I’m taking a course on epistemology this semester, and I thought I would post here a paper I wrote for the course (with minor modifications based on instructor feedback). I’ve been wanting to post some “real” philosophy on here, so to speak (actually, I’ve been wanting to get something, anything, posted here), so I thought this would be a good opportunity to do so. Expect more content here related to epistemology, ethics, and their intersection in the future. Without further ado, I’ll tackle the question: are there really two kinds of skepticism?
It is often thought that there are two distinct kinds of skepticism: Pyrrhonian skepticism and Cartesian skepticism. The most well-known facet of Pyrrhonian skepticism is the Agrippan Trilemma (also known as the Münchhausen Trilemma),1 which in its strongest form states, in brief, that there are only three (though Sextus Empiricus names five)2 possibilities when justifying a proposition: either the reasons (and reasons for reasons, etc.) go on to infinity, they include the proposition to be proved (that is, they are circular), or they stop at some statement held to be self-evident. But all of these possibilities seem to be unsatisfactory ways of gaining knowledge, implying that we cannot have any knowledge (or justified belief). Cartesian skepticism argues that in order for us to know a proposition p (paradigmatically, some seemingly-clear proposition about the external world), we must know that some skeptical hypothesis s (e.g. that we are dreaming) does not obtain. But we cannot know that said skeptical hypothesis does not obtain. Therefore we cannot know p, and because p is held to be representative of most or all of our knowledge, it follows that we cannot know (or have justified beliefs about) anything (at least about the external world).3 Leaving aside the the question of whether or not these arguments succeed, it seems interesting to ask whether there are, in fact, two distinct forms of skepticism at issue here, or only one. Are these arguments independent, or do they in fact depend on each other? Do they stand or fall independently, or together? That is, is it possible to refute both of them simply by refuting one of them? To the best of my knowledge, they are usually held to be two distinct forms of skepticism, which stand or fall independently of each other. But I’m not so sure. In what follows, I shall explain each form of skepticism in more detail, and in so doing endeavor to discover the extent to which they are, or are not, intertwined.
The argument for Cartesian skepticism as it is elaborated in Stroud (1984) can be expressed as follows:
- Suppose that p is a proposition (e.g. the proposition that I have two hands) such that if I know anything at all about the external world, then I know that p.
- If some skeptical hypothesis s, such as that I am dreaming, that some evil genius is deceiving me, that I am a brain in a vat, etc., obtained, then I would not know that p.
- Therefore, in order to know that p, I must know that not-s.
- I do not know that not-s.
- Therefore, I do not know that p.
- Therefore I know nothing at all about the external world.
I will not discuss this argument’s soundness here, except to note that, as Stroud correctly points out, it turns on the truth or falsity of premise (3). Since, as Stroud argues, seems obvious both that (2) is true and that if (3) is true then (4) is true, much hangs on premise (3), which is less clearly true, yet far from obviously false.4 To see its force, consider Stroud’s example of the goldfinch in the garden. If I look out the window and see a yellow bird, which I claim to know is a goldfinch, someone else might ask how I know it’s a goldfinch. If I were to respond “because it’s a yellow bird,” they might object that, since canaries are also yellow, based on that evidence it could just as easily be a canary. The idea is that in order to know that the bird is a goldfinch, I must first know that this alternative possibility, which is compatible with all the evidence I’ve thus far put forward, does not obtain. This is analogous to the case of the skeptical hypothesis: s is such that it is compatible with all evidence for p, and is constructed in such a way as to be impossible to prove false. Since, therefore, I can’t know that s does not obtain, it seems as though, by analogy with the canary case, I cannot know p. And since p is taken to be representative of most or all of my beliefs, it seems I can’t know anything at all.5
The “five modes of suspension of judgment” which Sextus Empiricus enumerates, and which Diogenes Laertius attributes to Agrippa,6 are: 1.) “the mode deriving from dispute;” 2.) “the mode throwing one back ad infinitum;” 3.) “the mode deriving from relativity;” 4.) “the hypothetical mode;” and 5.) “the reciprocal mode.” The modes which form the Agrippan Trilemma are the second, fourth, and fifth, which correspond to the infinite regress of reasons, stopping at some self-evident reason, and the circular chain of reasons respectively. The mode deriving from dispute seems to me, rather than being a horn of the trilemma in itself, to be the motivation to search for justification for one’s knowledge. Disagreement is what highlights our need for justification: as long as we disagree with one another, we will continue to be compelled to search for reasons to justify our beliefs.7Though the argument from relativity is interesting in its own right, it is not relevant to our discussion here, so I will leave it aside.
Put more formally, the Pyrrhonian argument can be expressed as follows:
- Any belief b must be supported by reasons (e.g. arguments or evidence).
- But each reason for b, being itself a belief, must itself be supported by reasons (which must in turn be supported by further reasons, and so on).
- This results in one of the following three possibilities:
- The chain of reasons go on for infinity.
- The chain of reasons circles back on itself at some point.
- The chain of reasons stops at some “foundational” belief that is self-evident (i.e. a belief which does not stand in need of justification, or which one is justified in believing simply in virtue of understanding its contents).
- But each of these possibilities does not result in knowledge (or even in justified belief).
- Conclusion: Since the possibilities in (3) were exhaustive, and by (4) all are insufficient for knowledge, we cannot have any knowledge.
Premise (4) requires further argument, because I have not stated reasons why each of the three possibilities listed in premise (3) do not result in knowledge.
First let us consider the possibility of an infinite chain of reasons. As finite beings, it seems impossible for us to grasp an infinite chain of reasons. We have a difficult enough time understanding the concept of infinity itself, and grasping it in other domains, so the idea that we could grasp an infinite chain of reasons seems absurd. The time it would take even to list such a chain of reasons would, by hypothesis, be infinite, such that there is not even enough time in our finite lives to be presented with them all, let alone to understand them (supposing that the idea of “all” of an infinite series of reasons is even sensible). Stronger still, some might say that an infinite series of reasons cannot even in principle provide justification, because it never rests on anything solid, so to speak. There are perhaps objections to these considerations, but to consider them would take us too far afield; my aim here is not to refute skepticism, but to understand it, and for that purpose these considerations will suffice.
Next let us examine the possibility that the chain of reasons circles back on itself, including the belief to be justified as a premise in its own derivation. This also doesn’t seem as if it can lead to knowledge. As Aristotle argues in the Posterior Analytics, “the premisses of demonstrated knowledge must be … better known than and prior to the conclusion[,]” and since anything can be, at most, as-well-known as itself, and cannot be prior to itself, nothing can serve as a premise in a proof of itself. More basic than this, it seems odd to suppose that we could prove something on the basis of a premise which we didn’t already know, and since, by hypothesis, we don’t yet know the proposition to be proved, we can’t prove it based on itself.8
Finally, let us consider the strategy of stopping the chain of reasons at some “foundational,” self-evident belief. What reasons could be given for why this fails? The reason seems simply to be our premise (1), that any belief b must be supported by reasons. But this raises the question: why must this be so? If we don’t say something more, our argument would be circular, which as we’ve already established cannot get us anywhere.
Sextus Empiricus argues that,
[i]f to avoid [infinite regress and circularity] our interlocutor claims to assume something by way of concession and without proof in order to prove what comes next, then the hypothetical mode is brought in, and there is no way out. For if he is convincing when he makes his hypothesis, we will keep hypothesizing the opposite and will be no more unconvincing, and if he hypothesizes something true, he makes it suspect by taking it as a hypothesis rather than establishing it; while if it is false, the foundation of what he is trying to establish will be rotten. Again, if hypothesizing something achieves anything towards making it convincing, why not hypothesize the object of investigation itself rather than something else through which he is supposed to establish the object about which he is arguing? If it is absurd to hypothesize the object under investigation, it will also be absurd to hypothesize what is superordinate to it.9
However, Sextus Empiricus seems to be missing an important concept: self-evidence. Modern foundationalists do not simply “assume” their foundations as “a hypothesis”; rather, they claim that their foundations can be known to be true without proof, or at least that we can be justified in believing them to be true without proof. An example (in the context of moral epistemology) is Michael Huemer’s principle of Phenomenal Conservatism: “Other things being equal, it is reasonable to assume that things are the way they appear.”10 He explains that, while appearances (which he takes to include, among other things, both perceptions and intuitions, including ethical intuitions) are fallible, they provide some initial justification for beliefs based on them, without having to be justified themselves.11 I am not concerned to argue for the existence of self-evident propositions here, only to point out that the skeptic must provide an argument for why such things cannot exist, and that Sextus Empiricus’ reasoning is therefore incomplete.
And here we find the first hint of an answer to the question with which we began, for the argument of Cartesian skepticism seems to be just such an argument. Why is this? Consider what some might put forward as a paradigm case of a self-evident, foundational belief: a belief derived from a piece of sense data acquired under ideal sensory conditions (e.g. the perceived object is viewed close-up, under bright, natural light, with no distortions in the air, by an observer with perfect 20:20 vision, etc.). This is precisely the type of belief that Cartesian skepticism is designed to undermine.
Perhaps all such an argument establishes is that appearances are defeasible. Huemer’s theory admits this much, and claims that it is compatible with appearances nevertheless providing foundational justification. But it seems to establish more than this. If appearances are defeasible, then in order to derive knowledge from them, it seems like we would need to establish that they are not, in fact, defeated in whatever situation is at issue. Recall the example of the goldfinch from above. The very existence of a defeater seems to remove our claim to knowledge until we can establish that the defeater is not, in fact, in play. And of course, in order to establish that the defeater is not in play, we would need a reason(s), which begins the chain all over again.
Thus, it seems as if Cartesian skepticism plays a role in the argument for Pyrrhonian skepticism; namely, it completes one horn of the trilemma. If this is so, and if there is no argument that can fill that role as well as Cartesian skepticism, then we have established a dependence relationship between the two kinds of skepticism: if Cartesian skepticism falls, then so does Pyrrhonian skepticism. This raises two questions. First, does such an alternate sub-argument exist that can fulfill the same role as Cartesian skepticism in the Pyrrhonian argument? Second: does a similar dependence relationship go the other way? That is, does Cartesian skepticism depend on Pyrrhonian skepticism as well?
To the first question, in one sense the answer seems to be a clear “yes”, because the argument one gives against foundationalism will defend on which foundationalist arguments one’s interlocutor advances. So of course there are many arguments to be given against foundationalism. But this isn’t enough to show that Cartesian skepticism isn’t a necessary component of Pyrrhonian skepticism. There are better and worse arguments for foundationalism, and if one’s opponent gives an argument to the effect that intuitions are always correct, one can simply point out that intuitions are sometimes wrong, and that will serve as an argument against that form of foundationalism.12 But it may be the case that there are one or more foundationalist arguments against which only Cartesian skepticism succeeds, and in that case it would play a necessary role in Pyrrhonian skepticism as I suggested above. To find out whether this is the case would require a survey of many foundationalist arguments and possible counterarguments to them, which would vastly overstep the limits of this paper. But for the sake of argument, suppose it turns out that each and every foundationalist argument can be refuted without recourse to Cartesian skepticism. If this were the case, it would not be possible to refute Pyrrhonian skepticism simply by refuting Cartesian skepticism, but in the event that Cartesian skepticism were refuted, the Pyrrhonian skeptic’s job would still be made much more difficult, because they would need to rely on other arguments to fight back against foundationalists. And as long as we don’t know whether such an argument exists or not, a refutation of Cartesian skepticism would be a major blow to Pyrrhonian skepticism as well, given the existence of sophisticated foundationalist accounts that defy easy refutation by means other than Cartesian skepticism.
One consideration that suggests Pyrrhonian skepticism is not dependent on Cartesian skepticism is that Pyrrhonian skepticism seems to establish a more global skepticism than Cartesian skepticism: Pyrrhonian skepticism casts doubt on any proposition p, while Cartesian skepticism cannot establish skepticism about propositions of the form “it appears to me that p.” Since even in a dream or in the case of an evil genius one would still know how things appeared, those propositions can’t be cast in doubt by Cartesian skepticism. But if one branch of Pyrrhonian skepticism (the anti-foundationalist branch) only establishes a more limited form of skepticism, it’s unclear how Pyrrhonian skepticism could establish the more broad form of skepticism it claims to establish. Thus, the arguments must be independent.
However, even Pyrrhonian skepticism cannot establish skepticism about how things appear. Pyrrhonian skepticism can only get started if it makes sense to ask the question “but why do you believe that p?” That is, if a particular proposition doesn’t stand in need of justification, then Pyrrhonian skepticism can’t cast doubt on it, because Pyrrhonian skepticism only establishes that a proposition p cannot be justified. But it in fact doesn’t make sense to ask the question “but why do you believe that it appears to you that p?” If asked that question, it’s appropriate to respond with surprise at the absurd question and answer “because it just does appear to me that p.” Therefore, Pyrrhonian skepticism does not establish a broader skepticism than Cartesian skepticism.
To the second question, one might wish the answer to be “yes”, because that would invite the charge of circularity on skepticism as a whole; if each argument depends on the other, then they both seem to lack justification, undermining the skeptical project as a whole. But Cartesian skepticism does appear to be independent of Pyrrhonian skepticism; its arguments function perfectly well, even if Pyrrhonian skepticism were refuted. The only premises that might be thought to depend on Pyrrhonian skepticism are premises (3) and (4). But premise (3) can be established, even in the absence of Pyrrhonian skepticism, simply by analogy with other cases of putative knowledge such as the goldfinch case. And premise (4) is true even in the absence of Pyrrhonian skepticism by virtue of the fact that any appropriate skeptical hypothesis s is usually constructed such that it is unfalsifiable; thus, even without Pyrrhonian skepticism, we can establish that we don’t know ~s.
What does all this imply about the answer to our original question? Are the two forms of skepticism really united? Does one form stand or fall with the other? In one sense, the answer to both questions is simply “no”; it’s not the case that the two are in fact a single, unified argument, and it’s not the case that we can refute one by refuting the other. That would be too strong. As I said above, Cartesian skepticism does not require Pyrrhonian skepticism, and Pyrrhonian skepticism likely doesn’t, strictly speaking, require Cartesian skepticism either. However, Cartesian skepticism does nevertheless make an important contribution to the justification of Pyrrhonian skepticism; if it were refuted, Pyrrhonian skepticism would be considerably weakened. So while they are not united, they are in that sense linked.
1 Wikipedia contributors, “Münchhausen trilemma,” Wikipedia, The Free Encyclopedia, https://en.wikipedia.org/w/index.php?title=M%C3%BCnchhausen_trilemma&oldid=752016424 (accessed September 10, 2017).
2 Sextus Empiricus, Outlines of Scepticism (tr.: Julia Annas & Jonathan Barnes), Cambridge: Cambridge University Press 2000. pp. 40-41.
3 Descartes, René, Meditations on First Philosophy (tr.: Donald A. Cress; 4th ed.), Indianapolis: Hackett 1998.; Stroud, Barry 1984, The Significance of Philosophical Scepticism, Oxford: Clarendon.
4 Stroud (1984).
5 Stroud (1984), pp. 24-25.
6 Sextus Empiricus (2000), pp. 40-41.; Ibid, n. 168.
7 Though strictly speaking we would need justification as a prerequisite for knowledge, even in the absence of disagreement, disagreement is what drives us to begin the search and to continue it. In fact, according to the Argumentative Theory of Reason advanced by Mercier and Sperber (Mercier, Hugo and Dan Sperber. “Why do Humans Reason? Arguments for an Argumentative Theory.” Behavioral and Brain Sciences. (2011) 34, 57–111. doi:10.1017/S0140525X10000968), our reasoning faculty evolved for the express purpose of allowing us to resolve the problem of disagreement which was posed by the evolution of language and cooperation.
8 This argument is drawn from an essay which I wrote for an Aristotle seminar in Fall 2015, titled “Socrates Really Did Know Nothing: The Hard Problem of Knowledge.” The relevant passage of Aristotle is Posterior Analytics, I.2 71b21-22, in the edition: Aristotle. Posterior Analytics. The Basic Works of Aristotle. Ed. Richard McKeon. Trans. G.R.G Mure. New York: Random House, 2001. Print.
9 Sextus Empiricus (2000), pp. 42.
10 Huemer, Michael. Ethical Intuitionism. New York: Palgrave MacMillan, 2005. pp. 99. Huemer develops the view in more detail as a response to global skepticism in his Skepticism and the Veil of Perception. Lanham: Rowman & Littlefield, 2001. However, not having read this, I will be relying on his account in Huemer (2005). This runs the risk of misrepresenting Huemer’s view, but this is not terribly important for my purposes here; I simply use his view as one example of a foundationalist view.
11 Huemer (2005), pp. 99, 106-107.
12 I give this argument against one of Aristotle’s two arguments for foundationalism in “Socrates Really Did Know Nothing: The Hard Problem of Knowledge.” Aristotle’s argument can be found in Posterior Analytics, II.19 100b8-14.