Some Counsel on Humean Relations

48 SOME CQUHSEL ON HUMEAN RELATIONS In a paper published eight years ago I tried to bring out a neglected feature of Hume's theory of relations, namely the difference between philosophical and natural re1 2 . lations. Now Ijnlay , without referring to my work , has expanded some of its themes in an extremely interesting and, I think, important way. At least he has made me rethink the whole distinction between philosophical and natural relations and, what goes with relation theory, Hume's analytic-synthetic distinction. The strengths of his paper are, embarrassingly, the weaknesses of my own. I did not articulate clearly the connection between relation theory in Hume and the analytic-synthetic distinction. I shall turn to these strengths presently. My focus in this paper, however, will be on a central weakness in Imlay 's; and the point, it turns out, is one I gave full credence to earlier. Imlay does not distinguish what in Hume can and must be distinguished if we are to make any sense (let alone full sense) of his theory of causation. The distinction is be3 tweer, necessity and certainty . In trying to explicate Hume's distinction between intuitive and demonstrative inference on the one hand, and probability on the other, Imlay finally reaches an impasse. For, if the absence of the corresponding impression does not deprive mathematical necessity of... existence, why should it do so in the case of causal necessity? This, as Imlay recognizes, is at least related to standard criticisms of Hume, most forcibly articulated by Prichard. Imlay fortunately does not recommend—as Prichard does — throwing out the baby with the bath. If I understand him correctly, what he advocates is that we simply admit that Hume's attempts to ground necessity are inadequate. Before proceeding to details, let me issue two brief caveats. 49 First. It is of course not surprising that Hume fails to make a clear analytic-synthetic distinction. Has anyone? Lacking the logical tools of a Russell, or the mathematical understanding of a Leibniz, Hume labors under insuperable difficulties which, in many instances, are selfcreated . He is not clear as to whether necessity attaches to 'objects' or ideas (let alone sentences or propositions); he confuses, by invoking pseudo-psychological criteria involving conceivability, analytic with synthetic a priori truths; he has what many, including myself, consider to be a ludicrous theory about mathematics and mathematical objects . With these horrors to work with it is small wonder that necessity never gets grounded. Indeed, it is miraculous that the discussion even gets off the ground. But, Prichard ' s sneeringly just comments notwithstanding, both he and Imlay miss something, and I hope to show precisely what that is. Second. Hume is notoriously obscure in his discussion of what we would call ordinary physical objects. Sometimes (when he is in a psychologist or even ordinaryperson mood) he speaks as if physical objects cause impressions . Sometimes (when in the philosopher's closet) he speaks as if it makes no sense to speak of anything but impressions and collections of them, ordinary physical objects being relegated to the fictions of the vulgar. His theory of relations rather obviously suffers from the same disease; it is not clear what the relata are. Are impressions related temporally? Spatially? Is it relations between objects 'out there' of which he speaks when he discusses causation? Imlay is quite aware of the confusion ; he just lets it go. For now, I shall more or less follow suit. Later, as we shall see, one's choice makes a difference. In the long run, I think any coherent view of Hume will have to treat his analysis of physical objects in a way similar to Berkeley's · I. Hume, Prichard, and Imlay At first glance it would seem easy to allay at least Prichard' s fears. What, after all, is the connection between Hume's theory of relations, his attempt at distinguishing the necessary from the probable or possible, and his theory of causation? Hume claims that causation is a relation that, unlike some others, eg., resemblance, cannot be 'seen' to hold between its relata. But all of us believe that causal connections are necessary. Why do we all believe what...