Assumptional Epistemology

I’m working on Duncan Pritchard’s book, writing a substantial notice of it for PPR. One of the central elements in Duncan’s thinking is the idea he gets from Wittgenstein that there are bedrock propositions that fall outside standard epistemological assessment (he calls them “hinge” propositions). They don’t count as known or justified or rational, or something in that neighborhood.

It struck me, in noting this feature, the growing sentiment in this direction in recent epistemology. Gil Harman is defending a version of assumptional epistemology, and one way to understand Schaffer’s contrastivism is along the same lines (Jonathan uses the neat encapsulation phrase “what is presupposed rather than proved” in describing his view). And, of course, these views share a lot in common with the denials of closure found in Dretske’s work and in Nozick’s as well. I haven’t seen what Duncan has to say about closure yet, but I expect something akin to Dretske’s idea that closure is fine so long as the entailed claim isn’t a “heavyweight” one (though Duncan would likely call it a “hinge proposition” instead).

Contextualists like to claim that they hold the middle ground between skeptics on the one hand and Mooreans on the other, but they also can claim to hold the middle ground between assumptionalists and non-assumptionalists (of which Moore is a paradigm example). They are not alone on this middle ground, of course–there are also the invariantists who espouse pragmatic encroachment into the epistemic realm (e.g., Stanley, Hawthorne, McGrath&Fantl).

For those of us not the least inclined toward being sucked into the black hole of skeptical epistemology, this taxonomy is more congenial to our philosophical temperaments and not discussed as much as it deserves. There is the literature on closure, which may be the Achilles’ heel of assumptional epistemology, but the jury is still out on this question. But there is little direct discussion in mainstream epistemology addressing the primary and substantive commitment of assumptional epistemology. Or perhaps I’ve missed a body of literature on the subject?


Assumptional Epistemology — 17 Comments

  1. In an intro text to epistemology, Stephen Cade Hetherington (Knowledge Puzzles, ch.24, p159) discusses contextualism and the possibility that the contextualist may hold that there are ajustified beliefs. These would seem to have the character of assumptions.

    It might help if you gave your meaning of the term “assumptionalism”. The word brings up no results on JStor or Philosopher’s Index, and is resistent towards being taken at face value, given the way this post has described it, and given the accuracy of Hetherington’s characterization.

  2. Jon, when assumptionalists make a claim like this,

    “bedrock propositions that fall outside standard epistemological assessment. .”

    how should I understand that? It’s a kind of dogmatism, no doubt. Let p be such a proposition.
    Do they mean by the claim that ‘p falls outside standard epistemic assessment’ that denying p is meaningless? Or is it something like denying p is not practically possible? Or something else altogether? In any case it seems that we are being asked to take the position that (i) I might be wrong about p and (ii) p is beyond epistemic assessment. But then (i) itself seems to be an epistemic assessment of p. But they cannot be denying (i), can they? I just received Pritchard’s book, but no time to review it.

  3. Hi Mike, a quick gloss from Duncan is that the hinge propositions are special because one can’t give reasons for them. His claim is that one can’t give reasons for them because anything that one might cite would be less certain for you than the hinge proposition.

    My claim that such claims fall outside epistemic assessment is a little misleading here–the claims can’t be justified by reasons, known, or rational, but they aren’t unjustified or irrational as a result.

    I’m betting you’ll want to question the claim that nothing can be a reason for believing a given claim when it is less certain than that claim… 🙂

  4. Benjamin, nothing I wrote implied that contextualists couldn’t be assumptionalists, but only that they needn’t be. The standard versions of DeRose and Cohen, for example, don’t posit assumptions of the sort posited by Duncan. There is, of course, Schaffer’s contrastivism, which can be thought of as a version of contextualism, and it has a place for assumptions or presuppositions, though not in the way employed by either Harman or Pritchard (since he denies the epistemic claims in question when the contrast is merely presupposed rather than proved).

  5. Hi Jon,

    I’m not sure exactly what you intend by ‘assumptionalist’, but one group of views that are at least interestingly related to what you’re after and are currently getting some attention in the literature are default entitlement views. Burge has a version of this sort of view, Wright has another (and he explicitly traces a link to Wittgenstein’s notion of a hinge proposition), and Field’s evaluatism puts an interesting new spin on the idea. Pryor’s dogmatism is, I guess, also in the vicinity.

  6. A propos Carrie’s post: I guess there’s a difference between a hinge proposition and a Burgian default entitlement, namely that the latter is a special status that belongs to certain propositions whereas the former are propositions that have a special status.

    This is not a quibble: Certain propositions, e.g. I remember being a young boy, have the Burgian status, but aren’t hinge propositions – or what?

  7. Carrie and Andreas, the entitlement views are interestingly related to Duncan’s view. I’m not sure of the following, but I think for Wittgenstein, it would be a mistake to ascribe positive epistemic status to bedrock propositions. For Duncan, that is not the case, though there is something epistemic we can’t say about them. We can’t, for example, defend them successfully by appeal to other claims. I’m not sure what relevance this point has to the epistemological theory, however, since it is so obviously related to rhetorical context rather than to theory. What it takes to be justified and what it takes to show, say, the skeptic, that one is justified are quite different, and there’s no obvious reason why the second implies anything about the first.

  8. This —

    (a) They don’t count as known or justified or rational, or something in that neighborhood.

    — is hard to square with this:

    (b) anything that one might cite would be less certain for you than the hinge proposition.

    I would have thought that certain is in the known/justified/rational neighborhood. So (a) would seem to imply hp’s can’t be certain (which isn’t to say they’re uncertain). But (b) seems to be saying that they are very certain indeed.

  9. Yes, Keith, that’s exactly right. I think Wittgenstein denied positive epistemic status to bedrock claims–that’s what he didn’t like about Moore. If we read ‘certain’ in the epistemic sense, then we don’t preserve this Wittgensteinian point.

    I’m not sure what Duncan’s view is here. One option is that ‘certain’ is to be understood psychologically, but I don’t think he means that. And once we distinguish the rhetorical context of answering the skeptic and other questioners from the non-rhetorical context of the structure of knowledge and justification, it is hard to see why the inability to cite something more certain or as certain has any skeptical bite to it at all. And yet Duncan wants hingehood to worry us a bit about skepticism.

  10. Jon,

    I suppose this follows on Keith’s post. You wrote:

    “His claim is that one can’t give reasons for them because anything that one might cite would be less certain for you than the hinge proposition.

    My claim that such claims fall outside epistemic assessment is a little misleading here–the claims can’t be justified by reasons, known, or rational, but they aren’t unjustified or irrational as a result.”

    (1) I’m unsure as to whether just you, just Duncan, or both make the claim that follows the hyphen.

    (2) I suppose it is uncontroversial that

    (i) “one can’t give reasons for them because anything that one might cite would be less certain ”


    (ii) “the claims can’t be justified by reasons”

    But it is not at all obvious that (i) entails either of the following:

    (iii) “the claims can’t be known”

    (iv) “the claims can’t be rational”

    Indeed, I think it entirely possible for one to rationally believe and know a claim that one cannot give reasons for of which one is more certain.

    (3) I suppose something like a rock is not irrational (and of course is not rational either) and that one would not make an epistemic assessment if one were to say “That rock is not irrational.” (Don’t ask for a context, outside a discussion of philosophy, where this might seem a natural thing to say.) But it sure seems to me that one makes an epistemic assessment when one says of a belief, “That is not irrational.”

  11. Mike, you’re right, too. I’m still trying to sort out what the connection is for Duncan between hinge propositions and worries about skepticism. There is supposed to be one, but it is not clear what it could be. Gotta go back and read the rest of the book!

  12. A purely exegetical point; I have been talking about this with experts of Wittgenstein, and I was regretfully convinced that, for Wittgenstein, hinge propositions are not strictly speaking even true (or false). This is for broadly verificationist reasons; if you cannot give reasons for P, it is not clear what it means to say that P is true. If, instead, he had considered them true, he would have probably seen the difficulty (or the opportunity) DeRose and DePaul point to.

  13. Daniele, that’s an interesting point, both in this regard and in the context of the knowability paradox. If bivalence holds for such claims, however, the epistemic options are still open. Keith and Mike are right that one will have to be careful how such claims are identified, though: better not identify them in terms of being more certain than all else, and then deny positive epistemic status to them. That problem isn’t terribly difficult to avoid for W., I would think. He could talk of the fundamental claims involved in the language game of reason-giving. I doubt such a move will help with the regress problem, but that is a different matter.

  14. Pingback: Certain Doubts » June Stats

  15. Hi Jon,

    I’m really pleased to hear that you’re writing a critical review of the book for PPR. I won’t try to summarise my view here, but just offer a few quick clarifications which might help with some of the issues raised above. First, I do think that the hinges are known, so there’s no tension with closure. Moreover, I don’t try to motivate scepticism about knowledge in terms of an appeal to hinges (though I do think that there is some sceptical import to the thesis). What I take from Wittgenstein is simply a point about the dialectical force of reasons and the bearing this has on the propriety of certain knowledge claims.

    I’m in the process of writing the entry on ‘Wittgenstein on Scepticism’ for the new Oxford Handbook to Wittgenstein, and should have a working draft in the next few weeks, so if anyone wants to hear more about (my take on) Wittgenstein’s view of hinges then I’d be happy to send this on.


  16. Quick question — actually a couple of question s — about this summary (from Jon K) of Duncan Pritchard’s position on hinges:

    “[O]ne can’t give reasons for them because anything that one might cite would be less certain for you than the hinge proposition.”

    1. Is this property supposed to define hinge propositions?

    2. What counts as something one might cite?

    The point of question 2 is that often even a single line of reasoning for some claim X are quite complex and involve both things we’re quite sure of and things we’re somewhat less sure of. And in a lot of the most interesting cases, the reasoning is non-deductive. So is the reason, in such a case,

    (a) the whole set of relevant considerations bearing on X,
    (b) some function on the whole set of considerations that takes into account the non-deductive nature of the reasoning,
    (c) the most plausible claim that forms a non-redundant premise in the reasoning,
    (d) the least plausible claim that forms a non-redundant premise in the reasoning (what we might call the weakest link), or
    (e) something else?

    In any event, it certainly seems that there are cases where a proposition might have, exogenously, a higher probability than at least the conjunction of each cluster of relevant evidential considerations but where those clusters, taken jointly, could raise its probability asymptotically towards 1, e.g., when a large number of moderately trustworthy independent witnesses concur that the outcome of rolling a well-balanced icosahedral die down corrugated cardboard was not a 17.

    I’m not suggesting that “The die didn’t fall 17” is a hinge proposition in Pritchard’s sense. But this consideration does suggest that the classification of anything that is not absolutely certain as a hinge proposition might be problematic.

Leave a Reply

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