However the honours are finally shared between active and passive, we have asserted the existence of a domain of rules, intentions and reasons and of actors who try to do the rational thing. But we must not assert in ontology what, as epistemologists, we could not possibly know to be true. It is time to pay our epistemological debts by honouring those promissory notes on the bank of Reason.

The debts will be paid with a defence of rationalism. So let it be clear that rationalism has three distinct senses. The blandest is marked by a broad belief that there is the sort of order in experience which makes science possible. J. S. Mill was once described as ‘the high priest of rationalism’ in this sense, without impugning his robust empiricism. It would be a motley alliance got by lumping together all who thought science possible and their disagreements about the kind of order required would be more instructive than their agreement on the fact of order. Only sceptics, romantics, mystics and a few other champions of the random, the fragmentary, the spontaneous and the ineffable would be debarred. But I mention this sense of rationalism not from mere nostalgia for the rationalist societies who used to sneer at the Trinity in damp, gaslit halls. There is also a present danger in radical critiques of orthodox social science, in the heady relativism brought on by intoxication with paradigms and in recent fulminations against method. It is the danger of so undermining claims to objective scientific knowledge that not even rational preference for one conclusion against another is left. I shall try briefly to stave it off later in the chapter.

Secondly ‘rationalist’ is sometimes used to label theories of action which rest on assuming rationality in men. There are many ways to construe the assumption and this book endorses one of them. Such theories place some sort of divide between natural and social sciences and we shall ask where and how deep it should go. They also prompt the thought that the social sciences are somehow normative. But it is one thing to say that norms are an ineliminable part of the subject matter, another to deny that the social scientist can have a value-free standpoint and yet another to detect normative implications in the findings of social science.