Unpossessed Evidence: What’s the Problem?

Pat is a precocious 16 year old; her reading varies widely, and when she is not devouring War and Peace or the Autobiography of Alice B. Toklas, she picks up scientific works that are listed in a great books program. One of these books is B.F. Skinner’s Science and Human Behavior. It is the first work on psychology Pat has read all the way through and she is fascinated by Skinner’s ambitions, struck by his confident tone, and carried away by his arguments. Upon reading it, Pat becomes enthralled with the idea of behaviorism; she spends months reading and reflecting daily on Skinner’s book, looking for whatever flaw she can find in the argument. At the end of several months' time, she becomes a convinced behaviorist.

Is Pat’s belief in radical behaviorism justified? There is good reason for saying that it is. After all, Pat has understood Skinner's arguments, and, taken alongside Skinner’s authority, they provide prima facie reasons for accepting behaviorism. Furthermore, Pat can’t see what’s wrong with these arguments. Every objection she forms in the midst of her reading is soon dispatched by Skinner in the following paragraphs and chapters or by herself on further reflection.

Suppose, however, that one day Pat hears that a scientist named Noam Chomsky stridently attacked Skinner’s behaviorism, and some psychologists and linguists have been convinced by these criticisms. Pat does not know the content of Chomsky’s criticisms. What she does know is that there is evidence that she doesn’t have that some scientists take to refute behaviorism. Awareness of this unpossessed evidence seems to defeat the reasonableness of Pat’s belief in behaviorism.

In his 2015 paper, “The Significance of Unpossessed Evidence” and again in his 2019 book Knowing Our Limits, Nathan Ballantyne draws attention to the ways in which our awareness of unpossessed evidence can threaten the rationality of our beliefs. According to Ballantyne, it can be difficult to square rational belief with the recognition that our evidence is only part of a much larger whole. This is what Ballantyne calls the “problem of unpossessed evidence”. Ballantyne expresses the problem of unpossessed evidence in the form of three different arguments: the meta-defeater, overlooked defeater, and doubtful fairness arguments. These arguments are supposed to describe how reflecting on unpossessed evidence can defeat the rationality of our beliefs.

If successful, Ballantyne’s arguments would greatly limit the domain of rational belief. This is because we normally lack most of the evidence that bears on the truth of a given question. This is obviously true for lay people when it comes to technical issues; but even experts might doubt whether they have in their possession most of the evidence bearing on an issue. On the one hand, hyper-specialization ensures that general experts will not have most of the evidence readily available to hyper-specialists. On the other hand, oftentimes the evidence bearing on a question will be effectively unlimited. For instance, each instance confirming (or disconfirming) a theory can be taken as evidence speaking in favor of (or against) the theory; however, there seems to be an effectively unlimited number of such instances for general theories.¹

A number of philosophers have noted Ballantyne’s treatment of unpossessed evidence. However, nothing has been published critically assessing Ballantyne’s arguments. This paper fills this gap. I will give reasons for thinking that Ballantyne’s arguments expressing the problem of unpossessed evidence do not threaten the rationality of our beliefs, even our controversial beliefs. Nevertheless, I think that Ballantyne is right that the problem of unpossessed evidence is a real problem for anti-skeptics. I give a new argument, which I call the argument from respectable counter-evidence that I think better expresses the problem of unpossessed evidence. Anti-skeptical philosophers who are interested in protecting the rationality of controversial beliefs should focus on this argument.

The structure of the paper is as follows. In the first section, I briefly lay out the problem of unpossessed evidence in an intuitive way. In the next three sections, I lay out Ballantyne’s arguments expressing the problem of unpossessed evidence and criticize them. Finally, in the fifth section, I lay out my own argument, the argument from respectable counter-evidence, highlighting some of the strengths of this argument, and considering some possible responses to it.

1 The Intuitive Problem of Unpossessed Evidence

Let’s call the case of Pat’s reading Science and Behavior and adopting radical behaviorism, the PRECOCIOUS READER case. While Pat has some prima facie reasons for accepting behaviorism, when she reflects on the evidence she is lacking that bears on the question, her outright belief in behaviorism seems to be defeated. Once she is aware that there is relevant evidence that she does not possess, it seems that it would be irrational for her to believe outright that behaviorism is true.

Lacking so much of the total evidence is not a predicament that is particular to youth. People of all ages can learn that they lack large amounts of evidence relevant to determining whether a given proposition is true. Ballantyne motivates the problem of unpossessed evidence by two such cases, which he calls the LIBRARY and WWW cases respectively.

LIBRARY: Sammy is wandering among rows of bookshelves at the library. These books concern matters about which she holds views. But Sammy has read only a few of them. Let’s imagine she thinks that free will and determinism are compatible, having studied a dozen journal articles and a couple of books years ago in graduate school. Scanning the shelves here, Sammy appreciates that there are several dozen titles relevant to the question of whether compatibilism about free will is true. Some books contain arguments against her view. Sammy hadn’t considered this mass of work until now and she hasn’t yet looked at it. (modified from Ballantyne 2019, p. 173)

WWW: Fifteen years ago, Holden thought carefully about economic ideas and arguments. Then life changed. With a family and busy job, he hasn’t kept pace with recent discussions. Now he wonders what has happened in the intervening years, so Holden searches on Google and JSTOR with some relevant keywords (‘government spending economic growth’) and his searches return thousands of results. As he quickly recognizes, there are hundreds of articles and books, all potentially relevant to figuring out what to think about this one economic issue, about which Holden once had carefully considered views. He knew the arguments, replies, and counter-replies, and he had a good rationale for his favored positions. It’s evident some of the recent discussion challenges Holden’s thinking. But he has not studied any of it. Rehearsing his earlier rationale, it still seems perfectly right to him, but he now knows of new evidence he does not have. (modified from Ballantyne 2019, p. 173)

In the LIBRARY and WWW, as in the PRECOCIOUS READER case, we have individuals who have prima facie reasons for holding some belief. Furthermore, at the beginning of the cases they are aware of no defeater for their beliefs. Thus, there is some reason for thinking that their beliefs at the beginning of these cases are justified. Things seem to change, however, once these individuals begin to reflect on evidence they don’t have. Somehow it seems that reflecting on all of the evidence they don’t have makes it irrational for them to hold their beliefs, at least outright. But how exactly does it do this? Ballantyne gives three arguments that are supposed to show three different ways in which reflecting on unpossessed evidence can defeat belief. These are the meta-defeater argument, the overlooked defeater argument, and the doubtful fairness argument. In the next section I will critically examine each of these arguments.

Before doing so, however, I should note that I will be focusing on whether Ballantyne’s argument provide full defeaters for outright belief.² Full defeaters for outright belief are set in contrast to partial defeaters. If I have a partial defeater for my outright belief that p, then I am less justified than I otherwise would be in having my outright belief that p. Perhaps, I should hold my belief with a little bit less confidence. Nevertheless, my outright belief that p might still be justified simpliciter. However, if I have an undefeated full defeater for my outright belief that p, then my outright belief that p is not only less justified, it is not justified simpliciter. That is, it is no longer rational for me to outright believe that p. While Ballantyne presents his arguments as being ambiguous as to whether they are referring to partial or full defeaters, I will be focusing on his arguments as referring to full defeaters. This is because if we understand the problem of unpossessed evidence as the problem of squaring rational belief with the recognition that our evidence is part of a much larger whole, then partial defeaters seem to be besides the point. If reflecting on unpossessed evidence only gives me a partial defeater for my outright beliefs, then there is no difficulty in squaring rational belief with the awareness that there is evidence that I don’t possess. Given the definition of a partial defeater, I am still rational in holding my outright belief, even if perhaps I should hold it with a little less confidence.

It might be objected that if Ballantyne’s arguments are successful in providing partial defeaters, they still make it problematic to hold onto an outright belief that p. This is because we might take these arguments to apply recursively. Suppose that I start out with a set of evidence e for p. Make d be the fact that I possess so little of the total evidence bearing on p. Suppose that I reflect on d, so that my evidence for p is now e + d. Given Ballantyne’s arguments, d is a partial defeater for my belief that p on the basis of e. Thus, I lose some justification for my outright belief that p, though it is still justified simpliciter. Suppose that I accordingly adjust my degree of confidence in p. I might reflect again, however, that the evidence at my disposal, e + d, is still only a small part of the total evidence bearing on p (call this fact, d*). Given Ballantyne’s arguments, however, d* is a partial defeater for my belief that p on the basis of e + d. Thus, I lose some justification for my outright belief that p (though it may still be justified simpliciter). It seems, however, that I can repeat this process until I have lost justification simpliciter for my outright belief that p.³

In response, I note that this kind of recursive application of Ballantyne’s arguments is highly suspect. To see this, compare it to the following skeptical argument from possible hypoxia. In a hypoxic state one lacks sufficient oxygen. A possible symptom of lacking sufficient oxygen is that one’s thinking becomes irrational without one being aware of this. Take a random proposition p that I believe on the basis of evidence e, say that I am currently sitting down in my office, writing a paper on unpossessed evidence, which I believe on the basis of introspection and sensory experience. Let d = There is some very small, but non-zero chance that I am in a hypoxic state and that my evidence e is misleading. It seems that d is true. After all, there are known cases of people’s environments becoming unexpectedly oxygen deficient, and I have no internal monitor that would let me know if my environment was becoming oxygen deficient. If d is part of my evidence it seems I should believe that p with slightly less confidence than if it isn’t. This is because, if I were in a hypoxic state I could easily believe on the basis of introspection and sensory experience that I was in my office, writing a philosophy paper, all while I was typing nothing but gibberish. But suppose I realize that even after I lessen my confidence, there is still a small but non-zero chance that I am in a hypoxic state, and that my evidence e + d is misleading. Again this seems to be a partial defeater for my belief that I am in my office writing a philosophy paper. As a result, I should believe this with ever so slightly less confidence. And suppose I note, that again, there is still some small but non-zero chance that I am in a hypoxic state and that my evidence e + d + d* for p is misleading. Since this process of thinking is recursive, I can repeat it until I lose justification simpliciter for holding the outright belief that p.

It should be clear that there is something wrong with this line of reasoning. The mere fact that there is a non-zero possibility that I am in a hypoxic state should not threaten the reasonableness of my outright belief that I am in my office writing a philosophy paper on unpossessed evidence. What is wrong is this: once I take into account the remarkably small chance that I am in a hypoxic state, further reflection on this fact does not make an epistemic difference for me. To treat the argument from possible hypoxia as applying recursively is a mistake. Similarly, once I take into account that I possess only a small part of the total evidence bearing on a proposition, further reflection on this fact does not make an epistemic difference for me. Thus, it would be a mistake to recursively apply Ballantyne’s arguments from unpossessed evidence.

As a result, if we suppose that Ballantyne’s arguments provide only partial defeaters for our controversial beliefs, they do not make outright belief problematic. Because of this, in what follows I will be focusing on full defeaters for outright belief. As I hope to show, even though Ballantyne’s arguments fail, there are other ways of motivating the idea that reflecting on unpossessed evidence provides a full-defeater for our controversial beliefs.

2 The Meta-defeater Argument

Ballantyne’s first argument that is supposed to express the problem of unpossessed evidence is the meta-defeater argument. The meta-defeater argument goes as follows.

• M1: Evidence of the existence of a defeater for believing p relative to some body of evidence is a prima facie defeater for believing p relative to any body of evidence.

• M2: I have evidence of the existence of a defeater for believing proposition p relative to some body of unpossessed evidence.

• M3: I have no defeater for that (prima facie) defeater for believing p.

• Therefore,

• MC: I have an undefeated defeater for believing p. (Ballantyne 2019, p. 180)

Reflecting on unpossessed evidence is supposed to give us reason for believing M2. For instance, in the LIBRARY case, reflecting on the unpossessed evidence contained in the many books Sammy hasn’t read, gives her evidence that there is a defeater for accepting compatibilism, at least relative to some bodies of evidence (for example, relative to those bodies of evidence which contain mostly arguments against compatibilism). But according to M1, if we have evidence that there is a defeater for a proposition relative to some body of evidence, this is a prima facie defeater for believing this proposition relative to any body of evidence. The idea, then, is that reflecting on unpossessed evidence in the library case gives Sammy a prima facie defeater for her belief in compatibilism, by giving her evidence that there is a defeater for believing in compatibilism relative to some body of evidence.

I do not think that the meta-defeater argument explains how awareness of unpossessed evidence can defeat our beliefs. Most importantly, premise M1 of the meta-defeater argument seems to be false. To show this, I first need to lay out some terminology. To begin with, consider the idea of a defeater. A standard way of thinking about a defeater d relative to a set of evidence e, is as follows. D is a defeater for e, if e makes the belief that p rational, but e + d does not. This can happen in two ways. First, d can give us sufficient reason for thinking that not-p is true. In this case, d is a rebutting defeater. Second, d can give us reason for thinking that our evidence e is not good evidence for p. In this case, d is an undermining defeater.⁴

There doesn’t seem to be any standard way of thinking about prima facie defeaters in the literature, but we can think of prima facie defeaters as analogous to prima facie reasons as presented by Pollock (1987, p. 484). On this view, d is a prima facie defeater for e, if d is a defeater for e, but d itself could in turn get defeated by some defeater defeater d*. In such a case, while e would make rational the belief that p, e + d would not; however, e + d + d* would.

Given this background, we can now make the case against the meta-defeater argument. Suppose that we become aware that there exists a defeater d relative to some body of evidence e*. We can call this bit of evidence m. M1 implies that for all bodies of evidence e, m is a defeater for e. We have good reason to believe that M1 is false.

First, the idea of a prima facie defeater d relative to every body of evidence seems to be incoherent. Suppose that d is a prima facie defeater relative to all bodies of evidence. Given this, d must defeat the belief that p relative to the body of evidence e. But since it is a prima facie defeater, there must also exist a defeater defeater d* in relation to the body of evidence e + d. We can now derive an inconsistency as follows. If there is some possible body of evidence e + d + d* that justifies p, then d is not a prima facie defeater for e + d* and thus not a defeater for p relative to every body of evidence. However, if there does not exist some body of evidence e + d + d* that justifies the belief that p, then there is no defeater defeater for d, relative to e, and thus d is not a prima facie defeater relative to every body of evidence. But it must either be the case that there is some body of evidence e + d + d* that justifies p or not. Thus we have a contradiction to the supposition that d is a prima facie defeater relative to any body of evidence.

Second, even if the idea of a prima facie defeater relative to all bodies of evidence is coherent, it does not seem that m is this sort of defeater. Remember m is the claim that there is a defeater for p relative to some body of evidence. On the one hand, it is clear that m does not give us evidence for not-p that is strong enough to rebut any body of evidence for p. For it to do this, m would have to be very strong evidence for not-p. But m is surely not very strong evidence for not-p. Rather, it seems to be rather weak evidence for not-p. For instance, no one would think much of the proof that incompatibilism is true, because there exists a defeater for compatibilism relative to some body of evidence. Nor does the mere fact that there is a defeater for the belief that p relative to some body of evidence e, give us reason to believe that every body of evidence for p is not good evidence for p. Suppose as a not particularly informed non-specialist I encounter a defeater for my belief that p (where this is a belief about the economic dangers of deficit spending by the US government). I then adopt the belief that not-p. But then, I find a defeater for my new position. I am reduced to skepticism. Does the fact that relative to my evidence at time t1 I had a defeater for p, and and relative to my evidence at time t2 I had a defeater for not-p, give me reason for thinking that no one could have good evidence about whether p? Surely not. But there is nothing special about my case, and so it should generalize to others. If this is right, the mere fact that there is a defeater for p relative to some body of evidence is neither a rebutting nor undermining defeater relative to all other bodies of evidence. But if m is neither a rebutting nor an undermining defeater with regards to all bodies of evidence then it is simply not a defeater relative to all bodies of evidence. We can conclude that M1 is false.

Since M1 is false, the meta-defeater argument fails.

3 The Overlooked Defeater Argument

Ballantyne’s second argument is the overlooked defeater argument. This argument goes as follows.

• O1: If I have reason to think I have very likely missed defeaters for believing p, then I have a (prima facie) defeater for believing p.

• O2: I have reason to think I have very likely missed defeaters for believing p.

• O3: I have no defeater for that (prima facie) defeater for believing p.

• Therefore,

• OC: I have an undefeated defeater for believing p. (Ballantyne 2015, p. 325).⁵

Reflection on unpossessed evidence is supposed to give us grounds for believing O2. So for instance, reflecting on all of the books and papers that have been written and gone unread by Holden are supposed to give him reason for thinking that very likely there are defeaters for Holden’s economic beliefs. But given O1, this is supposed to give Holden a prima facie defeater for his economic beliefs.

The overlooked defeater argument relies on the following inference: I lack a lot of the evidence bearing on p, therefore it is highly likely that I have overlooked a defeater for my evidence that p. It is only through this inference that we are able to arrive at O2. However, we have good grounds for doubting the cogency of this inference. Note that this inference is good only on the following supposition:

• SUPPOSITION: It is very likely that the total body of evidence bearing on p contains a defeater for my evidence e for p.

To see this, suppose SUPPOSITION is false, and that it is not very likely that a defeater for your evidence that p exists in the total body of evidence. In such a case, the fact that you lack so much of the evidence bearing on p does not make it very likely that you have overlooked a defeater for your evidence. After all, in this case it is not very likely that there is such a defeater for you to overlook.

The problem for the overlooked defeater argument is that reflecting on all the evidence I lack does not give me reason for accepting SUPPOSITION. This can be shown in two ways. First, we can show this by way of example. Consider the following cases:

• SMOKING IS BAD: Isa knows that heavy smoking is linked to lung disease. But Isa is a philosopher who specializes in Carnap and the history of logical positivism, and is not a health professional. As a result, she lacks most of the evidence concerning the link between smoking and lung disease. She is aware that this is the case.

• FIRST PRESIDENT: Gonzo knows that George Washington was the first president of the United States. Gonzo is not a historian specializing on George Washington; rather he is a lawyer. He is aware that he does not possess the greater part of the evidence bearing on George Washington and the presidential succession in the US.⁶

In the SMOKING IS BAD case, Isa knows she lacks most of the relevant evidence. Nevertheless, Isa has no reason (not even a prima facie one) for thinking that the total evidence very likely contains a defeater for her belief that heavy smoking is linked to lung disease. Likewise, Gonzo knows that he lacks a lot of the relevant evidence. Nevertheless, he has no reason for thinking that the total evidence very likely contains a defeater for his belief. What these cases show us is that in general, awareness of a lack of evidence does not give us reason for thinking that the total evidence very likely contains a defeater for our beliefs.

We can also show this more abstractly. Consider all the sets of evidence e that a person could have that bear on a proposition p. Some of these sets of evidence will have defeaters for them in the total body of evidence; some of them won’t. We can suppose that the different sets of evidence will be more or less comprehensive, all of them will only be a small part of the total evidence that bears on p. After all, there is no clear limit to the kinds of considerations that may bear on the truth of a given proposition; and no one can be expected to have at their disposal the greater part of such considerations. But in this case, the fact that my evidence e for p is only a small part of the total evidence bearing on p, gives me no reason for thinking that there are defeaters for my evidence in the total body of evidence bearing on p. After all, every body of evidence I could possess, both bodies of evidence for which defeaters exists and those for which defeaters do not exist share this trait. So just as the fact that someone has a digestive system is not a reason for thinking that she is very likely to develop heart disease (since everyone in the relevant population has a digestive system), the fact that some body of evidence for p is only a small part of the total evidence, is not a reason for thinking that there very likely are defeaters for it (since every body of evidence under consideration is a small part of the total evidence.)

If this is right, reflecting on unpossessed evidence does not, by itself, give us reason for thinking that SUPPOSITION is true. What reflecting on unpossessed evidence does bring home for us is that we are unable to rule out that there are no defeaters for our beliefs based on defeasible evidence. Given all the evidence we don’t possess, we can’t be absolutely sure that there isn’t a defeater out there, after all. Perhaps this should make us have less than absolute certainty regarding our belief. But this shouldn’t be a problem for us if we reject views that identify outright belief with having credence 1 and infallibilist accounts of justified outright belief.

For the overlooked defeater argument to work, then, we need some reason for believing SUPPOSITION. However, reflecting on unpossessed evidence does not give us reasons for accepting SUPPOSITION. Thus, absent a further independent argument for SUPPOSITION, the overlooked defeater argument fails.

4 The Doubtful Fairness Argument

Ballantyne’s final argument expressing the problem of unpossessed evidence is the doubtful fairness argument. The argument goes as follows.

• F1: If I believe proposition p on the basis of some evidence e and I have either (i) (prima facie) reason to disbelieve that e is a fair sample or (ii) (prima facie) reason to suspend judgment whether e is a fair sample, then I have an undermining defeater for believing p.

• F2: I believe p on the basis of some evidence e and I have either (i) reason to disbelieve e is a fair sample or (ii) reason to suspend judgment whether e is a fair sample.

• F3: I have no defeater for the undermining defeater for believing p.

• Therefore,

• F4: I have an undefeated undermining defeater for believing p. (Ballantyne 2019, p. 186)

To understand this argument we need to understand Ballantyne’s idea of a fair sample of evidence. Ballantyne defines a fair sample of evidence as follows:

A fair sample of evidence of a particular size is drawn from a body of evidence by a process that makes it equally probable that each possible sample of evidence of that size will be selected. (Ballantyne 2019, p. 184)

Reflecting on unpossessed evidence makes it clear to us that we rarely have a fair sample of the total evidence. Take the LIBRARY example. Suppose that all of the evidence bearing on the question of compatibilism is the totality of philosophical writings dealing with free-will and determinism. It is obvious that Sammy’s sample of evidence in this case is not a fair sample. Sammy was not equally likely to read every set of 12 articles and 2 books dealing with these themes. Clearly Sammy was much more likely to have certain sets of evidence than others. The question is, does awareness of this fact give her a defeater for her belief?

The idea of a fair sample of evidence is analogous to the idea of a (simple) random sample of a population in statistics. A simple random sample from a population of a particular size will be such that every possible sample of that size has the same chance of being selected. Inferences based on random samples are more reliable than inferences that are based on non-random samples. Suppose that someone surveys all the students in his intro to philosophy class for the purpose of testing folk intuitions about unpossessed evidence. This is not a simple random sample of the general population. Only individuals who have completed high-school and are enrolled in university will be represented in the sample. As a result, there is greater likelihood that this sample is not representative of the population as a whole. Perhaps, for instance, university students are much more likely to view unpossessed evidence as a problem for holding religious beliefs than the general population. In this case, the sample will make it seem that the folk in general view unpossessed evidence as problematic for religious belief, when this is not so. Having a simple random sample makes this sort of misrepresentation much less likely. Perhaps university students skew a certain way in their attitudes, but university students will tend to not be overrepresented in a simple random sample.

The benefits of having a random sample when making statistical inferences are clear. There is good reason, however, to doubt whether having a fair sample of evidence is beneficial.

It seems that Ballantyne is concerned with fair samples of evidence because of the link between having a fair sample of evidence and one’s evidence being representative of the total body of evidence. Ballantyne defines a sample of evidence (to a first approximation) as being representative of a body of evidence in the following terms.

[A] sample of evidence represents a body of evidence with respect to a proposition p only if the sample captures enough of the body’s evidential features, so that the sample makes reasonable the same type of attitude to p that the total available evidence makes reasonable. A sample is not representative if the attitude to p that it makes reasonable is an entirely different type of attitude than the one made reasonable by the total evidence. (Ballantyne 2019, p. 188)

Thus, if a sample of evidence makes the outright belief that p reasonable, it is representative of the total evidence if, and only if, the total evidence makes the outright belief that p reasonable. If, for example, the total evidence made it reasonable to believe that not-p, or if it made it reasonable to only believe that p is very probable but not to outright believe that p, then the sample of evidence would not be representative.

The problem is that there are two distinct senses of a sample being representative at play. In statistical inferences, a sample of the total population is representative if it has the same distribution of relevant characteristics as the total population. (For instance, a sample of individuals will be representative of the population with regards to attitudes towards unpossessed evidence, if the distribution of attitudes towards unpossessed evidence is the same in the sample as it is in the population.) In this case, having a fair sample makes it more likely that this sample is representative of the total population. But when we think of a representative sample of evidence, we aren’t interested in a distribution of relevant characteristics. Rather, we are interested in the sample of evidence supporting the same doxastic attitudes as the total body of evidence. But the extent to which a body of evidence supports a given doxastic attitude does not seem to be determined by the distribution of certain characteristics in the pieces of evidence that make up this body. Maybe most of the evidence that exists favors the idea that government spending beyond a certain rate leads to inflation; nevertheless, one bit of defeating evidence could make it such that the total body of evidence favors the denial of this claim. Or again, maybe most of the evidence regarding e.g. compatibilism and incompatibilism is so weak that it favors suspending judgment about this issue. Nevertheless, it could be that the total body of evidence favors accepting incompatibilism.

Furthermore, it seems that having a fair sample of evidence can seemingly make us less likely to have a representative sample of evidence. Consider all the philosophical writing that has been done (and will be done) on free-will and determinism. Supposing that this is all the evidence on the issue of free-will, who could think that we would be better off reading 12 random articles and 2 random books on the issue, than reading 12 “classic” articles and 2 “classic” books recommended by specialists working today from opposing points of view? Having a fair sample of evidence in this case would most likely have us reading through mediocre (or worse) works of philosophers who have rightly been forgotten (or rightly will be forgotten). This fair sample of evidence would be less likely than an unfair sample of evidence to be representative of the total body of evidence in the relevant sense. And this seems generally to be the case. What we want is a subset of evidence containing the best evidence for and against a given claim, and having a fair sample of the evidence is a bad method for obtaining such a sample of evidence. Because of this, there is good reason to reject F1. Even if we have reason for thinking that our sample of evidence isn’t fair, this, by itself, does not threaten the rationality of our beliefs.

Since F1 is false, the doubtful fairness argument fails.

5 The Argument from Unpossessed Respectable Evidence

In the past sections I have argued that the meta-defeater argument, overlooked defeater argument, and doubtful fairness argument fail. How then does reflecting on unpossessed evidence defeat justification? Or does the anti-skeptic have nothing to worry about from the problem of unpossessed evidence?

I believe that the problem of unpossessed evidence is a real problem for anti-skeptics, at least for those philosophers who wish to maintain that we can have justified outright belief concerning controversial claims. To state my argument, I will appeal to the idea of a respectable inquirer. A respectable inquirer with regards to a proposition p is someone who has a reasonable chance of correctly interpreting his evidence for p; and thereby has a reasonable chance of being justified in his belief about p, if he takes his evidence to justify believing that p. Furthermore, a respectable inquirer will have a real chance of getting things right with regard to p because he has correctly interpreted his evidence. In a nutshell: you should pay attention to respectable inquirers. We can contrast respectable inquirers with negligible or despicable ones. A negligible inquirer is just as likely as not to misinterpret his evidence and to not be justified in his belief. We can safely ignore negligible inquirers. Despicable inquirers, on the other hand, are such that they are either likely to misinterpret their evidence and to have unjustified beliefs, and/or have no real chance of hitting upon the truth. We should ignore despicable inquirers.

It is worth noting that respectable inquirers need not be epistemic peers. Epistemic peers have equal epistemic standing with regards to a proposition p (cf. Frances and Matheson 2019). But two respectable inquirers with regard to p might have different epistemic standings with regard to p. One respectable inquirer might be more likely to get things right about a given matter than another; still, both merit being taken seriously, since both are likely to have justified beliefs with regard to p, and both have some real chance of having true, justified, beliefs with regard to p.

I take truly controversial claims to be claims that are both accepted and rejected by prima facie respectable inquirers. In some sense, there is a controversy as to whether the earth is roughly a globe or whether it is flat; there are some flat earthers. But that the earth is a globe is not truly a controversial claim, since both sides of the controversy are not respectable inquirers. Flat-earthers are epistemically despicable and should be ignored.

The mere fact of disagreement about truly controversial claims need not imply that one should suspend judgment about controversial issues. It might be that one knows the arguments and evidence in favor of the other side’s position, but is able to adequately respond to this counter-evidence. However, there is good reason for thinking that awareness of unpossessed counter-evidence with regards to a controversial claim gives us a defeater. In particular, learning of evidence that respectable inquirers take to justify disbelieving (or seriously doubting) that p, seems to give us prima facie reason for thinking that our evidence for p is not good enough to justify outright belief.

Here it is helpful to return to the PRECOCIOUS READER case. Pat starts out with a justified belief that behaviorism is correct; she then learns that Chomsky has offered a refutation of behaviorism. Here, we may assume that Chomsky is a respectable inquirer, and Pat thinks of him as such. Thus, if Chomsky thinks that his arguments refuting behaviorism justify his disbelieving behaviorism, they probably do. Furthermore, there is a non-trivial chance that Chomsky will be right about the merits of behaviorism because of this evidence. Pat doesn’t know exactly what Chomsky’s evidence is, but the mere awareness that Chomsky takes this evidence to justify disbelieving behaviorism seems to deprive her of justified outright belief in behaviorism.

To see this note that when one outright believes that p, he takes p for granted; he assumes p either explicitly or implicitly in all his theoretical and practical reasoning. Thus, if one is justified in outright believing that p, then he is justified in taking p for granted; i.e., he is justified in assuming p in all theoretical and practical reasoning (cf. Fantl and McGrath 2009, p. 99; Wedgewood 2012, p. 312; Ross and Schroeder 2014, pp. 267–68). It follows from this that if one is justified in outright believing that p, then one is justified in dismissing out of hand the supposed counter-evidence to p that one does not possess. This is because one may reason as follows: p. But if p, then any supposed counter-evidence against p is misleading evidence. But misleading evidence should be ignored. Thus, this supposed counter-evidence against p should be ignored. We can call such reasoning dogmatic reasoning.

Sometimes we can correctly engage in dogmatic reasoning. Most importantly, we can do so when we are dealing with evidence brought forward by negligible or despicable inquirers. Suppose for instance that Willy stops into his local used bookstore. On the new arrivals shelf he sees Samuel Birley Rowbotham’s Zetetic Astronomy: An Experimental Enquiry into the True Figure of the Earth; Proving it is a Plane without Axial or Orbital Motion, and the Only Material World in the Universe! (1865). It has an impressive title, and even more, it is filled with arguments, putative facts, figures, and a number of charts all of which are supposed to provide evidence that the earth is flat and immobile relative to the sun and moon. However, Willy dismisses all of this supposed evidence in favor of flat-earth without even looking at it. Willy reasons as follows: “We know that the earth is roughly a globe, and that it moves on its axis and in an orbit around the sun; whatever supposed evidence to the contrary that this Rowbotham has collected is specious at best. Most likely the majority of what he has to say is patently absurd.” (We can call this case the FLAT-EARTH case.) In the FLAT-EARTH case, Willy’s dogmatic reasoning is fully acceptable. Rowbotham and his successors are despicable inquirers.

But things aren’t like this for Pat. It seems that she cannot rightly engage in dogmatic reasoning when it comes to Chomsky’s arguments. For Pat to reason that since behaviorism is true, thus, Chomksy must be wrong, would be to effectively deny that Chomsky is a respectable inquirer. Similar comments hold for Holden and Sammy. Holden starts out with a hard won justified belief concerning the merits of some economic policy. But then Holden learns of the numerous articles and book chapters that have been published against his position. Presumably, there are a number of respectable inquirers that take these papers and book chapters to contain evidence, evidence that Holden does not have, to justify doubting, or even outright disbelieving Holden’s position. But since these are respectable inquirers, Holden cannot dismiss their supposed counter-evidence by dogmatic reasoning. After all, they are reasonably likely to be correct that their evidence does justify disbelieving Holden’s position; furthermore, there is a non-trivial chance Holden’s position is false given that these respectable inquirers are justified in disbelieving it. Likewise, we may suppose that Sammy starts off with a justified outright belief about compatibilism; but when she becomes aware of evidence that she is ignorant of, which respectable inquirers take to justify disbelieving or at least doubting compatibilism, she cannot reasonably dismiss this unpossessed evidence by means of dogmatic reasoning. To do so would be to treat these respectable inquirers like negligible or despicable ones.

Given that one is not justified in engaging in dogmatic reasoning to dismiss the counter-evidence put forward by respectable inquirers, it follows that one is not justified in having the relevant outright belief. Recall, to be justified in outright believing that p, one must be justified in assuming p in all practical and theoretical reasoning. This is exactly what Pat, Holden, and Sammy are not justified in doing. So it seems that becoming aware of unpossessed evidence taken by respectable inquirers as justifying disbelieving or doubting that p is a defeater for one’s outright believing that p. Since the phrase, “unpossessed evidence taken by respectable inquirers as justifying disbelieving or doubting that p” is a mouthful, sometimes I will replace it with the phrase “respectable unpossessed counter-evidence”. So we can say, that becoming aware of respectable unpossessed counter-evidence is a defeater for outright belief.

This is not true for all unpossessed evidence. Simply becoming aware of unpossessed evidence does not seem, in itself, to defeat our outright beliefs, as the SMOKING IS BAD and FIRST PRESIDENT cases show us. In these cases there is an awareness of unpossessed evidence but no defeater. Nor is it even true for unpossessed counter-evidence if this supposed “counter-evidence” is put forward by negligible or despicable inquirers. Remember, in the FLAT-EARTH case Willy does not have a defeater for his belief that the earth is roughly a globe, though he does become aware of a large amount of supposed counter-evidence to this claim. In the PRECOCIOUS READER, WWW, and LIBRARY CASES, however, Pat, Holden, and Sammy all do have defeaters. The relevant difference between these cases seems to be that in the PRECOCIOUS READER, WWW, and LIBRARY cases, Pat, Holden, and Sammy become aware of unpossessed evidence that respectable inquirers have taken to justify either disbelieving or doubting the relevant claims, whereas this is not the case in the SMOKING IS BAD, FIRST PRESIDENT, or FLAT-EARTH cases.

We are now in a position to give the argument from unpossessed respectable counter-evidence.

• URC1: If I am aware that respectable inquirers have evidence, which I do not possess, and that they take this evidence to justify either disbelieving or doubting p, then I have a defeater for outright believing p.

• URC2: I am aware that respectable inquirers have evidence, which I do not possess, and they take this evidence to justify either disbelieving or doubting p.

• URC3: I have no defeater for this (prima facie) defeater for outright believing p.

• Therefore,

• URCC: I have an undefeated defeater for outright believing p.

URC1 is justified by the preceding paragraphs. We can justify URC2, by noting that it will tend to be true whenever we think of the state of play in truly controversial areas of thought. To see this, note that there doesn’t seem to be anything special about the PRECOCIOUS READER, WWW, or LIBRARY cases. The position of Pat, Holden, and Sammy is one that will be occupied by most individuals if they stop and think about how things stand with regards to some truly controversial issue. When there is a real controversy, there will be respectable inquirers that disagree with us; furthermore, these respectable inquirers will have evidence, not at our disposal, that they take to justify doubting or disbelieving our position.

The justification for URC3 is more complicated. Why think that it is generally the case that we lack defeater-defeaters for unpossessed respectable counter-evidence? One reason is that it seems that awareness of unpossessed respectable counter-evidence is an undermining defeater for our beliefs. Recall that while rebutting defeaters defeat our belief that p by giving us evidence for not-p, undermining defeaters give us evidence for believing that our evidence for p is not good evidence. It seems that there are various ways that undermining defeaters can work (cf. Melis 2014, p. 438), but one way is by providing reasons for thinking that one’s original evidence is insufficient for justifying one’s belief. So for instance, suppose that at your local art museum there is an exhibition of conceptual art; in this exhibition a number of indistinguishable scenes are presented throughout the museum, but some small number of these scenes involve fakes or illusions. So for instance, a number of matching pink tables are placed throughout the museum, as well as some indistinguishable white tables that are lit up by pink lights. Suppose that Kieran is at the museum and is looking at what seems to be a pink table, and on the basis of his visual experience, Kieran forms the outright belief that there is a pink table in front of him. If Kieran learns about the conceptual art exhibition, however, he will have a defeater for his outright belief. This is not because he now has evidence that his original evidence does not support his belief. His visual experience does support the claim that there is a pink table in front of him, given that there are only a few fakes out of many. Rather, Kieran now has reason to believe that his visual experience does not provide him sufficient evidence for having a justified outright belief that there is a pink table in front of him. (We can call this case the MUSEUM case.)

My suggestion is that when we learn about the existence of unpossessed respectable evidence, we gain reasons to believe that our original evidence is insufficient for justifying an outright belief. But instead of doing this by directly weakening the connection between the truth and our evidence, as in the MUSEUM case, it does so by amplifying the amount of evidence that we must have in order to be justified.

Sometimes, in different situations, there are different amounts of evidence that we need in order to be justified. So for instance, oftentimes someone’s testimony is sufficient for outright belief, without taking into account further evidence; however, sometimes it is not. Consider the following cases.

• NEWSPAPER: Austin reads in the paper that eye-witnesses saw Mr. X, committing a double murder. Austin forms the outright belief that Mr. X committed a double murder, without taking any other evidence into consideration. Austin’s belief seems to be justified.

• JUROR: Austin is a juror in a criminal case. Austin forms the outright belief that the defendant Mr. X is guilty on the basis of eye-witness testimony, without taking any further evidence into account. Austin’s belief seems to be unjustified.

In the NEWSPAPER case, and in the JUROR case, we might suppose that the strength of one’s evidence is the same. However, only in the NEWSPAPER case is Austin’s belief justified. This is because it is not just the strength of our evidence that determines whether or not we are justified. The amount of relevant evidence we possess also matters. If we possess too little of the relevant evidence, we will not be justified in our beliefs.

What determines what is relevant evidence, how much of it we need to be justified is a complicated issue. I don’t propose to give a theory about this. But it does seem that, in general, once we are aware of respectable counter-evidence to our beliefs, we must somehow take this evidence into account in order to have a justified outright belief. Becoming aware of respectable counter-evidence, then, gives us evidence that our evidence is insufficient to justify our beliefs because we are lacking this counter-evidence and the further evidence that would neutralize it.

If this is right, then becoming aware of unpossessed respectable counter-evidence against p provides us with an undermining defeater for our belief that p. In this case, we can no longer rely on our original evidence for p to dismiss the unpossessed counter-evidence.⁷ Another way of putting this is that, we cannot rightly engage in the following semi-dogmatic reasoning: e, therefore p; therefore this unpossessed evidence against p is misleading evidence; therefore, it may be ignored. In general, semi-dogmatic reasoning in response to undermining defeaters is unjustified. To see this, suppose that in the museum case Kieran were to reason as follows: it seems to me that there is a pink table in front of me; therefore, there is a pink table in front of me; therefore, it is not a white table illuminated with pink light. Given Kieran's circumstances, such reasoning is absurd. Likewise, Pat reasoning from the premises provided by Skinner to the truth of behaviorism to the conclusion that Chomsky’s arguments (whatever they might be) are unsound is also absurd. Analogous remarks hold for Holden and Sammy.

This means that in order to have a defeater-defeater for unpossessed respectable counter-evidence against p, we need more than just the evidence that originally justified our belief that p. We need independent reasons for believing that the unpossessed respectable counter-evidence is misleading or in fact is not actually unpossessed respectable counter-evidence. For example, Pat might have testimony from a reliable source that Chomsky’s arguments ultimately fail – that there are defeaters for the defeaters presented by Chomsky. This is not part of Pat’s original evidence for believing that behaviorism is true; it’s independent evidence that the evidence cited by Chomsky is in fact misleading. Likewise, Pat might have some reason for thinking that while she has not read Chomksy’s “Review of B.F. Skinner’s Verbal Behavior” that whatever evidence he presents there is already part of her evidence; or again Pat might have some independent reasons for thinking that Chomsky is not a respectable inquirer after all, at least with regard to the question of behaviorism. For instance, Pat might have evidence that Chomksy has a long track record of offering “refutations” of behaviorism that are anything but, or that he lacks the relevant expertise to correctly evaluate arguments for or against behaviorism. Again, this wouldn’t be evidence that Pat relied on in forming her outright belief in behaviorism. Instead it is independent evidence that what she thought was unpossessed respectable counterevidence actually wasn’t.

However, as we have constructed the PRECOCIOUS READER case, it is clear that Pat doesn’t have any such evidence. She lacks independent reasons for thinking that Chomsky’s arguments fail. Likewise, she lacks independent reasons for thinking that Chomsky is not a respectable inquirer after all. In sum, Pat lacks a defeater for the defeater given to her by her awareness of unpossessed respectable counter-evidence.

Similar comments apply to the WWW and LIBRARY cases. We can imagine that Holden starts out with a justified belief regarding the role of government spending in economic growth. But now Holden is confronted with an overwhelming amount of new studies and arguments bearing on the question; furthermore, he knows that many of the arguments and studies point away from his position. Additionally, Holden is aware that many of the studies and arguments against his position have been published by respectable inquirers. Finally Holden supposes that the authors of these studies believe their evidence to justify either outright disbelieving Holden’s position, or at the very least doubting it.

Holden’s awareness of this unpossessed respectable counter-evidence provides a prima facie undermining defeater for his outright belief concerning the role of government spending in economic growth. As a result, Holden cannot rely on his original grounds for belief to dismiss this unpossessed counter-evidence.

However, Holden does not have independent reasons for dismissing this unpossessed counter-evidence either. On the one hand, we can suppose that Holden has no independent reason for thinking that all this unpossessed counter-evidence fails; he lacks, for instance, expert, reliable testimony that there are defeaters for all of this new evidence. Nor can Holden appeal to a track record argument as evidence that this new unpossessed evidence is misleading. The most promising track-record argument would go as follows: everytime in the past that Holden has come to examine respectable counter-evidence against his view, he has found defeaters for this counter-evidence. Thus, Holden has reason for thinking that this will be true for all of these new cases of unpossessed respectable evidence as well. The problem with this line of reasoning is that Holden’s ability in the past to respond to respectable counter-evidence justifies a belief that he will be able to do so in the present, only if he has worked through a large enough sample of counter-evidence, and he has reason for believing that the unpossessed counter-evidence will be similar in the relevant respects to the counter-evidences he has encountered. Neither of these conditions seem to be met. On the one hand, Holden’s exposure to the relevant counter-evidence does not seem to be sufficiently ample, given the impressive amount of studies that have been put out; on the other hand, he lacks reason for thinking that the unread studies are like the studies he has seen in the past. For all that Holden knows, the unread studies could be radically different than what he dealt with when originally forming his beliefs.

Nor does Holden have reason for thinking that this unpossessed counter-evidence is not respectable counter-evidence after all. He lacks independent reasons for thinking that the economists who have put forward this evidence as grounds for rejecting Holden’s position are incompetent in interpreting it; nor does he have reason for thinking that somehow he already possesses the evidence in these new papers and books. In sum, Holden has no independent reasons for thinking that the relevant unpossessed respectable counter-evidence is misleading, nor does he have reason for thinking that it ultimately is not unpossessed respectable counter-evidence. Holden lacks a defeater for the defeater given him by awareness of unpossessed respectable counter-evidence.

Similar remarks apply for the LIBRARY case too. And it seems that they will hold for us with regards to the majority of our controversial beliefs. We are aware that ever more evidence and arguments (which we do not possess) are being produced in these areas of controversy. We are also aware that respectable inquirers find some of this evidence to justify either outright disbelieving or doubting our position. Since awareness of this unpossessed evidence provides us with an undermining defeater for our outright beliefs, to have a defeater-defeater we will need independent reasons for believing that there are defeaters for this unpossessed respectable evidence, or we need some independent reasons for believing that it is not unpossessed respectable evidence after all. Oftentimes we will not have this. Like Pat, Holden, and Sammy we will not be able to rely on expert, reliable testimony that this unpossessed counter-evidence fails; nor will we be able to rely on track record arguments to show that this respectable unpossessed counter-evidence fails or is not really respectable unpossessed counter-evidence. Oftentimes, it seems, we will have no way of defeating the defeater given to us by our awareness of unpossessed respectable counter-evidence.

A couple of points are worth stressing here. First, the argument from unpossessed respectable counter-evidence is not just a skeptical argument from disagreement. Skeptical arguments from disagreement tend to rest on the following sort of principle.

CONCILIATION: If S believes that p and acknowledges that S’s epistemic peer, T, believes ~p and S has no independent reason to think that S is more likely than T to be correct about p, S thereby has a defeater for her belief that p. (See King 2013, p. 193).

The argument from unpossessed respectable counter-evidence makes no appeal to CONCILIATION; nor does it make any appeal to the notion of an epistemic peer. Thus one could reject CONCILIATION and still be moved by the argument from unpossessed respectable counter-evidence. This is a plus for the argument from unpossessed respectable counter-evidence. As critics have noted, the condition of being an epistemic peer with someone is highly idealized; because of this the real-world skeptical implications of CONCILIATION are unclear. The idea of being a respectable inquirer on the other hand is not overly idealized. There are many people who are reasonably likely to interpret their evidence correctly and as a result have a non-negligible chance of believing correctly about an issue. When we are aware that such people possess evidence we lack that they take to justify outright disbelieving or doubting p, this gives us a defeater for our belief that p.⁸

The argument from unpossessed respectable evidence, then, lies somewhere between conventional arguments from disagreement and skeptical arguments that appeal solely to the notion of unpossessed evidence. As the SMOKING IS BAD and FIRST PRESIDENT cases show, it is unlikely that merely reflecting on unpossessed evidence provides defeaters for our beliefs. Nevertheless, reflecting on unpossessed evidence can give us defeaters that are distinct from defeaters that we have from reflecting on peer disagreement. As I have argued, being aware of unpossessed respectable counter-evidence gives us a defeater by giving us reason to believe that our original evidence is insufficient for justifying outright belief.

Second, it is worth stressing the force of the argument from unpossessed respectable evidence. This argument attacks only the rationality of outright beliefs when we are aware that we lack respectable counter-evidence. One might think that, in many cases, this does not present a skeptical problem, but represents good epistemic sense. For instance, in the midst of scientific controversy it seems dogmatic to have an outright belief while one has yet to master the evidence that respectable inquirers have against one’s favored hypothesis. However, in a wide range of cases in ethics, religion, philosophy, history, politics, and economics, we have convictions in the face of controversy. At the same time, in most of these cases, there is prima facie respectable counter-evidence that we do not possess. If the argument from unpossessed respectable evidence is correct, there is something wrong with this. Either our convictions are unjustified, or contrary to appearances there is no respectable counter-evidence to our beliefs that we don’t possess. Either way, the argument from unpossessed respectable counter-evidence requires us to revise a common way of thinking about convictions in controversial areas.

Furthermore, the skeptical power of the argument is increased if we suppose that we can have knowledge only if we have justified outright belief. In this case, the argument from unpossessed respectable counter-evidence would call into question our ability to have knowledge in the face of real controversy. While some philosophers might welcome this implication, many others would not.

Furthermore, the argument from unpossessed respectable counter-evidence is immune to certain common moves made against other better known skeptical arguments. Consider two common responses to versions of Cartesian skepticism. The first common response is fallibilism. The Cartesian skeptic, it is often claimed, is supposed to demand conclusive evidence for justified belief. But according to fallibilism this demand is too strong: pace the Cartesian skeptic, we can have justified belief and knowledge on defeasible, non-conclusive evidence (see e.g. Brueckner 2005, pp. 384–85).

While fallibilism promises to provide a response to Cartesian skepticism, it is no help with the argument from unpossessed respectable counter-evidence. In fact, accepting that we have only defeasible evidence in favor of our beliefs allows the argument from unpossessed respectable counter-evidence to get off the ground. If we were to have conclusive evidence for our controversial beliefs, then anyone who disagreed with us could be considered a despicable inquirer; while someone who took themselves to be justified in believing that not-p may in fact be justified, given that we have conclusive evidence for p, they have no chance of having a true justified belief. Thus, we can ignore them, and whatever counter-evidence they supposedly have. But once we have to rely on fallible, defeasible evidence, the fact that other reasonable people take themselves to have evidence (not in our possession) that refutes our position, gives us reason for thinking that there is unpossessed respectable counter-evidence against our belief.

The second common response to Cartesian skepticism is dogmatism. The Cartesian skeptic, it is claimed, demands that we give prior, independent reasons for believing that we are not in the grips of an illusion in order to be justified on the basis of e.g. our perceptual evidence. According to dogmatism, however, this demand is illegitimate. The mere possibility that we are under the grips of an illusion is not a defeater for our beliefs, even if we lack (perceptually) independent reasons for ruling this possibility out. However, the dogmatist allows that if we have positive reason for suspecting that we are under the grips of an illusion, this can be a defeater for our beliefs (see Pryor 2004, pp. 353–54). A generalized version of this kind of dogmatism would hold that the possibility of a defeater obtaining for a belief is not itself a defeater. Thus, even if there is the possibility that a defeater for a belief p based on evidence e obtains, we do not need e-independent reasons for ruling this possibility out.

Again, while adopting dogmatism is a promising strategy to deal with a wide range of skeptical problems, it does not seem to help with the argument from unpossessed respectable counter-evidence. This is because the argument from unpossessed respectable counter-evidence does not appeal to the mere possibility of a defeater obtaining. Rather it claims that awareness of unpossessed respectable counter-evidence against p gives us a defeater for our belief that p. Put differently, the argument from unpossessed respectable counter-evidence does not argue that our evidence might be insufficient to justify our belief, and that we need to rule out this possibility. Rather, it argues that the existence of unpossessed respectable counter-evidence gives us positive reason to believe that our original evidence is insufficient to justify outright belief.

There seems to be three main strategies that we might use to deal with the argument from unpossessed counter-evidence. First, we might claim that in fact we have conclusive evidence for our controversial beliefs. In this case, we can safely ignore any unpossessed counter-evidence, since the evidence at our disposal guarantees the truth of our belief. The problem with this approach is that there are difficulties in spelling out how we have conclusive evidence for our controversial beliefs. For instance, neither Pat nor Holden seem to have anything that could be construed as conclusive evidence for their beliefs. One possibility, inspired by Williamson, is that everything we know is part of our evidence, and all evidence is evidence for itself (see Brown (2018, pp. 26–44)). Thus, if we have knowledge about some controversial issue, we will have conclusive evidence for it as well. However, this approach seems to license an unacceptable dogmatism. Suppose Holden were to reason the following way, when he becomes aware of the thousands of articles and books contradicting his economic beliefs “I know that p. Therefore, p is part of my evidence. But I thus have conclusive evidence for p, and so I don’t need to worry about all of the counter-evidence I don’t possess.” Prima facie, if our response to the problem of unpossessed evidence licenses such a line of reasoning, something has gone wrong.

A second way of responding to the argument from unpossessed respectable counter-evidence is to deny that there are undermining defeaters. On this view, if one’s evidence sufficiently favors p, then one is justified in outright believing that p, even if one has reason to suspect that one’s evidence is not good enough to justify outright believing that p. This sort of skepticism about defeaters has been independently argued for by different philosophers (for example Lasonen-Aarnio (2014), and Weatherson (ms.)). I do not think, however, that this is a satisfying response to the argument from unpossessed respectable counter-evidence. On the one hand, defeater skepticism is highly controversial, and seems to have a number of implausible consequences (for an overview of these consequences see Steel (2019)). On the other hand, even if defeater skepticism is correct, it seems that there is something wrong with outright believing that p, while having reason to think that one’s evidence for p is not good enough to justify one’s outright belief.

Finally, one might try to argue that we do have independent reasons for dismissing the putative counter-evidence we do not possess. Prima facie, a general argument for this conclusion seems hopeless. However, I do not rule out that for the particular controversial beliefs of particular individuals such a case could be made. Admitting this, however, is simply to admit that the awareness of unpossessed respectable counter-evidence gives us a defeasible defeater for our outright beliefs.

6 Conclusion

In this paper, I have argued that Ballantyne’s meta-defeater, overlooked defeater, and doubtful fairness arguments fail. Nevertheless, I agree with Ballantyne that the problem of unpossessed evidence is a real problem. I have attempted to express this problem in the argument from unpossessed respectable counter-evidence. According to this argument, awareness that respectable inquirers take evidence we do not possess to justify doubting or disbelieving our beliefs gives us an undermining defeater for these beliefs. Furthermore, it seems that in most cases we do not possess a defeater for this defeater. Given this, philosophers who hold that we properly have convictions regarding controversial matters should focus on responding to the argument from unpossessed respectable counter-evidence.