Monday, July 24, 2006

No Contradictory Beliefs?

I accept a dispositional account of belief. That is, I think of believing as being disposed to act and cogitate and feel in certain ways under certain conditions. An alternative approach holds that to believe is to have some sentence-like representation stored in one's mind. (See my forthcoming Stanford Encyclopedia entry on belief and my Phenomenal, Dispositional Account of Belief for more on this.)

One point of contrast between these two approaches: On a representational warehouse approach to belief, it seems perfectly natural to suppose that sometimes we have baldly contractory beliefs. One might believe both P and not-P simultaneously; one might believe, say, both that God exists and God doesn't exist, or that lighting the match is safe and that lighting the match is not safe. All that's required is that one has the sentences or representations of both P and not-P stored (in the right way) in one's mind.

Dispositional approaches generally don't countenance baldly contradictory beliefs. The dispositions to act in accord with one belief will tend to flatly contradict the dispositions to act in accord with its negation. One can't, for example, simultaneously be disposed both consistently to act as though God exists and consistently to act as though he does not exist. At most, one will have a muddled, mixed-up, in-betweenish set of dispositions, such that it's not quite right to say that you believe that P and not quite right to say you don't.

Is it ever compelling to suppose that a person genuinely has simultaneous and baldly contradictory beliefs?

Consider the self-deceived person who sometimes and in some respects acts as though she perfectly well knows that her son is a drug dealer and who at other times and in other respects perfectly sincerely denies that he is. We might say part of her believes it and part of her doesn't -- but that's merely metaphorical. She doesn't literally have different compartments in her mind, with divergent judgments, alternating control over her behavior, does she? Surely it's at least as plausible to say that she's in a confused state somewhere between believing it and failing to do so -- perhaps somewhat like H.H. Price's "half belief" (compare the case of Geraldine in my Phenomenal, Dispositional Account of Belief).

Or take the man who lights the match to see in the dark, momentarily disregarding the fact that there's a gas leak -- an example sometimes cited by philosophers as a case of baldly contradictory belief. Is it really compelling to say that he both believes that lighting the match is safe and that it isn't?

Can you think of a better case? (If the "paradox of the preface" comes to mind, I'll deal with that in a future post.)


  1. I think it might be reasonable to say that people don't have contradictory beliefs at a given moment, but there are other situations where that seems to be the most natural conclusion. For example in one circumstance a person may assert P, and in another not-P, just not at the same time. Also for the sake of argument here I am assuming that they flip back and forth depending on the circumstances, not that one of the circumstances caused them to change their mind permanently. I agree that we could explain this with a dispositional model, such that circumstance A + disposition D yields P, and that B + D yields not-P. However it seems that there is something missing from this account, specifically that to assert P is to believe that it true period, not just true at the moment (assuming that P is a objectively fixed claim, like “God exists”, not like “it is hot outside”). So under the dispositional model we seem to be saying that in neither case the person really believed P or not-P, since belief implies that one thinks P or not-P is objectively true, i.e. all the time. This might be consistent but it seems to do away with belief in some cases, and thus it seems slightly simpler and more intuitive (to me) to say the they believe both P and not-P ,and that it is simply one or the other that gets expressed at a given time. This can still be reconciled with a dispositional account, but in this case there are two dispositions, one for P and one for not-P, that coexist in the mind.

    Another example of conflicting beliefs is where a person says one thing (and let us allow that they consciously believe it), but then acts in a way that seems to assert the opposite belief, such as a person who swears to you that they aren’t afraid of spiders, but then jumps away when they see them. Again the most natural explanation seems to believe that they consciously believe P, but unconsciously believe not-P, two beliefs in one mind.

  2. There is also a more trivial case where the person doesn't realize the two ideas are contradictory, say complicated theorem A and complicated theorem ~A. However, because of they way they are presented, the person thinks that they both have valid proofs, and doesn't realize one is the negation of the other. Thus they may act and think A and ~A simply because they don't realize that they are contradictory.

    From this and my previous comment I worry that doing away with contradictory belief will result in doing away with the concept of belief as it is commonly understood altogether. For example given any idea X there is probably some obscure logical consequence Y. Thus people who believe X may be persuaded to believe and act as if ~Y, simply because they don't realize that Y is implied by X (or don't believe that it is a consequence). I know, from this post and earlier ones, that you would be inclined to say that they only partially believe X. This conclusion leads to my concern that to accept the idea of partial belief as a substitute for the possibility of contradictory belief leads us to conclude the people believe most things only partially, which seems to undermine what we mean by the statement "I believe X", and to deny that my experience of really believing X 100% is accurate; that I only thought I believed X, but that I was wrong. However isn't a belief a reflection of a certain state of consciousness, as well as a disposition? How can we as outside observers claim that someone doesn't really have a belief, that they only think they do. It seems like denying that someone is really in pain, or is really conscious, and claiming instead that they only think that they are.

  3. Hi Eric,

    I don't see why the dispositional account is supposed to make contradictory beliefs seem less plausible, but I suppose a lot of this hinges on what "acting as if P" is supposed to mean. If it means "acting in ways only possible when P is true" then acting as if P and acting as if not P simultaneously would be impossible. But certainly that can't be the kind of action crucial to a dispositional account of belief since it would rule out the possiblity of false belief.

    So maybe "acting as if P" means "acting in ways consitent with P being true". But then acting as if P and acting as if not P should be a piece of cake. (Pizza lovers who can't wait for their pies to cool before snarfing but say "ouch" between bites may very well be acting as if the pie is too hot to eat and not too hot to eat simultaneously.)

    I suppose, then, that "acting as if P" is supposed to be closer, while not identical, to the first of the two extremes I detailed above. But even then, it wouldn't suffice to make contradictory belief impossible, would it?



  4. I don't mean to post-mob you, but you have really gotten me thinking about belief. I think that if we said that beliefs were unconscious dispositions that gave rise to conscious thoughts such as "X is true" some of my objections fall away. This conscious thought "X is true" is not properly speaking a belief at all but an expression of a belief. A belief expression probably contains the "feeling" that it is a constant and universal truth, but that is not necessarily part of the unconscious disposition. Then your no contradictory beliefs hypothesis might again become the simplest solution, now all we have to account for are the unconscious dispositions, and since they are unconscious there is no reason not to "sum" them into a single disposition. Well, this doesn't solve all my questions, but certainly some of them.

  5. Hi Eric,

    I'm a little confused as to how someone who holds a representational theory of belief has to admit contradictory beliefs. Surely there has to be more than just a condition of representability for a state to be a belief (even on their account)?

    There seem to be lots of different suggestions for a regulative norm for belief. For instance, Velleman's: a belief is correct iff it is true. Adler's being (quite strong): S believes that p iff S has sufficient reasons for p. From the perspective of either of these normative accounts is not the possibility of contradictory belief is ruled out.

    For instance, taking Adler's proposal, S can only believe that p if he has sufficient evidence that p. What would count as sufficient evidence for a true contraditiction? If there is sufficient evidence for one side of contradiction the other side,a dn their conjunction, is ruled out as something that one can believe.
    So for S to believe that p, where p contains a contradiciton, entails that either S doesn't really believe that p or that S has deceived himself about the warrant for p (say that it actually doesn't contain a contradiction). When such regulative norms are introduced as necessary conditions for belief the possibility of believing contradictions passes.

    What do you think? I'm not familiar with the representational accounts you refer to. Do they refer to regulative norms for belief or is it meant to be an account 'beneath' such norms?
    Best wishes,

  6. Thanks for all the comments, folks!

    Peter: You say in your third post what I was getting all primed to say in response to your first two! I'd prefer to think of belief as a long-term, dispositional state. When you suddenly have the thought, "P!" or "Not P!" -- those I'd call "judgments". An important part of the dispositional structure relevant to the belief that P is the disposition to reach occurrent judgments of the form P. As you suggest, putting weight on the belief vs. judgment/thought distinction can obviate some of your concerns -- though probably not all of them!

    Pete: Consider these examples: Central to the dispositional structure relevant to the belief that God exists is the disposition to express verbal assent to the claim "God exists" and to express dissent from "God does not exist". Central to the dispositional structure relevant to the belief that God does not exist are the opposite dispositions. Or: The believer will (stereotypically) be disposed to reflect on the fact that God is watching him; the atheist will be disposed not to so reflect. It's not logically possible simultaneously to possess these contradictory dispositions (at least if they're bald and unqualified).

    I'm not sure how exactly these examples stand with respect to the distinction you note -- I don't think I quite want either of the two alternatives. Maybe your concern arises in part because it's unclear how to flesh out what I call the "dispositional stereotype" for believing that God exists (or for any other belief). I think the necessary flesh isn't just a matter of consistency with P or inconsistency with its negation but involves contingent facts of human psychology. (But even in my published work on belief, I confess, I leave this aspect of the approach underdeveloped.)

    (I have an uncomfortable feeling here, though, that I'm not getting to the root of your concern.)

    Jack: I don't mean to suggest that a representationalist has to accept the possibility of contradictory belief. Perhaps having P stored in the right way makes it logically or nomologically impossible also simultaneously to have not-P stored in the right way. (But if so, as story needs to be told about why.)

    Regarding Adler in particular: I'll have to refresh myself on the details, but here's perhaps the core issue. Can one regard oneself as having adequate evidence for P and also simultaneously regard oneself as having adequate evidence for not-P? It depends on what's involved in "regarding oneself as having adequate evidence". If all that's required is the disposition to think "yes there's adequate evidence for that" when asked about P and also the disposition to think "yes there's adequate evidence for that" when asked about not-P -- well, those dispositions are simultaneously compossible. But if "regarding oneself as having adequate evidence" is more of an occurrent thing, or if it involves a broader dispositional base, then having the two contradictory beliefs might not be simultaneously compossible.

    The archetypal representationalist about belief, to my way of thinking, is Jerry Fodor: To believe that P is to have a representation with the content P inscribed in one's "belief box".

  7. Eric, what about the case of the committed dialetheist who proudly and frequently announces, as dialetheists are wont to do, both that the Liar sentence is true and that the Liar sentence is false. There's a pair of behavioural dispositions right off the bat. Suppose also that he's got a host of surrounding dispositions to discourse, which lead him under the right circumstances to produce detailed explanations of his justifications for each of his two contradictory belief, and also that he's disposed to accept anything that follows in a suitable substructural logic from either or both of the beliefs. What more could one want on a dispositionalist account in order to attribute both sides of the contradictory coin? With something like the Liar, you're going to be hard-pressed to find much in the way of dispositions to action beyond these sorts of acts of theoretical discussion.

  8. That's a very clever example, Bargaining Sheep -- thanks!

    I think I have to allow that some abstract beliefs mainly involve verbal dispositions. And the dialetheist has good reason not to have (or manifest?) the otherwise stereotypical disposition to deny the negation of her belief in a case like the liar.

    I will have to ponder the case further!

  9. Hi Eric,

    Let me try a different angle: I'm not quite seeing what it is about a dispositonal analysis of belief that would, in virtue of being a dispositional analysis, rule out the possibility of contradictory beliefs without tacitly assuming something about beliefs, namely that they cannot be contradictory. So, when I try to keep an open mind about what the dipsositions in question might be, it seems open to me that on a dispositonal analysis of belief contradictory beliefs are possible.

    I must admit I don't find your examples about God compelling, and I suspect that it hinges on what the actualization of the dispositions consists in (and not anything specifically about God). In the assent case, if assent is merely uttering "yes", then a machine can be built that is disposed to say both "yes" and "no" to the query "Does God exist?". Of course you and I can probably agree that such a machine may not have beliefs, but a serious question remains as to how much must be added to it until it does.

    My problem with your reflection case is slightly different from the assent case, and highlights my worry about circularity. Reflection on the fact that P may not be the same as belief that P, but may nonetheless be an assertoric propostional attitude (like thinking that P, suspecting that P, knowing that P, etc) and I worry about what the dispositional analysis of those are going to be such that contradictory beliefs wind up being ruled out by something other than fiat.

    Does this help clarify where I'm coming from?



    P.S. I am disposed to believe both that it is clarifying and that it is not.

  10. Pierre believes that London is not pretty. Mais il croit que Londres est jolie.

  11. Do we have contradictory beliefs? What comes to my mind is that the question has an obvious relation to the discussion about the proper objects of belief. If you think that what we believe are propositions understood in terms of possible worlds, contradictory beliefs will be legion! I wouldn't say that this approach is particularly appealing - but there certainly is a connection here to the issue of contradictory beliefs.

  12. Thanks for taking another pass at this, Pete. I'm afraid I'm still not entirely getting your points, though! Forgive me if I'm just being dense! If a machine is disposed sometimes to say "yes" to P and sometimes to say "no" (and has whatever other structures are necessary to be a believer), we can't simply say it is disposed to affirm P when asked. Or is your thought that we can? On your second point, maybe it would be helpful to say that I'd like to distinguish dipositional propositional attitudes from occurrent ones. Will it break the threat of circularity if I don't attempt to analyze "reflection on P" or "the occurrent judgment that P" in terms of further dispositions?

    Anonymous: Regarding Kripke's famous puzzle about Pierre, I'm inclined to think the whole force of the puzzle comes from Kripke's (and the reader's) reluctance to countenance the possibility that it's not quite right to say that Pierre believes that London is pretty and not quite right to say he fails to believe that (or disbelieves it). He's in an in-between state with a mixed set of dispositions. Grant this and the puzzle vanishes. After all, as Kripke suggests, there is something pretty unsatisfactory-seeming about saying that Pierre does fully and genuinely believe both that London is pretty and that it's not.

    Mike: I think you're absolutely right that one's sense of the structure of the objects of belief will inform (and be informed by) one's sense of the plausibility of ascribing contradictory beliefs. But maybe you could explain more to me why a sets-of-possible worlds analysis will lead to legions of contradictory beliefs? One way to go with an analysis of that sort is to say that we really only have one belief, defined by the set of worlds to which we think it epistemically possible the actual world belongs. Now, of course, a problem with this (related to Stalnaker's difficulties with the "problem of equivalence") is how to handle what we might ordinarily think of as inconsistent sets of beliefs, since that grand set will have to be self-consistent -- maybe that's part of what you have in mind? Logical inconsistency, of course, isn't the same as bald self-contradiction....

  13. Eric, I think you may underestimate the availability of the "in-between" response to Pierre cases. Imagine the following small variant. Pierre reads the travel brochures, but misreads the title, so that he thinks it says "Berlin" (or whatever the French equivalent is). He is then transported to London, and comes to believe that London is not pretty. Surely we want to say here that he's not in an in-between state - this looks like a perfectly normal belief state. One day he happens on his old travel brochures, and notices that they are actually titled "Londres". So he drops his old assent to "Berlin est jolie", and instead comes to assent to "Londres est Jolie". Other than that, nothing much changes (he'll now advise people to go to Londres for vacation, and little things like that). Is it really plausible to say that his previous belief that London is not pretty has been somehow downgraded to a mere in-between state just by the new dispositions regarding the lexical item "Londres"?

  14. This comment has been removed by a blog administrator.

  15. That's a nice adaptation of the example, Bargaining Sheep! Now, what changes in your case when Pierre sees that the brochure really says "Londres"? A small change, sure, but a vast number of dispositions shift as a result. He'll say all kinds of things now about "Londres" that before he would have said about "Berlin"; he'll react differently when someone tells him she's from Berlin or (in French) from Londres; he'll fantasize about different travel arrangements. So I'll bite the bullet and say this is enough of a shift to say that he has moved from genuine belief that London is ugly to more of an in-between state.

    Perhaps part of what I need here is negatively characterized dispositions. Part of the dispositional profile for believing that London is ugly is not to be disposed to say "Londres est jolie". Although Pierre in your case retains most of the dispositions it's most natural to characterize positively, he loses many dispositions it's natural to characterize negatively.

    Maybe negatively-characterized dispositions seem problematic in some way -- or problematic in the role I'd like to have them play in the structure of belief? (Could a concern of this sort be partly behind Pete's comments above?) If so, I don't see why. I've never been a fan of the idea that there are logically important systematic differences between negative and positive properties.

  16. Hi Eric,

    I feel like I'm making progress in understanding this, so thanks for the patience. Regarding your first question, the one regarding the machine, I'd be inclined to say, "yes" it has contradictory beliefs. Regarding your second question, the one regarding non-dispositional occurent judgments, I'd say "yes" that would probably break the regress I was envisioning.

    Regarding your speculation as to my motives in your reply to B. Sheep, I'd say "no" that's not where I'm coming from. I don't have much of a problem with negative properties. My not having a probelm with them is one of the properties that I have!



  17. So maybe we're isolating our disagreement here better now, Pete. I'd like to say that your machine is in an in-between state, not quite accurately describable as believing that P and not quite accurately describable as failing to believe that P. It doesn't fully match the dispositional profile for believing that P (because sometimes it says not-P or refuses to say P) and it doesn't fully match the dispositional profile for believing that not-P (for analagous reasons).

    I think that's a coherent and attractive view. Is there some reason to prefer attributing two genuinely held contradictory beliefs?

  18. Hi Eric,

    Admittedly, my claim that contradictory beliefs will be legion under the possible worlds account was rather unclear. What I had in mind was this: If p and q are true at exactly the same worlds, but I nevertheless have a disposition to act as if p, but not as if q, then (given an oversimplified dispositional account of belief)I do and don't count as believing that q. (I do believe that q here simply because 'p' and 'q' denote the same object of belief.) You are right, this seems to be a version of Stalnaker's problem of equivalence.

    I wouldn't say that this amounts to a problem for the dispositional account of belief. Obviously, the possible worlds account is not in harmony with our common practice of belief ascription.

  19. Eric, by "negatively characterized dispositions", do you mean "not being disposed to phi", or "being disposed to not phi"? I have no particular objection to either one, but if you're appealing to the first, some people may think that your characterization of the disposition-belief link is begging the question.

    It's not clear to me that the second kind of negative dispositions were there in the first place (since LEM is going to fail with the narrow-scope negation), but even if they were, I think their loss still leaves you with a sizeable bullet to bite. This change in generally linguistic behaviour just seems like a narrow shelf on which to balance the undermining what seemed like a perfectly good and robust belief.

    One final thought: do you think that the Pierre puzzles can arise already in the dispositions? So could Pierre be disposed to (roughly) act in Londres-pretty ways, but also disposed to act in London-ugly ways? I realize I should just go RTFA, but I'm lazy about that sort of thing.

  20. I see your point now, Mike. Thanks for clarifying it for me! Yes, I take it as stemming principally from the problem of equivalence on possible-worlds views of propositional content; so it may not reflect much on dispositional models.

    Bargaining sheep: I'm not inclined to think that Pierre puzzles can arise in the dispositions, though they can arise in the dispositional attributions if the attributer thinks London and Londres are distinct cities. There's a related question whether disposition-attributing contexts are transparent or opaque, which is a bit tricky. I'm inclined to think that, largely, they're transparent unless they import opacity from somewhere else -- e.g., by involving an opaque term (he's disposed to occurrently judge that the local newspaper reporter is a weakling) or from implicitly referring to a mode of presentation (he's disposed to ride the fastest subway [on a certain disambiguation]). It's a mess if you try to get picky about it!

    As for the scope of the negation: I see two obvious spots for negation as in following examples. (1.) He's not disposed to {ride in taxis}. (2.) He's disposed to {ride in not-taxis}. I'm inclined to think (3.) He's disposed to {not ride in taxis} collapses into either (1) or (2), I suspect. By "negative dispositions" I meant type (1). I'm not sure what's potentially question-begging about incorporating them into the dispositional profile relevant to a belief, though. Kripke's Pierre accords with the stereotypical profile of a London-is-ugly believer in being willing to say "London is ugly" but he does not have the negative disposition of not saying "Londres est jolie". (That sentence has too many negations to parse easily, I know, but I need them all to put the point precisely.) I'd be interested to hear if you have more to say about your question-beggingness concern

  21. Here is a fun case:

    Paul says, repeatedly, "I can't buy a motorcycle. Pat believes that they are too dangerous."

    It looks as though Paul believes that Pat believes that M.

    However, Paul is an eliminativist. He believes that no one believes anything. We ask him: don't you and Pat believe that eliminativism is true? And isn't that pretty funny?

    He replies: no, neither I nor Pat have any beliefs.

    It looks as though Paul believes that Pat has no beliefs. Thus, his second-order beliefs are inconsistent. Changing the story so that his beliefs are contradictory seems a bit forced to me.

    If this is right, if he has inconsistent beliefs, it is a strange case. For one thing, there is no mode-of-presentation confusion (as in Kripke's puzzle). Also, he would vigorously deny that he has any beliefs, including the belief that Pat believes that M. He would say that he was speaking loosely when he said that she believed M. But if he has the second-order belief anyway, then his vigorous denial about his own mind is mistaken.

    And what about metaphysicians like van Inwagen or Merricks? They believe that there are no chairs; they've said just that on many occasions. But can they stop themselves from acquiring beliefs like "There are only two chairs in my office"?

    A lot depends on how automatic it is to acquire second-order beliefs. What do you think?

  22. Brace yourself for what is probably a very lame question: What is the difference between in-between believing and having contradictory beliefs? I mean aside from your disposition with respect to the application of those terms, which I think you would like to see rub off on other people, thereby making that difference at least a social reality. Is there such a difference?

  23. Ah, those wacky eliminativists! It always helps me understand them sympathetically to compare the case of saying "the sun is going down" -- something we find it convenient to say, though we know of course that the sun isn't literally going down. More accurately, the horizon is rotating so as to occlude it. There's nothing any more terrible in Paul saying Pat thinks motorcycles are dangerous than there is in his saying, "Come out, honey, and watch the sun go down!"

    Now let's suppose that current astronomical theory is incorrect and the sun really does go down in the most literal possible sense. When Paul asks Pat to see the sun come down does he believe it's going down or not? I take it this is an in-between case. He has some but not all the relevant dispositions. In some respects and in some contexts it seems appropriate to ascribe him the belief; in others not. It depends on what aspects of his dispositional structure we're interested to highlight. (Compare: Does Ortcutt believe the mayor is a spy? He sees a man whom he takes to be a spy, but doesn't realize that man is the mayor.)

    I know that's only a sketch of an answer, Bryan. Maybe I should work up an entire post on this. That could be interesting. Thanks!

    Thanks for your question, too, Daril. It's always nice to hear from a former student! Here's my take: It's a difference in how we describe the cases. I see no reason to ascribe baldly contradictory beliefs; I think they're more helpfully conceived of as in-between cases. So I take what I'm doing to be recommending a particular way of thinking and speaking about certain types of cases. In general, I think metaphysics is mostly linguistic and conceptual recommendation of this sort!