Naturalism on the Rocks

The philosopher Michael Devitt defines naturalism as the epistemological thesis that all knowledge “is empirical, ‘justified by experience’…[so that] there is only one way of knowing” (“There is No A Priori,” 105). According to Devitt’s colleague Alex Rosenberg, naturalism “is now a dominant approach in several areas of philosophy — ethics, epistemology, the philosophy of mind, philosophy of science and, most of all, metaphysics” (“Why I Am a Naturalist,” New York Times, 17-Sep-11). And if it turns out that naturalism circumscribes reality, “the implications will be grave” for what human beings value. We might even have to do away with God.

No A Priori?

Let’s ask first how one might go about defending naturalism thus construed. In a recent paper “There is No A Priori,” Michael Devitt offers a shiny (if unoriginal) defense. He notes, at the start, that the claim that some knowledge is empirical is “overwhelmingly plausible” (105). And with that it is hard to disagree. The question, though, is whether all knowledge is like that. For Devitt’s part, he finds the thesis “attractive.” However, he is quick to recognize that there is a legitimate motivation for flatly rejecting it. Here’s the problem. Consider necessary truths—such true propositions as

(1)  The square root of 16 is 4.

(2)  Red is a colour.

(3)  Anything that has a property exists.

(4)  If a proposition p is necessary, then it is necessary that p is necessary.

These propositions, one thinks, are not only true, they’re necessarily true. They’re true and couldn’t have been false. Furthermore, it seems that we can directly intuit or see that this is the case without the aid of sense experience or fancy epistemological theories. If that’s right, we have a priori reasons for thinking (1)-(4) are necessary. This would show that naturalism is false, since then some knowledge would be non-empirical (a priori).

"Occult" Connections

According to Devitt, this is wrongheaded. First, he says, rational or a priori intuition is an “unexplained,” “obscure,” “mysterious,” “occult” notion. (We suppose that means Devitt doesn’t favour it.) We have no idea what it is. If (1)-(4) are objectively true, then each is part of the furniture of the world. If they’re also necessary, then necessity is part of that world too. And therein lies the problem, says Devitt:

What sort of link could there be between the mind/brain and the external world, other than via [sense] experience, that would make states of the mind/brain likely to be true about the world? What non-experiential [i.e., non-sensory] link to reality could support insights into its necessary character? (114).

In other words, if necessity were part of the furniture of the world, it would be an externally perceptible fact that could be known via sense experience. That’s just the way human knowledge works. As Devitt says, “beliefs are related to the experiences that the world causes” (ibid). But then his reason for thinking a priori intuition is an “obscure” notion is really just that it isn’t a sense experience. Well of course that is true; that’s just what an a priori intuition is: one not based on the senses. Here Devitt wears his naturalism on his sleeve, and begs the question in the process. For obviously, you can’t rightly puncture the motivation behind anti-naturalism by simply declaring in a loud voice that all knowledge is caused by external, physical facts. For that just is (or at least entails) naturalism itself.

Holistic Justification

There is a second reason Devitt looks askance at knowing necessities by rational (non-empirical) intuition. It’s simply unneeded. Here’s another way of thinking of the matter. The justification of a belief—even entire theories—is assessed holistically, that is, against the backdrop of “the whole of science” or our total “web of beliefs.” There are no foundational or immediately justified beliefs. Justification doesn’t transfer from basic to non-basic beliefs; rather, a belief is justified by virtue of its inclusion in an evidential system according to certain criteria (e.g., coherence, explanatory adequacy, ontological conservatism, and so on).

Now according to Devitt, even scientific laws are confirmed in this holistic way. So why should mathematical or logical laws be any different? If Devitt is right, the justification for a necessary truth will be (a) empirical (since it is tested ‘against the system’ like any other scientific claim), and (b) indirect (because it derives from the evidential system of which it is a part). And therefore, each of (1)-(4) will also be revisable. Thus, as Quine taught us, given holism

no statement is immune from revision. Revision even of the logical law of excluded middle has been proposed as a means of simplifying quantum mechanics; and what difference is there in principle between such a shift and the shift whereby Kepler superseded Ptolemy, or Einstein Newton, or Darwin Aristotle? (“Two Dogmas,” 43).

No mathematical or logical law enjoys any privileged position inside the system. Each must be prepared to ‘give way’: to be modified or even eliminated in the face of new experience.

Against Naturalism

Problems loom on every horizon. Here’s just a few of the more critical ones. Consider, first, Devitt’s general assumption that unless there is a connection between our beliefs and reality, we aren’t justified in believing what we do (and thus don’t know). “If those connections are not via [sense] experience,” he says, “they do indeed seem to be occult” (114). The implication here is that we can only be justified in believing something if we have come into causal contact with it. (Let’s call this the ‘Causal Theory of Justification’ [‘CJ’, for short].)

What I’d like to know is whether Devitt thinks he’s justified in believing (CJ). We may suppose that he does. (What good is an unjustified belief in (CJ)?) However, on the theory itself, justified belief in CJ results from causal contact with physical facts. Now how is that supposed to work? Approximately as follows. It appears that Devitt wants to infer (CJ) from his experience. He begins with a factual premise:

(F)  Every justified belief results from causal contact with external, physical facts.

That Devitt believes (F) is evident from his answer to the question, “In virtue of what is any belief justified and hence likely to be true?” His answer is that we “look at the way in which the beliefs are related to the experiences that the world causes” (114). In short, the belief that p is justified when (via sensation) it is caused by the fact that p. Devitt then draws this conclusion

(CJ)  No one can be justified in believing anything apart from causal contact with external, physical facts

which he quickly deploys against the idea that mathematical and logical truths are known by a priori intuition.

Problems with Causality

I have three issues with this argument. First, whether Devitt is justified in believing (CJ) depends on whether he is justified in believing (F). Is he so justified? Well, if (CJ) is true, then there must be a physical fact that has caused Devitt to believe that (F) is true. Now (F) is a universal generalization, but we dare say you won't find the universally generalized, physical fact that causes it. And that's because there is no such fact. (If there were, wouldn't naturalists try to put it squarely before us? In that way, they could guarantee that we all become just like them--naturalists who have had belief in (F) causally installed in us by the Grand Universal Irresistible Fact.

Here it is no use appealing to the many particular instances of (F) in which an individual’s justified belief has been caused by the facts. For how many of them would you have to be in contact with before the universal belief (F) would be caused in you? Answer: all of them; otherwise, the belief wouldn’t be a universal generalization. Alas, however, life is short—too short, anyway, to get in touch with the sea of facts needed to underwrite (F).

Problems with Revisability

So that’s the first problem. The second is that the argument from (F) to (CJ) is simply invalid as it stands. Just because as a matter of fact justification has always supervened on contact with physical facts, it hardly follows that it is necessary that justification thus supervene. You might as well argue (to borrow Tom Morris’ example) that since every human being that has ever existed has been born on the surface of the earth, it’s not logically possible for a human being to be born elsewhere (say, on a space station).

We mustn’t forget either that the necessity involved in (CJ)—cf. “No one can…”—is not the sort of thing we can be empirically acquainted with on Devitt’s view. For on holism, necessity isn’t an objective feature of reality; it is entirely system dependent. Hence, what (CJ) really tells us is that (F) is necessary, which is only to say that (F) is included in Devitt’s evidential system (call it ‘S’). (F) isn’t absolutely necessary; it’s relatively necessary. But then if Quine is right (and Devitt follows him more or less slavishly), (F) will be revisable.

If we were savvy non-naturalists (I like to think I am), we might well invoke Lord Quine and ‘revise’ (F) right out of S. That shouldn’t be that big a deal. If we can drop Excluded Middle to make room for Quantum Mechanics, we can surely drop (F) to accommodate the a priori reasons we claim to have for (1)-(4). Surely, this sort of shift will be far less dramatic than the one “whereby Kepler superseded Ptolemy.” About the only move Devitt has available to him is to reiterate his complaint about “occult” connections. But as we’ve already seen, that accomplishes little apart from spectacularly begging the question.

Problems with Circularity

Maybe one final point. Let’s say we manage to construct a valid argument for (CJ), one proceeding from premises (/beliefs) caused in us by physical facts. There’s still a serious problem. For what shall we say about the rule of inference used to put that argument through? Suppose it’s our old friend modus ponens (MP). The non-naturalist can say that there are a priori reasons for thinking MP necessarily true. Devitt can’t. MP is only justified, on his view, if it ‘fits into’ his evidential system S. But of course there wouldn’t be an evidential system like S for MP to ‘fit into’ unless MP were already a part of S. For MP, along with the other valid rules of deduction, is what forges the connections between parts of S to begin with. So we have the part being justified in terms of the whole, which already includes the part. It’s circular justification.

Here Devitt concedes the point, but says it’s not “reprehensible” circularity (110). Premise-circularity, where the conclusion appears as a premise in an argument, is a fatal defect. But we’re only looking at rule-circularity, says Devitt, where the conclusion “asserts the goodness of the rules used in that very argument” (110). That’s not reprehensible. But surely that is incorrect. Here’s a little example from Susan Haack that makes the point (for details, see her “The Justification of Deduction,” 76). Suppose you’re trying to prove the validity of modus ponens, and offer the following argument:

(5)  P is true.

(6)  If P is true, then MP is valid.


(7)  MP is valid.

Should we accept your argument? Well, (7) doesn’t appear as a premise; so at least it’s not premise-circular. Let’s also agree that (5) and (6) are true, so that the truth of the premises isn’t in question. Is the argument thereby free from defect? Surely not. For you can scarcely hope to prove that MP is valid if your proof presupposes that it is. For the argument from (5) and (6) to (7) is quite clearly a modus ponens argument for modus ponens. And it’s obvious (isn’t it?) that if you have to presuppose that MP is valid to prove the validity of MP, then you’re just spinning your wheels. You haven’t given anything like a proof of MP. You've only succeeded in assuming it. The argument you’ve given is indeed rule-circular, but it's nonetheless reprehensible for that. It’s rule-reprehensible, we might say.

The evangelists for naturalism are not averse to preaching. If epistemological naturalism is true, they announce, you might have to face the fact that there is no God. Unfortunately, in the case of Devitt, we get scarcely more than his 'witness' for naturalism. He proclaims his “attraction” to the doctrine, while tossing in some question begging arguments for anyone who disagrees. For my part, I wouldn’t call that a proof of naturalism; indeed, I can’t even say it’s a good ‘witness’.

— Richard Brian Davis

Works Cited

1. Michael Devitt, “There is No A Priori,” in Contemporary Debates in Epistemology, eds. Steup and Sosa (Blackwell, 2005).
2. Susan Haack, “The Justification of Deduction,” in A Philosophical Companion to First-Order Logic, ed. R.I.G. Hughes (Hackett, 1993).
3. “W. V. O. Quine, “Two Dogmas of Empiricism,” in From a Logical Point of View, 2d. ed. (Harvard, 1961).
4. Alex Rosenberg, “Why I Am a Naturalist” The Stone, The New York Times, September 17, 2011.