It may seem that once one knows something, nothing could change that. After all, if one knows that p, does one not thereby know that there can be no non-misleading evidence against p? And does knowing this not entitle one to ignore any evidence against p? As against this, we have the strong intuition that ignoring evidence is always wrong. One should believe something on the basis of evidence and cease to believe it if one finds evidence against it. As Hume (Reference Hume1748) puts it, “the wise man proportions his belief to the evidence.” If one clings to a belief in the face of counter-evidence, one is a dogmatist of the worst sort.

Some have seen a paradox here. Harman formulates it as follows:

Harman's own response is to say:

Here is Sorensen's gloss on this:

As an illustration of the “intransigence” to which commitment to the hardiness of knowledge can lead, Sorensen offers the following argument:

However, it is not clear that the argument shows what Sorensen thinks it does. First, one may worry that two different senses of ‘evidence’ are being run together. We describe something as evidence if it is intended to result in someone's adopting a belief on being presented with it. This is perhaps the way the term is most commonly understood, and on this understanding, misleading evidence counts as evidence. But there is another notion of evidence on which misleading evidence is merely purported evidence – ‘alleged evidence,’ as Kripke (Reference Kripke2011) calls it – and not really evidence at all. ‘Purported evidence’ is not, as it is sometimes put, “logically predicable” in the way ‘red car’ is (Geach Reference Geach1956). We use the latter expression to say that the thing in question is both a car and is red. We do not use the expression ‘purported evidence’ in that way. Saying that something is purported (or alleged) evidence leaves open whether it is really evidence or merely appears or pretends to be, even if it is intended to persuade someone and even if it succeeds in doing so. If we know (or at least have reason to believe) that things are not the way the purported evidence appears to show, we do not regard it as evidence.Footnote ¹ We do not think that the evidence the witness we found to be lying gave was evidence of the accused's guilt. Since C₁ is true, what Doug provides is evidence in the first sense, but not in the second. And if I know (or am justified in believing) that my car is in the parking lot, I know (or I am justified in believing) that what Doug provides is only purported evidence. Sorensen agrees. He takes his argument to show that I am justified in believing that “Doug's evidence is misleading evidence.” But he sees this as a problem, because he thinks that ignoring misleading evidence, even if one is justified in thinking that it is misleading, is ignoring evidence.

This is a symptom of a deeper problem with Sorensen's argument. It equivocates between what is the case and what one is justified in believing to be the case. C₃ does not follow from my being justified in believing C₁. It follows only from C₁'s being the case. But C₁ says nothing about my degree of confidence in C₁. What would follow from my being justified in believing C₁ is that I am justified in believing C₅. But that can be so even if Doug's report is correct, as it may well be for all the justification I may have for believing C₁. Being justified in believing C₁ does not stand in the way of considering new evidence, and thinking oneself justified need not lead to dogmatism.

Suppose, though, that instead of C₁, we have C_j = I am justified in believing that my car is in the parking lot. To be justified and to think oneself justified are obviously different things. Whether one is being dogmatic has to do not with what one is justified in believing or knows but with what one believes oneself to be justified in believing or to know. One may believe that one is justified in believing something without being justified in believing it, and one may think that one knows something even when one does not.

As Veber recognizes, “The problem is that human subjects often take themselves to know things they do not” (Veber Reference Veber2004: 567). But what he takes this to show is that we should reject a principle (which he regards as essential for generating the alleged paradox) that says that “if a piece of evidence is known by S to be misleading, S is in a position to disregard it.” The principle “is objectionable not because of what it says but because of the cost and likelihood of misapplying it … Adopting [it] will result in our disregarding evidence which we merely think is misleading” (Veber Reference Veber2004: 567). Veber is right that those who think we have a paradox here are relying on such a principle, since they think that knowing something entails knowing that any evidence against what one knows is misleading.Footnote ² But even if it does, it does not entail that one thinks that the evidence (in the broad, non-factive sense) is misleading and second, and more important, it does not entail that one has any attitude to such evidence. One may know something and not realize, at least for the moment, that one knows it, be completely open-minded about it, even seek evidence concerning it.

Furthermore, the principle Veber identifies as essential for generating the so-called paradox is not one anyone could adopt. If, as near-universal agreement has it, one knows only if the proposition one believes or thinks one knows is true, whether one knows depends (in part) on the truth-value of a proposition one believes or thinks one knows, something that is inaccessible to one. But one's epistemic conduct cannot be governed by something to which one has no access. It is the fact that the principle is formulated the way it is and not in the first person that obscures this. Put in the third person, it makes it seem that it is, or should be, the fact that S knows that governs his epistemic attitudes. But this is just an appearance, and a misleading one. One has no access to such facts. Thus while Veber is right in calling attention to the distinction between knowing something and thinking that one knows it, he does not see that the former cannot play any role in governing one's epistemic policies. If so, it cannot have any bearing on whether those policies are dogmatic or not.

Of course, there is a difference between being dogmatic and being open to evidence. Unless we assume that knowing entails knowing that one knows, one may know something without realizing that one does. If that is the case, what could be more rational than to take seriously evidence that seems to run counter to what one believes? It is whether one thinks one knows, rather than whether one knows, that matters in shaping one's attitude to (purported) evidence. If evidence comes my way that seems to tell against what I believe but do not think I know, should I always take it seriously? Generally, yes, even if I am justified in believing something, whether I in fact know it or not. Sometimes, though, I have good reason to think that the evidence seeming to tell against something I believe is misleading. If so, refusing to consider it is not dogmatism. There is no dogmatism in dismissing out of hand evidence that appears to call into question something I am confident I know, as long as I am entitled to that confidence. I know that the magician is not pulling a rabbit out of an empty hat, and it is not dogmatism to ignore the apparent evidence to the contrary even while admiring the skill in producing it.

What makes for confusion is conflating knowing something and thinking that one knows it. If knowing something entails that it is true, there cannot be evidence against it, if that means evidence that shows it to be false. But, first, it does not follow from this that there cannot be evidence that tells against it, and, second, it follows even less that there cannot be evidence that seems to tell against it.

We can now also see why Harman's proposed solution to the alleged paradox misfires. He says that even if one knows that p, one should be open to evidence against p, since that evidence may result in one's no longer knowing p. In Sorensen's words, he denies that knowledge is ‘hardy.’ But this cannot be, if one can know that p only if p. Then if I know that p and thus p is true, it does not stop being true because I have evidence against it that purports to show that it is not. Harman is right that if I take such evidence that shows that p is not true at face value, I no longer know that p, as I am no longer (doxastically) justified in believing that p and, if rational, no longer believe it. Does this mean that one should be dogmatic when presented with purported evidence against p? Not if the argument above is sound. Even if we concede that sometimes one has access to knowing that p, it is still the fact that one thinks one knows, and thus thinks that one is justified in believing, that p that determines one's attitude to new evidence. And thinking that one knows is not hardy: one can stop thinking that one is justified in believing that p and thereby stop thinking that one knows that p even if p is true.

To say that there is no paradox here is not to endorse dogmatism. It is only common sense that often, though not always, if something comes one's way that seems counter to what one believes or thinks one knows, one should examine it and re-examine the justification one believes one has for one's belief.Footnote ³ The point is that dogmatism is not forced on one by the facts about what one's epistemic states are. These facts can play no role in shaping one's epistemic attitudes – these can be governed only by one's beliefs about one's epistemic states. These are, of course, themselves epistemic states. However, the supposed paradox rests on taking the first-order epistemic states they are about as determining one's attitude to evidence, something they cannot do.

Whether one is being dogmatic depends on whether one's second-order beliefs about one's first-order states are justified. We should also bear in mind that one can have different beliefs about one's epistemic state. Obviously, one can think that one does not know. But the belief that one does know can have varying strength. One can think that one knows, one can be confident that one knows, one can be certain that one knows and adopt different attitudes to evidence that may come to light depending on which one thinks. If it is the first, not only is it reasonable to be open to more evidence, it may be so to set out to find some. But even if it is the last, one is not always entitled to refuse to consider putative new evidence against what one is certain one knows. Whether one is or not depends on the nature of the putative evidence and its relation to everything one thinks one knows. I may be certain that I am alone in the house with the doors locked, but I cannot refuse to believe my eyes if I find you in the kitchen in a way I can refuse to believe my eyes when the magician appears to pull the rabbit out of the empty hat. Doing so is common sense, not dogmatism. So with Sorensen's Doug. Suppose I think I know that my car is in the parking lot but Doug tells me that he has just seen it being towed away. What should I do? It depends. What is Doug's track record? Is he usually reliable? Do I, absent-minded professor that I am, sometimes forget to move the parking sticker from one car to another? I take his testimony seriously. If he is a prankster, and I am sure I moved the sticker this morning, I am not being dogmatic in ignoring it. Whether I actually know what I think I know makes no difference to any of this. The putative paradox of dogmatism stems from not seeing this.Footnote ⁴