Browsing in the library I recently stumbled across Edward O. Wilson‘s ambitious book Consilience, in which the famous biologist tries to propose a unified account of knowledge – one in which the natural sciences take a supreme role. I have a certain sympathy with Wilson’s project – his attempt to unite the different realms of human knowing is not so different from my own long-term goal of constructing a philosophy that draws from many (and preferably all) traditions of inquiry. It’s unfortunate, then, that Wilson’s attempt falls flat in a wide variety of ways – but they can, at least, be highly instructive ways. They do very well, especially, at illustrating the limits of an empiricist account of knowledge.

The example that first struck me is Wilson’s critique of Immanuel Kant’s metaethics. Wilson characterizes Kant’s ethics as follows:

Nature, Kant said, is a system of cause and effect, while moral choice is a matter of free will, for which there is no cause and effect. In making moral choices, in rising above mere instinct, human beings transcend the realm of nature and enter a realm of freedom that belongs to them exclusively as rational creatures. (Wilson 248-9)

Wilson isn’t wrong in his characterization so far. But he goes off the rails when he says Kant’s view is wrong because it “does not accord, we know now, with the evidence of how the brain works” – and then adds “I find it hard to believe that had Kant, Moore and Rawls known modern biology and experimental psychology they would have reasoned as they did.” (249) (There’s no room here to talk about Wilson’s placing of G.E. Moore in the same category as Kant.) If there is any doubt that Kant would laugh heartily at such a characterization, it would only be because Kant didn’t laugh much in general.

Kant’s system is a system of transcendental idealism, which means that, like Anselm, most of his arguments proceed a priori, independent of sense knowledge. The whole idea is that they’re not arguments that can be proven true or false by empirical evidence; rather they are part of the conditions that make empirical knowledge, and action in the world, possible in the first place. Wilson’s argument in the preceding pages against a transcendentalist view, alas, shows no awareness of any of this; it is merely an argument against a popular kind of religious belief, one which he admits is “the one I first learned as a child in the Christian faith.” (Wilson 248)

On the matter of free will, Kant already knew, three hundred years ago, that if we study the patterns of human behaviour (whether using the methods that would become biology, psychology or sociology), we can make modestly reliable predictions. His point is that from such a standpoint our own behaviour is unintelligible. The very idea of my making an action, making a choice or decision, makes no sense whatsoever unless I consider my action to be free of causal determination, and consider it instead to be my own. This is all true even though we can study human behaviour in ways that make it predictable; all such study approaches our behaviour from a third-person standpoint, but for our own actions to be intelligible we must approach them from a first-person standpoint.

There are reasonable ways to rebut Kant’s argument here; I don’t think the argument is ironclad. But Wilson never suggests that he has understood it, or even attempted to do so. And I think Wilson’s lack of understanding is telling: he just doesn’t seem to get the idea that something might be true without being empirically testable. He’s inheriting an empiricism from David Hume, who famously said that if any statement is neither a definition nor empirically testable then we should “commit it to the flames”; but he doesn’t seem to realize that Kant’s whole philosophy was a reply to Hume’s, an attempt to show Hume wrong.

One can try to defend Hume against Kant, but to do so one must take Kant on on his home turf of logical argument; one can’t refute him with empirical evidence, because the whole point is to go beyond empirical evidence. There is a drastic flaw in the basic idea that only empirically testable statements or definitions are meaningful: that basic idea is itself not empirically testable. Therefore, by its own standard, either it’s a mere definition of “meaningless” (and therefore a semantic quibble) or it’s meaningless. It therefore is quite straightforwardly self-contradicting, and is not worthy of sustained consideration as a candidate for truth. We need standards for truth and meaning beyond empirical verifiability or falsifiability. Kant knows this, and his whole philosophy proceeds from this understanding.

Wilson is correct to say: “Sometimes a concept is baffling not because it is profound but because it is wrong.” In this case, however, Wilson was clearly baffled not by Kant’s wrongness, but rather by his profundity.