective time, but in the past in my subjective time, something we can mark by pointing out that the headache is something I remember, and am glad to have gotten over, but am not anticipating or dreading. The conclusion, then, is that subjective time can differ from objective time. This seems to imply that whatever may be said about objective time, subjective time is A-theoretic, since subjectively my headache is past, even though objectively the event of the onset of my headache may be later than my gladness that the headache is gone. Pruss disagrees, however, on the grounds that the A-theorist must hold that there is an objective present, and the view under cannot give us an objective present—at best, it could only give us a subjective present, and that is something the possibility of which Pruss denies. Before going on to that point, a brief aside about the form of Pruss’s argument.

His suggestion that subjective time is B-theoretic seems to go through the excluded middle. Here's a sloppy schematization (hopefully I have it right):

(1) Time is either A-theoretic or B-theoretic.
(2) A-theorists claim there is an objective present.
(3) Subjective time lacks an objective present.
(4) Subjective time is not A-theoretic. (from 2&3)
(5) Therefore, subjective time is B-theoretic. (from 1&4)
This won't do, since the B-theory is not just the fall-back position for anyone who doesn't think there is an objective present. After all, one can reject the present while still rejecting that all relations between events are reducible to later than/earlier than relations, and someone who rejects this is certainly not a B-theorist. So by the same token, we might argue that a theory that insists on the reality of past and future, even if it leaves out the present, is still an A-theory. At least, it is clearly not a B-theory, and although it differs from most A-theorists in leaving out the present, by insisting on the reality of past and future it retains some A-theoretic status.

But why should we reject the present? One problem, I think, rests on Pruss’s taking subjective time to be a variety of internal time; in fact, he argues that all objects have internal time, and subjective time just seems to be the form of internal time that applies to conscious entities. If we understand subjective time on this model, however, I do not think we are really talking about anything subjective; only something individual. If I pin a chronometer to myself, leave another one on my desk, and then go time traveling (without the desk), then the two chronometers will diverge (if I jump a hundred years into the future, spend five minutes there, and then return to ten minutes after my initial jump, my personal chronometer will have measured five minutes; the one on my desk, on the other hand, will have measured ten; immediately after the first jump, in the future, the chronometer on my desk will show a hundred years, while my personal chronometer will show no time has passed). But neither chronometer, as far as I can tell, is showing subjective time: both show objective time, it’s just that one is measuring objective time relative to my time-traveling body, while the other is measuring objective time relative to the non-time-traveling world. The very distinction between internal and external time, as Pruss presents it, seems to me to collapse: external time just is the internal time of the universe taken as a whole.

My point, then, is that Pruss’s view of subjective time doesn’t seem to be subjective: it seems to be objective time within a frame of reference that is only a part of the universe (my body) rather than the universe itself. It is on these grounds that Pruss thinks that we should reject the notion of the present. His arguments against the idea of a subjective present are, in this regard, very telling. In response to some prodding from Heath White, with whose comments I think I am in agreement, Pruss writes,

Surely the right answer to the question: "At noon, what time is subjectively present to x" is noon. (The subjective present isn't just the time one believes to be present--one can be wrong about what time it is!) What sense could there be in saying that at noon, it is 1 pm for x? Would that mean that x is literally now in the future? No. So, at every time t, the time that is subjectively present to x is t.
And Pruss correctly concludes that this is not a subjective present. But of course if we think of the subjective present in terms of “noon” or “1 pm”, then we’ve already set up the problem in such a way as to exclude the subjective present: after all, clock-time is an intersubjective convention, designed to co-ordinate the activities of human beings according to the regular order of objective time. And so it would not make any sense to speak of a subjective present as involving a point on a clock-face that differs from the objective present point on the clock-face.

But if we stop modeling subjective time on objective time, this difficulty goes away. Pruss grants, after all, that there might still be a subjective past and a subjective future. If so, then clearly we need not dispense with the idea of a subjective present: the standpoint I occupy, from which I have a subjective past and a subjective future, just is my subjective present. I doubt very much that we could have such a subjective present without reference to a subjective past and a subjective future, but this isn’t overly problematic, since “past”, “present”, and “future” are interrelated concepts; they come bundled together. Whenever we correctly apply one of them, we must be able to apply the other two as well. And this is effectively what distinguishes the A-series from the B-series. For on the B-series, no point in time logically implies an earlier or a later point; every point on the timeline has its own independent reality. But on the A-series, every point is interdependent. This, of course, is roughly why McTaggart thought that the A-series could not be real. But his attempt to cleave off the subjective from the real is not one we are bound to accept.

Continue Reading...