Fallibilism and its Discontents

by Scott F. Aikin and Robert B. Talisse

JTxEnx5BcFallibilism is a philosophical halo term, a preferred rhetorical mantle that one attaches to the views one favors. Accordingly, fallibilists identify their view with the things that cognitively modest people tend to say about themselves: I believe this, but I may be wrong; We know things but only on the basis of incomplete evidence; In the real world, inconclusive reasons are good enough; I'm open to opposing views and ready to change my mind. But there are different kinds of epistemic modesty, and so different kinds of fallibilism. Let's distinguish two main kinds of fallibilism, each with two degrees of strength:


Weak: It is possible that at least one of my beliefs is false.

Strong: Any one of my beliefs may be false.


Weak: It is possible that I know something on the basis of inconclusive evidence

Strong: All I know is on the basis of inconclusive evidence

Belief-fallibilism is a commitment to anti-dogmatism. It holds that one (or any!) of your beliefs may be false, so you should root it out and correct it. The upshot is that one should hold beliefs in the appropriately tentative fashion, and face disagreement and doubts with seriousness.

Knowledge-fallibilism is a form of anti-skepticism. It holds, against the skeptic, that one does not need to eliminate all possible defeaters for a belief in order to have knowledge; one needs only to address the relevant defeaters. The knowledge-fallibilist contends that the skeptic proposes only the silliest and least relevant of possible defeaters of knowledge. We rebuke the skeptic by rejecting the idea that all possible defeaters are equally in need of response. Again, the knowledge-fallibilist holds that knowing that p is consistent with being unable to defuse distant skeptical defeaters; knowing that p rather requires only that the relevant defeaters have been ruled out.

Although these two varieties of fallibilism are propositionally consistent, they prescribe conflicting intellectual policies. Belief-falliblism yields the attitude that, as any of one's beliefs could be false, one must follow challenges wherever they lead. But knowledge-fallibilism holds that one needn't bother considering certain kinds of objections; it thereby condones the attitude that a certain range of challenges to one's beliefs may be simply dismissed.

When we think about knowledge, we often toggle between wildly different viewpoints. The anti-skeptical viewpoint holds that we know many things, some quite easily; it therefore maintains that any theory of knowledge that has skeptical implications is untenable. Thus if a theory of knowledge says you don't know that you have hands, you should simply toss the theory. However, we also are attracted to an anti-dogmatic viewpoint, one that prizes relentless questioning and values doubt. Hence we tend to recoil at the dogmatic thought that one should evade criticism of one's ideas. Call this vacillation between anti-skepticism and anti-dogmatism the shifting problem.

Fallibilism is typically proposed as a solution to the shifting problem. Roughly, it enjoins us to question everything that's worth questioning. However, recalling our descriptions above, we can see that fallibilism is not a solution to the shifting problem, but rather a restatement of the problem! In its two forms, fallibilism simply restates the conflicting inclinations; belief-fallibilism manifests anti-dogmatism, while knowledge-fallibilism embraces anti-skepticism. We still lack advice on how the resulting conflicts between these inclinations should be managed.

Perhaps we've moved too quickly. Let's return to knowledge-fallibilism. The crucial thought, again, is that one can know that p without being able to defeat all of the possible skeptical scenarios that would undermine one's knowledge. In other words, one could know that p even if one cannot entirely rule out the possibility that we're all living in the Matrix or under the influence of an omnipotent Cartesian demon. To capture the knowledge-fallibilist's thought, imagine the possibility that every zebra in every zoo has been secretly replaced with a painted mule. The knowledge-fallibilist thought runs as follows:

Sam sees what seems to be a zebra at the zoo. It is a zebra, but on the basis of what he sees, he can't eliminate the possibility that he's seeing only a painted mule. Nonetheless, Sam knows it's a zebra.

Knowledge-fallibilism thus makes wide room for concessive knowledge-attribution. And this is where knowledge-fallibilism has a connection with a wider pragmatist tradition, as our attributions of knowledge also function as endorsements of actions. So Sam's communicative action of telling his kids, for example, that the zebra just ate some grass is acceptable. Knowledge-fallibilism saves a good deal of the knockabout usage of the term knowledge and thus vindicates our actions in light of that usage.

That's the good news about knowledge-fallibilism, but there's bad news too. It comes in two stages. The first is what we call the infelicity problem. It sounds weird to say, “I know that p, but I may be wrong.” Suppose Sam says, “I know that there's a zebra over there, but I may be wrong about that, and it may be a painted mule or something…” Surely his daughter, Geraldine, is well within her rights to say, “So you don't know, then, right?” David Lewis called statements of the form “I know that p, but I might be wrong,” mad, and he termed this trouble the madness of fallibilism. That may be a little over the top, but it's on the right track. (Lewis, by the way, ended up endorsing fallibilism because he took it to be less mad than skepticism!)

The second problem with fallibilism is what we call the epistemic ascent problem. It is simply this: even in cases of appropriate fallible knowledge-attribution, one must nonetheless stipulate that the proposition known is true and the defeaters are false. The problem is that fallible knowers are not warranted in making those additional stipulations.

To see this, consider that, by hypothesis, Sam has knowledge, because it was stipulated that he's really looking at a zebra, and he's not looking at a painted mule. But notice that his claiming that he sees a zebra is appropriate in both circumstances. But Sam can't claim that he knows which circumstance he's in. This is because the claim to know in these cases requires that we've stipulated that what the purported knower holds is true and the defeaters are false. But Sam can't do that. So knowledge-fallibilism saves knowledge for those who are in the right circumstances, but doesn't make it a requirement to ascertain that you're in fact in the right circumstances. The result looks strange:

Sam: That's a zebra over there, Geraldine!

Geraldine: Neat! But do you know it's a zebra? It could be a painted mule.

Sam: Huh. That's a funny thought. I hadn't even considered that. Well… if it's not a painted mule but a zebra, then I know it's a zebra.

Geraldine: Ok, if it's a zebra and it's not a painted mule, then you know.

Sam: Yep, and it's a zebra, alright! Now, who wants ice cream?

Geraldine: Me!

Fallibilism works for knowledge attribution when we know the truth values for the known content (as true) and that for the potential defeaters (as false). The trouble is that this strategy for third-personal knowledge attribution doesn't work in the first-personal form in which skeptical scenarios are posed. When Sam says, “Yep, and it's a zebra, alright,” he's covertly switched between what's appropriate for him to say within the situation to the position of assessing what the situation is. It'd be appropriate for him to say that it's a zebra, even were it not a zebra but a painted mule. But his first-order perspective on what he sees at the zoo doesn't warrant the claim that he's in a zoo containing zebras rather than painted mules.

Fallibilism is a positive impediment to successful anti-skepticism and anti-dogmatism, because in order to attribute knowledge in the first-person, we need both first-order and second-order information of the circumstances. Otherwise we're just claiming we know without any basis. That's a little better than yelling it or typing it in ALLCAPS, we think, but not much.

There is, without a doubt, more to say. But, equally without a doubt, the shine is off fallibilism's halo.