What criterion for knowledge?

Modern skeptics try to respond to the ancient ones on an age old problem

Image from differencebetween.com.

There is a problem in epistemology—the branch of philosophy that deals with the nature of knowledge—that has been raised more than two millennia ago and just doesn’t seem to go away. Here is how the problem, in a nutshell, is rendered by Andrew Cling in his contribution to the edited collection Skepticism—From Antiquity to Present: To know a proposition, we must first know a criterion of truth. To know a criterion of truth, we must first know a proposition. Therefore we cannot know any proposition or any criterion of truth.

To put it differently, the so-called problem of the criterion comes about because (i) whatever answer we give to the question “what do we know about X?” presupposes an answer to the underlying question “how do we know about X?” But (ii) we cannot answer the second question without answering the first one. Which means we can’t really answer either. Ergo, (iii) we don’t know crap, unless we are willing to (iv) engage in circular reasoning in which a proposition is justified by a second proposition, which is then justified by a third one, and so on, until we encounter a proposition that can only be justified by a previous one; or (v) we are okay with an infinite series of justifications, in which the first proposition is justified by a second one, which is justified by a third one, and so on and so forth, forever.

If you are tempted to just dismiss the above as philosophical hair-splitting, pause for a moment and reflect that really smart people have considered the problem of the criterion seriously for over twenty centuries. What are the chances that they were all fooling themselves?

The word “criterion,” Cling informs us, comes from the Indo-European word for sifter, so criteria for truth are tools by which we sift propositions so to separate the true ones from the false ones.

There are five modern responses to the problem of the criterion: reject claim (i), a position known as particularism; reject claim (ii), a school of thought referred to as methodism; accept (iii), which is the skeptic position; accept (iv), known as coherentism; or accept (v), known as infinitism.

Let us start with particularism, the rejection of (i), the idea that in order to answer “what do we know?” we have to answer “how do we know”? Particularists say that sometimes it is acceptable to believe something, that is, to make a particular claim to knowledge, even though we have no epistemic justification for that belief. For instance, some propositions may be arrived at by way of direct awareness of their truth. I know that it is day because I’m looking outside the window right now.

The problem with that, of course, is that the skeptic can (and should!) reasonably ask us on what grounds we think that whatever means of apprehension we are using (say, vision) are, in fact, reliable. Moreover, particularists will still acknowledge that some methods of belief production are not, in fact, reliable. But how do we distinguish those that are from those that aren’t?

If we reject (ii) above, that we cannot answer “how do we know about X?” without answering “what do we know about X?” we are methodists (not in the religious sense of the term). This essentially says that a criterion for truth can have epistemic value even though there is no good evidence in its favor. One example would be reliance on intuition.

The issue here is similar to the one just encountered in the case of particularism: it seems arbitrary to say in which cases a given non-epistemic criterion like intuition is reliable or not, unless we use empirical evidence, thereby opening ourselves up again to the skeptic challenge. Either way, it seems that the “how” and “what” questions are inextricably connected and that skepticism (option iii) may, after all, be the way to go.

We have two more possibilities to consider. The first one is coherentism, which says that sometimes it’s okay to deploy what is normally a no-no in logic: a circular argument. The second one is infinitism, which endorses a different kind of no-no: an infinite regress of reasons.

You may have guessed where this is going next. Coherentism faces yet another version of the usual problem of arbitrariness. Since some examples of circular reasoning, according to the coherentist, are acceptable and others are not, how do we distinguish between the two cases? Coherentists also owe us an account of why they are okay with a procedure—circular reasoning—that is usually brought up as a standard example of bad reasoning.

Perhaps not surprisingly at this point, similar issues affect infinitism, beginning again with the fact that it lacks an explanation of why something that is usually considered bad—infinite regress—might be acceptable, and if so, under what circumstances. The more we dig around, the more it seems like skepticism is really difficult to seriously challenge.

There is one more, desperate option that we haven’t considered, however, based on the distinction between epistemic and non-epistemic reasons for believing something. To grasp such difference consider that knowledge of statistics indicating that few people will pass a bar exam and become lawyers is a good epistemic reason for me to believe that I will likely not become a lawyer. However, convincing myself that I will, in fact, pass the bar exam may be a good pragmatic reason for me to endorse that belief, no matter how statistically unlikely it may be.

This appears to be a promising anti-skeptical venue, but it immediately splits into a logical fork that comes with its own issues. On side of the fork leads to the observation that perhaps the pragmatic reason for a given belief is itself based on an epistemic one. To continue with the example of the bar exam, it may be that I try to convince myself that I will pass against the odds because empirical evidence shows that belief in oneself does, in fact, increase one’s odds. But now the way is open anew for the skeptic to question the epistemic value of such evidence and we have to begin daccapo.

The second side of the fork is more subtle but equally lethal to the anti-skeptic: if I accept non-epistemic reasons for a belief I hold then we are no longer talking about knowledge. Knowledge, after all, is justified true belief [1], and if I forgo the epistemic aspect I don’t have justification, which means I don’t really have knowledge. But the skeptic never claimed that people don’t have beliefs. The claim is that people don’t have knowledge (or, at least, that they don’t know whether they have it). [2]

If you have read my recent article on Agrippa’s trilemma, you will have noticed that particularism, methodism, coherentism, and infinitism are all modern attempts to get around the trilemma (or, more broadly, the famous five modes). But there doesn’t seem to be a way to do so. Agrippa laid down the gauntlet in the first century of the modern era. It is refreshing and humbling that two millennia of attempts to prove him wrong have so spectacularly failed.

_____

[1] Yes, yes, the epistemologically savvy among my readers may have noticed that I am using Plato’s classical definition of knowledge, which does not account for so-called Gettier problems. We’ll discuss those another time, okay?

[2] A similar scenario arises if one opts for William James’ infamous “will to belief,” which I have criticized elsewhere.