The structure of moral encroachment

According to moral encroachment facts about epistemic justification can vary with moral factors that are unrelated to the truth of the belief. Most of the literature on this topic has focused on how beliefs can wrong, and whether the data that moral encroachers offer in support of their view can be explained within a purist framework. A largely neglected question has been what kind of consequences moral encroachment would have for epistemic justification if the thesis were true. Here I remedy this shortcoming by examining what kind of structural implications moral encroachment has for epistemic justification.


Introduction
According to moral encroachment facts about epistemic justification can vary with moral factors that are unrelated to the truth of what is believed. There are two versions of this view. Moderate moral encroachers hold that epistemic justification is sensitive to the moral status of the actions and options that beliefs license. Radical moral encroachers hold that epistemic justification is sensitive to the moral status of the beliefs themselves. 1 The central idea behind radical moral encroachment is that 1 3 moral and epistemic requirements are coordinated in that they cannot issue conflicting demands. In slogan form: moral mistakes in the realm of belief make for epistemic mistakes. My main task in this paper is to examine what kind of structural implications moral encroachment would have for epistemic justification if the thesis were true, and how they differ from the ones that pragmatic encroachment on justification entails. To anticipate, I demonstrate that both moral and pragmatic encroachment entail failures of epistemic closure principles and that an unjustified belief can become justified due to a change in practical or moral factors. I also show that radical moral encroachment, but not pragmatic or moderate moral encroachment, entails that there are some true propositions that cannot be justifiably believed due to their content, and that moral considerations have lexical priority over epistemic ones when it comes to epistemic justification. I also argue that despite first impressions it might not be easy to reconcile pragmatic and moral encroachment. While I think that the structural features that both moral and pragmatic encroachment entail are implausible, encroachers might think otherwise. I leave it to the reader to decide whether the structural consequences that the encroachment views entail are a boon or a curse.
The structure of this essay is the following. In Sec. 2 I lay out the core claims of moral and pragmatic encroachment. In Sec. 3 I examine what kind of implications moral and pragmatic encroachment have for epistemic justification. Comparing the two encroachment views shows that the purported truth of radical moral encroachment has more revisionary consequences for epistemic justification than pragmatic encroachment has. In Sec. 4 I argue that despite first impressions, it might not be easy to reconcile moral and pragmatic encroachment. I conclude in Sec. 5 by offering an alternative explanation of the kind of cases that encroachers offer in support of their view. According to this proposal we are either morally or practically required not to believe the target propositions in putative cases of encroachment, but our beliefs are nevertheless epistemically permissible, and hence epistemically justified. But epistemic normativity doesn't give rise to positive obligations to believe, and therefore given that we are either morally or practically required to not believe the target propositions, we should, all things considered, suspend our judgment.

Moral and pragmatic encroachment
Radical moral encroachment is often motivated by the following kind of examples, where the subject's evidence seems to license believing the target proposition, but the belief seems to be morally questionable.
Tipping Prediction: Spencer's shift has just ended, and he sees a black customer, Jamal, entering the diner where he works. Earlier today Spencer learned that black customers tip on average substantially lower than white customers. Based on this statistical information Spencer predicts that Jamal will tip lower than the average. 2

3
The structure of moral encroachment Wine Stain: Laura's partner, who has struggled with an alcohol problem for many years but been sober for a few months now, came from a departmental reception with a big wine stain on their shirt. Laura knows that her partner has fallen off the wagon many times in the past, and that it is very likely that they have done so again. Based on this, she believes that her partner drank alcohol at the reception. 3 Radical moral encroachment holds that even though Spencer's and Laura's beliefs are well supported by their evidence (or at least as well as most of our justified beliefs are), their beliefs are epistemically unjustified in virtue of the fact that they either wrong, or risk wronging those whom they are about. Therefore, moral factors that are unrelated to the truth of what is believed can influence whether a belief is epistemically justified.
Moral encroachment is usually presented as a corollary of pragmatic encroachment, according to which facts about justification can vary with practical factors, such as the practical stakes involved. 4 The key motivation behind pragmatic encroachment is the idea that epistemic justification is intimately connected to action: you are justified to believe that p only if you can (rationally) act as if p were the case (Fantl & McGrath, 2002, p. 78). This view is often made vivid by the following kind of pairs of cases, adapted from Stanley (2005, pp. 3-4).
Bank Low: It's Friday and Hunter ought to deposit a check within two days, but it won't be a big deal if he fails to do so. As Hunter drives past the bank he sees that the lines are quite long and he really would prefer not to queue. Hunter recalls that the bank was open last Saturday, and there were no lines back then. He drives home in order to return tomorrow and justifiably believes that the bank is going to be open on Saturday. Bank High: It's Friday and Hunter must deposit a check within two days or disaster will strike. As Hunter drives past the bank he sees that the lines are quite long and he really would prefer not to queue. Hunter recalls that the bank was open last Saturday, and that there were no lines back then. He drives home in order to return tomorrow but his belief that the bank is going to be open on Saturday is unjustified.
The pragmatic encroacher holds that although Hunter's evidence is the same across the cases his belief is justified only in Bank Low, since he can act as if the bank was open only in Bank Low. But whether it is permissible to act as if p is the case can depend, not only on purely practical factors, such as the disutility of having to return home after walking to the office and only then realizing that one forgot the office keys at home (happened to me more than once), but also on the distinctively moral costs of the error, such as forgetting to write a letter of recommendation without which your student's application isn't even considered (happened to a colleague). Hence pragmatic encroachment, and perhaps moral encroachment, gives rise to the following test on justification: Practical Adequacy: S is justified to believe that p only if the difference between S's epistemic position with respect to p and being epistemically certain that p does not make difference as to what S ought to do. 5 Drawing on Anderson and Hawthorne (2019) we can make this idea more precise. First, 'epistemic positions' with regard to specific propositions are represented by epistemic probability functions. The strength of an epistemic position with respect to p is the epistemic probability of p. What a subject S ought to do is determined by the expected utilities of her options, which are calculated with epistemic probabilities and the utilities of S's options. A proposition p is then practically adequate for S if, and only if, the ranking of the top-most option doesn't differ between S's actual epistemic position and an ideal epistemic position with respect to p (i.e. being epistemically certain that p). 6 If moral factors can (partially) determine the utility of options, as seems plausible on a rational preference ranking, moral encroachers, and moderate ones in particular, ought to subscribe to the practical adequacy test.
While some moral encroachers emphasize the apparent link between practical and moral encroachment (Fritz, 2017;Worsnip, 2021), others deny it. For instance, Basu (2019a) holds that moral stakes are not just a straightforward extension of practical stakes, and that in the case of moral encroachment it is impossible to construct the kind of low-stakes/high-stakes pairs of cases that are traditionally used to motivate pragmatic encroachment. Furthermore, many moral encroachers hold that beliefs themselves can wrong irrespective of whether the subject acts, or even could act, on those beliefs (Basu, 2021;Basu & Schroeder, 2019;Moss, 2018), thus de-emphasizing the connection between justified belief and action. 7 Fritz (2020) disagrees, and 6 There is good reason to think that a practical adequacy test has skeptical consequences. For suppose that you believe that a Big Mac is the best food you can get (p) on the basis of good, but not conclusive evidence. Given your current epistemic position with respect to p you ought to buy a Big Mac. You are then offered the following bet. If you accept the bet and p is the case you earn $1. But if not-p is the case you will be tortured for the rest of your life. Clearly you shouldn't take the bet given that it is epistemically possible for you that not-p is the case. So you should buy a Big Mac and not take the bet. But conditional on being epistemically certain that p you should take the bet (and buy a Big Mac), because you would gain $1. Therefore, no matter how epistemically probable p is, short of certainty, the justificatory status of p can be destroyed by offering you bet that only those who are epistemically certain about p should take. 7 Enoch and Spectre (forthcoming) argue that beliefs cannot morally wrong.

3
The structure of moral encroachment argues that norms of epistemic justification are sensitive, not to the moral status of beliefs, but to the options and actions that the beliefs license.
In what follows I will focus on radical moral encroachment and treat it as the thesis that beliefs that morally wrong cannot be epistemically justified. According to radical moral encroachment moral and epistemic standards on belief are coordinated in that beliefs that wrong cannot be epistemically permitted (Basu & Schroeder, 2019, p. 197). 8 The way in which moral factors influence epistemic justification closely mimics the way in which practical factors can influence facts about justification according to pragmatic encroachment. On both encroachment views, the relevant practical or moral factors can raise the bar on how much epistemic support outright justification requires (Basu & Schroeder, 2019;Fantl & McGrath, 2002;Pace, 2011). Moral factors that raise the bar of sufficient epistemic support are ones that increase the risk that the belief morally wrongs or harms someone. For instance, by believing that a person is staff on the basis of their race, one risks wronging the person.
When does a belief morally wrong a person? Basu and Schroeder hold that a belief morally wrongs someone when it constitutes a doxastic wrong. Basu holds that "(1) doxastic wrongs are directed; (2) doxastic wrongs are committed by beliefs rather than the consequences of acting on a belief; and (3) doxastic wrongs are wrongs in virtue of the content of what is believed (Basu, 2021, p. 108)." 9 But this cannot be exactly right, since there is nothing morally amiss with Laura if she believes that her partner drank alcohol at the reception one the basis of her partner's testimony. Similarly, if Jamal tells Spencer that he is fallen on hard times and will tip a bit less than the average customer, then Spencer does not doxastically wrong Jamal in taking his word for it. So it cannot be the case that "doxastic wrongs are wrongs in virtue of the content of what is believed". Rather, it seems more plausible that doxastic wrongs are wrongs in virtue of believing the content in a certain kind of moral context. 10 It is only in a specific moral context that believing the content has the kind of moral implications that it has. So Spencer's belief does not doxastically wrong Jamal simply because he believes that Jamal will tip less than the average. 8 Most moral encroachers subscribe only to this negative thesis, and hold that moral factors can only ever render a belief impermissible. But holding some beliefs might be morally good, such as believing our loved ones even when their claims are not very well supported. Indeed, in so far as Laura wrongs her partner by believing that they fell off the wagon in Wine Stain, she would likewise seem to wrong her partner if she suspended her judgment on the matter when the partner tells her that they abstained from drinking at the reception. Pace (2011, p. 259), with whom the term 'moral encroachment' originates, argues that moral factors can make beliefs that are more likely true than not epistemically justified, due to the fact that some such beliefs are good from a moral point of view. In my mind it is quite puzzling how moral factors could only ever render beliefs epistemically unjustified if they cannot make them justified. For instance, Basu and Schroeder (2019) argue that moral and epistemic requirements/permissions are coordinated because moral factors help to determine the bar of sufficient epistemic support. But this is exactly how Pace (2011) argues that moral factors can make beliefs epistemically justified and unjustified. For the present purposes I set this issue aside. 9 See also Basu and Schroeder (2019, p. 181). 10 See Suni (2022, pp. 207-9) for a similar idea. Thanks to Ninni Suni and Antti Kauppinen for discussion.
Rather, Spencer wrongs Jamal because believing that proposition in his moral context is an instance of racism.
It is important to bear in mind that radical moral encroachers want to avoid saying that the way in which the belief is formed explains why it is a doxastic wrong. There is a good reason for this, since if radical moral encroachers maintained that the way in which the subject believes explains why they doxastically wrong, their opponents could argue that those ways are epistemically inapt, irrespectively of their moral properties. Indeed epistemic purists like Gardiner (2018) have argued that the cases that moral encroachers put forth can be explained by the fact that the ways of believing are bad from a purely epistemic perspective, and Traldi (forthcoming, pp. 8-9) has argued that if doxastic wrongs occur in virtue of a problematic way of believing then subjects should not be sensitive to the evidence they possess and hence epistemic and moral norms on belief can conflict.
In a given moral context beliefs can wrong others via various mechanisms. For instance, a belief might falsely diminish the person (Schroeder, 2018), or it might betray the fact that the believer does not treat the person as an end, but merely as a means, and therefore that the believer fails to hold others in personhood (Basu, 2021). 11 The belief might be racist, contribute to structural injustices, or undermine the kind of relationship one is committed to being in with another person.
Some moral encroachers hold that only false beliefs can wrong others (Moss, 2018;Schroeder, 2018), while others maintain that even true beliefs can do so (Basu, 2019a;Bolinger, 2020). This dispute is somewhat inconsequential for our purposes since even a true belief might carry the risk of wronging someone, and moral encroachers hold that this is sufficient to render the belief epistemically unjustified (Bolinger, 2020(Bolinger, , p. 2422Moss, 2018). 12 Given that it is impermissible to pose unnecessary risks on others, merely by invoking the risk of wronging someone a belief is morally wrong (though it doesn't perhaps necessarily wrong any individual or group). For the sake of the argument I accept that we can morally wrong others in thought, irrespective of the actions and options that those doxastic states license. Finally, although the ideas of doxastic wronging and radical moral encroachment are logically distinct (Basu, 2021), I will examine the structural consequences of radical moral encroachment when it is coupled with the idea of doxastic wronging. The reason for this is two-fold: First, radical moral encroachers in fact embrace both ideas. Second, radical moral encroachers hold that epistemic justification is sensitive to the moral status of the beliefs themselves, but they have not provided any other mechanisms than doxastic wronging that could alter the moral status of a belief. Having laid out the core claims of radical moral encroachment I now turn to the main task.

3
The structure of moral encroachment

Structural consequences
The structural properties of epistemic justification determine what kind of inferences are valid for justification. A widely debated question about the structure of justification is whether justification is closed under competent deduction. While many doubt the validity of multi-premise closure principles for justification, few would wish to deny the following kind of single-premise closure principle: SPC: If S has a justified belief that p and competently deduces q from p while retaining her justified belief that p, then S's belief that q is justified.
If we denied this principle it is hard to see what kind of way of believing would suffice for justification. If competent deduction from a justified belief is not a good enough process to extend our body of justified beliefs, what could be? Competent deduction is, after all, the most reliable process there is. But this is not the proper place to argue for SPC. Let's turn to examine whether SPC is valid if the encroachment views are correct.
Pragmatic encroachment invalidates SPC. 13 This is because it is possible that p is practically adequate for a subject, while p ˅ q is not. Therefore, a subject S, who justifiably believes that p, and competently deduces p ˅ q from p while retaining her justified belief that p throughout, isn't necessarily justified to believe that p ˅ q. To prove that p might be practically adequate for S while p ˅ q is not, suppose that the following matrixes represents S's preferences and the strength of her epistemic positions with respect to relevant propositions. Epistemic probability p = 0.9 q = 0.05 Given S's epistemic position, p is practically adequate for her. This is because what she ought to do given her actual epistemic situation doesn't differ from what she ought to do if her epistemic position with respect to p was ideal. The expected utility of A on her actual epistemic position is 90 and the expected utility of B is 60, and the expected utility of A, conditional on p is 100 and the expected utility of B, conditional on p is 60. Therefore, since what S ought to do doesn't differ between her actual epistemic position and an ideal epistemic position with respect to p, p is practically adequate for S. Suppose then that S believes with justification that p and competently deduces p ˅ r from p while retaining her justified belief that p throughout. But p ˅ r is not practically adequate for S. Given S's actual epistemic position 1 3 S ought to choose option A, but conditional on p ˅ r S ought to choose B, since the expected utility of B conditional on p ˅ r is 60 and the expected utility of A conditional on p ˅ r is 34. Therefore, according to pragmatic encroachment S cannot believe with justification that p ˅ r via competent deduction from p since p ˅ r is not practically adequate for S. 14 It's worth to emphasize that our epistemic position with respect to a disjunction is often better than our epistemic position with respect to either disjunct since the probability of a disjunction is always greater or equal to the probability of either disjunct, whichever is more probable. Therefore, it's not just the case that a subject can fail to believe with justification a proposition she competently deduces from a premise she believes with justification -she can fail to do so even if her epistemic position with respect to the conclusion is better than her epistemic position is with respect to the premise.
The above argument demonstrated that if there is a practical adequacy test on epistemic justification, then SPC is invalid. Moral encroachers ought to subscribe to the practical adequacy test if the utilities of options are partially determined by their moral status. And there seems to be no reason why utilities of options should not be partially fixed by the moral status of those options (Worsnip, 2021, p. 540). Moderate encroachers straightforwardly endorse the practical adequacy test since they think that epistemic justification of a belief depends partially on the moral value of the options it licenses (Fritz, 2020, p. 3056). But radical moral encroachers ought to subscribe to the thesis also. After all, it is hard to see the motivation for a view according to which the epistemic standing of a belief can be affected by the belief's moral properties, as long as those moral properties do not hinge on the moral standing of the options that the belief licenses. Indeed, it should come as no surprise that radical moral encroachers like Schroeder endorse both moderate and radical moral encroachment (Schroeder, 2012(Schroeder, , 2018.
There are also concrete cases that do not rely on the practical adequacy test that can be used to demonstrate that radical moral encroachment invalidates deductive closure principles. While it is hard to give a natural case where something is deduced from a single premise, since nearly all actual inferences contain more than one premise, it is easy to demonstrate that competent deductions that proceed from just a few justified premises can fail to preserve justification if moral encroachment is true. Consider the following alteration of Wine Stain.
Wine and Deduction: Yesterday Laura's partner, who had struggled with an alcohol problem for many years but been sober for a few months, was at a departmental reception where wine was served. Laura's partner came from the reception with a big wine stain on their white shirt, but at the moment Laura thought nothing of it. During lunch Laura is told by her good and reliable friend that there was only one person with a white shirt at the reception, and that they spilled a glass of red wine on their shirt while drinking. Laura justifiably believes that [her partner was at the reception wearing a white shirt] (p), and that [the only person at the reception with a white shirt was drinking wine] (q). From p and q Laura competently deduces that [her partner drank alcohol at the reception] (r).
Assuming that we can acquire justified beliefs through perception and reliable testimony, we must grant that Laura is justified in believing p and q. Since Laura competently deduces r from p and q Laura ought to be justified to believe that her partner drank alcohol at the reception. But radical moral encroachers must deny this, since believing r in Laura's moral context constitutes a doxastic wrong. We shouldn't believe the worst of our loved ones. Therefore, justification is not closed under competent deduction.
Both pragmatic and moral encroachment entail failures of single-premise closure. What other structural implications do these views have for justification? Perhaps it is somewhat natural to think that pragmatic and moral factors can raise the bar for justified belief, and that such factors can therefore turn a belief that was justified unjustified. Both moral and pragmatic encroachment entail that justification can be defeated by extra-evidential factors. But it is much less plausible that pragmatic or moral factors can turn an unjustified belief into a justified one. However, if pragmatic factors can raise the bar for justification they can also lower it. And it is easy to see that a belief can be practically inadequate at t 1 but practically adequate at t 2 without the subject gaining any relevant evidence for her belief. Consider for instance the following case: Car Keys: Claire borrowed her car to Matt yesterday and told him to leave the car keys on the tire next to the driver's seat. Claire believes that the keys are on the tire (p) and is just about to leave home and take the three-kilometer walk downtown to the car, when she recognizes that p is practically inadequate for her. She should take the spare keys with her just in case, and not act as if p was the case. Having taken the keys with her, she recognizes that nothing hinges on p anymore, and that she can act as if p. As she walks downtown, she justifiably believes that the keys are on the tire.
In Car Keys Claire turns her unjustified belief into a justified one simply by picking up the car keys. So it is not just the case that justification can be defeated by pragmatic factors, as when you suddenly learn that it is extremely important that the vegan ice cream you served does not contain nuts since one of your guests is allergic to nuts. Rather, you can also gain justified beliefs when the disutility of an option decreases as time passes. 15 Radical moral encroachment has a similar consequence. To see this consider the following case: Old Wine in New Relations: A few years ago Laura's partner, who had struggled with an alcohol problem for many years, but been sober for eight months, came from a departmental reception with a big wine stain on their shirt. Ever since that evening Laura thought that her partner fell off the wagon. Unbeknownst to Laura, her belief doxastically wronged her partner since we shouldn't believe the worst of our loved ones. But Laura broke up with her partner long ago (largely due to the fact that her partner started mistreating her a year ago) and nowadays there's little love lost between them. As a result her present belief (that her partner fell off the wagon) doesn't doxastically wrong the ex-partner and is epistemically justified.
If moral encroachment is true then Laura's belief was unjustified when she formed it, but after things turned sour her belief became justified, even though she didn't gain any new evidence in favor of the belief. 16 But it is puzzling that the epistemic standing of Laura's belief could change in such a way. The fact that changes in moral factors can turn unjustified beliefs justified also entails the truth of weird counterfactuals. For instance, if Laura believed that [I am not justified to believe that she fell off the wagon, but if we broke up I would be justified to believe it] while still in the relationship, her belief would be true, and it is hard to see why it could not be believed with justification. But perhaps radical moral encroachers could claim that Laura's belief in the counterfactual would doxastically wrong her partner, and hence the belief would be unjustified. 17 If that is the case then there would be true propositions that cannot be justifiably believed according to radical moral encroachment. I will return to such cases below. Now I turn to consider structural features that differentiate pragmatic and radical moral encroachment. I argue that if radical moral encroachment is true then there are true propositions that cannot be justifiably believed due to their content. The argument is straightforward.
According to radical moral encroachment beliefs that morally wrong cannot be epistemically justified. Recall that radical moral encroachers hold that "doxastic wrongs are wrongs in virtue of the content of what is believed" (Basu, 2021, p. 108). Earlier we noted that this statement must be qualified to reflect the fact that the moral implications of believing a certain proposition depend on the moral context in which one believes that proposition. Suppose then that S believes a true proposition p in a moral context C in which believing that p constitutes a doxastic wrong. Perhaps p is a true racist generalization, and in believing that p S contributes to structural injustices. Given that believing that p constitutes a doxastic wrong in the moral context C, p cannot be believed with justification due to its content in context C. This is a striking consequence since it is generally accepted that epistemic rationality does not 17 I would like to thank an anonymous reviewer for this suggestion. 16 I assume here that if friendship is, or involves a kind of commitment, as Stroud (2006, pp. 511-512) holds, then that commitment should be discharged once the friendship is over. I might owe it to my friend to pick them up at the airport if they were to ask me to do so, but if they do things that make the friendship end I owe them no such thing. I would like to thank a reviewer for inviting me to be more explicit here. Thanks to Max Lewis for discussion.

3
The structure of moral encroachment forbid forming specific beliefs due to their content, if that content isn't contradictory. True enough, some have argued that we cannot make rational mistakes about the requirements of rationality, and these mistakes need not involve believing any contradictions. 18 But at least these mistakes involve believing propositions that are necessarily false, and there are many theories of epistemic justification on which such beliefs are never justified. 19 So while there might be principled reasons to hold that some propositions that are not contradictions are such that we could never be justified to believe them due to their content in a given context, radical moral encroachers cannot invoke similar reasons, since according to their view there are some true propositions that cannot be justifiably believed in virtue of their content in certain moral contexts. 20 Note that nothing analogous holds of pragmatic encroachment or moderate moral encroachment. There are no true propositions, such that one's epistemic position with respect to them couldn't be practically adequate due to the content of those propositions. This is because whether p is practically adequate for a subject depends partially on how epistemically probable p is. If S is epistemically certain that p, then her epistemic position with respect to p is trivially practically adequate since the ranking of options between S's actual epistemic situation and an ideal epistemic situation with respect to p is the same, since S's epistemic situation with respect to p is ideal. Mutatis mutandis for moderate moral encroachment in so far as their view is captured by the practical adequacy test.
According to radical moral encroachment beliefs that doxastically wrong are thereby epistemically unjustified, no matter how much epistemic support the belief has. This entails that moral considerations have lexical priority over epistemic considerations when it comes to determining whether a subject's belief is epistemically justified. That is, moral requirements on belief must be satisfied at the expense of epistemic requirements if they stand in conflict. 21 This feature of radical moral encroachment has the potential to re-orient our understanding of how inquiry ought to be conducted in certain cases. For instance, suppose that you know that believing p in your moral context would constitute a doxastic wrong. Therefore you know that if you were to inquire into the question whether p, your inquiry could not settle on p since believing that p would be unjustified and wouldn't constitute knowledge either. 22 So prior to evaluating the evidence for p and ¬p you can foreclose p. Notice that this doesn't entail that you could settle on ¬p since the evidence might actually favour p to the extent that it renders believing ¬p epistemically unjustified. This leaves us in an uncomfortable spot. A subject's inquiry regarding whether p might terminate in the following position: She knows that she is not justified to believe that p, since by believing that p she would doxastically wrong, and she knows that she is not justified to believe that ¬p, since the epistemic probability of p is so high. It seems to me that a flawless inquiry into whether a racist generalization is true would lead to such a predicament if the generalization and radical moral encroachment were true.
It is worth pointing out that it would be quite surprising if moral considerations had lexical priority over epistemic considerations when it comes to epistemic justification. If our beliefs should first answer to moral requirements, and only then to epistemic requirements, moral factors would arguably be the most important factor when it comes to determine whether a belief is epistemically justified. How is it then that philosophers have failed to recognize this feature of epistemic justification for several millennia? One would suppose that everyone has not until now ignored the most important factors when it comes to determine whether a belief is epistemically justified. This of course does not entail that radical moral encroachment is false, since philosophy is hard. But it is fair to ask radical moral encroacher for an explanation why we have failed to realize that moral considerations have lexical priority over epistemic ones when it comes to epistemic justification.
Again, no analogous problem arises for pragmatic encroachment. There is no pressure to think that practical considerations would have lexical priority over epistemic considerations, since epistemic factors directly influence whether a proposition is practically adequate for a subject. No matter the stakes involved in believing that p, your epistemic position with respect to p will be practically adequate if you are epistemically certain that p. Mutatis mutandis for moderate moral encroachment in so far as their view is captured by the practical adequacy test.

Embracing both moral and pragmatic encroachment
Before offering my own explanation of the kind of cases that encroachers have used to motivate their views, I wish to explore whether pragmatic and moral encroachment are at odds with each other. As we will see, encroachers cannot just accept pragmatic and moral encroachment as stated, for that kind of view is crucially 1 3 The structure of moral encroachment underdeveloped. Encroachers must tell us how the two views interact with each other. The following case illustrates how pragmatic and moral encroachment can at least seemingly deliver conflicting recommendations.
Escape from Cosmos: 007's mission has taken her to the Cosmos Club. Having received the intelligence report before the operation, she knows that all of the staff are black and that the chance that there is another black guest at the club is very low. As she is drinking a Martini, she spots the villain who is just about the leave the premises. It is extremely important that 007 catch the villain and the only way to do it is to get her coat from the coat check (the villain would be alerted is they saw someone leaving the club without their coat). A black man is standing near the coat check. She gives her coat check to the man, and receives her coat in return.
Escape from Cosmos is adapted from a case much discussed in the moral encroachment literature, in which radical moral encroachers hold that the protagonist doxastically wrongs the man. 23 Since 007 believes on the basis of the man's race that he is staff, her belief doxastically wrongs the man. 24 But given what is at stake, 007's epistemic position with respect to her belief is practically adequate. She ought to act as if the man was staff, since that is the only way she can get her coat in time. So according to pragmatic encroachment 007's belief might be justified, while radical moral encroachers must hold that 007's belief is unjustified. Moderate moral encroachers, like Moss (2018, p. 198) and Bolinger (2020Bolinger ( , p. 2426 hold that by adopting certain beliefs on the basis of statistical generalizations we immorally risk acting on those beliefs, and this renders the beliefs epistemically unjustified even if they are true. Hence some moderate moral encroachers should feel the pull that 007's belief is unjustified due to the actions it licenses.
Of course Escape from Cosmos doesn't show that moral and pragmatic encroachment are inconsistent views. No example could do that, since both views give rise only to necessary conditions for justification. But Escape from Cosmos does put pressure on the encroachment views nonetheless. After all, the key motivation behind pragmatic encroachment is that action and epistemic justification are connected. Many pragmatic encroachers subscribe to justification or knowledge-action links according to which one can treat a premise as a reason in practical deliberation only if one knows or justifiably believes the premise. And since 007 can act as if the man is staff, it seems that she can treat it as a reason in practical deliberation. So pragmatic encroachers ought to feel the pull of accepting that 007 is justified to believe that the man is a member of the staff.
If the stakes are high enough, as they often are in movies vaguely based on Ian Fleming's books, 007 can be practically required to believe that the man is staff. But no matter the purely practical stakes involved, 007's belief would wrong the man, and hence must be unjustified on radical moral encroachment. So moral and pragmatic norms on belief can issue conflicting demands. This fact is hard to swallow for someone who would like to endorse both moral and pragmatic encroachment, since radical moral encroachers hold that epistemic and moral requirements on belief cannot conflict. For instance, Basu and Schroeder (2019, p. 197) write: "If morality forbids beliefs that epistemic rationality does not, then there will be cases of belief that are rational in every way required for knowledge, but are nevertheless morally wrong. We believe that there are no such cases." Basu and Schroeder also state that it would be puzzling if a belief could be morally impermissible while being epistemically permissible (ibid.). But I contend that the opposite is true on the condition that there are moral requirements on belief. It would be puzzling if epistemic and moral requirements on belief could not conflict.
We have already seen that pragmatic and moral requirements on belief can conflict, and this gives us reason to suppose that epistemic and moral requirements on belief can also conflict. Moreover, everyone agrees that a belief might be permissible from a pragmatic perspective, while being impermissible from an epistemic perspective and vice versa. Believing at gunpoint that 2 + 2 = 5, and that your resolution to quit smoking will not hold, being the case in point. So why would the interplay between epistemic and moral norms on belief be special, in that these two domains cannot give rise to deontic conflicts?
There are concrete cases where they conflict. Fritz (2020, p. 3051) discusses cases featuring moral bribes in which something morally good would happen if you formed a belief on the basis of insufficient evidence. Plausibly we have a conflict between moral and epistemic requirements on belief in such cases. Moral and epistemic norms can also conflict outside the doxastic domain. For instance many hold that there is an epistemic norm on assertion, which requires that one must assert that p only if one knows that p (Williamson, 2000, p. 243) or if it is reasonable for one to believe that p (Lackey, 2007, p. 608). These putative epistemic norms of assertion specify when it is epistemically proper to assert something (Kauppinen, 2018). But sometimes we are morally required to lie, and this plausibly involves asserting something that one believes to be false (Stokke, 2013), and hence asserting something that one (plausibly) does not know, nor is reasonable for one to believe. Therefore, we might be morally required to violate the epistemic norm of assertion.
Furthermore, it seems that moral normativity in itself can give rise to conflicting demands. It is far from obvious that there doesn't exist genuine moral dilemmas: cases in which we face conflicting moral requirements neither of which is over-ridden in any morally relevant way. 25 If moral normativity is not free from dissonance,

3
The structure of moral encroachment why should we think that moral and epistemic norms on belief are always in harmony? And the possibility of genuine epistemic dilemmas shouldn't be ruled out by fiat either. Epistemic norms such as the truth norm, which states that "one ought (epistemically) to believe only truths", and the rationality norm, which states that "one ought (epistemically) to be epistemically rational" plausibly conflict in cases featuring misleading evidence. According to epistemic dilemmism, defended by Hughes (2019), both the truth and the rationality norms are binding, and there exists genuine epistemic dilemmas, cases in which we end up violating a genuine epistemic norm no matter what. If there are genuine dilemmas of either moral or epistemic variety, then it is hard to see why moral and epistemic requirements on belief would always be in harmony. After all, for all epistemic dilemmas, it applies that one epistemically should Ф and that one epistemically shouldn't Ф, where Ф-ing is believing or not believing some proposition. If there are no conflicts between moral and epistemic requirements on belief, moral requirements must be silent regarding whether it is permissible to Ф. But if epistemic dilemmism is true, then one can be in an epistemic dilemma with respect to practically any proposition, including those that are such that believing them would constitute a doxastic wrong. Therefore, moral and epistemic requirements can conflict. 26 Finally, there is a version of Escape from Cosmos where 007 is morally (all things considered) required to believe that the man is staff, since it is morally imperative that she catches the villain, but she will still doxastically wrong the man by forming that belief, and hence radical moral encroachers must hold that her belief is epistemically unjustified. Basu and Schroeder themselves open up this possibility, since they hold that one can wrong others by φ-ing, even though one is all things considered required to φ. They (2019, p. 198) state that "some actions wrong someone but are not wrong all-things considered […] strikingly, apology does still seem owed in such a case". And "if you owe someone an apology […] then you must have wronged them" (ibid.), and "there cannot be beliefs that are both rationally epistemically permissible and also constitute doxastic wrongs" (ibid., p. 199). From these remarks it follows straightforwardly that there can be cases where one is morally permitted (and even required) to believe that p, but believing that p is epistemically impermissible. But if that is the case then moral and epistemic requirements on belief can stand in conflict.

3 5 A purist explanation
In this section I offer a purist explanation of the kind of cases that encroachers have used to motivate their views. 27 By purism I refer the idea that what one is epistemically is permitted or required to do can only be influenced by purely epistemic matters (matters relating to truth, such as evidence). 28 The intuitions encroachers wish to elicit can be explained while holding that epistemic normativity is insulated form moral and practical normativity. The explanation is straightforward.
Let us accept for the sake of argument that in putative cases of encroachment the subject's beliefs enjoy sufficient epistemic support, so that their beliefs are justified. 29 Therefore, Hunter, Spencer, Laura and 007 are epistemically justified in their beliefs. The challenge a purist faces is to explain why our protagonists should not believe as they do in the 'high stakes' scenarios.
The first step to meet this challenge is to accept that there are no positive, purely epistemic duties or obligations. 30 If there are no positive, purely epistemic duties, then no one is required to form specific beliefs solely in virtue of their epistemic position. Therefore, even though it is highly probable for instance on Spencer's evidence that Jamal will tip substantially lower than the average customer, he is not epistemically required to believe so. Hence the purist account that I advocate doesn't require that subjects in putative cases of encroachment believe what they believe from an epistemic point of view, since according to the view no one is required to believe anything from an epistemic point of view.
The second step is to notice that in the 'high stakes' cases that encroachers put forth, the protagonists have either practical or moral reasons to not believe as they do. I think it is clear enough that certain beliefs can be impermissible from a practical point of view. From a practical point of view, you should not believe that there is a spider in the room because you suffer from arachnophobia, and believing that there 27 I should note that my proposal is one among many and I don't claim that it is the one we should necessarily adopt. A nice feature of my solution is that it covers cases of pragmatic encroachment as well. For other purist explanations of the kind of cases that moral encroachers have used to motivate their views, see for instance Osborne (2021) and Gardiner (2018). 28 Fantl and McGrath (2007, p. 558) give a similar definition of purism regarding knowledge. Note that the way in which I understand purism is consistent with the idea that pragmatic factors can influence what one ought to believe, and hence it is compatible with pragmatist views such as those espoused by Quanbeck and Worsnip (forthcoming) and Paul (manuscript). 29 Some have argued that subject's like Spencer and 007 are not epistemically justified in their beliefs because epistemic justification is not a matter of evidential probability, but of something else, like ruling out relevant error-possibilities (Bolinger, 2020;Gardiner, 2018). I am inclined to agree with these epistemologists, since I too think that bare statistical evidence is insufficient for epistemic justification (Hirvelä 2019(Hirvelä , 2022. However, for the sake of argument I will work with the assumption that bare statistical evidence sometimes suffices for epistemic justification. The purist explanation I offer is therefore dialectically stronger than alternative explanations, since I accept (for the sake of the argument) that the protagonists in encroachment cases are prima facie epistemically justified in their beliefs. 30 Nelson (2010) has argued that we have no positive epistemic duties. I endorse his arguments. It is worth to note that Basu and Schroeder (2019, p. 196) cite approvingly the same arguments. My aim here is not to convince the reader that we don't have positive epistemic duties, but merely to demonstrate what can be achieved by accepting the claim. is a spider in the room would cause you to scream and wake up your toddler that you've just managed to put to sleep. Similarly, we can accept, at least for the sake of argument, that there can be distinctively moral reasons not to believe certain propositions, because it would morally wrong others to do so. If in 'high stakes' cases subjects have no epistemic duty to believe, but have either a moral or a practical duty to not believe, then barring some special circumstances in which they are permitted to violate the moral or practical pro tanto duty to not believe, they shouldn't believe in such cases. Subjects in 'low stakes' cases, however, don't have moral or practical reasons to not believe, and hence are permitted to believe as they do.
Some will no doubt object to this explanation, and maintain that we have positive epistemic duties. Sometimes, form a purely epistemic point of view, we ought to believe certain propositions, and to do otherwise would be epistemically irrational. Hughes, for instance, discusses a case where you are searching for your favourite mug and see it in the dishwasher. Your visual experience is as vivid as it could be, the mug has a distinctive pattern and so you couldn't easily have mistaken it for another one. If you now failed to believe that the mug is in the dishwasher you wouldn't be "a paragon of rationality" (Hughes, 2019, p. 4061). I agree with this sentiment. If you failed to believe that your favourite mug is in the dishwasher, you would not be a paragon of rationality. But I don't think that you would be epistemically irrational, rather you would manifest practical irrationality. Given your desires and epistemic situation, you ought to believe that the mug is in the dishwasher. But this ought is not purely epistemic. Rather, you only ought to believe that your favourite mug is in the dishwasher given that you want to drink your morning coffee from it. You have a practical reason to find your mug, and if you fail to believe that the mug is in the dishwasher in your epistemic situation you fail to achieve something that you could easily achieve in your current situation.
The fact that your failure to believe isn't an epistemic one is evidenced by the reactive attitudes that it would be appropriate to have towards you given your failure to believe. Kauppinen (2018) has argued that the proper reactive response to someone who violates an epistemic norm, such as the (epistemic) rationality norm, is to lower one's credence in what they would say in the future about similar issues. But it isn't at all clear why I should lower my credence in what you say in the future given that you simply didn't form the belief that your favourite cup is in the dishwasher. While you might be a bad informant regarding this matter, you are not more likely to supply me with false than true beliefs due to your practical irrationality. Rather, it seems appropriate for me to not count on you in action, and to place a lower credence in your success in pursuing our shared goals regarding kitchenware. This seems to imply that you violated a norm of practical rationality. 31 Furthermore, while I reject the idea that there could be positive, purely epistemic duties, I do think that there can be positive practical and perhaps even moral duties to believe certain propositions. Climate change deniers and anti-vaccinationists can rightfully be criticized for their failure to believe what the evidence supports. From a practical, and perhaps from a moral point of view, they ought to believe in 1 3 accordance with the evidence. And since they are also epistemically permitted to believe what the evidence supports, they ought believe, rather than suspend their judgment.

Conclusion
I explored the structural implications that pragmatic and moral encroachment would have for justification if the theses were true. I demonstrated that both encroachment views invalidate single-premise closure principles, and that both views allow for the possibility that factors irrelevant to the truth of what is believed can turn an unjustified belief into a justified one. Furthermore, I argued that radical moral encroachment, unlike pragmatic or moderate moral encroachment, entails that there are some true propositions that cannot be believed with justification due to their content, and that moral considerations have lexical priority over epistemic ones. I also explored the interconnections between moral and pragmatic encroachment, arguing that it is far from obvious how the two views can be reconciled. Finally, I offered a purist explanation of the kind of cases that encroachers have used to motivate their views. The central premises of this explanation were that we have no positive epistemic duties to believe, and that we can have pragmatic, and perhaps even moral reasons to believe and suspend.