Mary Coleman

Mary Coleman (Bard College), "Holistic Directions of Fit and Smith's Teleological Argument," with commentary by Michael Smith (Princeton).  The paper, commentary, and reply can be found here.

Comments

[uriah]

Thank you both for an amiable and marvelously clear exchange. The way I see it, the endpoint of the exchange is a disagreement on the following proposition:

(T) If compresent belief-like state B and desire-like state D are modally separable, then their compresence does not constitute a unitary psychological state.

Coleman says "nay," Smith says "aye." Coleman claims that her DOF Holism entails the falsity of (T), whereas Smith claims that it does not. If (T) is false, then Smith’s argument for the impossibility of besires, and therefore against at least one way of combining internalism and cognitivism in moral psychology, doesn’t work.

What I want to suggest is a separate way of rejecting (T), one that does not rely on appeal to DOF Holism.

Right now I’m sitting across from the St. Kilda pier near Melboune, glad that the weather is nice. Being glad that the weather is nice is a mental state that involves believing that the weather is nice and desiring that the weather be nice. Moreover, my current belief and desire that the weather is/be nice are modally separable. Yet there is a strong intuition that my gladness is a single unitary psychological state. What accounts for this intuition, it seems to me, is the fact that there is what we might call a "unitary phenomenology" of being glad that the weather is nice.

If we accept this description of the case, we might be tempted to reject (T) in favor of the following:

(T*) If compresent belief-like state B and desire-like state D are modally separable, and there is no unitary phenomenology associated with B+D, then their compresence does not constitute a unitary psychological state.

Note that someone who accepts this may still reject the relevant version of internalist cognitivism, namely by denying that there is a unitary phenomenology associated with moral judgments.

Whether there is such a unitary phenomenology is a question in what Horgan and Timmons call Moral Phenomenology. Given their recent "cognitivist expressivism," I think Horgan and Timmons would claim that there is indeed a unitary phenomenology associated with moral judgments. I tend to agree, but could be swayed.

[Jussi Suikkanen]

I wonder if the 'holistic fix' creates more problems than appears to solve, and thus I'm not certain how much work it can be put to do.

According to Coleman, mental states can have two kinds of dispositions. There are first the mind-to-world direction-of-fit constituting dispositions - tendencies to go out of existence when faced with incompatible state of the world. There are also the world-to-mind direction-of-fit constituting dispositions - tendencies not to go out of existence when faced with incompatible state of the world and tendencies to bring about the relevant actions.

Now, the first problem is that all mental states seem to have both types of dispositions. Beliefs don't go out of existence when we face weak conflicting evidence and desires do not always bring about actions and may go out of existence in certain circumstances when the world is not in the way the desire presents them as being.

On the first look then all mental states have both belief constituting and desire constituting dispositions and thus all mental states are besires (have both belief-like and desire-like features). Where once were useful differences, there are none now.

To avoid this consequence, one can say what both Coleman and Smith say - beliefs have mostly mind-to-world state constituting dispositions and desires mostly world-to-mind state constituting dispositions. And, if there are besires maybe they fit somewhere inbetween. But where do we draw the lines? More than 66.666...% of mind-to-world state dispositions and you have a belief, less than 33.333...% and you have a desire, between 33.333...% and 66.666...% and you have a besire? Why just those numbers and not 75%, 25%, and between 25% and 75%? Have some dispositions more weight than others?

If we solve the number problem, then we face one that is even more difficult. How do we identify such dispositions of mental states? 'No entity without identity' as the saying goes. What are the truth-conditions for the claims of particular mind-to-world DoF state constituting disposition? Does sizes of puddles matter or the loudness of the child's groaning? One problem is that any actual event, where a mental state does not react to some evidential state in the world, can either be seen as constituting a distinct disposition not to react to the relevant state (desire-like) or as an instance of a failure 'to fire' of a disposition to react usually to that type of states.

Without answers to such questions it's hard to assess what the holistic suggestion is. And, without a precise suggestion it's hard to say anything about the existence of besires. There is a temptation to think that these problems are close to a reductio against the dispositional view of mental states. One is tempted to ask why can't we use semantic terms like truth (where beliefs aim) or normative notions like ought (beliefs *ought* to go out of existence) to identify beliefs and desires.

[Anne Jacobson]

May I comment quickly about the following example and the claims made about it? It is from Uriah:

"Being glad that the weather is nice is a mental state that involves believing that the weather is nice and desiring that the weather be nice. Moreover, my current belief and desire that the weather is/be nice are modally separable. Yet there is a strong intuition that my gladness is a single unitary psychological state. What accounts for this intuition, it seems to me, is the fact that there is what we might call a "unitary phenomenology" of being glad that the weather is nice."

The claim above brought back my concern with the lack of attention to empirical studies that the original discussion of belief and desire had. For example, there is considerable evidence that we in fact have some dispositions for retaining beliefs and so filtering evidence in order to sustain them. In addition, the mind really seems to behave as though there is some modularity at least, and one result is that beliefs can get more or less isolated from counter-evidence.

In the present example, it seems difficult to know how to assess a claim about unitary phenomenology. There are scores of things about which a casual take on the phenomenology neglects a complexity. What we count as taste is really the product of two sensory components, for example. In the present case, one might have that sense of gladness and find the belief, but not the feeling, remains during a discussion of global warming and skin cancer rates in Australia.

[Mary Coleman]

Thanks to everyone who has written in about my paper. I just wanted to weigh in with a few quick replies.

To Uriah: Like Smith, I think that it would be a mistake to give an account of desires (solely or primarily) in terms of their phenomenology. Such an account does not allow the Humean to put his best foot forward, since it's clear that we sometimes do what we believe we have most reason to do as *opposed* to what we have the strongest felt desire to do. For that reason, I would rather not rely on phenomenology to settle the issue between Smith and I. (However, there might be a way to take up your suggestion without also taking up a phenomenological account of desires. That would be interesting.)

To Jussi: Several people have written about the relative merits of dispositional accounts vs. normative accounts of directions of fit, so you are raising an important and complex issue. I have obviously not offered an account of the truth conditions of statements about dispositions in this short paper, and I don't have anything particularly novel to say about that. I would simply point in the direction of whatever your favorite account is of the truth conditions of counterfactuals. Furthermore, it is not at all obvious that a satisfactory *normative* account can be given of the direction of fit of desires. (For example, it isn't always the case that the world ought to be changed so as to fit my desire. Suppose I am sitting in a particularly frustrating faculty meeting, and what I most want to do is to curse at the top of my lungs in response to the idiot who's talking.)

As for your worry about how to differentiate desires, besires, and beliefs on the holistic view, I agree with you that the boundaries between these states will not be clear, but I think that that is a virtue of the account, not a vice. I think that the boundaries between them really are not clear. (Peirce may have overemphasized the connection between beliefs and action, but not by much.)

To Anne: You write, "we in fact have some dispositions for retaining beliefs and so filtering evidence in order to sustain them." This is quite intriguing. Although I have followed Smith and offered a dispositional account of beliefs and desires, I have done this partly for polemical reasons, i.e., partly in order to respond to his particular argument against the possibility of besires. That said, I actually think that the set of dispositions that are constitutive of (ordinary, non-practical) beliefs are constitutive ideals of what it is to have beliefs. To count as having beliefs at all, you have to have most of the relevant dispositions in the relevant situations. What you write suggests that we are actually farther from the ideal than one might have thought.

This brings me back to Jussi's comment. Although I am skeptical about the possibility of a normative account of the direction of fit of desires, I actually favor a kind of combination account--normative/dispositional--of both ordinary beliefs and besires is probably right. (Again, I give a straight, dispositional account in the paper for polemical reasons.) To have beliefs is not simply to have mental states that, e.g., ought to go out of existence in certain circumstances. Rather, I would say: one does not have beliefs unless the candidate mental states actually do tend to go out of existence in those circumstances enough of the time.

[Michael Smith]

This is really just a footnote to Mary's post picking up on what she said about Jussi's and Anne's comments. My own view, as is I hope plain from my comment on Mary's paper and other things I've written, is that there are really three kinds of psychological state all of which are best understood in terms of their function, and all of whose functional roles need to be defined simultaneously in the manner of analytic functionalism: believing, desiring, and being rational. (Here I side with Hempel in 'Rational Action' and against Davidson in 'Hempel on Explaining Action.') No account of the nature of any one or two of these psychological states can be given independently of the other(s). The idea, very roughtly, is that being perfectly rational is a matter of functioning--that is, having one's beliefs and desires related to each other and to perception and action--in perfect conformity to all of the norms of rationality. Of course, since none of us are perfectly rational it follows that the psychological state that plays a role in the explanation of our actions is defective form of this psychological state: our being rational to some limited extent. As I see things, then, the normative account that Jussi mentions, which is apparently the account that tempts Mary, isn't a competitor to the sort of dispositionalism I embrace. The normative account is rather a corollary of that kind of dispositionalism.

[Allen Coates]

Coleman suggests two ways for a rationalist to respond to the teleological argument:

(1) Practical beliefs have both directions of fit (i.e., are besires), and

(2) Practical beliefs have only mind-to-world direction of fit, but the desire that motivates can be wholly explained by the same rationalizing explanation that explains one’s having the practical belief.

Coleman favors (1), but I think something closer to (2) is the rationalist’s best bet.

Smith uses the weak will / modal separability argument to reject (1). If I understand it, the argument holds that a state is unitary only if the elements it comprises are modally inseparable. But this is in tension with the sort of holism Coleman proposes. According to it, a state can be a belief even if it lacks some of the dispositions that constitute beliefs, as long as it possesses most others. But this suggests that these dispositions are modally separable. The modal separability argument would therefore imply, absurdly, that beliefs aren’t unitary states. So if Smith accepts Coleman’s holism (as he should), he should reject the modal separability argument.

The modal inseparability criterion of unity seems too strong, but I do not have an alternative to offer. But for the sake of argument, I'll assume one is available, and that besires count as unified states.

So what is wrong with (1)? Call it the modal attachability problem. A belief may be paired with various different desires to form a besire (and a desire may be paired with a variety of different beliefs). These pairs may not always be rational. Smith, following Stocker, notes that we can desire what we believe is bad. If there are such things as besires, perhaps there could be a perverse besire: the desire to A paired with the belief that one shouldn’t A. Or perhaps there could be, say, an arachnophobic besire: the belief that there is a spider in front of me paired with a desire to scream and flee.

What these examples show is that the conditions under which the belief portion of a besire is rational and those under which the desire portion is rational can come apart: the sheer fact that a desire is paired with a rational belief does not guarantee that it is rational. Besires, then, are of little help in the rationalist’s project of getting motivation under rational control. Of course, a rationalist could reply that besires are helpful when the belief and the desire components share the same rationality conditions. But in that case, why not just go with (2)?

[uriah]

Thanks for responding, Mary. Note, however, that neither (T*) nor the thesis that moral judgments have a unitary beliefish-desirish phenomenology entails the claim that we can give an account of desires (solely or primarily) in terms of their phenomenology. I may be wrong, but the way I understand the dialectic, you and Smith are (or could be) in agreement that when X judges that she ought to phi, X has a belief and X has a desire - however we account for belief and desire. The disagreement is on whether the belief and desire X has form a "unitary psychological state." This is a question about how to individuate mental states rather than about how to account for desires (or for that matter beliefs). What (T*) commits to is the idea that phenomenological unity of the belief and the desire is a sufficient condition for the treating them as parts of a single state rather than two separate states.

I do appreciate, though, that phenomenology is scary and better avoided when possible. Like you say, there might be a different (non-phenomeological) way to explain the intuition that gladness that p is a unitary psychological state. At the end of the day, what matters is that gladness that p has (i) both thetic and telic direction of fit and (ii) modally separable components for each, and yet it intuitively counts as a single state. If gladness can be such, the thought is, maybe moral judgments can as well.

I suspect one could argue directly from modal separability to the falsity of internalist cognitivism without going through the issue of unitary statehood. After all, if the two components of moral judgment are modally separable, such that one part accounts for the objective pretension of the judgment and the other for its motivational oomph, then it is not the objective that is oomphy (if you will), and we can still ask whether the normative authority of a moral judgment is to be associated with its objective component or with its oomphy component. Complicated.

[Michael Smith]

This is a comment on Allen Coates's post. Apologies for its length!

The modal separability argument assumes that we can test the truth of the claim that having a certain bundle of dispositions is constitutive of having a particular psychological state by asking whether that state could exist and the bundle not exist. Call this the 'modal separability test.' If the answer is that it couldn't, then the constitutive claim stands. If the answer is that it could, then the constitutive claim falls: having the dispositions in the bundle is not constitutive of having the psychological state in question.

Against this Allen says: "But this is in tension with the sort of holism Coleman proposes. According to it, a state can be a belief even if it lacks some of the dispositions that constitute beliefs, as long as it possesses most others. But this suggests that these dispositions are modally separable. The modal separability argument would therefore imply, absurdly, that beliefs aren’t unitary states."

I am not persuaded. The problem, I think, is that Allen equivocates when he talks of "the dispositions that constitute beliefs." On the one hand there are the dispositions that constitute the agent's having that particular belief at that time. On the other hand, there are the dispositions constitutive of having that particular belief. Let me explain why it is important to make this distinction.

Consider the set of dispositions that a particular agent has at some particular time in virtue of which he counts as having some particular belief at that time. If we accept Coleman's holism, the proper conclusion to draw is that having the dispositions in that set is not constitutive of having that particular belief. This is the point Allen notes in his post. In other words, even though the dispositions constitute the agent's having that belief at that time, they are not constitutive of his having that belief. This is also the answer delivered by the modal separability test. So there is no tension between Coleman's holism and the modal separability test so far.

Next consider the very large set of dispositions, most of which are possessed by any agent who counts as having a particular belief. If we accept Coleman's holism then this very large set obviously isn't constitutive of having a particular belief either. For, once again, the belief may exist and that large set of dispositions not exist: all anyone is required to have is most of the members of this very large set. But this too is the answer delivered by the modal separability test. So there is still no tension. (Moreover, just for the record, note that, short of imagining a God-like figure, it is also unlikely that having all of the members of this set constitutes anyone's having that particular belief either.)

Rather, if we accept Coleman's holism, what's constitutive of having a particular belief is having some bundle of dispositions that comprises the majority of the large set just mentioned. This is a huge disjunction of bundles. But the modal separability test delivers this result too. For it isn't possible for someone to possess the belief in question and yet lack the dispositions in one or another of these bundles. The dispositions that constitute one agent's having the belief may be different from the dispositions that constitute another agent's having the belief, to be sure, but both agents count as having the same belief because they both have some bundle of dispositions that comprises the majority of the large set.

Does it follow that belief is or is not a unitary psychological state? Belief turns out to be a unitary psychological state, albeit a unitary psychological state whose principle of unity is stated in the holistic way Coleman proposes. Once we see that this is so we see that the unitary nature of belief is not in tension with the modal separability test.

[Allen Coates]

Thanks for responding to my post, Michael.

In my first post, I claimed that the modal separability argument is in tension with Coleman’s holism. In response, you write, “The modal separability argument assumes that we can test the truth of the claim that having a certain bundle of dispositions is constitutive of having a particular psychological state by asking whether that state could exist and the bundle not exist. Call this the 'modal separability test.'” You then show that the modal separability test is compatible with Coleman’s holism.

I agree that this test is compatible with her holism. But I don’t yet see how that test relates to the modal separability argument. As you explain it, the test is used to determine whether a bundle of dispositions is constitutive of having a given psychological state. But what the modal separability argument needs is a test to determine whether a given psychological state is unitary, and I don’t quite see how the test you describe can do that.

I can imagine supporters of besires arguing as follows. “Consider a given besire. Since the dispositions that constitute it are modally separable, one after another can be peeled off, so to speak. Smith’s modal separability test can tell us when it is no longer a besire, but is instead, say, merely a belief that p. The remaining dispositions are likewise modally separable, and can be peeled off one after another. Again, Smith’s modal separability test can tell us when it is no longer a belief. (Perhaps at this point it is merely the state of supposing or hypothesizing or guessing or accepting that p.) This, of course, doesn’t show that the belief isn’t a unitary state. And by the same token, we haven’t shown that the besire isn’t a unitary state.”

On the other hand, something like Uriah’s (T) would be a test of whether a state is unitary. Here’s a more generalized test:

(T**) A state constituted by a package of dispositions is unitary iff the dispositions in the package are modally inseparable.

This would be a modal separability test for unity, but given Coleman’s holism, beliefs would fail it as well as besires.

Of course, (T**) presumably isn’t the only way to use modal separability as a test for unity. And perhaps there is another test which beliefs would pass but besires would fail. So I’m not sure that my argument in my previous post is successful. But I’m not yet convinced that it isn’t.

[Antti Kauppinen]

I'm a little puzzled by Michael's suggestion that a normative account of direction of fit is a corollary of his kind of dispositionalism. Let's say that the basic motivation adopting a normative account is the possibility of systematically recalcitrant beliefs (leaving desires aside for the time being), that is, beliefs that won't go out of existence no matter what evidence there is against them (such as 'Saddam Hussein supported al-Queda'). So we say that beliefs are those psychological states that *ought* to go out of existence in the face of evidence that the world isn't as they represent it as being, as opposed to saying they're the ones that *do* tend to go out of existence, as Michael originally said. So, first, it seems to me that the normative fix makes the holistic fix unnecessary (though they are obviously compatible).

My main worry, however, concerns Michael's present position, particularly the role that being rational plays in it. Here's what would seem to me an obvious way to put the sort of normative dispositionalism he suggests in his intriguing comment here:

(Bel) A psychological state S(p) has a mind-to-world direction of fit iff, inter alia, it tends to go out of existence in the face of an appearance that not-p, insofar as the agent is rational
(Des) A psychological state S(p) has a world-to-mind direction of fit iff, inter alia, it tends to stay in existence in the face of an appearance that not-p and dispose the agent to bring it about that p, insofar as the agent is rational
(I'll bracket worries about 'appearance', perception, or whatever here, though I think they are very serious)

The addition of 'insofar as the agent is rational' makes the functional definitions normative in the relevant sense. Being rational, in turn, is for Smith a matter of conforming to norms of rationality. (Earlier, he has talked about having a disposition toward coherence.) But what are these norms of rationality? Suppose one of them is that one rationally should not believe something that doesn't seem to one to be the case (call this Adjusting to Appearances, or AA). (Again, AA is false without qualification, but let as assume for the sake of argument that one can be found.) Let's see how this would work. Imagine someone, x, who is in S(p). (Bel) combined with AA tells us that S(p) has a mind-to-world direction of fit iff it tends to go out of existence in the face of appearance of not-p if x is disposed to give up beliefs that are contrary to appearances. Or, shorter, S(p) is a belief if, among other things, it would go away in the face of evidence, given that x gives up beliefs (and only beliefs) in the face of evidence. (I'm not worried about circularity here – I take it that's inevitable if all the three functional roles are to be (inter)defined simultaneously.) We could also take another norm of rationality, say that one should not believe contradictory things (No Contradictions, or NC), and with some more complexity (since we'd have to involve a number of states) and suitable modification of (Bel), end up with the same result.

Now, suppose it is a norm of rationality that one rationally should do what one judges one ought to; call this Rational Internalism, or RI. RI seems plausible, and Smith has argued for it in detail. But for now, let's just see what follows from RI with respect to the rationalistic functionalism of (Bel) and (Des). Our question is: what direction of fit does the psychological state of morally judging I ought to do something have? Well, it doesn't tend to go out of existence if it appears to me that I'm not doing what I ought to, even if I conform to AA. Instead, insofar as I conform to the norm of rationality RI, it disposes me to bring it about that I do what I judge I ought to. In other words, in terms of the rationalistic functionalism of (Bel) and (Des), given that we assume that AA and RI are norms of rationality, it has a world-to-mind direction of fit; it meets both criteria of (Des). So is it a desire? Well, NC and surely other belief-specific norms of rationality that don't apply to desires seem to apply to it, so it has belief-like features as well. So, is it a besire? I must have gone wrong somewhere, for that would seem to follow from my reconstruction of Michael's remarks above. It would be nice to know where, and what would be the correct way to capture the notion that normative accounts of direction of fit aren't in competition with his dispositionalism.

Apologies for the length.

[Jussi Suikkanen]

Antti,

I'm probably missing something but I'm quite puzzled about this:

'Our question is: what direction of fit does the psychological state of morally judging I ought to do something have? Well, it doesn't tend to go out of existence if it appears to me that I'm not doing what I ought to, even if I conform to AA.'

In terms assessing which direction of fit the a mental state has, wouldn't the relevant question be how the state responds to the evidence that is imcompatible with the content of that state. If a it's a belief, then change the state, if a desire, then change the world. In this respect, that I'm not doing what I think I ought to do doesn't seem to be the relevant kind of 'thing' that could be incompatible with the judgment of what I ought to do. Only evidence that it's not what I ought to do would.

And certainly even though the state may bring me to do something, it won't bring me to do that this is what I ought to do. That would be analogical to the situation where my desire (that I take a shower) makes me to take a shower.

I'm not sure whether that makes sense at all. Good question! The way of figuring out directions of fit doesn't seem to extend to normative judgments.

[Michael Smith]

Thanks for yours, Allen.

I take it that everyone agrees that some psychological states may be constituted by other psychological states. It is hard to think of uncontroversial examples, and harder still to state examples with any precision, but some possible candidates would be: the instrumental desire to do x (this might be thought to be constituted by an intrinsic desire to do something and the belief that that can be done by doing x); hoping that p (this might be thought to be constituted by a desire that p and a belief that p is unlikely); and so I could go on (I'm sure you can come up with other, and perhaps better, examples yourself).

The challenge for the besirist is to convince us that besires aren't another example of such a psychological state. They must convince us besires aren't simply constituted by independent beliefs and desires. They must convince of this because if they are then the Humean can happily admit their existence. (No Humean takes issue with the existence of instrumental desires on the grounds that they are both belief-like and desire-like: for, they say, instrumental desires are belief-like and desire-like because they are constituted by a belief and a desire.) This is where a principle of unity is needed, and, as I understand it, a principle of unity was in effect proposed precisely to rule out this possibility by McDowell.

McDowell claimed that we mustn't think that besires--though of course he didn't and perhaps wouldn't use the word 'besires'--are constituted by independent beliefs and desires because virtuous people have a distinctive way of conceiving of the situations that they confront. To conceive of situations in this way, McDowell tells us, is to be in a cognitive state whose "possession entails a disposition of the possessor’s will" ('Are Moral Requirements Hypothetical Imperatives?' p.18). In other words, it is to have a besire.

McDowell argues for the claim that virtuous people have besires, not psychological states that are constituted by independent beliefs and desires, by way of an example. "[In].. urging behaviour one takes to be morally required, one finds oneself saying things like this: ‘You don’t know what it means that someone is shy and sensitive.’ Conveying what a circumstance means in this loaded sense, is getting someone to see it in the special way in which a virtuous person would see it. In the attempt to do so, one exploits contrivances similar to those one exploits in other areas where the task is to back up the injunction ‘See it like this’: helpful juxtapositions of cases, descriptions with carefully chosen terms and carefully placed emphasis, and the like... No such contrivances can be guaranteed success, in the sense that failure would show irrationality on the part of the audience. That, together with the importance of rhetorical skills to their successful deployment, sets them apart from the sorts of thing we typically regard as paradigms of argument. But these seem insufficient grounds for concluding that they are appeals to passion as opposed to reason: for concluding that ‘See it like this’ is really a covert invitation to feel, quite over and above one’s view of the facts, a desire which will combine with one’s belief to recommend acting in the appropriate way." ('Are Moral Requirements Hypothetical Imperatives?' pp.21-2)

According to McDowell, then, we must think that virtuous people have besires, and we must think of these besires as unitary psychological states, not as built up out of independent beliefs and desires, because if they were built up out of independent beliefs and desires then we could imagine someone who had the component belief--someone who had the very same conception of their circumstances as the virtuous person has of his--but who simply lacked the desire that the virtuous person has about what is to be done in that situation. But, says McDowell, we cannot imagine this. To see things as the virtuous person does--to besire what he besires--is to be moved in the way the virtuous person is.

What's crucial for establishing the existence of besires, then--that is, what's crucial for establishing that besires are unitary psychological states, not psychological states that are constituted by independent beliefs and desires--is establishing that there are some ways things are which are such that, though someone can besire that things are that way, noone can believe that they are that way. A failure of the will must constitute a failure of the understanding. The modal separability test enters the scene in the attempt to adjudicate debates over various alleged contents for besires.

This brings me to your post, for in your post you seem to be working with a conception of besires that is very different to this. You seem quite happy to admit that one person may have a besire that p and that someone else, someone who doesn't have the desire-like dispositions possessed by the person who besires that p, may have a belief that p. But the whole argument for the claim that there exist besires, where besires are unitary psychological states, not simply psychological states that are constituted by independent beliefs and desires, is that this is impossible. And the whole point of the modal separability argument is to challenge this alleged impossibility. The modal separability argument purports to prove, on a case by case basis, that there are no "conceptions of situations" that meet McDowell's criteria for being the conceptions of situations possessed by those who have besires, but not by those who merely have beliefs.

[Mary Coleman]

Thanks to everyone for the interesting discussion that's been going on for the past few days. Here are a few thoughts in response.

Michael, you remind us that, on your view, beliefs, desires, and being rational are three different functional states. You also claim that "no account of the nature of any one or two of these psychological states can be given independently of the other(s)." I have always been puzzled by this view of the relationship between beliefs and rationality. It seems to have the strange implication that someone could have beliefs and, yet, be entirely irrational with respect to them, i.e., not form or revise any of them in accordance with the relevant norms of rationality. In fact, this implication looks worse than strange; perhaps it's downright inconceivable. But I'm not sure how you can avoid it if beliefs and being rational are really *different* states. (Another way to put my question is to ask what it means to say that (1) you can't define what it is to have a belief without using the notion of being rational, but to insist that (2) having a belief and being rational are two different mental states.) On my view, by contrast, beliefs are essentially rational. One does not have beliefs at all unless (and here is my holism again) one forms and revises enough of them in accordance with the relevant norms.

Allen, re: the arachnophobic besire . The reason why I wrote about practical beliefs in the paper rather than besires was to emphasize the importance of *these* beliefs in particular in the rationalist conception of motivation. This is not something that I talked about in the paper, but, on my view, it is a constitutive norm of having practical beliefs that (1) one form and revise them in accordance with (something like) what Uriah calls Adjusting to Appearances (in particular, appearances regarding what one has most reason to do); it is also a constitutive norm of having practical beliefs that (2) one act in accordance with them. (To go even further beyond the bounds of the paper, I actually think--for reasons I won't go into here--that the state which has both directions of fit is not actually a belief but rather a state of having decided to do something. One constitutive norm of making decisions says that one ought to act on one's decisions; another constitutive norm says that one ought to decide to do either what one has most reason to do or--in those cases in which there is no single thing that one has most reason to do--one ought to decide to do one of the things that one has sufficient reason to do. I discuss this view in an unpublished work in progress called "Motivation by Decision.")

This is to Allen also: you suggest that the rationalist ought to take the second of the two options left open by the teleological argument. I argue in the paper that a desire which is amenable to rationalizing explanation actually has both directions of fit. So this second option is not much different from the first, but, rather, the same option described in a different way. In the first option, a belief is said to have two directions of fit; in the second, a desire is said to have two directions of fit.

Uriah, you write, "if the two components of moral judgment are modally separable. . .we can still ask whether the normative authority of a moral judgment is to be associated with its objective component or with its oomphy component." This is an interesting suggestion. However, if we think about a parallel suggestion regarding ordinary beliefs, we can see that there is something odd about it. There is a *sense* in which ordinary beliefs--i.e., beliefs that are not about what one has most reason to do (or ought morally to do)--have both an objective component and an oomphy component. The oomphy component of these beliefs doesn't "push" the believer toward acting but, rather, towards forming and revising other beliefs. Suppose that for the last three months my son has been scheduled to come home from Iraq on May 5. So, for these three months, I have believed that he will come home that day. But I get a visit from the Notification Officer on May 2 telling me that he was killed. And suppose that I believe the officer and I really do "take in" what he said; I truly believe that my son was killed. But, irrationally (out of grief, say), I also keep believing that he will come home on May 5. (I discussed an example somewhat like this in the paper.) In that case, the oomphy component of the belief that my son was killed has failed to "fire" or has been blocked by something. Nonetheless, I actually don't think that it's illuminating to think of beliefs as having two separate components: an objective component and an oomphy component. If beliefs are unitary states--and I think we all agree that they are--then they *really* are unitary states, even if there is more than one constitutive norm (or constitutive disposition) associated with them. And the same goes for practical beliefs (or for decisions). Or so it seems to me.

[Jussi Suikkanen]

Mary,

I wonder if Michael means that even though beliefs, desires and being rational are all separate functional states and psychological states, still only beliefs and desires count as individual, atomic mental states of the agent whereas being rational is a complex state of the agent or her psychology consisting of coherence and unifiedness of the beliefs and desires. It is these atomic states in certain relations to one another which can constitute the state of being rational. If something like this was right, then it can be the case that these three states are conceptually interdefinable but one can have beliefs and desires without being in the rational state.

[Allen Coates]

Mary, I agree that both rationalist options are, in effect, the same in the following sense: if the argument of your paper is correct, then whichever of the two options you pick, you end up with the same results: rationalism and practical beliefs. But the argumentative route to these results is different in each of the two options. The first, if I understand it, argues from the existence of besires to the truth of rationalism. The second argues in reverse order, from rationalism to the existence of besires. I think the second argument stands a much better chance of success than the first.

The problem is that it would be circular to use rationalist norms to explain what constitutes practical beliefs, then use practical beliefs to defend rationalism. So in order to take the first option and argue from practical beliefs to rationalism, you need a non-normative account of practical beliefs.

But in that case, I don't see how you can rule out arachnophobic and other irrational besires. What these besires show is that the desire-like dispositions of the besire don't inherit the rationality of the belief-like dispositions with which they're united. Of course, rational desire-like dispositions could be united with rational belief-like dispositions. But even then, the former don't inherit their rationality from the latter. In short, the psychological connections between belief and motivation won't establish normative connections. So a non-normative account of practical beliefs won't establish rationalism.

On the other hand, I don't have any objections to the second option.

[Mary Coleman]

Allen,
As I say in the conclusion of the paper, my holistic objection to Smith's argument is not meant to be an argument for rationalism. I have not given (nor did I intend to be giving) an argument for rationalism in this paper. I think rationalism is true, and I argue for it in other places. (e.g., the work in progress I mentioned in another post, "Motivation by Decision"). But in this paper I intended only to show that Smith's teleological argument does not succeed, i.e., he does not succeed in showing that practical beliefs could not possibly have both directions of fit. As I say in the conclusion of the paper, it's a further question whether they actually *do* have both directions of fit. I think that they do--or rather, that that the state of having decided to do something has both directions of fit--but I haven't argued for that view in the paper.

[Allen Coates]

My mistake, Mary. Thanks for the interesting and insightful paper!

[uriah]

That makes sense, Mary.

[Mary Coleman]

Jussi,

This is in response to your post on May 5 suggesting what Michael might mean about the relationship between having beliefs, having desires and being rational. I can see, given what you propose, it follows that what it is to be rational could not be defined except in terms of having beliefs and desires in certain relations to each other. However, I don't see how it would follow that what it is to have beliefs and desires could not be defined without reference to what it is to be rational. Or, to put it another way, I still don't see why it would be impossible for someone to have beliefs and desires that were entirely irrational. Help!

[Jussi Suikkanen]

I guess the answer could be that the beliefs and desires are identified by the functional roles which the states would counterfactually have in the psychology of a rational agent (that it, what would happen to them on the face of conflicting evidence were the agent rational). Still, even though the states are thus identified they can exist in the psychology of a non-rational agent where they do not correctly relate and react to one another. At least that's how I understand Smith who is probably a better authority on himself.

[Antti Kauppinen]

Jussi, here's an overdue response to the point you made a while earlier. You suggest that the relevant question is "how the state responds to the evidence that is imcompatible with the content of that state". This is a little unclear - presumably you mean that question is how the agent responds to evidence that the world doesn't match the content of the state. When we're trying to determine the direction of fit characteristic of a psychological state with the content that p, we obviously can't assume it has a particular direction of fit. So we use 'S' as a placeholder for whatever kind is in question. Now, I took it that the following states share the same content: I believe that I will visit my sick friend at t, I desire to visit my sick friend at t, and I judge I ought to visit (or I judge it right to visit) my sick friend at t. That content, obviously, would be the proposition that I visit my sick friend at t; call it 'p'. Now, it seems that I can adopt various attitudes toward p, one of them being that I ought to bring it about. (Does it make sense to see the following as analogous: obtains(p), goal(p), required(p)?) What is the direction of fit of this complex state? Well, if I believe that not-p (it is not the case that I (will) visit my friend at t) and I'm rational, then it seems I will be disposed to make it the case that p, rather than give up the ought-judgment. If the functional role of a state is defined in terms of what the agent would do if she were rational, then this state would have (at least) a world-to-mind direction of fit. (In other words: perception(not-p) is rationally incompatible with obtains(p), but not with goal(p) or required(p).)

On reflection, I suppose Michael would deny that the three states I mention above have the same propositional content. That is, when I judge I ought to do something, I have a belief about what my ideally rational, informed and maximally coherent self would advise me to do. This is not an attitude toward p, but toward another proposition about myself, say q. This would avoid the question I raised. Maybe that's the right way to go. But I don't think it's far-fetched to think that an ought-judgment and a desire, say, can share a content. Maybe this is a step that's implicit in how expressivists think about the issue.

I hope this makes some more sense.

[Jussi Suikkanen]

That makes more sense, thanks. What you said about what Michael would deny is pretty much what I had in mind. One way of putting this would be that in the case above p (visit sick friend at t) is the content of what I maybe ought to do, i.e., it could be that I ought(p). But if that's the case then the content of my judgment about this matter must be something different from p, namely 'ought(p)'. My state could thus be formulated as S(ought(p)). You are right that expressivists need to deny that there is such a content and assume that it is merely p. Hereabouts people would think that there really is such a sui generis content and it's about the 'oughts' in the world. That's kind of interesting that you might have to settle the content issue before you get to ask which direction of fit the state has.