People often resist engaging with some evidence because it conflicts with their preconceived notions. It is widely acknowledged that evidence resistance contributes to sexism and racism, vaccine hesitancy, and climate change denial and that it poses a serious threat to democracy (Strömbäck et al., 2022: 2–5; Glüer & Wikforss, 2022). For democracy to function, citizens must be reasonably informed about politics, society, and science, which, in turn, requires that they engage seriously with the available evidence.

Consider for instance vaccine hesitancy. Skepticism about, and refusal to, take vaccines can have disastrous consequences. The World Health Organization lists vaccine hesitancy as a top ten global threat, sharing the distinction with known killers like air pollution, climate change, and population displacement due to conflict and war.Footnote 1 Sometimes, vaccine hesitancy is due to people refusing to accept scientific evidence indicating that vaccines are safe and effective. By shielding themselves from available evidence, they can cultivate false and/or unfounded fears of adverse side effects (Taylor et al., 2014; Yaqub et al., 2014).

The problem with vaccine hesitancy and vaccine refusal is not only practical but also moral. Those who are hesitant to getting vaccinated or who outright refuse to do so promote their individual interest by benefitting from community’s protection from infection (herd immunity) without contributing to the community’s protection by becoming vaccinated themselves (e.g., Betsch et al., 2013; van den Hoven, 2012). This problem of free riding has not only a moral but also an epistemological dimension (John, 2011). Those who refuse to be vaccinated because they are worried about adverse side effects set their epistemic standard higher than the normal level accepted in the expert community. An epistemic standard “specifies the kind and degree of evidential support that some kind of proposition must enjoy before agents should accept that proposition” (John, 2011: 502). For instance, the epistemic standard for determining guilt in an American criminal trial is “beyond all reasonable doubt.” When the epistemic standards are raised, data which the experts take to be evidence for the safety and effectiveness of vaccines lose their probative force. The vaccine skeptic wrongs the community by refusing to abide by the epistemic standards set by the relevant experts. John calls this “epistemological free riding.” It is a special type of evidence resistance. He writes:

Ceteris paribus, agents act wrongly if they raise the epistemic standards which they use to guide their acceptance of some kind of proposition above the level normally agreed to be sufficient in their community, and this standard-shifting is premised on predictions about the costs and benefits of false acceptance in the context which assume that others will not shift their epistemic standards beyond the community accepted standard. (John, 2011: 513)

Given that evidence resistance can pose a serious threat to democracy and public health, it is not surprising that the phenomenon is also studied by epistemologists. In epistemology, evidence resistance is also goes by the name of “closed-mindedness” and “dogmatism.” Even though epistemologists usually disagree with one another, everyone agrees that evidence resistance (dogmatism) is a bad thing. It is bad not only because it promotes group polarization which undermines democracy but also because it is a paradigm of irrationality.

Cassam (2019: ch. 5) lists dogmatism among the epistemic vices, alongside prejudice, wishful thinking, overconfidence, and gullibility. Epistemic vices are defined as traits, attitudes, or ways of thinking that systematically obstruct knowledge. Even though a dogmatic attitude can protect you from misleading evidence, it obstructs knowledge in a systematic fashion and therefore qualifies as a vice. Battaly too thinks that closed-mindedness is usually an intellectual virtue. She characterizes closed-mindedness as “an unwillingness or inability to engage (seriously) with relevant intellectual options” (Battaly, 2018a: 262). Dogmatism, for her, is a kind of closed-mindedness: it is an unwillingness to engage seriously with relevant alternatives. Finally, Simion (2023) defines evidence resistance as a form of epistemic malfunctioning. Resistance to evidence amounts to violating the duty to believe, that is, “the duty to form a belief that p if there is sufficient and undefeated evidence for S supporting p” (Simion, 2023: 3).

Presumably, the reason evidence resistance is seen as an epistemic malfunctioning is that it flies in the face of the popular total evidence principle. There are two versions of the total evidence principle (Roush MS, n.d.). One version states that when forming beliefs, we should take into account all of the evidence that we currently have, not just some of it. As Carnap says ,

if a judge in determining the probability of a defendant’s guilt were to disregard some relevant facts brought to his knowledge; … then everybody would regard such a procedure as wrong.Footnote 2

The other version of the total evidence principle says that, by and large, we are epistemically better off if we acquire and assimilate more new evidence than we are if our evidence base is small. In other words, we may not only not ignore evidence we currently have, but we should try to consider as much of the available evidence as possible. Resistance to available evidence constitutes a breach of both versions of the total evidence principle. This may be the reason why evidence resistance is taken to be a paradigm of irrationality.

This paper attempts to do two things. First, it provides an analysis of the notion of evidence resistance. To this end, I first discuss the concept of evidence and then that of resistance. The proposed analysis of evidence resistance states that if S inquires as to whether p is the case, S resists some evidence e if (i) e is available to S, (ii) e constitutes a defeater for the belief that p, and (iii) S either fails to countenance e or is insensitive to e’s probative force. Second, I explore the possibility of evidence resistance being justified. The aim is to develop a principled reason for distinguishing justified from unjustified instances of evidence resistance. I argue that S is justified in resisting counterevidence e if and only if e is irrelevant.

1 The Notion of Evidence in “Evidence Resistance”

Everything that happens around us is evidence of something. Yet the fact that we are aware of only a fraction of what goes on around us does not mean that we are constantly resisting some evidence. The notion of evidence resistance presupposes that the evidence in question is relevant for some questions we are trying to answer. Evidence resistance is resistance to topic-specific evidence where the topic is of concern to the subject. S resists evidence e for p, if S does not engage seriously with e even though S inquires whether to believe that p, disbelieve that p, or suspend judgment about p.

For us to resist some piece of evidence e, e must not only be relevant to the inquiry we are engaged in but must also be available to us. We cannot resist evidence that is not available to us. For instance, before Ignaz Semmelweis developed the germ theory in 1846, the fact that puerperal fever in obstetrical patients occurred much more often when the physician had previously conducted an autopsy was not regarded as evidence for the importance of hand disinfection. Physicians were not resisting the evidence for the importance of hand disinfection since the evidence was not available to them. It was not available to them because they could not see it as such. Likewise, I am not resisting the evidence that a childhood friend of mine that I have not seen in years is called Patricia. Since I completely forgot her name, evidence to the effect that she is called “Patricia” is not available to me.

We can distinguish three dimensions of the availability of evidence: a qualitative, quantitative, and environmental dimension (Simion, 2023: 6). For evidence to be available to us, it has to be the type of thing we can access and process, it has to remain within the number of things we can access and process at a particular time, and it has to be easily available in our internal or external epistemic environment. The external environment is the physical and social surroundings; the internal environment is our mind.

In the context of the Gettier debate, epistemologists distinguish between possessed and unpossessed evidence.Footnote 3 In classical Gettier cases such as Gettier’s coin/job case (Gettier, 1963), the subject does not know that p because that the evidence on the basis of which they believe that p is false or is not suitably connected to the reason why the belief is true.Footnote 4 In failed threat cases such as Goldman’s (1976) fake barn scenario, the reason the subject does not know that p is that there is defeating evidence that is available but not possessed by the subject. If the subject became aware of the defeating evidence, the rational thing for them to do would be to disbelieve that p or to suspend judgement about p.

The notion of evidence in “evidence resistance” refers to possessed and unpossessed evidence alike. One can fail to engage seriously with evidence one is aware of. Depending on whether belief is taken to be a categorical or a graded state, this can happen in either of two ways: one can fail to form the attitude vindicated by the possessed evidence, or one can fail to adopt the confidence level vindicated by the evidence. Failure to engage seriously with unpossessed evidence amounts to failing to do due diligence in gathering sufficient evidence before forming a belief. The unpossessed evidence can support the proposition in question, but it can also take the form of an undercutting or rebutting defeater. In the fake barn case, for instance, the unpossessed defeater to the effect that the environment is littered with barn facades is an undercutting defeater for the belief that there is a barn in front of me.

Resistance to relevant yet unpossessed evidence constitutes a violation of Clifford’s principle: “It is wrong always, everywhere, and for anyone to believe anything on insufficient evidence” (Clifford, 1999). This principle expresses the core insight of evidentialism about the ethics of belief. The evidentialist thesis is that we are obliged to form beliefs always and only based on sufficient evidence in our possession. (Ignoring unpossessed evidence is permissible if we have reason to believe that it is very likely misleading.) How demanding the principle is depends, among other things, on how we understand the notion of evidence. If evidence is wholly constituted by mental states, as advocates of mentalism claim, Clifford’s principle states that one may not believe that p if there is no experience, memory, or belief in one’s mental repertoire (belief box) that speaks in favor of p. Mentalism is advocated by Conee and Feldman (2001). The mentalist claims that one’s beliefs are justified only by factors that are internal to one’s mental life is similar to access internalism but may not be confused with it. Access internalism states that the factors that determine propositional justification are reflectively accessible states. Mentalism also allows for reflectively inaccessible mental states (e.g., repressed beliefs) to qualify as evidence. It is of course an open question to what extent mental states that are not reflectively accessible qualify as being possessed.

Clifford’s principle is much harder to live up to if evidence is not restricted to what goes on in the subject’s head. Propositionalism about evidence, defended among others by Williamson (2000) and Neta (2008), states that all evidence consists in propositions. Williamson writes, “evidence is the kind of thing that which hypotheses explain. But the kind of thing that hypotheses explain is propositional. Therefore, evidence is propositional.”Footnote 5 A version of propositionalism is Arnold’s (2011) factualism, which claims that evidence consists in epistemically usable facts.

I will assume propositionalism about evidence. Evidence, the way I use the term, consists in true propositions or facts. To say that evidence is propositional is not to deny that mental states can function as evidence. Yet it is not the state of believing that q that is evidence for my belief that p, but rather it is the fact that I believe q or q itself that is the evidence.

How does propositionalism deal with misleading evidence? Consider the famous Tom Grabit case by Lehrer and Paxson (1969). I saw someone who looked like Tom Grabit steal a book from a library and thus formed the belief that Tom Grabit stole a library book. The man I saw was indeed Tom Grabit, and he did steal the book. Yet Tom Grabit’s mother claims that on the day in question Tom was out of the country but that Tom’s identical twin brother John was at the library. Thus, there is misleading evidence to the effect that I am wrong in believing that Tom stole the book. Given that propositionalism claims that evidence consists in true propositions, it does not seem to be able to qualify the mother’s testimony as rebutting evidence. The way around the apparent problem is to say that the rebutting evidence is not what the mother said but the fact that she said it. The misleading evidence is the true proposition Tom Grabit’s mother testified that on the day in question Tom was out of the country, but that Tom’s identical twin brother John was at the library.

Evidence must not only be true, but it must stand in a support relation to the target proposition. Evidence must have a bearing on the likelihood of the target proposition by either increasing, decreasing, or confirming the likelihood. The support relation in which the evidence stands to the target proposition can be probabilistic (Pr(p/e) > Pr(p)) or non-probabilistic (e entails p).

Does the notion of evidence resistance apply both to confirming and disconfirming evidence? In the context of evidence resistance, only disconfirming evidence is of interest. Resistance to evidence is always resistance to counterevidence. Failing to seriously engage with evidence that corroborates one’s point of view is not an instance of evidence resistance. Note that one’s point of view on the issue need not take the form of a belief. Evidence resistance does not require that one holds a belief about the subject matter in question. One can resist some piece of evidence while still being on the fence about whether to believe or disbelieve that p.

2 The Notion of Resistance in “Evidence Resistance”

Having analyzed the notion of evidence, we can now turn to the notion of resistance. Evidence resistance is a form of selective exposure. Selective exposure is the phenomenon whereby people choose to focus on information in their environment that is congruent with and confirms their current attitudes to avoid or reduce cognitive dissonance (Knobloch-Westerwick, 2014: 3). The causes of selective exposure can be conscious and intentional or unconscious and non-intentional. Typically, however, selective exposure is a result of confirmation bias, that is, an implicit tendency to favor information that confirms or supports one’s prior beliefs or values and ignore information that does not. Confirmation bias goes hand-in-hand with motivated reasoning which is the tendency to readily accept new information that agrees with one’s point of view and critically analyze that which does not.

What does the resistance to evidence amount to? As we saw above, Battaly defines closed-mindedness as “an unwillingness or inability to engage (seriously) with relevant intellectual options” (Battaly, 2018a: 262). The problem with this view is that evidence resistance seems to be fully compatible with someone seriously engaging with relevant intellectual options. Consider, for instance, a conspiracy theorist about 9/11 being a government job. Typically, a conspiracy theorist about 9/11 knows much more about the subject matter than the rest of us. They spent many hours researching the topic in an effort to come up with (deviant) explanations for all of the relevant data points. Although it would not be fair to accuse the conspiracy theorist of not engaging seriously with the subject matter, they are nevertheless guilty of evidence resistance. They are resisting the evidence by reinterpreting disconfirming evidence as confirming evidence. Conspiratorial beliefs are immune to revision by counterevidence because the counterevidence is reinterpreted as being corroborative evidence.Footnote 6

To resist some piece of evidence is to do one of two things—to not treat it as evidence or to not respond to it in the appropriate way. I call the former kind of evidence resistance evidence exclusion and the latter kind evidence insensitivity.

Evidence exclusion consists in the failure to acknowledge something as evidence. This can take two forms. One can fail to acknowledge that some proposition is true, or one can fail to acknowledge that it has an epistemic bearing on the belief under consideration. Both strategies have the same result, namely, the exclusion of selected propositions from one’s total evidence. The goal is the avoidance or reduction of cognitive dissonance.

Note that evidence exclusion contradicts the total evidence principle. If a belief of mine is challenged by some proposition, I can deny the status of the proposition as evidence or I can include the proposition in my total evidence and assign it low probative value thereby rendering it innocuous. The latter strategy is compatible with the total evidence principle, but the former is not.

Evidence insensitivity consists in the subject acknowledging the evidence as such but not responding to it in the appropriate way. This can take two forms. One can be insensitive to the evidence by not forming the attitude vindicated by the evidence (in the case of categorical belief) or by not adopting the confidence level vindicated by the evidence (in the case of graded belief).

In sum, if S inquires as to whether p is the case, S resists some evidence e if (i) e is available to S, (ii) e constitutes a defeater for the belief that p, and (iii) S either fails to countenance e or is insensitive to e’s probative force. This account of evidence resistance differs from that of Battaly in that it straightforwardly applies to conspiracy theorists. Another feature of the proposed account is that it does not make a value judgment. It is not part of the account that evidence resistance is an epistemic malfunction. This sets the proposed account apart from that of Simion.

3 The Dogmatism Paradox

Given the concept of evidence resistance developed so far, we will now explore the question of whether evidence resistance is necessarily an epistemic malfunction or whether it can be rationally permitted. I will argue that one is justified in resisting counterevidence if and only if it is irrelevant.

Many cases of permissible evidence resistance discussed in the literature are such that what is good about it is not epistemic, but moral or legal. For instance, evidence resistance is clearly called for when the appropriation of the evidence would violate someone’s right to privacy. A right to privacy is a right to limit public access to oneself and to information about oneself. Just as evidence resistance can be used to treat others unjustly, it can be used to respect their dignity (cf. Driver, 1989; Manson, 2012; Matheson, 2013). Yet what makes it permissible to resist private information are not epistemic but moral considerations.

Next, consider an example where evidence resistance is justified legally. Suppose there is conclusive evidence for the accused to be guilty and suppose that the jurors know about the evidence, but since there has been a break in the chain of custody, they are asked by the judge to not use (bracket) the evidence in their deliberations. Here, evidence exclusion is mandated by the legal system but does not serve an epistemic goal.Footnote 7

Can evidence exclusion be justified on epistemic grounds only? Battaly (2018b) gives an example that is supposed to do just this. She imagines a knowledge-possessing agent who finds herself in an epistemically hostile environment, that is, an environment that is utterly saturated with intellectual options that are false, unreliable, or aimed at misdirection. According to Battaly, the agent has epistemic reasons to be closed-minded—to be unwilling to engage seriously with relevant intellectual options that conflict with what she already knows. “In an epistemically hostile environment, closed-mindedness is an effects-virtue.”Footnote 8

The reason why in this case evidence resistance is epistemically justified is that the subject knows that the information environment is corrupted. The subject knows in other words that the evidence she resists is misleading. What justified the resistance to the evidence is her knowledge of the misleading nature of the evidence. In this respect, Battaly’s example resembles Kripke’s (2011) famous dogmatism paradox.

The dogmatism paradox states that knowledge induces dogmatism because, if you know that p, you know that any evidence against p is misleading and can thus be ignored. By the same token, you know that any positive evidence would be superfluous since you already know that p. Hence, it is rational for you to close your mind to any new potential evidence regarding p. But this is paradoxical because “[one] is never in a position simply to disregard any future evidence even though [one] do[es] a great many different things” (Harman, 1973: 148).

The debate about the dogmatism paradox concerns the question whether the closed-minded dismissal of counterevidence causes the subject to lose her knowledge that p (Battaly, 2018b: 35). According to Cassam (2019: 114–7), the dismissal of counterevidence robs the subject of her justification for p and consequently of her knowledge of p. Fantl (2013, 2018), on the other hand, maintains that knowledge that p is compatible with the subject not being able to refute counterevidence to p. Sosa (2014) defends a position in the middle. The closed-minded dismissal of counterevidence is said to be compatible with animal knowledge but undermines reflective knowledge. I side with Fantl.

Fantl defends forward-looking dogmatism, which is the view that “knowledge can survive closed-minded dismissal of apparently flawless relevant counterarguments” (Fantl, 2018: 32). The reason for why knowledge is resilient in this way is twofold. The first reason is the principle of modesty which states that for any controversial proposition one knows, there could easily be an apparently flawless relevant counterargument (Fantl, 2018: 35). The second reason is that there is no relevant difference between the possibility of there being an apparently flawless counterargument and there actually being one. Moreover, an actual apparently flawless counterargument does not defeat one’s knowledge when the argument’s apparent flawlessness is unsurprising because “unsurprising arguments are often impotent” (Fantl 2019: 40).

I will not argue for forward-looking dogmatism. Others have done this already. Instead, I assume the truth of forward-looking dogmatism and explore a related issue. The dogmatism debate presupposes that the subject knows that p. This is why the counterevidence to p is misleading. But what about cases where the subject is not sure that they know that p? Can it ever be rational to resist counterevidence to p even when it is not clear that the subject knows that p? It is possible to reverse the direction of the argument: instead of inferring from the knowledge that p to the dismissal of counterevidence, can one infer from the dismissal of counterevidence to the knowledge that p? I will answer this question in the affirmative.

4 Resistance to Irrelevant Counterevidence

According to the relevant alternative theory, a belief counts a knowledge if it is true and justified (reason-based) in a way that eliminates all relevant alternatives. Irrelevant alternatives can be properly ignored. Which alternatives can and cannot be properly ignored depends on the context of the subject. In ordinary contexts, only ordinary alternatives are relevant. But in some contexts (e.g., skeptical scenarios), a broader range of alternatives counts as relevant. With respect to many propositions, we can rule out ordinary alternatives, but not some of the more exotic ones. Hence, we can know things in ordinary contexts that cannot be known in exotic contexts.

Consider Dretske’s (1970) famous zebra/mule case. At the zoo, you ordinarily know that the striped animals in the pen marked “zebras” are zebras. You know this because ordinarily the relevant alternatives are that the animals in the pen are giraffes, or elephants, or lions, or some other animal typically found in zoos. You have conclusive evidence that these animals are zebras rather than giraffes (or elephants, lions, etc.), so you know they are zebras. But suppose it became a relevant alternative that the animals were cleverly disguised mules made to look just like zebras. Since you have no evidence for the claim that the animals are zebras rather than cleverly disguised mules, you do not know that they are.

What makes an alternative relevant? It has proven quite difficult to answer this question with any degree of precision and generality. There is an objective and a subjective reading of relevance. In the objective reading, the relevance of an alternative has to do with the objective probability of its realization. If there are hoaxing zoos in one’s vicinity, then this could render it a relevant alternative that what looks like a zebra is in fact a disguised mule. It does not matter whether one is aware of the prevalence of hoaxing zoos. In the subjective reading, an alternative is relevant if the believer takes it to be probable. Here, relevance depends on the assumptions, purposes, and intentions of the subject who makes the knowledge claim and on the “community” in which he operates.Footnote 9

When the relevant alternative framework is applied to the question of when evidence resistance is epistemically justified, the following view emerges: S is justified in resisting counterevidence e if and only if e is irrelevant. Evidence resistance whereby you keep your belief that p is justified when all the relevant alternatives that not-p are ruled out by the evidence you already have. Irrelevant alternatives need not be ruled out. Evidence resistance whereby you suspend belief is justified if and only if some relevant alternatives are p and some are non-p. It does not matter if the irrelevant alternatives are all p or non-p. This account of justified evidence resistance applies both to evidence exclusion and to evidence insensitivity.

A feature of the proposed account of justified evidence resistance is that it can accommodate Fantl’s position of forward-looking dogmatism. If one knows that p, then one knows that counterevidence to p is irrelevant and thus can be properly ignored. But it is not necessary to know that the counterevidence is irrelevant to know that p; it is enough that the counterevidence is irrelevant.Footnote 10