Although it is widely recognised that human beings are fallible in various ways, the doctrine of fallibilism – according to which knowledge is compatible with fallibility – is a modern creation, typically credited to Charles S. Peirce. Peirce however appears to have equated the denial of infallibility with a denial of universality. I gloss this as saying not that we are unable to believe a proposition that is universally true, but only that we are unable to know such propositions. (I ignore the tricky question of whether the universality in question must be explicitly coded into the proposition in order to be unknowable.) I’ll compare this denial of universality with two other denials: the denial of absoluteness and of necessity. The former corresponds to ignorance of things as they are in themselves. The latter is evident in Peirce’s ambivalence towards the applicability of fallibilism to a priori truths, e.g. mathematical truths. To the extent that he does apply it, however, Peirce seems to think there is a special problem, specifically, he tends to infer contingency from fallibility, e.g. that 2 + 2 might not equal 4 – though it is unclear (as it often is in discussions of fallibility) what is meant by “could” here.
Peirce thought that fallibilism pressures us not to take mathematical truth at face value. I gloss this in terms of scepticism regarding knowledge of necessity. That is, it is the same thing to say a belief is not universally true (as far as we know) and to say that it is not necessarily true (e.g.: the non-universality corresponds to a possible world where the proposition believed is false). On this reading, non-universality does not affect contingent truths, i.e. there is no contingency of contingency (though it seems legitimate to me to wonder why not).
Peirce’s position – as I understand it – is interesting because it appears halfway between scepticism regarding knowledge of things as they are in themselves, and fallibilism, according to which the modal status of a belief, like its status as mind-independent or absolute, is completely independent of the fallibility or infallibility of its basis. Again, Peirce (on this reading) accepts that the fallibility of my justification or warrant for believing that p does not license the inference that the truth of that belief is relative to me or non-universal, except in the sense that p is thereby shown to be – as far as we know – contingently rather than necessarily true. (This puts him at odds with the contemporary fallibilist.) Scepticism regarding knowledge of things in themselves, and regarding necessity, partially overlap given the premise that (some) necessary truths characterise intrinsic/essential properties of things – the latter being properties of those things as they are in themselves. On my reading, Peirce shows more sympathy for the latter than the former, and this suggests that his fallibilism is transitional in nature. (As I said, Peirce wavered on whether fallibilism should be extended to a priori knowledge.)
Let’s put aside any exegetical intent and focus on the core idea of inferring contingency from fallibility, which can be understood as a way of modelling the relativisation of knowledge of necessary truths. Necessary truths become relativised as contingent truths. The question is then whether it makes sense to apply this contingency to itself. Of course, if we must represent every truth as metaphysically contingent as far as we know, then it seems self-refuting to say of this necessity that it too is metaphysically contingent as far as we know, since we would then have to say that something is metaphysically necessary after all (because if contingency is contingent then something is necessary), and that forces us to take back with one hand what we have just ceded with the other. (Alternatively, if relativism requires that for any proposition p relatively true for a subject S, ¬p is relatively true for a different subject S*, then relativism entails – and hence cannot consistently deny – contingency.) From the Peircean viewpoint – or at least the broadly Kantian one Peirce shows sympathy towards – this reasoning is plausible: there is an ubiquitous relativity that delimits the phenomenal realm from the inside, and we must represent the in-itself as in principle metaphysically distinct from this realm, which is another way of saying that the relation between the two realms (whatever exactly is meant by “realm” here) is metaphysically contingent. All of this rests, however, on the premise Langton discovers buried in Kant, to the effect that humility – that is, our ignorance of things as they are in themselves – requires a distinction, and thus a contingency, of this sort. However, if contingency is not essential in this way, it will be possible to relativise it without self-refutation.
Now, on the one hand, contemporary fallibilism is not a type of relativism – exactly the opposite, since it ignores the restriction to phenomena (see. e.g. Langton 2004). Moreover, it says that, as a general rule, true beliefs we form regardless of their modal status are, or can be, known. On the other hand, by freeing itself from the need to appeal to metaphysical contingency, fallibilism also shows us how relativism need not depend upon it either. That is, the relativist does not need to say that what is knowable is the subset of my beliefs that are contingently true when true at all (which, as I said, amounts to inferring scepticism about knowledge of necessary truths from the premise of fallibility). Nor does she even need the premise that the connection between my justification (i.e. the basis of my belief, whatever it is) and the truth-value of my belief is metaphysically contingent. Instead we have – the fallibilist claims – a purely epistemic fallibility, i.e. one that can be cashed out in terms of uncertainty, epistemic circularity (e.g. Stove’s Gem), epistemic probability, etc., without incurring any metaphysical commitment that would stand out like an Achilles’ Heel to be preyed upon. This means that the ball is squarely in the court of anyone who wants to show that relativity/scepticism and contingency are connected anhypothetically.
[last edited July 25 2013 – sorry about the late changes]
- Christopher Hookway, “Peirce and Skepticism”, in John Greco ed., The Oxford Handbook of Skepticism, (OUP: 2008), pp. 310-29.
- Rae Langton, “Elusive Knowledge of Things in Themselves”, Australasian Journal of Philosophy, 82 (2004), pp. 129-36.