Is knowledge easy or impossible? In an earlier post, I described epistemic circularity as the use of a particular faculty or source of warrant as sole justification for belief in the veracity of that faculty or source of warrant. Thus, for example, it is epistemically circular to give an inductive defense of induction. Now, a recurring motif of epistemic externalism – according to which there is some type of special access or privilege that is not necessary for justification or knowledge – is that our analysis of knowledge must either make it very easy to get, or else impossible. This motif (and indeed externalism itself) can be glossed in terms of epistemic circularity: if knowledge is compatible with epistemic circularity then it is very easy to get, whereas otherwise it is impossible. That would seem, however, to obliterate the distinction between knowledge and faith, or at any rate that between science and non-science. So is the underlying argument for this conclusion – that knowledge is either easy or impossible – convincing? And moreover, is there anything essentially religious about this argument? In this post I comment briefly on each of these questions.
(1) Is the externalist dilemma (as I shall call it) any good? I’ll focus on Van Cleve’s presentation of it. Van Cleve does not defend the view that knowledge is compatible with epistemic circularity, but rather the conditional claim that it (seemingly) must be if scepticism is to be avoided. Thus phrased polemically, the choice seems stark: either no knowledge at all or else an irrational leap of faith.
Van Cleve initially defines internalism as follows: first-order knowledge that p requires higher-order knowledge of the factors that make this first-order knowledge possible. This can be grasped heuristically via the familiar closure principle for knowledge: if I know that p and know that p entails q then I know (or am in a position to know) that q. Hence if p is true then this entails that my justification for believing p is not misleading me with regard to the truth-value of p. Hence by closure I am in a position to know that my justification is not misleading me with regard to p. This is a good approximation of what Van Cleve intends by his preliminary characterisation of internalism.
Given the ubiquity of epistemic circularity (the argument for which is given in the post linked to above), it follows that I am in a position to know that some method of belief-formation M, that features in the justificatory ancestry of my belief that p, is able to justify belief in its own veridicality, i.e. is able to do the very thing that strikes us as so implausible in the case of the inductive defense of induction. At the level of epistemically circular reasoning, Van Cleve observes, there is a strong tendency to regard this sort of thing as objectionably circular: for example, if I argue for the reliability of sense perception using premises derived from sense perception, then I am using premises that could only be known on the basis of the faculty or method which I am in the process of vindicating. Of course, from the viewpoint of the externalist, this is not really a problem, as we don’t need to vindicate the method before we are able to use it to acquire knowledge. Indeed it is not even necessary to reject closure: all of the additional higher order structure of my epistemic position can be understood externalistically. That is why the definition of internalism given above must be understood as requiring antecedent (i.e. epistemically non-circular) justification as a precondition of knowledge. It is because knowledge of this sort is impossible that Van Cleve concludes that we must choose between externalist (i.e. easy) knowledge, or no knowledge at all.
To reiterate, internalism is initially defined as the view that first-order knowledge requires higher-order knowledge. To get from here to the externalist dilemma, we add three further premises, each already featuring in my discussion so far.
- We need to know a method is reliable before we can know anything on the basis of that method (requirement of epistemic non-circularity).
- For some method M, we can only know that M is reliable on the basis of M (iteration).
- “Knowledge” is used univocally with regard to first-order and reflective knowledge.
We can now assign a label to the negation of each of (1)-(3).
Coherentism (¬1): I can know that p on the basis of M whilst simultaneously (in the epistemic sense of denying any antecedence or asymmetry) knowing that M is veridical or reliable. According to coherentism, the topology of justification/knowledge is curved, like Quine’s web of belief. Justification arises as a function of the correspondence between beliefs (judged by criteria of coherence, whatever these might be), rather than as a function of the correspondence between beliefs and something other than belief. Put differently, my justification for believing p is that q is true and my justification for q is r, etc., with the chain of justification eventually returning to my justification for p. The nodes in this chain are bound by “mutual dependence without mutual priority”. As for the entire web, it is said to be self-supporting. (Note: coherentism about justification has historically been accompanied by coherentism about truth, but each can be defended independently of the other.)
Reidianism (¬2): What Van Cleve calls “Reidianism” is not the denial of 2 as I have formulated it, since Reid (on his reading) only asserts that it is possible to know that some faculty is reliable on the basis of something other than that faculty, not that this is possible for every faculty. Indeed it is the restriction of scope that best expresses Reid’s position, since, much like the coherentist, Reed needs an element of self-support in his account. Van Cleve glosses Reid’s view as follows: knowledge of reliability is not derived in any way, but epistemically basic. Indeed, although the coherentist officially rejects basic knowledge according to Van Cleve, it is tempting to think that the two views shade into one another at this point. Thus taking each direction in turn, if we iterate Reid’s denial then it matches up with the denial of 2 as I have formulated, and this brings us back to the coherentist denial of 1. Conversely, Van Cleve argues that the coherentist cannot preserve her distinction from the Reidian: either she gains reflective knowledge in the same way as the externalist – in which case the knowledge is still easy – or else she requires for such knowledge the unrealistically high standards of the coherentist epistemologist, in which case scepticism looms once again. Either way the externalist dilemma remains.
The two-levels approach (¬3): This strategy, championed by Ernest Sosa (and in a different way by Sellars and Brandom), is prima facie susceptible to the same objection just given, though obviously I’d like to think the jury is still out.
(2) The overall gist of Van Cleve’s argument is this: coherentism and Reidian foundationalism are ultimately indistinguishable, and no attempt to escape this result by analysing reflective knowledge differently to first-order knowledge is likely to be successful. Perhaps the key upshot of Van Cleve’s argument is that it hollows out the space between radical scepticism and radical non-scepticism. What is lost here is precisely the critical spirit, as found most famously in Kant. We end up with the impossibility of a principled critique of fideism, of pre-critical metaphysics, etc. This is a distasteful result. Indeed, the association between externalism and the reformed epistemology of Plantinga, Bergmann, and others tempts us to simply dismiss it as guilty by association, for the simple reason that anything capable of justifying a revival of religiously inspired epistemology and metaphysics cannot be right. This however strikes me as too easy and convenient a dismissal. The externalist dilemma is stubborn because it is backed by philosophical considerations that are at least prima facie plausible. And analyses of knowledge that make it easy are given by several prominent (atheist, as far as I know) epistemologists, e.g. Hetherington, Goldman, and most recently Foley. (See also Williamson and the new Oxford realists.) It thus strikes me as a still open problem: to pass between the horns of the externalist dilemma, to make knowledge hard without making it impossible.
- James Van Cleve, ‘Is Knowledge Easy – Or Impossible? Externalism as the Only Alternative to Skepticism’, in Steven Luper ed., The Skeptics: Contemporary Essays, (Ashgate: 2003), pp. 45-59.