Does anyone hold the following view? If so, who? Where? When? Why?
The notion of desire is to be understood/analyzed in terms of the notion of preference, perhaps along the following lines:
S wants that P =df. S prefers P to not-P.
(The basic idea is that what one wants is (i) determined by one’s preferences, where preference is a contrastive relation that links a subject to two other entities, and (ii) we should accordingly analyze desire in terms of preference.)
Thanks!
I certainly think that the adjective ‘desirable’ should be analyzed in this way.
x is desirable =df. x is preferable to ~x.
Desirable is a gradable adjective, one that can be modified with degree adverbs like ‘more’, ‘less’, ‘very’, etc. And it seems plausible to suppose that gradable adjectives (e.g., ‘tall’, ‘fast’, and ‘smart’) must be analyzed in terms of their comparatives (e.g., ‘taller’, ‘faster’, and ‘smarter’). (I follow Mark Schroeder here — see p. 30 of his “Teleology, Agent-Relative Value, and ‘Good’,” which is online at his web site.) Now although ‘desire’ is a verb, not adjective, maybe we can think of it as gradable in the same way that we do ‘desirable’. Thus I can desire x more than y (which is just to say that I prefer x to y), and I can desire z the most (which is just to say that z is what I prefer to all of its alternatives).
So do you have some worry about the analysis of ‘desire’ that you give? It seems plausible to me. It seems that we have three things to analyze: (i) ‘desire’, (ii) ‘prefer/desire more’, and (iii) ‘desire the most’, and it seems that we need to take (ii) to be basic, and analyze (i) and (iii) in terms of (ii).
My ear disapproves of the above proposed analysis. If I prefer being dipped into a vat of bugs to being dipped into a vat of feces, I would not say that I want to be dipped into a vat of bugs. I do want that more than the alternative, but I do not want it simpliciter.
David,
The analysis doesn’t say that if S prefers that P to that Q, then S desires that P. It says that if S prefers that P to that ~P, then S desires that P. So how does what you say count against the analysis?
Ah, sorry. Maybe that takes care of my worry. I’ll think about it.
The view seems implausible.
I can 1) want that P, and 2) want that not-P.
But I can’t 1) prefer P to not-P, and 2) prefer not-P to P.
(Suppose that P = “My mother visits me for Christmas.”)
Eric,
Do you think that that the following proposition, E, can be both true and not true?
E: Eric, at t, desires that his mother visits him for Christmas.
To think that it can be seems really implausible, but isn’t that what follows from what you said.
Of course, we say things such as: “A part of me wants my mother to visit, and a part of me doesn’t want her to visit.” But isn’t this just a way of saying that my mother’s visiting would fulfill some of my intrinsic desires and thwart others? In any case, it seems to me that whether I desire that she visits depends on what I want, all things considered.
Hi Eric,
I don’t see why one can’t have inconsistent preferences; one shouldn’t, of course, have them.
Doug:
Eric isn’t committed to saying that E is both true and false.
S wants that ~P
does not imply:
~ S wants that P.
Just as
S believes that ~P
unfortunately does not imply that
~S believes that P.
Kris,
You’re right. Thanks for setting me straight. Of course, now that I see things more clearly, I’m with you in not seeing why one can’t have inconsistent preferences.
I’m with Eric — Doug, Eric clearly isn’t committed to the possibility of E being both true and not true! Rather, he thinks there are circumstances in which E is true and so is E’, namely, that Eric, at t, desires that his mother not visit him for Christmas.
I suspect that desire talk is loose. Here’s a hypothesis: whenever we say someone desires that S, there is some implicit contrast with some other state of affairs, S’, such that the person prefers S to S’. In the absence of any salient contrast, we’ll suppose that S’ is not-S. But this may not handle Eric’s example very well. Hm.
Hi Jamie,
Are S and S’ incompatible in some sense?
Jamie,
Yeah, I already conceded the point a couple minutes before you hit post.
We can define obviously technical terms as we want but I’m worried that the suggested definition fails to catch what we commonly mean by desires, wanting and preferring. Maybe it is rather a good definition for wanting*.
Here’s one problem. If the definition was right, then unwilling addicts would be conceptually impossible. Take the chain smoker who wants to quit but cannot get himself to do it. His evaluative judgment is that not smoking is much better state of affairs than smoking. After all he knows that he is trading his health for instant gratification. Therefore, he prefers not smoking to smoking. Does this imply that hasn’t got a desire to smoke? For me that suggestion sounds implausible even though this would be implied by the definition. After all, the guy really graves for a smoke. He wants it so badly that he cannot resist doing what he prefers not to do.
Of course, this is not a problem for Doug’s view which is not about psychological states but rather about evaluative properties of things – desirability and preferability.
Need to add that the addict would according to the definition want to not quit smoking if he prefers not smoking (and not only want to smoke). But, even this seems plausible – due the addiction it can be that the addict finds no part of him desiring to do what his evaluative judgment prefers, that is , not to smoke.
My cent and three-quarters:
I agree with Doug, as predicted, that ‘more desirable’ must be primitive with respect to ‘desirable’. And although there is more than one choice, here, I think that ‘more desireable’ is probably correctly understood as ‘correctly desired more’, rather than as, say, ‘more correctly desired’.
I part ways from Doug, however, over the initial question, about the relative priority of desire and of preference. My naive suspicion about human psychology is that preferences are not a fundamental explanatory feature of our psychological makeup, but that desires are, and this comports with what I think is common sense. Preferences are a mathematically tractable idealization that are useful for many modeling purposes, but mathematical tractability is not evidence of psychological reality.
My own naive view is that preferences are probably derivative from desires, but that things are complicated. Desires are complicated and multi-dimensional things, and not reducible to simple vectors in motivational space. Moreover, their effectiveness in motivation is highly subject to framing effects, and this is what I suspect leads to ‘inconsistencies in preference’.
So ‘correct to desire more’ does not reduce, on my view, to ‘correct to prefer’, as Doug suggested; rather, it is the other way around.
In answer to Kris’s original question, I think that my colleague Jake Ross has suggested Kris’s original analysis in conversation, but I need to double-check to be sure.
Sorry – just to be clear, I don’t agree with Doug that
x is desirable =df. x is preferable to ~x.
Even if ‘preferable to’ is the comparative of ‘desirable’, this wouldn’t follow. We wouldn’t think, for example, that
x is tall =df. x is taller than ~x.
Rather, to be tall is to be taller than sufficiently many in some comparison class. Similarly, to be desirable is presumably to be preferable to sufficiently many in some comparison class.
Mark,
I get that the following is false: “x is tall =df. x is taller than ~x.”
And I agree that “to be tall is to be taller than sufficiently many in some comparison class.” But why think that ‘desirable’ is relevantly similar to ‘tall’. What’s tall is clearly relative to some relevant comparison class. Thus what counts as tall for a giraffe is very different than what counts as tall for a human being. But why think that ‘desirable’ is relativistic in this way. It seems to me that x is desirable iff there is sufficient reason (of the right sort) to prefer x to ~x. I do, however, think that it is plausible to analyze ‘desirable’ in the attributive sense as ‘tall’ is to be analyzed. That is, x is a desirable F iff x is preferable to sufficiently many F’s. But I think that ‘desirable’ in the predicative sense should get the sort of non-relativistic analysis that I’ve proposed.
To sum up, then, why think that ‘desirable’ should be analyzed just as ‘tall’ is? And why do you object to my proposed analysis? It seems that just saying “Well, we wouldn’t analyze ‘tall’ like that” isn’t a good objection.
As an aside, would you object to the following:
S has a reason to desire that W (a possible world) is actualized iff S has a reason to prefer W to all other available alternatives.
Kris,
Are S and S’ incompatible in some sense?
Huh, I’m not sure. I’m inclined to say no, but (i) the ‘in some sense’ frightens me, and (ii) I can’t think of good counterexamples. Well, how about this. Which book do you want to read: Great Expectations or Little Women? I want to read Great Expectations.
The options are not incompatible, but I think the question is about my preference between the options.
I am not sure if this is a real issue or not and it may be a version of what has already been rejected above, but in any case here it goes: I may prefer A to ~A under circumstances C but prefer ~A to A under circumstances C’. Let’s say that right now, I find myself at C and I prefer A to ~A; but I would much rather be in C’ and highly dislike the fact that I am in C. Do I desire A? The definition says that I do and yet it seems to be possible that I don’t really desire A.
Example: If it is raining, I prefer to take the bus; while if it is sunny, I prefer to walk. It is now raining though and so I prefer to take the bus than walking. Do I desire to take the bus? I am tempted to say that it is possible that I don’t desire to take the bus but that I take the bus because it is the best available alternative. I prefer to take the bus under the current conditions which I may not really want to be in — maybe I don’t like rainy days or maybe I don’t like taking the bus. The definition would seem to say that if I prefer to take the bus then I desire to take the bus.
So anyway, I asked and Jake Ross does remember offering the account Kris originally asked about, but he also offered Eric’s counterexample to the ‘only if’ direction, and also had some examples he thought challenged the other direction, as well, but which I’m less sure about.
In answer to Doug – I take it we’re agreed that ‘desirable’ should be analyzed in terms of its comparative. I had a model in mind about how such analyses go, but maybe it was an overgeneralization.
Compare ‘seaworthy’, which is gradable. Some ships are more seaworthy than others. But perhaps being seaworthy full-stop is just being more seaworthy than, say, a typical rowboat (or some other threshold). I’m not really convinced this is right. But such a model would lead to an idea like:
x is desirable =df x is more desirable than a typical rowboat
(Or whatever threshold gets substituted for the rowboat.) That’s still not quite the same as your model for ‘desirable’, though. This might be wrongheaded, but I’m just expecting that if ‘desirable’ is a gradable adjective, we should be able to look to other gradable adjectives for models as to how it works, and yours didn’t fit models I was thinking of.
Even if inconsistent preferences are possible, I have the suspicion there are actual cases like this:
(1) A desires that P
(2) A desires that not-P
(3) A prefers not-P to P
(4) A does not prefer P to not-P
Ex:
(1) Andy desires that [his team loses and his love interest & coach of the other team, Bob, is happy].
(2) Andy desires that [his team wins and he and his players are happy].
(3) Andy prefers [his team winning and his and his players being happy] to [his team losing and Bob being happy].
(4) Andy does not prefer…
If some such case is possible, then the left-to-right conditional (‘if S wants/desires that P then S prefers that P to that ~P’) is false.
If we went with the comparative analysis of desire/want then I suppose 1 & 2 might modified to include comparison with the game being a tie, being rained out at half time, etc.
Sorry if I missed this being mentioned above – it might be like the smoker case but here it seems hard to deny he desires and has reason to desire each option.
Brad,
[his team wins and he and his players are happy] is not equivalent to ~[his team loses and his love interest & coach of the other team, Bob, is happy], so I don’t see how your second list of (1)-(4) is an instance of your your first list of (1)-(4).
The original question was, does anyone analyze desire in terms of preference. Wayne Davis, from “A Causal Theory of Intending,”:
“I believe desiring p is equivalent to preferring p rather than not-p.”
He does not defend this view here, but I’m sure he defends it somewhere.
About Doug’s suggestion: It would help if we clarified whether ‘prefers’ is an extensional or intensional relation. I hear it as extensional. But ‘desirable’ strikes me as an intensional predicate. So I don’t think you can analyze an intensional predicate in (very simple) extensional terms. For example:
the Janjaweed (the bad guys in Darfur) is after me, and I can let them kill me painfully, or I can jump off a cliff to my certain death. I prefer jumping to my certain death, to not jumping. But that doesn’t make “jumping to my certain death” desirable; if anything is desirable here, it’s “avoiding death at the hands of the Janjaweed”.
Continuing on what David Sobel said…
x – not being dipped into a vat of bugs.
I prefer x to not-x. (i.e. I prefer not being dipped into a vat of bugs than being dipped into a vat of bugs).
But I don’t desire not to be dipped into a vat of bugs.
This would be silly:
-What are the things you desire?
-Well, among the things I desire the most is not to be dipped into a vat of bugs.
Heath,
My analysis was in terms of what is preferable, not in terms of what is preferred. Again, the analysis is: x is desirable iff x is preferable to ~x.
Now a flourishing person’s jumping to her death is not, other things being equal, preferable to her not jumping to her death, and there’s an implicit other-things-being-equal clause in my analysis. So my analysis gets the right result: a flourishing person’s jumping to her death is not desirable. Of course, your jumping to your death is preferable to your being tortured and killed by the Janjaweed, but this does not imply, on my analysis, that your jumping to your death is desirable.
Tanasije,
As Parfit has pointed out, the term ‘desire’ is sometimes used such that one desires that x obtains only if one finds the prospect of x’s obtaining attractive or appealing, but philosophers often use ‘desire’ in a wider sense, such that to desire that x obtains is just to be motivated to bring it about that x obtains. On this view, desire is a kind of functional state: one desires that x obtains if and only if one is disposed to act so as to bring it about that x obtains. On this understanding of ‘desire’, it is not at all absurd for you to say, “among the things I desire the most is not to be dipped into a vat of bugs,” as I’m sure that you would be motivated to ensure that you are not dipped into a vat of bugs. So I take it that desire is ambiguous, and that the proposed analysis is in terms of ‘desire’ in the dispositional sense.
Doug,
Thanks for the quick response.
I am not clear that I need to defend that equivalence.
What is wrong with this?
P= Andy’s team wins
(P1) Andy desires that P
(P2) Andy desires that his team loses
(P3) Andy’s team losing constitutes Andy’s team not winning
(P4) Andy knows that P3 is true
(C1) Andy desires that Andy’s team not win
(C2) Andy desires that ~P
Hmm…not so sure about the way I put P3.
More specifically, I’m not sure we always desire that Q if we desire that P and know that P constitutes Q.
I need to capture the tight connection between losing and not winning. ‘Team A loses’ does not mean ‘Team A does not win’ so I can’t put that.
But the relation between losing and not winning seems so tight that if you are competent with the concepts of winning and losing, then you can’t help desiring that the team not win if you desire that they lose. And if you desire that they not win, then you desire that ~(they win).
Brad,
Okay, but I was worried about the right-hand conjuncts. His players being happy does not not constitute it’s not being the case that his love interest & coach of the other team, Bob, is happy. Right?
Brad,
If there are actual cases, where
(1) A desires that P
(2) A desires that not-P
(3) A prefers not-P to P
(4) A does not prefer P to not-P,
then just tell me what “A” and “P” stands for in one of those actual cases.
Talking about desiring is sometimes a bit odd but it is easier with wanting. I lost the dialect at some point so this might fail completely but here’s a possible case.
I’ve got a cup of really bad coffee in front of me now. I know it tastes really, really bad. I also know that it would wake me up.
I want to drink that coffee.
I want not to drink that coffee.
I prefer drinking the coffee to not drinking it.
I don’t prefer not drinking the coffee to drinking it.
Oh, I see; right. My first example was too quick, conflating objects of desires and reasons for desires – thanks.
How about this:
A: Andy
B: Bob
P: Andy’s team wins
These causal conditionals can be the reasons for Andy’s desires that P and that ~P respectively:
(R1) If P, then A is happy
(R2) If Andy’s team loses, then B is happy
Not sure this will work but…
I can, of course, imagine that Jussi desires not to drink anything that tastes as bad as that cup of coffee in front of him even though he doesn’t prefer not drinking that cup of coffee to drinking it. (Maybe he wants to wake up more than he wants to avoid the bad taste.) But I have a hard time imagining that Jussi desires not to drink that cup of coffee, all things considered, and yet he doesn’t prefer not drinking that cup of coffee to drinking it. I have the same problem with imagining that Andy desires that his team not win, all things considered, and yet he doesn’t prefer his team not winning to his team winning.
But I do desire not to drink it especially *all things considered* just as well as I desire to drink it *all things considered*. I think you are looking for what I desire more to do in the situation. As far as I know I desire to drink and not to drink just as much. But, I make the evaluative judgment that waking up is more important than avoiding the bad taste and proceed.
Jussi,
I never doubted that you do desire not to drink it, all things considered. But why should I not think that you also prefer not drinking that cup of coffee to drinking it?
I agree. Comparison with reasons might help & show that these are not cases of irrationality.
I can see there is reason to bring it about that P even if, all things considered, I think there is most reason to bring it about that not-P; outweighed reasons need not be silenced. (Cf. Dancy)
Similarly, outweighed desires need not be silenced.
Sometimes desiring what you do not prefer is a way of responding to the value of an option you cannot choose because you are in a forced trade-off btwn two good options.
A similar view arises when we think about regretting punishing even when that it the best option overall (we desire it more than the alternatives). Our regret records our sensitivity to the dis-value of causing others pain, but we still are right to prefer punishing because it leads to the most overall value (or the analogous retributivist line).
Cause I say so Doug. Hand over heart, I really did prefer drinking the coffee over not drinking it but did not prefer not drinking it to drinking it. Waking up is much more important. And, so I drank the coffee. Very bad. Truth be told it did not even work, had to go take a nap, and now I’m back…
Jussi,
I take your word for it that you really did prefer drinking the coffee over not drinking it. Indeed, there’s evidence for this, namely, that you drank the cup of coffee. But what evidence is there that you did not also prefer not drinking it to drinking it. Indeed, I bet that there was some evidence to the contrary. That is,
weren’t there times when you thought: “What I’m doing? I’d rather not be drinking this awful stuff.”
So I’m suggesting that just as you had incompatible desires, you may very well have had incompatible preferences as well, which is exactly what the proposed analysis implies given (1) and (2). And I don’t see that there’s any convincing evidence that you lacked this other preference, that is, you haven’t provided any convincing evidence that (4) is true. That you say (4) is true doesn’t convince me, because sometiems we have preferences that we’re not consciously aware of.
Doug,
OK, so your analysis of “desirable” is only supposed to apply to outcome/action-types, rather than outcome/action-tokens? Because it seems to me that in a concrete situation, NONE of the options available to me would be desirable, full stop, though obviously some might be more desirable than others and more preferable than others.
I guess I’m basically making a point made above: “desirable”, to my ear, sounds like it applies to outcomes/actions above a certain threshhold, whereas in a given situation perhaps nothing is above that threshhold.
Doug,
about this:
‘So I’m suggesting that just as you had incompatible desires, you may very well have had incompatible preferences as well, which is exactly what the proposed analysis implies given (1) and (2). And I don’t see that there’s any convincing evidence that you lacked this other preference, that is, you haven’t provided any convincing evidence that (4) is true. That you say (4) is true doesn’t convince me, because sometiems we have preferences that we’re not consciously aware of.’
Notice that you start with a weak modality ‘may have the incompatible preference’. I grant you that I may have that. However, the analysis requires that I always necessarily have also that preference. Even in this particular case, in introspection I did not find the opposite preference from my psychology. If we want to attribute that state to me against my first-personal authority we would need rather strong reasons. I can think of two. First, is that some kind of naive emotivism is true. The talk of preferences is just direct expression of desires. Second, I’m necessarily rational in so far as my desires and preferences match by some infallible process. I find neither of these explanations appealing. Therefore, the grounds for attributing the preference for me against my introspection seems rather thin.
Hi Kris,
I don’t think that one can have inconsistent preferences. That is, I don’t think it can be true that Mary prefers p to not-p, and that at the same time Mary prefers not-p to p. Of course, there will be cases where Mary does not prefer p to not-p, nor does she prefer not-p to p–either because she is indifferent, or hasn’t made up her mind, or whatever.
But maybe I am overlooking something. Can you think of a realistic counterexample?
It may be easier to think of examples of inconsistent preferences with more than two things:
I might at the same time prefer a ham sandwich to a tuna sandwich, a tuna sandwich to a plate of spaghetti, and a plate of spaghetti to a ham sandwich.
This is inconsistent if we require consistent preferences to be transitive and anti-symmetric (p over ~p and vice versa only requires anti-symmetry.) But transitivity seems like a good requirement to me for consistency of preferences.
On another note… This may be just changing the subject, but if the original question was about connections between preferences and desires, how about this:
A desires P more than Q iff A prefers P to Q
This requires a graded concept of desire, but, as mentioned above, that’s a desirable thing anyway.
Jake, I agree that transitivity failure is more plausible than asymmetry failure. Transitivity in your example would add not only a violation of asymmetry but a violation of reflexivity, which to me is impossible — incomprehensible.
I’m interested to hear (read) if anyone disagrees with your biconditional. Mark thinks it might be an analysis of ‘preferable’ — I don’t understand why, exactly — but in any case it does seem completely compelling that it’s a conceptual equivalence.
I disagree still with the biconditional. The unwilling drug addict desires more the get another hit than not but prefers not have one over going down that road again. Failure of the biconditional in both directions.
S wants that P =df. S prefers P to not-P
Hum. Suppose I want to drink a cup of tea. On this account that means I prefer to drink a cup of tea to not drinking one. But now what does that mean?
It doesn’t mean that I would prefer to find myself in any of the possible worlds where I have a cup of tea (assume relevant time indices in place) to any where I don’t. For that would mean I prefer the world where I have my tea then die a horrible and immediate death to the world where I miss out on my tea then win the lottery.
SO maybe there is a ceteris paribus clause missing: I.e. for any pair of words in one of which I have my tea and in the other I lack it but they are otherwise the same, I prefer the one where I have the tea to the one where I don’t. For these pairs include the pair at both of which tea tastes vile.
Or consider the pair of world at both of which tea is deadly poisonous. This analysis might seem to have the consequences that I should regard my having the tea as making no difference to whether I live or die. (If the worlds are “otherwise the same” I either survive in both – albeit miraculously in one – or die in both.) That would be daft of course – how would we handle the thisworldly case where I want to drink the tea for its health benefits or prudently don’t want to drink cyanide. So perhaps nomological proximity of some sort has to be factored in to the similarity metric.
Maybe proximity is the place to start. My desire to drink tea means I prefer the closest-to actual worlds where I drink the tea to the closest-to-actual where I don’t. But that is tricky too. If my desire is based on false information (suppose tea is very bad for me and I believe it is good for me), do we take the closest worlds relative to the facts as they are or the facts as I believe them to be? Or what if my desire is a gamble based on uncertainty…?
Perhaps it means that an ideal world – one I prefer to any other – is a tea-drinking-by-me-now world. But that’s implausible. Maybe while I want tea I’d sooner have champagne if it were available. Or I only want tea as a means to help me concentrate on a tedious task that in an ideal world I wouldn’t need to do at all. (Compare the desire that some bad person be punished where in an ideal world that person would not be bad.)
Someone must have written properly about these headaches. I air them just to cast a little doubt on any notion that maybe the notion of preference has a perspicuity that of desire lacks.
I think Dave Sobel gave up on his counterexample too quickly.
Think about it decision-theoretically. We have two propositions:
(B) You are dipped into a vat of bugs.
(F) You are dipped into a vat of faeces.
Suppose you prefer B to F, and that you’re rational in the sense of decision theory. So we have U(B) > U(F), where U is an expectational utility function representing your preferences. Suppose further that you’re certain that you’ll be dipped in a vat of bugs iff you’re not dipped in a vat of faeces. That is, we have Cr(B iff ~F) = 1, where Cr is a credence function representing your degrees of belief. It follows that U(F) = U(~B), and hence that U(B) > U(~B). That is, you prefer B to ~B. Yet, as Dave points out, it seems natural to say you don’t want to be dipped in a vat of bugs. You don’t want to be dipped in either vat; you prefer the vat of bugs only because it’s “the lesser of two evils”.
This seems to show, not only that it’s possible to prefer P to ~P without wanting P, but also that it’s rational to do so (at least in one sense of rational).