Inference to the best explanation

With the passing of Peter Lipton, who formulated and coined the term “inference to the best explanation“(IBE) I was reminded that I once “controversially” (at least to some, though not philosophers of history) deployed the term in a paper to do with the philosophy of history. When I spoke to Peter about it, he couldn’t see any objection to my usage. I restate the relevant excerpt here. (The dry-wall analogy which I refer to is explicated here).

If we consider how Oakeshott conceives, in his famous phrase, ‘the activity of being an historian’, we see a non-coherentist account of justification and truth at work. To avoid the problems of coherentism let’s try a different interpretation albeit a somewhat controversial one: the anticipated objections will be considered later. C. Behan McCullagh (The Truth of History, London and New York: Routledge, 1998, p. 46) outlines what he terms the “correlation” theory of justification and truth. In it I find nothing with which Oakeshott would disagree. Now here is the controversial aspect: I take it to be a form of inference to the best explanation (IBE).

IBE holds that we have sufficient reason (i.e. justification) for accepting that hypothesis which, if true, would best explain x, where ‘x’ is some available evidence that presents a problem of intelligibility. Its logical form is:

X (evidence to be explained)Y (hypothesis which, if true, would best explain X)


Therefore Y

Note that IBE is a form of non-deductive inference; the premises probabilify and do not necessitate the conclusion. We accept Y because it is the best explanation of X available to us; it may still be false.

Now, of course, a whole set of questions immediately presents itself as to what constitutes the ‘best explanation’. The matter cannot be fully discussed here; elucidation can be found in Peter Lipton’s standard text (Inference to the Best Explanation. London and New York, Routledge,1991).

We infer to the best explanation regularly in science, history, and practice. It is formally elusive, indeterminate in its technical expression, but easily recognisable in specific examples. Jack has never liked Jill but suddenly becomes affable towards her. Jill starts to receive invitations to Jack’s parties; Joan also sends Jill the occasional solicitous email; Jack asks Jill her opinion on a range of matters and listens carefully to her views. How best to explain this turn of events? We discover that Jill is standing for election to a committee which is likely to be divided on her candidature and on which Jill is likely to have a casting vote. So we infer that Jack has become affable towards Jill in order to secure her vote. From our knowledge of all concerned, this is the best explanation. It may be wrong; perhaps Jack has undergone a moral conversion. But we have no evidence, outside this episode, of any such conversion. If further evidence becomes available, the best explanation may change.

So far as I can make out, this is very much Oakeshott’s approach to the nature of both historical and scientific explanation. It is hard to see how else, in science, he could explain why:

The image of a stationary earth is replaced by that of a stationary sun, iron dissolves into an arrangement of electrons and protons, water is revealed to be a combination of gases and the concept of undulations in the air of various dimensions takes the place of the images of sounds (Rationalism in politics, pp.504-505).

These images changed because they provided or supported, according to the evidence available, the best explanation of a range of problems. And the image of the dry-wall, invoked in his later accounts of historical explanation , is exactly apt for IBE. We infer the hypothesis that would, if true, provide the best explanation of the available evidence. We build the wall (infer the historical hypothesis) that best fits the stones together (explains the available evidence). (Oakeshott’s “dry-wall” analogy has some resonance with Haack’s crossword analogy of scientific justification – her so-called Foundherentism which allows the relevance of experience to empirical justification without postulating any privileged class of basic beliefs or requiring that relations of support be essentially one directional).

Two objections may be expected to this account of Oakeshott. The first is that it commits the fallacy of supposing that, because IBE fits well with (much of) what Oakeshott says, that therefore he accepts the model of IBE. The reply to this is that we know that Oakeshott cannot be a correspondence theorist about justification, at least with respect to historical explanation, because our historical explanations cannot correspond to an inexistent past. If Oakeshott does not subscribe to IBE, then it would be interesting to know what presents itself as a probable alternative, if correspondence is certainly out of the question and coherence were not in play.

Even if Oakeshott were an IBE theorist, relativism returns to haunt him. This is because such inference is indexed to a given subject – an individual mind or a collectivity of minds – at a given time. A dilemma arises for IBE. If it allows for ethical, political, and social justification, then:

(a) it must affirm the empirical and conceptual possibility that different minds or collectivities of minds – or let us say ‘different persons’, which is a more natural phrase here – may justifiably accept ethical, political, and social beliefs and activities which, when universalised, are inconsistent. That is the logic of the IBE model. Or,(b) it must exclude the idea of ethical, political, and social justification. This would certainly avoid relativism in these areas.

On a clarificatory point: in the cases of IBE justification considered above, we focused on justification in believing that X (believing that something is the case), which may yield knowledge that X. It is clear that, on Oakeshott’s account, justification will operate differently in ethical, political, and social action as involving justification in decision-making or practical reasoning, in deciding how to act, as well as justification in what to believe. This is why the dilemma refers to ‘beliefs and activities’. In the background is Ryle’s epistemological distinction by which it is widely agreed that Oakeshott was influenced. But the question of justifying practical reasoning still applies.  

From “Constructivism and Relativism in Oakeshott” in The Intellectual Legacy of Michael Oakeshott