R. G. Collingwood and the Historical Stance

I'm spending a lot of my time at the moment wondering what Statistical Learning Theory (see John Langford's informative blog Machine Learning) can teach philosophy. Here's a small sample of possibilities: (1) There's more to learning than finding the generative mechanism producing observed data; (2) Choosing a hypothesis and then testing it is not the only way to gain confidence in its accuracy; (3) Complicated averages of hypotheses are often the more accurate; (4) It is possible to factor quantitively into the expected accuracy of a hypothesis how well other hypotheses in the class under consideration perform during training.

Parts of philosophy concerned with empirical inference can expect to be transformed by integrating these ideas. Here, as elsewhere, this should give us pause for thought. What to do when science starts to deal with traditionally philosophical topics? Judea Pearl (Causality, CUP, 2000) integrates some of the research of philosophers (e.g., David Lewis) on causality with ideas from statistics and graphical modelling into a powerful calculus for treating causal predictions and counterfactuals. What next? Must the philosopher wishing to continue thinking about causality take on board Pearl's work? Consider other traditional philosophical topics, such as time, which have been treated by other disciplines for longer. Must a philosopher who wants to discuss time keep abreast of the latest physical theories, the latest neuropsychology of the perception of time, or the latest historical research on the transformation of the time of everyday life by, say, the time-keeping of the monasteries or the advent of national railways?

It seems to me that we need a conception of philosophy as inseparable from the historical study of forms of thinking and acting, which can be expected to continue to develop. A philosopher who worked out such a position was the Waynflete Professor of Metaphysical Philosophy at Oxford Univerity, R. G. Collingwood (1889-1943). Here he is on logic:

Logic an Historical Science
The aim of logic is to expound the principles of valid thought. It is idly fancied that validity in thought is at all times one and the same, no matter how people are at various times actually in the habit of thinking; and that in consequence the truths which it is logic's business to discover are eternal truths. But all that any logician has ever done, or tried to do, is to expound the principles of what in his day passed for valid thought among those whom he regarded as reputable thinkers. This enterprise is strictly historical. It is a study in what is called contemporary history = history of the recent past in a society which the historian regards as his own society...Logic as 'theory of scientific method' is in effect, at any given time, a fragment of a history of scientific method. (The Principles of History, pp. 242-3)

We might take the recent results of statistical learning theory to be such a fragment.

Collingwood extended this historical interpretation to the whole of philosophy, including metaphysics. Metaphysics flourishes today, not least through the attention of the late aforementioned David Lewis, in a way that Collingwood would have seen as wrong-headed. Modern metaphysicians aim to give timelessly true accounts of identity, time, cause, necessity, property, etc. They thus become targets for empirically-inclined philosophers. Let's consider a contemporary critic of metaphysics - Bas van Fraassen. Van Fraassen in his recent book outlines what he calls 'The Empirical Stance'. Now, a stance is something like a philosophical orientation. It isn't composed of a set of factual beliefs, but rather a kind of epistemic 'policy'. E.g., the 'Metaphysical Stance', the target of his first chapter, advocates:

M1: Accept demands for explanation in terms of things underlying the phenomena.
M2: Attempt to answer such demands for explanation by postulation.

The Empirical Stance has been summarised by Anjan Chakravartty as follows:

E1: Reject demands for explanation in terms of things underlying the phenomena.
E2: A fortiori, reject attempts to answer such demands for explanation by postulation.
E3: Follow, as a model of inquiry, the methods of the sciences.

Notice how E’s position is defined in terms of M's, as a reaction. In the battle between stances, the only resource van Fraassen seems to have is E3 which he uses to deny that metaphysicians are behaving like scientists. This is just what he does. But it's hardly a killer punch.

Van Fraassen describes the metaphysician as engaging in at least one of the following two tasks: (1) Postulating entities behind phenomena; (2) Reading off the ontological commitments of theories. But to set the term of the debate in this way is surely to reason ahistorically if it is the case that metaphysics was once a different activity. Consider again Collingwood:

I have tried to dispel certain misconceptions about it [metaphysics] which have led (and, had they been true, would have led with perfect justice) to the conclusion that metaphysics is a blind alley of thought into which knaves and fools have combined these many centuries past to lure the human intellect to its destruction. (Collingwood, Author’s preface, p. civ, An Essay on Metaphysics, OUP, revised ed. 1998)

His way of dispelling such misconceptions takes him back to Aristotle:

In writing about metaphysics it is only decent, and it is certainly wise, to begin with Aristotle.” (ibid., p. 1)

For Collingwood, Aristotle has two conceptions of metaphysics, the first he considers as misguided:

(i) Metaphysics is the science of pure being.
(ii) Metaphysics is the science which deals with the presuppositions underlying ordinary science.
Adopting the second of these conceptions, he can then claim

... there are no ‘eternal’ or ‘crucial’ or ‘central’ problems in metaphysics. (ibid., p. 72)

How can metaphysics become a science? By becoming more completely and more consciously what in fact it has always been, an historical science. (ibid., p. 77)

It is important to note that Collingwood uses the term 'science' with a wider scope than is the norm today. For example, history for him is counted a science. This is in contrast to van Fraassen who takes it in the contmporary sense of natural science.

The metaphysics of an Aristotle is altogether different from that of a David Lewis. For Aristotle,

...no science examines the principles which are a presupposition of its having a subject matter to study; e.g. geometry does not consider whether there are points nor arithmetic whether numbers exist. These are questions for another study, which Aristotle calls first philosophy (metaphysics). But he thinks of this higher study as delivering conclusions which the sciences subordinate to it can use as first principles. Whereas twentieth-century philosophy has usually thought of science and metaphysics as quite distinct kinds of inquiry (because in our world they usually are), for Aristotle natural philosophy is simply 'second philosophy' (e.g. Metaphysics 1037a 14-15). It is a less abstract and less general enterprise than first philosophy, because it deals with one part of the subject matter of first philosophy, and secondary to it, because first philosophy has access to the ultimate principles of explanation (Metaphysics E. 1). That is all. (Miles Burnyeat, 'The sceptic in his time and place', in R. Rorty et al. (eds.) Philosophy in History, CUP, 1984: 246-7)

The rise of disciplinary boundaries between philosophy and the natural sciences, and the increasing implausibility that philosophers could dictate first principles to scientists, has led to metaphysics becoming an inquiry which aims to tell a story of how the world must be for science to work as it does, but with no pretensions to affect science. Van Fraassen's complaint is that it models itself on science - employing inference to the best explanation, etc. - but without exposing itself to any of the same kinds of risk that a scientist does. As we have seen, an alternative view of metaphysics, e.g., that of R. G. Collingwood, is closer to Aristotle's in taking it to be the study of the presuppositions of the sciences (taken in the broad original sense). A philosopher's contribution to the metaphysics of a discipline is not restricted to their arguing against, or finding tensions amongst, current presuppositions, although this does typically take place in philosophy of physics. (Some contemporary work in philosophy of physics is very much closer in outlook to Collingwood's 'The Idea of Nature' than it is to analytic metaphysics.) One can also help by providing a scaffold to support discussions of fundamental issues, frame debates, provide historical insight, make comparisons to other fields, etc.

Collingwood extends his historical stance to ethics, currently a boom industry in philosophy

Ethics as an Historical Science
(a) Ethics as an account of the principles of action depends for its content on the structure of the moral world of which it tries to give an account. Thus ancient Greek ideas of conduct are different from Christian ideas and consequently Aristotle's ethics (say) differ[s] widely from any seventeenth- to twentieth-century ethics, without this implying error on either side. Any ethical theory is an attempt to state what kind of a life is considered worth aiming at, and the question always arises-by whom? (b) There are departmental ethical sciences like politics, economics. These, at any given time and place, describe the political and economic principles accepted at that time and place. For economics, this has been seen by the Marxists, and it has been admitted by J. M. Keynes, with the odd result that he has tried to construct a 'general' economic theory, stating the supposedly permanent general principles of which any 'special' economic theory, like Adam Smith's, is a special case. This of course is illusory. (c) Even the distinction between logic and ethics is an historical one and no more. As we inherit it from the Greeks, it certainly has no permanent validity: the Indians or the Chinese do not make the distinction between thought and conduct in any such way as that which we presuppose when we make it. (The Principles of History: 249)

And this brings us to Alasdair MacIntyre who agrees:

From a methodological point of view, it is today clear to me that while I was writing A Short History of Ethics I should have taken as a central standpoint what I learned from R. G. Collingwood: that morality is an essentially historical subject matter and that philosophical inquiry, in ethics as elsewhere, is defective insofar as it is not historical. (A. MacIntyre, 'An Interview with Giovanna Borradori' in The MacIntyre Reader, Kelvin Knight (ed.), Polity Press, 1998: 261)

This point of view he adopted in 'After Virtue', 'Whose Ratioanlity? Whose Justice?' and 'Three Rival Versions of Moral Enquiry'.

Is this Historical Stance plausibly the one to adopt as regards mathematics? Surely we don't need to think of '2 + 3 = 5' as something in need of historical treatment, but rather should consider it in timeless fashion in a way orthogonal to the interests of the mathematician by answering the questions: To what kind of thing do the terms of the proposition refer? What kind of truth does it express? Well, I would argue we should. Insofar as we are considering this statement in the context of research mathematics, there's an awful lot behind even something as simple as this, and yet more behind 2 + 3 = 3 + 2, as you can see from the first of John Baez's talks here. Our conceptual understanding of the integers is still changing. And if you really want your intellect to be adequate to the integers, you'd better gain enough background to understand this claim 'The sphere spectrum is the true integers', even if it is only to criticise it from another perspective within mathematics. For an inkling of what André Joyal means by this claim try week 102 of Baez's This Week's Finds.

I'll leave the last word to Collingwood. Writing in 1935 about the conception of nature worked out by philosophers such as Bergson, Alexander and Whitehead, he remarks:

As in the time of Descartes, so again today physics and metaphysics are working hand in hand, and one of the most remarkable features of this new cosmology is just this fact: the fact that the separation of philosophy from science, of which we have been so long been conscious that we have come to regard it as a necessary evil, has disappeared. The results of this new situation ought to be extremely fertile both for science and for philosophy, strengthening and enriching both of them: and perhaps it is not too much to hope that the alliance and cross-fertilization of these two streams of thought will have a beneficial effect on the future of civilization, which has suffered greatly in the last few generations from the fact that those who ought to be its intellectual leaders have spent much of their time in a mutual warfare, damaging to the prestige of both sides and of no advantage to either. (The Principles of History: 253)