I. Bergmann on evidence

Bergmann’s anti-skepticism takes epistemic intuitions as the basis for a justified rejection of skepticism. Epistemic intuitions, for Bergmann, are seemings: conscious mental states with propositional content whose distinguishing feature is that they present or reveal the world as being a certain way (Bergmann, Reference Bergmann2021, p. 133). Seemings can have a wide range of contents. To name a few examples: memory seemings present certain experiences to us in a way that seems to be connected to events in our past, perceptual seemings present certain experiences to us in a way that seems to be connected to our immediate environment, and a priori seemings present a priori propositions to us as true or necessary.

Seemings are not infallible. It can seem to me that my old apartment was painted beige when it was really wallpapered, or that my son is the most talented trumpet player in the school band when he is in fact average. Seemings are distinct from beliefs (since it can seem to you that P without your believing P, as when you know that you are experiencing a visual illusion), and from inclinations to believe (since you can be inclined to believe that P for reasons that are unrelated to whether it seems to you that P) (Bergmann, Reference Bergmann2021, p. 131).

Epistemic seemings (i.e., epistemic intuitions) are seemings about epistemic matters, such as epistemic principles and the epistemic statuses of individual beliefs (Bergmann, Reference Bergmann2021, p. 135). An example of the former: it seems to me that, when confronted with the disagreement of an epistemic peer who has the same evidence as I do and whose belief I judge to be rational, I should decrease my confidence. An example of the latter: it seems to me that my current (perceptual) seeming that there is a gold-colored floor lamp a few feet away from me is veridical and provides warrant for my belief with the same content.

Bergmann joins a camp of contemporary epistemologists who maintain that one’s seemings are evidence. There are good reasons for adopting this view. One argument that Bergmann endorses is as follows:

Seemings are evidence

1. 1. Evidence is that which is the basis for one’s beliefs.Footnote ¹

2. 2. Seemings are the basis for many of one’s beliefs.

3. 3. Therefore, seemings are evidence.

Premise 1 expresses a standard assumption about evidence shared by many epistemologists. Premise 2, that seemings are the basis for many beliefs, draws support from reflection on paradigmatic instances of ordinary belief formation. My belief that there is a gold-colored lamp a few feet away is based on my visual experience of the lamp, which itself is a seeming—a conscious mental state with propositional content that presents the world as being a certain way.

One might question whether seemings play the role just described, positing instead that my perceptual and memory beliefs are based on something else, such as true propositions or mind-independent objects. If this is right, then Premise 2 in the above argument lacks support. Against this objection, Bergmann (Reference Bergmann2021, pp. 11, 24–25) offers the following argument:

Beliefs are based on seemings

1. 1. Part of what it is for a belief to be based on something is for the belief to be responsive in the right way to that thing.

2. 2. Our beliefs are not responsive in the right way to true propositions or physical objects, but they are responsive in the right way to seemings.

3. 3. Therefore, our beliefs are not based on true propositions or physical objects, but on seemings.

What is it for our beliefs to be “responsive in the right way?” Why are our beliefs not responsive to physical objects or true propositions in that way? Bergmann elaborates:

In this case, your belief that there is a spherical object in front of you is not responsive to physical objects (i.e., the basketball) or true propositions (i.e., the proposition that there is a basketball in front of you), because you continue to hold the belief even when the relevant physical object is annihilated and when the relevant proposition ceases to be true. Instead, your belief is responsive to your visual experience. And because your belief is responsive to your seeming and not to physical objects or true propositions, it is based on the former and not the latter. Call this line of reasoning in support of the conclusion that seemings are evidence, which begins from the assumption that evidence is that which beliefs are based on, and proceeds to the conclusion that seemings, not true propositions or external objects, are evidence, the basing consideration. The basing consideration, if sound, rules out any conception of evidence according to which evidence is factive (in Nicolas Silins’ (Reference Silins2005) terminology, any externalist conception of evidence).

A second reason for endorsing the view that one’s evidence primarily consists in one’s seemings over an externalist conception, and one which figures powerfully in Bergmann’s motivation (Chapter 1, Section 1.3), is that doing so enables one to affirm the New Evil Demon (NED) intuition—i.e., the intuition that you and your demon-victim twin (someone who is exactly like you with respect to her non-factive mental states but who is the subject of radical deception by a powerful demon) possess the same evidence and are equally justified in your beliefs. Someone who believes that evidence consists in true propositions must deny that you and your demon-victim twin have the same evidence in this case since propositions that are true for you will be false for your twin. And given the status of evidence as that which justifies beliefs, it also seems to follow that your beliefs and your twin’s are not equally justified. But since this is an unintuitive verdict, it is more plausible to reject the externalist conception of evidence in favor of the view that evidence supervenes on one’s non-factive mental states, such as one’s seemings.

Conceptualizing seemings as evidence allows the intuitionist particularist to say that, in paradigm cases, one’s anti-skeptical beliefs are based on and justified by one’s anti-skeptical epistemic intuitions, which are themselves evidence against skepticism. Jill’s belief that she is not in the Matrix is justified by, for example, her intuition that it is irrational to take the possibility that she is in the Matrix seriously; or her intuition that her current perceptual seemings are veridical; or her seeming that, in general, her faculties of perception are reliable. For many people, these epistemic intuitions are both extremely strong (i.e., very compelling), and stronger than conflicting intuitions in favor of epistemic principles that entail skepticism. When that is the case, Bergmann (Reference Bergmann2021, pp. 124–126, 253, 258) argues that one is justified in believing in accordance with one’s stronger epistemic intuitions and rejecting skepticism. Strong epistemic intuitions thus form the evidential basis for a justified rejection of skeptical hypotheses in all their forms.

II. What counts as evidence?

One challenge for the view that one’s evidence is exhausted by one’s seemings is that it rules out the possibility of two individuals possessing the same evidence in all but the most contrived cases. Yet, it is a standard assumption of several epistemological research programs, such as the epistemology of disagreement and the permissivism/impermissivism debate, that peers sometimes possess the same evidence, or roughly the same evidence, concerning some matter. So, those who believe one’s evidence consists primarily in one’s seemings will have to accept that two people cannot share the same evidence (Jackson, Reference Jackson, McCain, Stapleford and Steup2023).

A further challenge for the view is that it is at odds with how we normally talk about evidence in scientific, legal, and everyday contexts (Kelly, Reference Kelly2008). In these contexts, it is normal to think and speak of facts, or even objects in the world, as part of the evidence that bears on a particular question. For example, the bending of light in gravitational fields is evidence for Einstein’s theory of general relativity; the CCTV footage is evidence that supports the defendant’s guilt; and the cereal boxes and spilled milk on the kitchen table are evidence that my son has already eaten breakfast. However, if one’s evidence consists primarily in one’s seemings, these claims about evidence are strictly speaking false. The proponent of seemings as evidence will have to endorse an error-theoretic explanation for why our common ways of thinking and speaking about evidence, when taken at face value, entail that facts and objects are evidence.Footnote ²

This mismatch between the theoretical commitments of the one who views seemings as evidence and how we ordinarily think and speak of evidence can be drawn out by considering cases in which one’s beliefs change without a corresponding change in the information one has that bears on those beliefs. Consider the following case, adapted from Julia Staffel (Reference Staffel2021a, Reference Staffel2021b):

Detective

What should the proponent of seemings as evidence say about the state of Sam’s evidence in this case? First, consider how he will characterize the evidence in Sam’s possession on Wednesday night. Sam’s evidence includes various perceptual seemings (e.g., of the murder weapon, the blood spatter report, the ballistics report, and the notes summarizing interviews with witnesses). It also includes memory seemings, for example, of his own experiences at the scene of the crime, of interviewing witnesses, and so on. And importantly, it will also include many inferential seemings—seemings that connect existing facts with likely conclusions (Bergmann, Reference Bergmann2021, pp. 103–105). In Sam’s case, these inferential seemings may take the form of conscious mental states that present to him as true the content that some hypothesis is the best explanation of his total information, or the content that certain pieces of information make some hypotheses more likely than others.

Now, consider what the proponent of seemings as evidence will say about the state of Sam’s evidence on Thursday morning, at the conclusion of his inquiry. His perceptual and memory seemings remain the same. However, many of Sam’s inferential seemings will differ. This follows from the fact that ex hypothesi, Sam’s continued evaluation of the information relevant to the crime has the result that it seems fairly clear to him that Ms. Y is the perpetrator. While some of his inferential seemings will remain the same (e.g., the seeming that the ballistics report makes it very unlikely that Mr. Z could have committed the murder), others will differ (e.g., his seeming that the hypothesis that Mr. X is the murderer best explains the total information concerning the case).

This case is one instance of a class of cases in which sustained reflection on some matter produces a change in opinion, although no new information is acquired. Cases abound. It sometimes happens that I interpret my physiological state as one of hunger, but after attending more carefully, I realize that I’m actually thirsty. I think about a tricky logic or probability puzzle and at first one solution seems most plausible, but on closer inspection, another solution seems right.

If one’s evidence supervenes on one’s seemings, then these are all cases in which changes in my beliefs are all based on and explained by changes in my evidence. And yet, intuitively, one’s evidence does not change in this case. In Sam’s case, his evidence consists of the crime scene reports, witness testimony, and so on. This fits well with the fact that, when explaining his change in view to a colleague, it would be appropriate for Sam to say something like: “After a closer look, I saw that the evidence actually supports Ms. Y’s guilt.” It would seem odd, or even misleading, for Sam to say instead: “The original evidence supported Mr. X’s guilt. But then, the evidence changed so that it supported Ms. Y’s guilt”; or: “…then, I got new evidence that supported Ms. Y’s guilt.” Yet, given the view that seemings are evidence, the latter utterances are more accurate.

Bergmann might respond to this argument by conceding that there is a mismatch between how we ordinarily speak and think about evidence and what our evidence consists in, but that accepting this mismatch is required in the light of powerful theoretical considerations against externalist conceptions of evidence, such as the basing consideration and the NED intuition. In the remainder of this section, I raise some concerns about the basing consideration. In the final section of this article, I will raise some questions about how Bergmann’s framework handles philosophical disagreements and its implications for the evidential role of the NED intuition.

The basing consideration begins from the assumption that evidence is that which our beliefs are based upon, and argues that our beliefs are not based on, because not properly responsive to, external-world objects or true propositions. However, a belief’s being based on something, or failing to be based on something, does not tell us definitively about that thing’s status as evidence. On one hand, a belief’s being based on something is not sufficient for that thing to count as evidence, because there are many instances in which a person’s beliefs are based on that which is not evidence. For example, a person’s belief that God exists can be based on her hope that God exists; or a person’s belief that Qijiamei is not a promising candidate for a PhD in philosophy can be based on his prejudicial assumption that non-native English speakers do not have an aptitude for philosophical reasoning in the contemporary analytical style. Yet, hopes and biases are not evidence.

Moreover, a belief’s failing to be based on something does not entail that that thing does not count as evidence. Consider Kornblith’s (Reference Kornblith1980) case of Alfred, who has strong evidence for believing P, and strong evidence for believing If P then Q. Alfred also believes Q, but he does not believe it because he sees the inference from his other beliefs via modus ponens, he believes it because he enjoys the sound of the sentence that expresses Q. In this case, Alfred’s belief that Q is not based on his beliefs that P and If P then Q, but that does not prevent P and If P then Q from being evidence Alfred has for Q. This reasoning should be acceptable to evidence internalists and evidence externalists alike.

What the foregoing considerations show is that a belief’s being based on something is neither a necessary nor a sufficient condition for that thing’s being evidence. The evidence externalist may accept that many of our beliefs are neither responsive in the right way to true propositions nor based on them, but maintain that, for all that, the true propositions are still evidence. If beliefs being based on something is neither a necessary nor sufficient condition for that thing counting as evidence, it weakens the probative force of the basing consideration for the conclusion that seemings are evidence.

It seems to me that these observations show that the basing consideration isn’t a particularly strong reason to favor internalist conceptions of evidence over externalist conceptions. This leaves us with the other major theoretical consideration in favor of an internalist conception of evidence: the NED intuition. Affirming the NED intuition must be weighed against the costs of accepting a seemings conception of evidence identified in this section: acknowledging a deep incongruity between how we speak and think about evidence and what our evidence actually consists in; and accepting that shared evidence is virtually impossible. I suspect that Bergmann may be happy to accept these costs. However, as I will argue in Section III, given his own commitments about the evidential role of epistemic intuitions, it is unclear why affirming the NED intuition should play a decisive role in adjudicating the best response to skepticism.

III. Disagreement

Although Bergmann argues that the anti-skeptic is justified in her beliefs because she believes in accordance with her strongest seemings, he offers some important caveats to the view that one is, in general, justified in believing in accordance with one’s strongest seemings. For instance, he acknowledges that epistemic intuition is not infallible, and that one is not justified in believing in accordance with one’s strongest seemings when those seemings “don’t survive critical reflection,” as when a belief based on my memory seeming of having left the car keys on the hook by the door is defeated by the recognition (itself based on other memory seemings) that in the past, similar memory seemings have frequently been misleading (Bergmann, Reference Bergmann2021, p. 217). In such cases, stronger intuitions ultimately override weaker intuitions.

What about conflicting epistemic intuitions held by others? Can these serve as defeaters? Bergmann argues that second-order evidence about whether one’s presumed peer formed their beliefs in a “reliable and non-misleading way” will be probative here. In cases where one has equally good reason to think that a presumed peer’s belief was formed reliably as one does to think that one’s own belief was formed reliably, then it is not rational to demote one’s peer and one has a defeater for one’s belief. In cases where one’s (second-order) evidence strongly supports the conclusion that one’s own belief was formed reliably, but at best weakly supports the conclusion that one’s presumed peer’s belief was formed reliably, the rational response is to demote one’s peer (Bergmann, Reference Bergmann2021, p. 241).

In the case of disagreement between the skeptic (or the skeptically inclined) and the intuitionist particularist, Bergmann argues that demoting the skeptic is the rational response. The intuitionist particularist will have both very strong evidence (in the form of her first-order epistemic intuitions) that many of her perceptual beliefs are justified, and very strong evidence (in the form of other second-order epistemic intuitions) that her first-order epistemic intuitions were formed in a reliable way. Further, when the intuitionist particularist considers whether the skeptic’s beliefs were formed reliably, it will seem to her that they were not. Rather, it will seem to her that the skeptic’s epistemic intuition that he lacks knowledge of perceptual matters, or that he should take skeptical hypotheses or principles that entail skepticism seriously, is much stronger than it should be. Therefore, the rational response is demotion (Bergmann, Reference Bergmann2021, pp. 244–245).

The concern is that this view of disagreement may make adopting a dismissive attitude toward those who disagree with you too easy in many realms. First, notice that the view that Bergmann defends makes the demotion of a presumed peer very easy in many cases. In each case of disagreement—not only in epistemic matters but in far more mundane matters—by following the formula that Bergmann describes for assessing the epistemic significance of disagreement from a presumed peer, I will be led to consider whether my first-order evidence, consisting of my own intuitions, is a reliable and non-misleading basis for my first-order beliefs. There will be some instances in which I conclude that my own intuitions are unreliable. However, the most natural candidates for cases in which I judge that my seemings were unreliably formed are ones in which there is a verifiable track record for the kind of seeming in question. When one’s memory seeming about the date of one’s brother’s birthday or one’s a priori seemings about whether certain logical sentences are tautologies have been shown to be frequently incorrect in the past, the rational response to disagreement in these quarters will usually not be to demote a presumed peer who disagrees.

However, many categories of seemings do not tend to generate the kind of information about track records that would count as a reason to think that a particular seeming in that category was formed in an unreliable way. Our seemings about philosophical, political, moral, and religious matters are sometimes very strong. But unlike the examples of seemings given in the previous paragraph, experience rarely provides us with the kinds of data about the accuracy of these seemings that would produce a rational belief in their unreliability or misleading nature.Footnote ³ It is not easy—perhaps it is even impossible, in many cases—to see when these kinds of seemings were formed in an unreliable and misleading way. Absent definitive evidence that a particular seeming of mine belonging to one of these categories was formed in an unreliable or misleading way, it will presumably seem to me that the seeming in question was formed reliably and in a non-misleading way. And if it seems to me that my own seemings on some matter were formed in a reliable and non-misleading way, then it will likely also seem to me that a presumed peer whose seemings conflict with mine has seemings that are misleading. Therefore, in cases of disagreement about philosophical, political, moral, and religious matters, the demotion of a presumed peer will nearly always be the rational response.

The fact that Bergmann’s view appears to entail that demotion is frequently the rational response to disagreement is not itself objectionable; perhaps we are frequently justified in demoting our peers. Rather, the concern derives from the fact that these many instances of rational demotion involve thinking of one’s presumed peer as having faulty and misleading epistemic intuitions, and hence faulty and misleading evidence (Bergmann, Reference Bergmann2021, pp. 238, 244–245).

This push to conceptualize a much larger proportion of disagreements as based on differences in evidence discourages fruitful dialogue and may encourage a dismissive attitude toward demoted peers. In cases where one demotes one’s peer on the basis of a rational belief that they have faulty and misleading intuitions, it is also usually appropriate to view one’s peer’s belief as a flawlessly rational response to her evidence. Where a commitment to skepticism is based on the skeptic’s epistemic intuitions in favor of certain epistemic principles that entail skepticism, the intuitionist particularist will view the skeptic’s evidence as flawed, but his skeptical belief as perfectly rational, given his evidence. This is an unfortunate consequence because it discourages mutual efforts to understand and persuade. Seeing someone as in the grips of a mistaken interpretation of the evidence, or as holding a belief on the basis of mistaken reasoning, opens the door for fruitful attempts at correction. Seeing someone as in possession of misleading evidence that consists of fundamentally flawed intuitions, by contrast, discourages dialogue, because, unlike argument and inference, seemings are (presumably) not the kinds of things that respond to attempts at rational persuasion. Moreover, it is unclear how to begin to go about getting another person to change their seemings, or even to settle on a mutually agreeable standard for evaluating which seemings should be trusted. It’s difficult to engage in fruitful dialogue when “that’s the way it seems to me” is justifying evidence.

Finally, there is some reason to think that the evidential role that epistemic intuitions play in Bergmann’s version of anti-skepticism undercuts his claim that an intuitionist particularist response to skepticism is superior to other non-inferentialist responses that have been developed. Bergmann argues that his intuitionist particularlist response to skepticism is superior to that of externalists such as Williamson (Reference Williamson2000) and Pritchard (Reference Pritchard2012), because externalist solutions underestimate the appeal of radical skepticism by being willing to deny the NED intuition. As Bergmann puts it:

Why should preserving the NED intuition be a requirement of any adequate response to skepticism? Seen from within the intuitionist particularist framework, the answer must be that it is a powerful epistemic intuition, strong enough to outweigh conflicting intuitions, and as such constitutes powerful evidence for Premise 4 in the Underdetermination Argument Against Perception. Yet, still, within the intuitionist particularlist framework, the NED intuition will only count as evidence for those who experience a powerful epistemic intuition that you and your demon-victim twin are equally justified in your perceptual beliefs. Those who lack a strong epistemic intuition about the justificatory status of the twin’s beliefs will also lack strong reasons for preferring the intuitionist particularist response to skepticism over externalist responses to skepticism. The objection is not that the intuitionist particularist cannot say anything to persuade the externalist of the superiority of the intuitionist particularist’s response (a point that is emphasized within Bergmann’s “autodidactic” approach [Bergmann, Reference Bergmann2021, pp. 146–147]). Rather, the objection is that, by the lights of the intuitionist particularist’s own framework, externalist responses to skepticism are just as evidentially supported as the particularist’s is, and therefore the latter version of anti-skepticism has no claim to superiority.