A Modest Critique of Quine’s Web of Belief
By Daniel Tippens
Most of W.V.O Quine’s landmark essay, “Two Dogmas of Empiricism” is devoted to a critique of the analytic/synthetic distinction. Quine defines an analytic statement as either a logical truth – “Bachelors are bachelors” – or a statement that can be turned into a logical truth by exchanging synonyms –“Bachelors are unmarried men.” One of the features of analytic statements is that they are devoid of factual content and are thus, immune to empirical disconfirmation. Synthetic statements, on the other hand – statements like “the barn is on fire” – do express substantive facts and consequently are prone to empirical disconfirmation. Logical Empiricists, like A.J Ayer, relied upon the distinction between analytic and synthetic in order to distinguish meaningful propositions from meaningless ones (so as to eliminate so-called “metaphysical” statements) and characterized analytic statements as entirely “linguistic,” insofar as they are about nothing more than the meanings of words.
Quine’s main point with respect to the analytic/synthetic distinction is that there can be no adequate, non-question-begging account of what an analytic statement is. The argument would appear to leave us, then, with nothing but synthetic statements. But in the last section of “Two Dogmas of Empiricism” Quine also challenges the Logical Empiricist idea that individual synthetic statements have their own conditions for confirmation and disconfirmation, in favor of a confirmation holism: the idea every statement’s verification is connected with the verification of other statements, all of which are ultimately answerable to experience. I’ll spell this out in a moment, but hopefully confirmation holism will sound a familiar bell since, in defending it, Quine invokes his famous metaphor of a “web of belief.” And one apparent consequence of it is that ultimately every belief – mathematical and logical beliefs included – is revisable in light of empirical evidence.
To see why, consider the question of how revision of a belief occurs on Quine’s picture. The idea is that when recalcitrant evidence presents itself, it puts pressure on us to give up any number of different beliefs. Take a simple example: if I believe that it is sunny, and I walk out my door and see that it is not sunny, then my network of beliefs is inconsistent with my observations. When I encounter this evidence, I have a choice. I could give up my belief that it is sunny or I can give up my belief that I am not currently hallucinating, that there isn’t a giant spotlight above my house, etc. The point is that no piece of empirical evidence determines which belief we give up. Instead, it just indicates that we need to give up some belief, such that our network, and the evidence, remain consistent. Empirical evidence doesn’t put pressure on any particular belief, but rather our whole network of beliefs. This point, of course, is reminiscent of Pierre Duhem’s argument that when we perform an experiment, we are never testing between only two competing hypotheses. We might think that only hypotheses X and Y are in question, but really we are also testing the hypotheses that our microscopes work, our eyes are functioning properly, etc.
With this in mind, Quine goes on to suggest that mathematical or logical beliefs are packed more closely towards the center of our network. What this means is that while these beliefs can be revised in principle, we only do so when we encounter a massive amount of recalcitrant evidence that places significant empirical pressure on our entire network of beliefs or a good portion of it. The reason for this is that revising a belief at the center results in the revision of many of the beliefs elsewhere – if we were to give up modus ponens, for example, we would be forced to revise the beliefs that we had deduced by way of it. Of course, Quine can’t simply assert that our “center-beliefs” can be given up in light of empirical evidence – he needs to provide an example. So he cites the way in which work in quantum mechanics generated several paradoxes that were taken as recalcitrant evidence, thereby putting significant pressure on our belief-network and causing us to abandon the principle of distributivity.
Now for my (modest) critique. It seems to me that Quine must admit that at least one belief is not susceptible to revision, in the face of empirical evidence, in order for his confirmation holism to work. Of course, if this is true, then it is not true that all beliefs are revisable in light of experience, as Quine wants to maintain.
What it is that prompts us, on Quine’s picture, to count certain observations as being problematic with respect to our network of beliefs, thereby causing us to feel the pressure to revise some of them? The most natural candidate is the law of non-contradiction, according to which propositions P and -P cannot be true simultaneously.
Consider our earlier example and suppose that I hold the belief that it is sunny. When I go outside and see that it is not sunny, I take the belief that it is not sunny into my network of beliefs. Now I hold two contradictory beliefs: that it is sunny and that it is not sunny, and so I must give up one of these beliefs, on pain of violating the law of non-contradiction. On this picture, then, the law of non-contradiction is the very thing that makes us feel that our network of beliefs is not consistent with our observations, leading us to undergo pressure to revise some of our beliefs.
Could we ever give up the law of non-contradiction, on Quine’s view? Imagine a scenario in which we have an overwhelming quantity of observations that act as recalcitrant evidence, placing empirical pressure on us, forcing us to reconsider the law of non-contradiction. If we are feeling this pressure, then the law of non-contradiction is causing it. So if we were to give up this law, it seems we would be using it in the process of giving it up, and this strikes me as incoherent.
One might respond by citing other principles to which we might appeal in giving up the law of non-contradiction. Quine, for instance, says that when we experience a large body of recalcitrant evidence at the periphery, we will be inclined to give up a center-belief for reasons of simplicity. Maybe we could give up the law of non-contradiction on these grounds.
I don’t think this reply can work. To see why, consider the distinction between a principle that makes us feel we need to give up some beliefs, and a principle that we use to decide which beliefs to give up. Simplicity is an example of the latter, for we consult it in order to decide which beliefs to keep, while the law of non-contradiction is an example of the former, for it is what generates our need to give up some beliefs in the first place. Only once the law of non-contradiction has done its work do we appeal to simplicity to facilitate our decision-making process. So, giving up the law of non-contradiction on grounds of simplicity would still involve appealing to the law of non-contradiction.
Someone might also want to point out that people have, in fact, considered giving up the law of non-contradiction. Indeed, that is exactly what dialetheists, like Graham Priest think. Perhaps this demonstrates that regardless of the concerns I’ve raised, the law of non-contradiction can be given up, shielding Quine from my criticism.
It is not enough to show that the law of non-contradiction has, or can, be given up. One must show that the basis for rejecting the law was, or is, empirical evidence, as Quine did with respect to the principle of distributivity. Furthermore, I think I can safely dig my heels in the ground, here. If I am right that, on Quine’s view, the law of non-contradiction is what generates empirical pressure, and that this fact renders giving up the law of non-contradiction — on empirical grounds — incoherent, then I have firm footing to claim that, in the case of dialetheism, giving up the law of non-contradiction was either not done on empirical grounds or was methodologically incoherent. If the former answer is correct, then Quine is still in trouble. If the latter, then Quine could still be right, but has to admit that in some cases, we decide to give up beliefs on the basis of empirical evidence for incoherent reasons, which would seem a Pyrrhic victory.
Quine’s From a Logical Point of View (in which “Two Dogmas of Empiricism” appears).
SEP entry on dialetheism.