Lewisian Humility

Rae Langton (“Elusive Knowledge of Things in Themselves,” Australasian Journal, 2004) and Jonathan Schaffer (“Quiddistic Knolwledge,” Phil. Studies, 2005) have recently argued that the ‘skeptical’ thesis defended by David Lewis in “Ramseyan Humility” is no different, when it comes to the applicability of standard anti-skeptical proposals, from Cartesian skepticism. It seems to me, though, that Lewis – at least on a charitable reading – could not have meant what Langton and Schaffer take him to mean.

On the face of it, of course, Lewis’ humility thesis (“We have no knowledge of the instrinsic properties of things”) looks like the same kind of claim as Cartesian skepticism (“We have no knowledge of the external world”). And Langton and Schaffer stress the similarities in the arguments that Lewis and the Cartesian offer – both arguments appeal to the idea that there are a plurality of possibilities which are all equally supported by our evidence. In the Cartesian case, this evidence is (among other things) our perceptual experience, which supports (so the Cartesian argues) the hypothesis that there is an external world just as much as it supports the hypothesis that we are decieved by an evil demon. In the case of Lewis, the evidence is our scientific evidence – what we know (or have sufficient reason to believe) about the world is that a certain Ramsified sentence (encompassing both our ordinary beliefs about the world, along with our best scientific theories) is true. And this evidence, so Lewis argues, supports the hypothesis that the fundamental, instrinsic properties of things are p1, p2, p3, etc. just as much as it supports the hypothesis that the fundamental, instrinctic properties of things are q1, q2, q3, etc. There are “multiple realizations” (as Langton puts it) of the Ramsified sentence, since we can imagine different permutations of fundamental, instrinsic properties underlying the same causal superstructure, i.e. (and more importantly) warranting the same belief in the same Ramsified sentence. And knowledge that that sentence is true exhausts our (scientific and ordinary) knowledge.
So we have no knowledge of the fundamental, instrinsic properties of things.

But wait a minute. In Descartes’ argument, it is claimed that although I believe one thing (that there is an external world), my evidence actually equally supports another incompatible thing (that I’m decieved). But that is not claimed in Lewis’ argument. It is not that I believe that the fundamental, instrinsic properties of things are p1, p2, p3, etc., and then Lewis points out that my evidence equally supports the hypothesis that the fundamental, instrinctic properties of things are q1, q2, q3, etc. What I believe is that the Ramsified sentence, described above, is true. And that not only exhausts my knowledge, it exhausts my belief. The Cartesian skeptic argues that we are overly credulous – believing in an external world, without sufficient warrant. But Lewis does not argue that we are overly credulous.

Consider the beliefs that I do have about the fundamental properties of things. I believe, for example, that some particles have charge, and I think that charge obeys Coulomb’s Law. But Lewis’ argument does not challenge these beliefs. If that were what Lewis was saying – that I do not know that charge obeys Coulomb’s Law, because for all I know it is spin that obeys Coulomb’s Law, and not charge – that would make his argument akin to Descartes’ argument. And I think it would be right to reject it, on whatever grounds we reject Descartses argument. But it doesn’t seem to me that that could be what Lewis was saying.

He writes, on p. 16-17 of my manuscript, that our ignorance of instinsic properties is ineffable, because we would like to express our ignorance by listing (by name) the possible properties which fulfill a certain causal role (described by the Ramsified sentence), but we can’t. We do not have names for these properties. The property-names that we do have – ‘charge’, ‘spin’, ‘mass’, etc. – are not the names we want. The problem of our ignorance is not, on Lewis’ view, the problem of knowing whether any bodies have charge, the problem is knowing what property ‘charge’ picks out. (Or, if you prefer, the problem is knowing what property it is that in virtue of having it, things have charge.) Consider an analogous case: we have no knowledge of who the murderer is. This is because, let us suppose, that although we have evidence enough to suppose that there is some person who fulfills the role of ‘the murderer’, we do not know who the murderer is. But this ignorance is not Cartesian ignorance – it is not as we believe that the murderer is Jones, but do not have sufficient warrant for this belief. The point here is just that ignorance need not mean unjustified belief – ignorance is sometimes a lack of belief. In the case of the ignorance of fundamental, instrinsic properties defended by Lewis, it seems to me that it is the latter kind of ignorance. (Furthermore, it seems that what Lewis’ argument shows, if it is correct, is that we couldn’t form the beliefs in question – we don’t have the ability to express those propositional contents, the ones of which Lewis says that we can’t know whether they’re true or not. But this means that our ignorance is to our epistmeic benefit! Our inability to so much as believe one of Lewis’ true, unknowable propositions ensurses that we won’t form any unjustified beliefs, when it comes to fundamental, instrinsic properties.)

This, it seems to me, is why the situation is not “ominous” – because Lewis has not uncovered a domain of belief where we have rashly gone beyond our evidence (as the Cartesian claims). The thesis that most of my beliefs are unjustified, and the possibility that most of them are false, is ominous. But the thesis that there are some things (about the fundamental nature of things) I won’t ever be able to know – that’s not ominous at all.

A caveat: I find Lewis’ essay quite confusing, and I certainly wouldn’t be that surprised if I was wrong about what he meant. But if he really did mean that I don’t know that charge obeys Coulomb’s Law, because it’s possible that spin obeys Coulomb’s Law – well, that seems to me to be a somewhat bad argument, and it really doesn’t seem to have anything to do with the fact that charge is a fundamental property, or with the nature of intrinsic properties, or whatever. On the reading I’ve offered, we at least get the result that Lewis’ argument really does offer us something new – not just Cartesian skepticism “writ small” (as Schaffer argues) – and something that really is a consequence of the nature of fundamental, instrinsic properties.


Lewisian Humility — 5 Comments

  1. Hi Allan,

    1. I find the Lewis essay confusing as well. As I read Lewis, he thinks that there are two epistemic deficits we suffer with respect to the quiddities:

    a. Doxastically, merely quiddistic differences are ‘ineffable.’ We cannot, in belief (or any other attitude) distinguish between possibilities that differ merely by permutation or replacement of quiddity.

    b. Evidentially, merely quiddistic differences are indiscriminable. None of our evidence can distinguish between possibilities that differ merely by permutation or replacement of quiddity.

    The doxastic deficit only comes out in the final two sections [7 and 8] of the paper. The evidential deficit comes on stage early on, in sec. 3, when Lewis says:
    “There is indeed a true contingent proposition about which of the possible realizations [of the true and complete final theory T] is actual, but we can never gain evidence for this proposition, and so can never know it. If there are multiple possible realizations, Humility follows.”

    So as I read Lewis, he is primarily concerned with the evidential deficit, and the doxastic deficit is tacked on at the end. What is confusing (to me at least) is what relation the evidential and doxastic deficits are supposed to stand in to each other. Are these deficits independent? Is one the basis for the other? Do they spring from a common cause? (Would a reply to one ipso facto answer the other?)

    2. Of course Allan is right that there are some disanalogies between quiddistic skepticism and Cartesian skepticism about the external world. He points to a doxastic disanalogy, characterizing Cartesian ignorance as the sort in which we are overly credulous, and Ramseyan humility as the sort in which we are (luckily!) unable to form the beliefs by which we could be overly credulous.

    All I meant to argue in my paper was that /the evidential arguments based on indiscriminability/ are parallel (see the start of my sec. 3).

    But there are two points to consider that might make the analogy even tighter:

    a. It is not obvious whether Lewis’s ineffability concerns should also apply to Cartesian demon scenarios. The issue is complicated by the fact that he helps himself to an interpretation of the O(ld)-language, and uses it to interpret the T-language of the true and complete final theory. Perhaps in the demon scenario some O-language sentences will go false?

    b. It is not obvious whether a solution to the evidential problem for quiddities will have the consequence of rendering them effable after all.

    3. As to what counts as ‘ominous’ I do not know. I agree with Allan that if you put the doxastic and evidential deficits together, they cover each other nicely, in such a way that there is no “domain of belief in which we have gone rashly beyond our evidence.” That’s a nice point. But still, we would remain in deficit. We would remain ignorant of certain features of the world. Indeed we would lack the means to even conceive of our ignorance. Is that ‘lucky’ or not? The options would be: (a) grasping a question but being ignorant about the answer, and (b) not even being able to grasp the meaning of the question. One might think that (b) is an even sadder epistemic state.

  2. Hi Allan and Jonathan,

    I use to think that (if Lewis’ argument for Ramseyan Humility is right), then our ignorance was indeed ineffable. I’m not sure exactly what role you played in changing my mind, Jonathan, but I certainly have changed it. I can’t tell from your last comment, but have we now switched places?

    In any case, my reason for thinking that our ignorance is NOT ineffable, is just that I think we can state Lewis’ thesis in the following way, provided that we are careful of how to interpret it:

    RH: We will not ever know that the actual realization of T is the realization of T.

    I use to think that this could not be an accurate paraphrase of Lewis’ thesis because OF COURSE we know that the actual realization of T is the realization of T! However, I think this response hinges on an ambiguity. RH is either

    RH1: We will never know that p, where p is the A-intension (primary intension) of `The actual realization of T is the realization of T’.


    RH2: We will never know that q, where q is the C-intension (secondary intension) of `The actual realization of T is the realization of T’.

    I think that what is `obviously’ false is RH1—that is, OF COURSE we know the A-intension of `The actual realization of T is the realization of T’ (provided we know that T is true). However, it is RH2, I claim, that paraphrases Lewis’ thesis.

    Similarly, I don’t think that there is any doxastic limitation involved. Allan, you write that what is lacking are “names” for the properties that realize T. But why can’t we just use rigidified descriptions? Why, that is, can’t I believe the C-intension of `The actual realization of T is the realization of T’, and thereby believe the proposition that Lewis claims we will never know?

    I think it is at least somewhat telling that although Lewis titles the relevant section “Ineffable Ignorance”, he nowhere in it actually says that our ignorance is ineffable (does he?). Of course, he does say that we cannot “properly ask” the question: Which property realizes such and such role? But it’s an open question whether this implies that we cannot state what is we don’t know. (It’s also an open question whether Lewis is right about not being able to properly ask the question.)

    That said, I agree with Allan that Ramseyan Humility is not akin to ordinary skepticism. In fact, the paragraph of Allan’s post that began, “But wait a minute” … ” sounds as if it were a transcription of something I once said to Jim Joyce.

  3. Jonathan:

    On (3) I think I agree. It’s omnious in one way, but not in another. On (1) and (2), I think you’re right that the argument for the ‘evidential defecit’ from Lewis is paralell to the Cartesian argument. My big question, I think, is with your idea that anti-Cartesian machinery will suffice as anti-Lewisian machinery as well.

    Consider reliabilism. The way the reliabilist deals with the Cartesian skeptic is to maintain that a belief is justified (and therefore may amount to knowledge) iff it is formed and sustained in a reliable way. My ordinary perceptual beliefs are so formed, and are thereby justified. So we can have knowledge of the external world. But this can’t work against Lewis. because I don’t have the relevant beliefs. The question of whether “they” are reliably formed doesn’t arise.

    Perhaps I’m appealing to the “ineffability” considerations here. Although I would like to say that I’m not – I’m just thinking about what I believe, and I don’t think I have any beliefs that need “saving” from Lewis’ argument on reliabilist grounds.

    Aren’t the anti-skeptical proposals you canvass all, essentially, ways of showing that beliefs, which fail to meet the skeptic’s demanding standard, actually meet a proper standard for justification/knowledge? If that is what anti-skeptical proposals are, then they can’t help against Lewis. It seems to me that the varieties of “Dogmatism” and “Anti-Closureism” that you canvass in your paper are all ways of showing that my beliefs about the external world are justified. I don’t see how those proposals, therefore, carry over against Lewis, because there’s no beliefs that one might attempt to show justified.

    I do think that contrastivism (of course!) can say something nice about Lewis’ argument (if I understand your view right): that I know that charge, rather than mass, obey’s Coulomb’s law, but I don’t know that charge, rather than some quiddistically indiscernable non-charge property, obey’s Coulomb’s law. But unless we understand ‘knowledge’ in this contrastive way, I don’t know what we traditional anti-skeptics can say against Lewis’ argument.


    So I typically make some laughable mistake when it comes to this primary/secondary intension stuff, but so if you believed q, where q is the secondary intension of ‘The actual realization of T is the realization of T’, wouldn’t you therby believe something true? And can’t we (now) recognize that you would surely believe something true, in so believing? So wouldn’t you then know that q is true (at least if you recognized that you would surely believe truly, in so believing)?

    If the answer to that is that you would know, then RH2 is false.

  4. Allan, I think that’s exactly the question to ask. What I think Lewis would say is something like this: we can know that we believe a true proposition, without knowing (or even being in a position to know) the proposition we believe.

    Consider the following example. Someone has stolen my chicken, and, upon discovering this, I come to believe that the man who actually stole my chicken is the man who stole my chicken. Now, you might say that I can only believe the A-intension of the embedded statement—that is, that I can only believe that the man who actually stole my chicken, WHOEVER he is, is the man who stole my chicken. However, I don’t see why this should be. Why can’t I believe that THAT MAN (rigidly) who actually stole my chicken is the man who stole my chicken? I think that I can—that is, I think that I can believe the C-intension of `The man who actually stole my chicken is the man who stole my chicken’.

    Now, as you point out, it also seems that I know (or that I am in a position to know) that the proposition I believe is true. However, I don’t think we want to say that I know that proposition. Suppose we did say that. Then, supposing Jones stole my chicken, it would follow that I know the C-intension of `Jones stole my chicken’ (for the C-intension of this statement, supposing `Jones’ is a rigid designator, is the same proposition as the C-intension of `The man who actually stole my chicken is the man who stole my chicken’). But it doesn’t seem that I know the proposition that Jones stole my chicken, and this it seems, is not knowing the C-intension of `Jones stole my chicken’.

    Of course, we might reply that when we say that I don’t know that Jones stole my chicken, what we are saying is that I don’t know the A-intension of `Jones stole my chicken’. If that’s so, then I think knowledge isn’t really about C-intensions at all, which Allan Gibbard has recently argued in an, I think, as of yet unpublished paper. In fact, Gibbard uses just this kind of case in his argument. But I think there are two ways to respond to this type of case: either we can deny that knowledge is knowledge of C-intensions, or we can deny that one always knows what one knows one believes truly. I think that Lewis would opt for the latter. As for myself, I’m not sure, but I am recently leaning towards Gibbard’s view (and thus leaning towards giving this kind of a response to Ramseyan Humility).

  5. Allan, in reading your most recent response to Jonathan (and your initial post), I’m beginning to believe more and more strongly that you somehow got your hands on an earlier draft of the paper I’ve been working on. Haha, not really, but the overlap is remarkable. The example of a typical response to skepticism you went with—reliabilism—is the very example I chose to illustrate the very same point.

    But here’s a worry: you say you don’t have the relevant belief. However, if I’m right, then you could COME to have the relevant belief. This is why I don’t think that the response you’re giving to Jonathan (the one I once wanted to give) is ultimately going to work. Provided that you can COME to have this belief, through a reliable process, and assuming reliabilism is true, then Lewis’ thesis is false.

    Nevertheless, I think that the stolen chicken case shows why reliabilism cannot be generally true: coming to believe that p, where p is the C-intension of `The man who actually stole my chicken is the man who stole my chicken’ by first coming to know that SOME man stole my chicken and then believing that HE (rigidly, THAT man) stole my chicken is, although it is surely a reliable process, not one that gives one knowledge of p.

Leave a Reply

Your email address will not be published. Required fields are marked *