Lipton’s Inference

Peter Lipton Inference to the Best Explanation in W.H. Newton-Smith (ed) A Companion to the Philosophy of Science (Blackwell, 2000) 184-193.

[Our] current understanding of inductive reasoning remains remarkably poor … .

The model of Inference to the Best Explanation is designed to give a partial account of many inductive inferences, both in science and in ordinary life. … Its governing idea is that explanatory considerations are a guide to inference, that scientists infer from the available evidence to the hypothesis which would, if correct, best explain that evidence.

There are two different problems that an account of induction in science might purport to solve.  The problem of description is to give an account of the principles that govern the way scientists weigh evidence and make inferences.  The problem of justification is to show that those principles are sound or rational, for example by showing that they tend to lead scientists to accept hypotheses that are true and to reject those that are false. …

What is wanted is thus an account that permits vertical inference without permitting absolutely everything, and Inference to the Best Explanation promises to fill the bill. …

The difficulty of articulating Inference to the Best Explanation is compounded when we turn to the question of what makes one explanation better than another.  To begin with, the model suggests that inference is a matter of choosing the best from among those explanatory hypotheses that been proposed at a given time, but this seems to entail that at any time scientists will infer one and only one explanation for any set of data.  Yet scientists are sometimes agnostic, unwilling to infer any of the available hypotheses, and they are also sometimes happy to infer more than one explanation, when the explanations are compatible.  `Inference to the Best Explanation’ must thus be glossed by the more accurate but less memorable phrase, `inference to the best of the available competing explanations, when the best one is sufficiently good’.  But under what conditions is this complex condition is satisfied?  How good is `sufficiently good’?  Even more fundamentally, what are the factors that make one explanation better than another?  …

[Should] Inference to the Best Explanation be construed as inference to the likeliest explanation, or as inference to the loveliest explanation? …

The model should … be construed as `Inference to the Loveliest Explanation’.  Its central claim is that scientists take loveliness as a guide to likeliness, that the explanation that would, if correct, provide the most understanding, is the explanation that is judged likeliest to be correct.

Unfortunately … Reports of past observation will never entail that future inferences to the best explanations will in fact select true hypotheses; and any argument that the reliability of inference to the best explanation would itself be the best explanation of what we have observed begs the question.

Even if the model is of no avail against a complete inductive sceptic, it might have a role to play in the defence of scientific realism, according to which there are good reasons to believe that well-supported theories are likely to be at least approximately true, against positions such as constructive empiricism, according to which we can only have reason to believe that our best theories are empirically adequate, that their observable consequences are true. …

Neither of the justificatory applications of Inference to the Best Explanation we have considered appears promising. …


Loveliness seems an unpromising criterion, even if some group happens to agree on it. It is also not clear to me why we should always select ‘the’ best, or why we should seem to assume that in the long-run the best will be among those considered.

Consider an earth-centric and sun-centric cosmologies. Which is ‘best’? Does it matter if their predictions of eclipses agree? When a relativistic theory comes along, do we have to select the best right away, or might we not wait for diagnostic observations?

It seems to me that we might often identify a favourite theory as ‘the’ best, but we should always give some credence to potential alternatives even if they do not seem entirely lovely to us. This view  resembles constructive empiricism (a variation on logical positivism).


Think of driving a car, or any other activity that is potentially dangerous. It seems to me that it would be very dangerous to form a single possible model of the situation and then act according to it, ignoring all possibilities that are not the ‘best’ in some sense, such as the loveliest or most likely. Given that accidents are fairly rare, I suppose that people consider possibilities, particularly the dangerous ones, that fit the observations, and then try to ‘hedge their bets’, e.g. by slowing down. Why is  a less thoughtful approach assumed to be adequate for scientific enquiry?

See Also

Stanford Encyclopaedia on abduction.

Dave Marsay

%d bloggers like this: