[Unfinished] Working Toward a Refutation of Skepticism


Evan Jack
15 min readNov 1, 2022


As I continue my journey in refuting skepticism I continually come to realize to a higher and higher degree the dire epistemic condition we all face. After going through so many responses, we must recognize that we must either find some way to garner non-epistemically circular justification, epistemically circular justification, or abandon justification altogether. If we could find some justification that was not epistemically circular and not infinitely regressive, then we would arrive at a certain type of foundationalism.[1] I believe Michael Huemer has outlined this version of foundationalism within his book Knowledge, Reality, and Value in such a way that it can resist skepticism, but only architecturally. By “architecturally,” all I mean is the pure skeleton of foundationalism he presents, the bare concepts of foundational belief-forming methods, foundational justification, foundational beliefs, etc., can resist skepticism. Let us quickly go through it.

Huemer’s Foundationalism

For Huemer (2021), foundational justification is, “by definition, justification that does not rest on reasons” (p. 105). Foundational knowledge is, for him, “knowledge whose justification is foundational” (p. 105). Foundational methods are “the belief-forming methods that generate foundational knowledge” (p. 109). Huemer holds that “if you’re using a foundational method, then you do not need to first verify that it is reliable. You just get to rely on it, unless and until you acquire specific reasons for doubting that it is reliable” (p. 109). Huemer explains that the idea you can just rely on foundation belief-forming methods is the case simply because “this [idea] just follows from the core tenet of foundationalism. If some method M generates foundational beliefs, then by definition [emphasis mine], the person using M does not need to gather any other evidence in order for M-generated beliefs to be justified. That’s just what ‘foundational’ means [emphasis mine]” (p. 109). This I believe, definitionally, gets around the skeptic and their demand for (meta-)justification. But, the reason why I said it resists skepticism in a purely architectural manner above is because of the fact that it resists it definitionally. In other words, foundationalism resists skepticism by its very design (its architecture), i.e., by definition, as was just shown.

The problem for foundationalists of the Huemerian type arises with the question “What belief-forming methods are foundational?” Once one tries to answer this it seems that they will have to use some basic doxastic practices which they consider foundational in order to prove that those basic doxastic practices are foundational. To be clear, what I am saying is that in trying to vindicate their idea that some belief-forming methods are foundational, they themselves will have to use those belief-forming methods, for if they do not then they aren’t foundational. This will only lead the foundationalist to epistemic circularity. Again, remember what Huemer said, “if [emphasis mine] you’re using a foundational method, then you do not need to first verify that it is reliable” (2021, p. 109). As I’ve said, the question then becomes how do we know if are using foundation belief-forming methods? The answer to this question seems to entail epistemic circularity on the part of the foundationalist.

I recently interviewed Michael Huemer and he did not disagree that this “problem” arose for the foundationalist. His response is just that it arises for all theories of justification for if one did not use their theory of justification in the question of its justification they would be abandoning the theory. Huemer seems to not really care for the question of meta-justification as explicitly stated and implicitly stated as well. For example, in Seemings and Justification, he says, “At some point, we must reject the demand for evidence of the reliability of our cognition. And if we are to reject that demand at some point, we might as well reject it right at the beginning and allow appearances to justify beliefs by default” (Huemer, 2013, p. 349).

Before moving on to epistemic circularity, I will note that there seems to be another response that comes to just reject the demand for meta-justification as Huemer does. This response comes from James Pryor and he raises it in regard to skepticism about his dogmatism. To me it seems that Pryor is simply confused about the meaning of justification. When BonJour speaks of the meta-justification requirement, Pryor holds that BonJour is trying to hold us to something more demanding than just providing justification. Pryor self-admittedly holds a more “liberal” conception of justification. I feel that Pryor, then, has not actually gotten around the demand for meta-justification by rejecting it. Rather, he is just confused about the concepts at hand.

Epistemic Circularity 1

I believe that, in the face of this, we must evaluate epistemic circularity and if it can provide justification. Just because it seems that it prima facie cannot provide justification does not mean we should not begin an investigation into if it can. So let’s conduct this investigation now.

I say that we should start with Michael Bergmann’s original defense of epistemic circularity found in his book Justification without Awareness and then go to his defense of epistemic circularity he greatly explicated in his recent book Radical Skepticism and Epistemic Intuition. From there, we will turn to other accounts and consider them. So, let us start now with chapter seven, “Epistemic Circularity,” of Bergmann’s Justification without Awareness.

First, his argument that he puts forward for those that believe in non-inferentially justified beliefs obviously only has force for those that believe in non-inferentially justified beliefs. As (meta-)skeptics, we have no such commitment and so his first argument doesn’t really do anything for us. As he says, “The first argument in support of epistemic circularity … concluded that all who support the foundationalist thesis, F, are committed to approving of epistemic circularity” (Bergmann, 2006, p. 193). Again, we, as skeptics, hold no support for F, therefore meaning that the argument he provides has no compelling force to believe that epistemic circularity is not epistemic inappropriate. Second, his argument that attacks all the alternatives to epistemic circularity only has force for those that propose an alternative to epistemic circularity. Again, as skeptics, we are agnostic as of right now and even if we have showed how all the arguments for epistemic circularity really have no force we will still suspend our judgment on whether or not epistemic circularity can confer justification. Let’s move on to what I believe to be a more interesting proposal from Bergmann (at least, compared to the last two arguments).

Bergmann, after going over the last two arguments that have no rational force for us, he takes as his goal to “explain why it (misleadingly) seems that epistemic circularity is, in itself, a bad thing” (2006, p. 197). Bergmann first explains that “[t]he sort of belief that can get infected with epistemic circularity is a belief such as that one’s belief source, X, is trustworthy or that one’s belief, B, wasn’t formed in an unreliable way” (2006, p. 198). Bergmann then makes a distinction between “two kinds of situation[s] in which a person can form EC-infected beliefs [beliefs infected with epistemic circularity] about a source X or a belief B” (2006, p. 198). There are first “QD-[S]ituations” (2006, p. 198). QD-Situations are “[s]ituations where, prior to EC-belief’s [the epistemically circular belief’s] formation, the subject is or should be seriously questioning or doubting the trustworthiness of X or the reliability of B’s formation” (2006, p. 198). Second, there are “Non-QD-Situations” (2006, p. 198). Non-QD-Situations are “[s]ituations where, prior to the EC-belief’s formation, the subject neither is nor should be seriously questioning or doubting the trustworthiness of X or the reliability of B’s formation” (2006, p. 198). Bergmann basically then goes on to say that in QD-Situations, epistemic circularity should be rejected because there is doubt, but in non-QD-Situations, epistemic circularity shouldn’t be rejected because there is no doubt. This doesn’t seem like a very stronger argument as it seems to suppose certain norms regarding belief that themselves would have to be justified. I believe Bergmann’s best move is his recourse to Thomas Reid because it seems to me that if epistemic circularity can confer justification the Common Sense Tradition is going to realize that (but more on this later). Let us move on to Radical Skepticism and Epistemic Intuition now.

I believe Bergmann’s move in Radical Skepticism and Epistemic Intuition to focus on the role of epistemic intuition is great. What I want to focus on is the fourth section of chapter nine, “Intuitions in Defense of Epistemic Circularity.” This section is important because “it isn’t only arguments that support certain instances of epistemic circularity. Intuitions about certain cases also lend direct support the possibility of appropriate epistemic circularity” (Bergmann, 2021, p. 182). Bergmann gives us a case of epistemic circularity which he holds is not intuitively problematic:

Tom has never had any questions or doubts at all about the trustworthiness of his sense perception. In fact, he has never explicitly considered whether his sense perception is reliable, though he relies on it all the time. Suppose Tom is talking with a friend, Becky, who wants to slyly lead him to see that a number of his beliefs inductively imply that his perception is reliable. Becky catches Tom in a cooperative mood and explains to him how inductive arguments for the reliability of an information source work. Tom understands the idea very well and is able to inductively infer that a metal detector is reliable by checking to see that it repeatedly indicates that metal is nearby whenever he brings metal near it, and not otherwise. The next day Becky gets Tom to recall that on many occasions in the past he was in the presence of objects he could perceive, in conditions that were good for perceiving them, and that whenever he formed beliefs on those occasions (on the basis of his perception) about whether such objects were present, he believed that those objects were present. Tom agrees that these things about his past are true. Becky then reminds Tom about the inductive method they discussed the previous day, something that Tom hadn’t yet been thinking about at all so far on this day. And Becky suggests to him that the things he has just acknowledged, about his past perceptual beliefs concerning objects around him, inductively imply that his perception is reliable. Tom hand’t thought about this before, but he agrees that the repeated accuracy of his past perceptual beliefs does inductively imply that his perception is reliable. And he now believes for the first time that his perception is reliable (a reliability claim he hadn’t considered before) on the basis of this inductive inference. Moreover, he is using inductive inference in just the sort of way he competently used it the previous day when he concluded that the metal detector was reliable. (2021, pp. 183–184)

This case has to do with the track record of perception. “The track record argument” is something that I may deal with later on.[2] It seems that the rest of what Bergmann has to say deals with epistemic intuitions. I’ve said in the past that the vindication of epistemic intuitions leads to a meta-regress. However, if we were to pose a faculty out of which the epistemic intuition came to judge that faculty itself and did not rely on another singular epistemic intuition to vindicate the epistemic intuition we just had then we would escape the meta-regress but also be caught up in a case of epistemic circularity that may be problematic. I believe that to figure out if it is problematic we must move beyond Bergmann for he only presents a bind (to Keither DeRose) concerning the rejection of epistemic circularity regarding epistemic intuition(s). Let us move on from Bergmann now.

Back to Huemer

Before we dive much, much deeper into the issue of epistemic circularity, as we have only covered the most recent defense of it (which, for a skeptic, seems to have no rational force), I want to go back to Huemer’s foundationalism as it is a possible candidate as we have gone over. As we know, a foundational-belief forming method are ones that generate foundational knowledge and one, by definition, does not have to give a reason for why these methods generate foundational knowledge. The problem I proposed for the Huemerian foundationalist had to do with how one would know that they had a foundational-belief forming method or had justification for their belief that they had a foundational-belief forming method. Let us take K to be the collection of all foundational belief-forming methods. If we do not use K to know and/or justify our belief that belief-forming method M is foundational, then we would enter into circularity or an infinite regress because we did not use K. But, if we use K to then vindicate M’s supposed foundational status which will be a part of K, then a problem can arise. So, let’s assume that M really is foundational and in fact is the only foundational belief-forming method. Operating off this assumption, M is K. Thus, in this case, if I were to use K to then vindicate M’s supposed foundational status, then I will have entered into a case of epistemic circularity. Is there any response to this? Could there be a way out of this epistemic circularity?

In the comment section of one of Huemer’s blog posts, Huemer is asked,

valuating wheter a proposition is foundational depends on whether it has “justification that does not depend upon one’s having justification for any other beliefs.” Wouldn’t this require a belief in the correctness of some account of justification, making all such beliefs non-foundational? If you don’t have to justify what justification is, the door to arbitrariness is open, even if no foundationalists have chose to walk through.

Huemer replies,

This sounds like what Alston calls a “level confusion”, a confusion between (in this case) the conditions for being justified in believing P and the conditions for being justified in believing that one is justified in believing P.

We will come to the idea of a level confusion in the next section, but I still nonetheless want to note here that I do not find what Huemer has had to say to be a good reason to not keep running with skepticism because it seems to me that the idea of a level confusion is predicated on completely nominal distinctions (I’m not a nominalist). So, let us turn to Huemer’s essay “We Can Know” found in the book Problems in Epistemology and Metaphysics.

In his essay “We Can Know,” Michael Huemer asserts that all regresses of reasons terminate at appearances. He takes it to be the case that appearances non-inferentially justify basic beliefs. However, he takes reasons to be a type of belief. Instead, if we simply take a reason to be something that points in favor of the likelihood of something being true, then we must ask how appearances do this. It certainly appears that my appearance that my hands are in front of me points in favor of my belief that my hands are in front of me, however, how can we know that it really does point in favor of my belief? I think Aaran Burns puts the question nicely in “Agrippa’s Trilemma”: “Why believe that its’ seeming to you that P makes it even so much as probable that P?” (2020, p. 80). Huemer holds that this just “appears to beg the question once again” (2003, “Arbitrary Foundations?,” p. 146). Let’s see why this would be question begging according to Huemer.

Huemer holds that the skeptics main argument against foundationalism is this:

P1: “If one lacks reasons for believing P, then P is arbitrary.”

P2: “Arbitrary propositions are not justified.”


C1: “If one lacks reasons for P, then P is unjustified.” (2021, Knowledge, Reality, and Value, p. 106)

Under my view of reasons, the argument would just be this:

P1: If one lacks reasons for believing P, then one has nothing that points in favor of it likely being true that P.

P2: Having nothing that points in favor of a belief being true renders one without justification.

C1: If one lacks reasons for P, then P is unjustified.

But would this not just be to say: If one lacks reasons for P, then one has nothing that points in favor of it likely being true that P? Would this not then beg the question? Yes, but that is not at all what we are saying. We must modify C1 to embody our semantic intention.

C1: If one lacks reasons for P, then one is rendered unjustified in holding P to be true.

This would just be to say: If one lacks reasons for P, then one’s holding of P to be true is rendered not likely true. It certainly can be the case that there is something that points in favor of P being true. It certainly can be the case that P is justified. However, what the skeptic is dealing with in the context of dialectical exchange is you and you’re beliefs. Nonetheless, I think that Huemer is wrong when he says, “it is unclear why anyone at all tempted by foundationalism would be moved to accept [my argument]” (2003, p. 147). I think this because, under my understanding of reasons, they do have a reason for thinking that certain propositions are arbitrary and others aren’t. No where do I care for arbitrariness. Unless arbitrariness entails being unjustified, I see no reason why they are mutual exclusive. Nonetheless, foundationalists have appearances regarding which of the two propositions, “1+1=2” and “Unicorns are on Mars,” are foundational. Appearances in my view are reasons in that, for the foundationalist of the Huemerian type, point in favor of their belief likely being true.

Huemer does try to respond to us by saying that all we are saying is that “In order for S to be justified in believing that P, S must have a reason for believing that S is justified in believing that P” (2003, p. 147). He says that this “is only plausible if one accepts [that] … [i]n order for S to be justified in believing that P, S must be justified in believing that S is justified in believing that P” (2003, p. 147). He then says that “[n]o clearer example of a level could be desired [than this]” (2003, p. 147). Thus, for our response to be a good one, we must dispel this notion of a level confusion. Though, I must say that if one really is a phenomenal conservatism or some other kind of foundationalist, please do ask yourself if it really seems to be the case that one can be justified in believing something without having anything that points in favor of that something being true. That really just seems absurd.

Before we move on to level confusions, I want to suggest one last move we can make. Simply dispensing with the language of justification. All we must is that what we mean by justified is having something pointing in favor of it being true. It seems that even Huemer meets this in that he says that foundational beliefs rest on foundational belief-forming methods and the fact that they rest on such is what makes them foundational which by definition is what points in favor of them being true. So, if Huemer just wants to say that he takes justification to mean having a seeming that P, then fine, but this is just a semantic move that does not actually meet the demands of what the skeptic is asking for.

Level Confusions

It seems that what those who bring up level confusions are objecting to is the idea that “no belief in a justification standard is epistemically justified” entails that “no belief is epistemically justified” (Sanger, 2000, p. 145). To parallel the view of Lydia and Timothy McGrew, this argument from Huemer, the level confusion argument, depends crucially on the presupposition that “there are no significant level connections left” when the JJ thesis, the view that one must be justified in their belief that they are justified, “has been rejected” (McGrew, 1997, “Level Connections in Epistemology,” p. 85).


[1]: I believe that the decisive question to raise to infinitists is “Why does an infinite chain of non-repeating beliefs confer justification?” To this question, they must either give an infinite chain of non-repeating beliefs or provide something other than an infinite chain of non-repeating beliefs. If they do the former move, it seems they have entered into a circularity (whether it be epistemic or logical circularity is not important). If they do the latter move, then they have abandoned infinitism. This is important because what it shows is that the demand for meta-justification cannot actually be met by infinitists as has been supposed by many such as Peter Klein. This is why I believe we will either have to turn to some case of epistemic circularity or a case of non-circularity (which, as just shown, infinitism cannot achieve). I must also note that I have brought up a case of a meta-regress against proponents of epistemic circularity (e.g., Michael Bergmann). I still think this objection holds some weight but only because Bergmann et al. use singular epistemic intuitions to judge those facilities they see as foundational-belief forming methods. The problem with this is then that one must have an epistemic intuition about the intuition they just had and so on. This problem, in my view, can possibly be solved by either bringing up William Alston’s idea of “level confusions” or one could argue that the epistemic intuition vindicates itself, so we would not need to move to another one. The former response may be stronger than the latter but only because of the limitations of the latter response, not because the former response has some particular strength to it (Timothy and Lydia McGrew have written a paper critiquing William Alston’s use of the idea of “level confusions”). The issue that may arise with the latter response is its seeming incoherence with how we introspectively observe our intuitions and then judgments arise. What I mean by this is that when I consider certain epistemic intuitions it does not seem to be the case that the intuition I am considering supports itself, rather I have a separate epistemic intuition judging it. So, even if, there is a case of epistemic circularity that seems intuitively non-problematic, an infinite regress regarding intuitive “judgments” will arise leading to a meta-regress, in that, the regress brings us to a higher and higher epistemic level each time a regression occurs. (When I come back to epistemic circularity later on in this paper I will attempt to address this objection of mine.)

[2]: If I don’t then see Lawrence Sanger’s dissertation “Epistemic Circularity.”



Evan Jack

How sweet terror is, not a single line, or a ray of morning sunlight fails to contain the sweetness of anguish. - Georges Bataille