Hume, science, and religion

Suppose you buy Hume’s famous analysis of causation, and thus deny that we can have any knowledge of objective causal connections in nature (either because there aren’t any – the traditional, “verificationist” interpretation of Hume – or because there are but the mind can never genuinely know or understand them – the newer “skeptical realist” interpretation). You shouldn’t buy it (for reasons set out in The Last Superstition), but suppose you did. It is understandable why, in that case, you’d reject First Cause arguments for God’s existence. If we can’t have any knowledge of objective causal connections between things, then we can’t have knowledge of a First Cause.

But how in that case could you fail to reject modern science as well? Wouldn’t theism and natural science – which seems to be in the business of discovering objective causal connections between phenomena – stand or fall together? A way around this might be to adopt some kind of non-realist interpretation of science. You could take the instrumentalist view that scientific theories don’t really tell us anything about the nature of things, but are merely useful tools for predicting experiences.

One problem with this response is that non-realist interpretations of science are just implausible. As Hilary Putnam famously put it, realism is “the only philosophy that doesn’t make the success of science a miracle.” Another problem is that it arguably wouldn’t really address the heart of the objection anyway, but only push the problem back a stage. For if you are going to treat scientific theories as tools for predicting experiences, then it seems you are presupposing that at least one thing in the world – the experiencing human mind itself, or your own mind anyway – manifests causal regularities that can be captured in scientific theories. Even as a Humean you will do so; for example, qua Humean you will suppose that ideas come from, and can only come from, antecedent impressions; that they are combined only in accordance with the three Humean principles of association; and so forth. These seem precisely to be causal regularities, whose existence guarantees that there is an orderly set of experiences for an instrumentalist science to describe and predict.

Of course, this sits poorly with Hume’s own analysis of causation. If (as Hume claims) any effect might in principle follow from any cause or from none at all, where does he get off telling us that every idea (apart from the famous “missing shade of blue,” anyway) must derive from an antecedent impression? If there are no objective “necessary connections” between causes and effects – the very idea of necessary connection being a mere projection of the mind’s subjective expectation of A on the occasion of B, a propensity produced by observed constant conjunctions of A and B in the past – then why exactly must ideas behave only in accordance with the principles of association? Indeed, even the notion of a “propensity” to expect such-and-such an effect, and of this propensity’s being “produced” in the mind by constant conjunction, are themselves causal notions. So does Hume mean to say that all of these claims about the mind and about causation are themselves mere projections based on nothing more than observed regularities in the order of our impressions and ideas, and have no objective validity? Presumably not; indeed, the very idea seems not only implausible, but utterly incoherent. All the same, consistently pushing through a radical Humean skepticism about causation would seem to require applying Hume’s analysis of causation to Hume’s own claims about the mechanisms that govern the mind. And while this would certainly undermine First Cause arguments, it would also undermine science too – precisely because it would undermine almost every claim to knowledge, including Hume’s own theory. The traditional, “radical skeptic” reading of Hume leaves us with a snake that eats its own tail.

A more promising strategy for the Humean who wants to accept science while rejecting natural theology would be to take a softer “New Hume” or “Hume as skeptical realist” line, and deny that the Humean analysis of causation really entails taking a non-realist approach to science in general or causation in particular. One could hold that Hume’s position merely entails that we cannot strictly know or understand objective causal connections, but not that there are no such connections. The idea would then be that Hume’s view that our belief in objective causal relations is impervious to rational criticism anyway, since we can’t help but cling to it given our nature, also entails that the science we base upon this belief is something we can hardly bring ourselves consistently to doubt. Yet if causation might in fact have an objective basis even if we can’t know or understand it, so too would science. Hence we need not dismiss science, any more than causation, as an illusion. We may be unable either to justify it (rationally speaking) or to doubt it (psychologically speaking), but it doesn’t follow that a Humean has to regard our belief in it as either false or strictly meaningless. (I am not claiming that all of this is plausible or even coherent all things considered, either as an interpretation of Hume or as a defensible position in its own right. The point is just that this is a more promising position for a Humean to take if he wants to reject First Cause arguments but accept science.)

OK, so where does the rejection of First Cause arguments fit in? Why wouldn’t they, like science, survive a less radical “skeptical realist” Humeanism? If the mere possibility of objective (though unknowable) causes coupled with our animal faith in them lets empirical science off the hook, why not natural theology? The answer would seem to be that in the view of the Humean, and indeed of Hume himself, though we cannot know or understand objective causal connections, we can lay down criteria for determining which particular purported causal relations are likely to exist, if any causal relations really exist at all. And while the theories enshrined in “our best science” conform to these criteria, arguments for a First Cause do not.

Thus do we arrive at that long and dishonest (or at least woefully ill-informed) skeptical tradition of treating the traditional arguments of natural theology as if they were essentially little more than second-rate empirical scientific hypotheses, feeble exercises in “God of the gaps” reasoning (a tradition given aid and comfort by William Paley and his successors). And thus can the Humean have his scientistic cake and eat it too. For science is just an extension of what we cannot help but believe in “common life,” outside the philosopher’s study. Hence, despite its being rationally unjustifiable, science is OK; and natural theology would be OK too if only it were good science. On this view, it wouldn’t be Hume’s theory of causation per se that undermines First Cause arguments. Rather, the claim would be that considerations of parsimony, empirical adequacy, etc. make theism a less “probable” “hypothesis” than atheism.

The trouble is that, as I show in TLS (and I am hardly the first person to show it), the classical First Cause arguments are not empirical quasi-scientific “God of the gaps” arguments at all, but rather attempts at metaphysical demonstration. And the relevant metaphysics is the Aristotelian kind, which claims precisely to be doing nothing more than extending what we already take ourselves to know in common life. In particular, Aristotelian-Thomistic First Cause arguments attempt to show that the existence of a First Cause is a necessary precondition of there being anything like what common sense understands as “causation” in the first place. So, if for the Humean (or “New Humean”) our “common life” beliefs about causation (a) may well be correct, and (b) are legitimately held by us despite their rationally unjustifiable status, why may we not also accept the conclusion of such First Cause arguments? We are back once again to asking: If science is OK, why not natural theology?

The Humean may at this point object that such arguments still go beyond what an appeal to “common life” can possibly justify, insofar as they presuppose (a) an a priori metaphysical methodology, and (b) that we have a transparent grasp of the essences of things (and in particular of their causal powers). But while such an objection may have force against rationalist natural theology of the sort practiced by Leibniz – I’m not saying it really does, mind you, just conceding this for the sake of argument – it has no force against Aristotelian-Thomistic natural theology. For A-T does not argue a priori, and does not hold that we have in general a complete and transparent knowledge of essences; like the empiricist, the A-T metaphysician insists that knowledge of real existence must be a posteriori, and recognizes limits on our knowledge of the essences of things. Where A-T differs from empiricism is in refusing to collapse the distinction between intellect and imagination, between concepts and mental images – the cardinal sin of modern empiricism, from which all its many other errors follow. Hence A-T also rejects the nominalism that follows upon this chief error (or underlies it, depending on how you look at it), and rejects the radical skepticism about causation, substance, essence, etc. that follows in turn upon imagism and nominalism.

For this reason, while there is a tension between common sense and philosophy even on a “New Hume” view of causation – for given Humean epistemology and metaphysics, how can our common life understanding of causation be even possibly true, since causal connections become not merely unknowable but unintelligible? – there is no such tension in A-T. A-T really is what P. F. Strawson famously called a “descriptive” metaphysics, which leaves common sense intact, while Humeanism is “revisionary” to the core, thoroughly undermining common sense implicitly even when it pays lip service to “common life.” A Humean “skeptical realist” has, when outside his study, to feign complete ignorance of what he learned within it. Even the bare possibility that common sense might be right, the very intelligibility of causation, is ruled out when his methodology is taken seriously. No such pretense would be necessary on A-T even if the A-T philosopher were to entertain doubts about whether genuine knowledge is really possible, for there is nothing in his position (as there is in the Humean position) which casts doubt on the very intelligibility, and not just the knowability, of causes.

If we are really to take “common life” seriously, then, we have to take seriously the metaphysics which makes it even minimally intelligible, namely something like A-T metaphysics. And that means accepting (assuming they are otherwise unobjectionable, as I argue in TLS that they are) the First Cause arguments that follow from it. Conversely, to reject those arguments entails rejecting after all the idea that common sense is right even about the very possibility of objective causal connections – which means in turn rejecting even an “animal faith” justification of our commitment to science.

And that returns us yet again to the question we started out with: How can a Humean consistently accept science and yet reject First Cause arguments for the existence of God? The unavoidable answer seems to be: He can’t. As far as a consistent Humeanism is concerned, science and theism must stand or fall together. But then there is no good reason to be a Humean in the first place, and many good reasons not to be. So perhaps the question is moot.