‘Metaphysics’, said Bradley, ‘is the finding of bad reasons for what we believe on instinct, but to find these reasons is no less an instinct.’ This idea that reasoning is both instinctive and feeble is reminiscent of Hume; except that reasons in Hume tend to serve as the solvent rather than the support of instinctive beliefs. Instinct leads us to play backgammon with other individuals whom we assume inhabit a world which exists independently of our own perception and which will continue to exist tomorrow in a similar fashion to today. However, when instinct leads us also to reason about these beliefs they are all subject to sceptical attack. Their defence provides a challenge, a challenge which in thumbnail histories of the subject is met by Kant. He does this by use of a powerful new form of argument which he calls transcendental argument and which, in my opinion, provides not only reasons but also good reasons for the defence of some of our most central instinctive beliefs. The strategy involved in this kind of argument is to reflect on the necessary preconditions for comprehensible experience. In this way, some beliefs which are subject to sceptical attack, such as that there is a causal order between objects which exist independently of our experience of them, can be found to be the essential preconditions for having comprehensible experience at all. The reason for accepting them is, therefore, that they are the necessary preconditions of having any beliefs at all; and this provides a good, rather than a bad, reason for accepting these particular instinctive beliefs.