There are many forms of radical skepticism: skepticism of the external world, skepticism of other minds, and skepticism of rationality, to name just a few. They arrive at skepticism via different channels, some more successful than others.
Agrippan skepticism is an ancient Greek variety. It is perhaps the hardest-hitting attack on the possibility of knowledge in the history of philosophy. I don’t know of any satisfactory solution.
Epistemologists agree on this much: in order for a belief to count as knowledge, it needs to be at least a justified true belief. What does it mean for a belief to be justified? It means we have a reason for believing it. If this reason will work as justification, it must be a reason that we know.
Of course, if we know this reason, it must be a justified true belief. So what is its justification? It has to be some other reason that we know. And we’re off on a regress.
The problem can be put this way: justification can only happen in three ways:
- Regress argument: belief A is justified by belief B, which is justified by belief C, which is justified by belief D, and so on.
- Circular argument: belief A is justified by belief B, which is justified by belief A.
- Dogmatic argument: belief A is axiomatic. It requires no justification.
None of these options succeed in justifying a belief. Regress arguments fail to justify because they never bottom out at some belief that is already justified. Circular arguments attempt to justify two otherwise unjustified beliefs via each other. Dogmatic arguments arbitrarily make some beliefs immune from the justification requirement.
If there is no fourth way and knowledge requires justification, then knowledge is impossible.
This argument is known as the Münchhausen trilemma, as well as the Agrippan trilemma. The first name refers to the fictional baron Munchausen who pulls himself out of a swamp by his own hair. The second refers to Agrippa the Skeptic, who – according to influential ancient skeptic Sextus Empiricus – first put forth a version of the argument.
Many attempts have been made to avoid the trilemma. One way is to say that knowledge of sense experience justifies itself. What is the justification for this view? Presumably, that sense experience is immediate and undeniable (or “clear and distinct“). Why do the immediate and undeniable not require justification for knowledge? Whatever the answer might be, the Agrippan can reply with another “why?”
The Agrippan problem is like the child who responds to all questions with “why?” It is just as impossible to pacify.
Another move is to loosen, or otherwise change, the requirements for what counts as knowledge. This responds to a proof that knowledge can’t exist by saying, “okay, but this other thing can exist!” It can work, but only by admitting defeat and changing the topic. Karl Popper’s fallibilism arguably does this.
Probably the best way “out” of the trilemma is to accept it. Of course, this must be done carefully. Unqualified radical skepticism is self-defeating. Does a skeptic claim to know that skepticism is true? If he does, he’s not really a skeptic; if he doesn’t, then why does he assert it?
It is possible to give in to the trilemma without embracing skepticism. For example, we can grant that knowledge is dogmatic. We know what we know only by granting immunity from justification to some axiomatic belief X. From X springs all justification. If X is true, then we really know what we think we know. If it isn’t, then we don’t know what we think we know.
Of course, if we have any doubt about X, then we don’t know it. This means it can’t justify anything without the justified belief also being called into doubt. This is the problem with arbitrarily making some belief immune from the justification requirement. We need an explanation for why X doesn’t need justification. Any such explanation is really a justification. If it’s to fly, it’s gonna need (another) justification.
Here’s another problem with accepting dogmatism: let’s say X justifies Y. We axiomatically know X, so then we know Y. Great! But there’s a problem. Let’s call the fact “that X justifies Y” Z1. How did we know Z1? How was Z1 justified? Was it justified by X? If so, let’s call “X justifies Z1” Z2. Did we know Z2? How was it justified? Etc.
This latter problem doesn’t defeat the accept-dogmatism move. Someone can consistently say that all Z’s are justified by X. It just means accepting that X axiomatically justifies an infinite series of meta-justifications for every belief that it justifies. It’s counterintuitive, but not more so than believing X without justification in the first place.
I’ll write more about approaches to the Münchhausen trilemma on other occasions. I’ll end here with an observation: the Münchhausen trilemma is a special case of the recurring problem of grounding. Consider a different problem: how do we know the meaning of a word? Well, we can look it up. How do we know the meanings of the words in the definition? Well, we look them up. A similar problem arises. The problems of Bradley’s regress and the unity of the proposition work the same way.
Problems with this similar structure are all over philosophy. Sometimes, problems with analogous structures have analogous solutions. In this case, I don’t know what those might be.