Internalism and externalism in epistemology

3. Justification externalism – pro and con

In the most general sense, externalism is simply a rejection of the internalist constraint on access. (Externalism can allow reflectively accessible factors, but it denies that what affects justificatory status is restricted to them.) Externalism in this generic sense does not include any particular positive view of justification, but it is able to consider various possibilities that internalism blocks. In fact, externalism is always associated with ‘strong position’ conceptions of justification (believing in such a way as to be in a strong position to thereby get the truth). Since it can recognize conditions of justification that are not reflectively accessible, it is able uninhibitedly to carry out the idea that justification entails truth-conducivity.

The most prominent form of externalism is reliabilism, and in its most common form it holds that a belief is justified if and only if it was produced and/or sustained by a reliable belief-forming process (Goldman 1986) (see Reliabilism §2). A particular perceptual belief, for example, is justified provided it was formed from sensory experience in such a way that when a belief with a content like that is formed from an experience related in that way to the belief content, it will usually be true. A belief formed by inferring a fact from observable indications (believing that a party is going on because one notes a lot of lights on and hears a lot of noise) is justified provided beliefs with that sort of content formed by indications like that will generally be correct. Since the truth-making tendency of a certain process is not something we can ascertain just on reflection, access internalists are not in a position to embrace reliabilism, but externalists are free to do so.

There are many problems one encounters in working out a reliabilist view in detail. For example, how does one assign a particular belief-forming process to a class or type of processes? (note the use of ‘like that’ in the above examples) (Feldman 1985). And is the reliability a matter of actual track record or a matter of what would happen in situations of certain kinds? I will just note here that some views of this kind are formulated in terms of probability rather than in terms of processes. Thus the second example could be put in terms of the probability of a party occurring given that there is a lot of noise, and so on (Swain 1981; Dretske 1981) (see Information theory and epistemology; Probability theory and epistemology §1).

Since externalism amounts to a denial of internalism, it is to be expected that its main supports come from the difficulties in internalism. The main attraction of externalism lies in the facts that: (1) it enables us to retain the truth-conducivity implication of justification, as internalism does not; and (2) it does not require us to make exaggerated claims for the powers of rational reflection.

The chief criticisms of externalism are based on internalist intuitions and are typically directed against reliabilism. First, there is the consideration of ‘demon worlds’, worlds controlled by an evil omnipotent demon who sees to it that our beliefs are generally false, even when they seem most obviously true. In such a world one’s perceptual, inductive, and mathematical beliefs are false even though one has the same bases for them that we have in the actual world. The denizens of this world believe that there is a tree in front of them when they seem to be seeing a tree in front of them. They believe that 3 + 2 = 5 because it seems self-evident to them, and so on. There is a considerable tendency to judge that these unfortunates would be justified in their beliefs, just as much as we are in ours, even though they are formed in a very unreliable manner. Since they have the same grounds as we do, how can our beliefs be justified and theirs not? But then, reliability of formation is not necessary for knowledge (Foley 1987: Ch. 3). Although this argument is couched in terms of a fantastic scenario, the underlying idea is simple and sober. Since it is conceivable that we have what are recognized as very strong grounds for a belief even though it is not formed reliably, reliability cannot be necessary for justification.

Second, consider someone who forms beliefs in a reliable way but who has no, or insufficient, reason for supposing it to be reliable. Again, the argument is frequently presented in terms of outré examples like clairvoyance (BonJour: Ch. 3), but more standard cases will do as well. Suppose Jim infers emotional states from outward demeanour and behaviour in the standard way, but has no reason for thinking this to be a reliable way of making such judgments. There is a strong tendency to think that he has no justification for the beliefs so formed, given that this mode of belief formation could be completely unreliable so far as he can tell. But then the mere fact that the way of forming beliefs is reliable cannot be sufficient for justification.

The best counter-move for the reliabilist is to deny the ‘intuitions’ adduced by the critic. No doubt the demon world denizen has something going for them, epistemically, but, says the reliabilist, it is not justification, so long as there is nothing about the belief and the way it is formed or held that makes it likely to be true. And since the reliability condition is satisfied for Jim vis-à-vis beliefs formed in the usual way about the emotional states of others, that will suffice to render the beliefs justified, despite the lack of higher-level knowledge (justified belief) concerning how reliable that way is. This is to take truth-conducivity as both necessary and sufficient for justification and, as we have seen, this is the heart of epistemic justification from an externalist perspective.

What if Jim not only is not justified in supposing this way of forming beliefs to be reliable, but is justified in supposing it to be unreliable (even though it is reliable)? In that case the reliabilist could recognize Jim’s beliefs about the emotional states of others to be unjustified – by taking reliability to be sufficient only for prima facie justified belief, belief that is justified provided the justification is not overridden in one way or another. In this case, Jim’s justification for supposing that way of forming beliefs to be unreliable would be a sufficient overrider. To modify reliabilism in this way would not involve abandoning the basic thrust of the position. Reliability of belief formation would still be at the centre of the picture. It is just that other considerations are given a secondary role.

Various blends of internalism and externalism have been suggested. The distinction between ‘justifier’ and ‘justificatory efficacy thereof’ provides an obvious basis for such a combination. Thus Alston (1989: Ch. 9) develops a view (‘Internalist Externalism’), according to which a condition can be a justifier only if it is the sort of thing that is typically available to reflection (this is the internalist part), but there is no such reflective access required for the justificatory efficacy of the condition. That efficacy depends on whether the alleged justifier is in fact a reliable indication of the truth of the belief in question (this is the externalist part).