The Unbelievability Argument

Although there are different ways in which one might arrive at the conclusion that normative error theory is unbelievable, Streumer focuses on the fact that believing normative error theory would seem to involve believing, at the same time, that there is positively no reason to believe it.

To elaborate: Streumer uses the term belief to mean explicit and occurrent beliefs that satisfy the following two conditions:

In what follows, I accept both B1 and B2. However, Streumer thinks that satisfying B1 and B2 involves satisfying these further conditions:

Again, for the sake of argument I accept B3. If B4 is true as well, then normative error theory is going to be unbelievable. For someone who believes in normative error theory will, if they are to satisfy B3, recognize that this involves believing that there is no normative reason to believe normative error theory, which violates B4. But I think that there are good independent reasons to think B4 is false. If I am correct, then the fact that believing normative error theory involves believing that there is no reason to believe in normative error theory will not, in and of itself, provide grounds for thinking the view unbelievable, and so Streumer's case will fail.

Because the matter under debate here is the believability of normative error theory rather than its truth, the claims I make about the nature of normative reasons need only be believable for my case to go through. They do not need to be true; they do not even need to be particularly plausible (though I believe they are). Believability is all I need.

Normative Reasons

Following many other writers on normative reasons, including Streumer himself, I characterize normative reasons as favoring relations (see Bedke Reference Bedke2010: 47–48; Dancy Reference Dancy2004: 14; Harrison Reference Harrison2018: 11–27; Raz Reference Raz1975: 186; Shafer-Landau Reference Shafer-Landau2009. Streumer says ‘I will use the term ‘“reason for belief” to mean a consideration that counts in favor of having this belief’ (Reference Streumer and Machuca2017b: 74n9). As I discuss below, if normative reasons are favoring relations then for a consideration to count in favor of a belief is for it to be the consideration in virtue of which we are favored believing something. That is, a normative reason is a favoring of doing or believing something, whatever else it may be. This certainly seems believable, anyway.

Relations have relata. Nothing in my case is going to turn on how many relata the favoring relations constitutive of normative reasons have, so I mention just three that they typically—and believably—possess. These are a, the agent—the object of the favoring (that is, the person who is being favored doing or believing something); b the belief, act, or attitude that the agent, a, is favored doing, adopting, or believing; and c, the basis upon which a is favored doing, adopting, or believing b. To illustrate: if it is true that I have a normative reason to believe that Napoleon won Waterloo because there are facts that indicate he did, then I am relatum a; believing Napoleon won Waterloo is relatum b; and a fact indicating that b is true or likely true is relatum c. The normative reason itself is neither a, b, nor c. I am the one who is favored believing something, but I am not the favoring itself. And the act or belief that I am favored doing or believing is not the favoring either. Nor is the consideration on whose basis I am favored doing or believing it. And though it is true that some talk of facts as favoring things, plausibly talk of reasons as facts should be taken to be shorthand for ‘the fact is the basis upon which I am favored doing or believing it’ (so, ‘the fact she is suffering is the reason to stop it’ means ‘the fact she is suffering is the consideration that is generating the reason to stop it’—a point made by Bedke Reference Bedke2010. After all, the idea that facts themselves can favor things seems no more credible than thinking they can hope for things or intend things or fall in love with one another.

This analysis of normative reasons, as well as seeming entirely believable, also allows us to distinguish different kinds of normative reason. For instance, instrumental reasons can be understood to be favoring relations in which c is the fact that doing the act (or believing the proposition) will serve some of a's ends. A moral reason can be characterized as a favoring relation in which c is the fact that doing the act or believing the proposition will (say) be an act or belief that a self-interested rational deliberator would have agreed to from behind a suitably obscuring veil of ignorance (the details will depend on which normative moral theory is correct, of course). And an epistemic reason will be a favoring relation in which b is a belief and c is a fact indicating b's truth or likely truth.

Note, in characterizing normative reasons in this way I am not assuming the truth of any specific metanormative theory. For instance, a subjectivist about normative reasons is going simply to add a relatum to the relation, namely the bearer of the favoring relation (the bearer being a subject of some kind). By contrast an objectivist naturalist about normative reasons is going to identify relatum c as the favorer (that is, its featuring as relatum c constitutes its being the favorer) and a nonnaturalist is going either to identify the relation itself with some nonnatural feature or maintain that there does indeed need to be a bearer, but that this role is performed by a nonnatural feature. In this way then, the relational analysis is neutral between subjectivist, objectivist naturalist, and objectivist nonnaturalist analyses of the normative.

It seems believable, then, that the normative reason is the favoring relation that obtains between its relata and that it is a major mistake to confuse that relation with one of its relata. Normative reasons are relations not relata. The tendency to conflate a normative reason with a relatum within the favoring relation operates to make B4 seem more prima facie plausible than it really is. If, for example, one conflates a reason to believe p with facts, then believing that one has no reason to believe p would involve believing that there are no facts about p's truth. But this is not what believing one has no normative reason to believe p involves; it involves believing that there is no favoring of one's believing p. Similarly, normative error theory is the theory that there are no such favorings tout court, but it is not the theory that there are no facts (something I stress again below when I deal with objections). The claim expressed in B4 then, is that believing p is incompatible with at the same time believing that one is not favored believing it.

In summary then, it is believable that normative reasons are not facts or considerations that favor, but are favoring relations in which facts and other considerations feature as relata. And as such, normative nihilism is not the view that there are no facts or other such considerations, but rather the view that facts and other considerations systematically fail to generate the favoring relations constitutive of normative reasons. Relatedly, believing that one has no normative reason to believe a given proposition is not of a piece with believing that there are no facts about its truth. Rather, to believe that there is no normative reason to believe a given proposition is to believe that the facts in question do not generate any favoring of one's believing it.

Holistic Ted

Ted believes all of the above and is additionally a holist about the normative aspect to reality. That is, he believes that any consideration—so, any relatum c—that generates a favoring of doing or believing something in one context, may generate a disfavoring of doing or believing it in another, or generate no favoring of doing or believing it in yet others.

For example, Ted believes that, typically anyway, the fact an act will cause someone to suffer will generate a moral reason not to perform the act in which it features. But he does not believe that it will invariably do so. For instance, he thinks that in some contexts a person may deserve to suffer and that sometimes it is right and good to give someone what they deserve. Thus, there can arise situations in which the fact an act will cause a person to suffer may operate to generate a moral reason positively to perform the hurtful act; that is, it generates a favoring of it, rather than a disfavoring.

And he can imagine yet other contexts, such as ones in which the person the act will cause to suffer is the agent of the act and no one else, where there seems no moral reason not to perform the act, just an instrumental reason not to. For example, voluntarily striking himself in the face seems, other things being equal, to be imprudent rather than immoral. So again, though normally the fact an act will cause a person to suffer operates to generate a moral disfavoring of performing it, in this context it seems not to: it generates an instrumental disfavoring of performing it instead.

He thinks this same kind of variability is going to be true for other normative reasons as well (after all, why would it not be?). Take instrumental reasons. Instrumental reasons are, by definition, reasons to do and believe things that have been generated by the fact that doing or believing them will serve some of one's ends. But although Ted believes that the fact an act will serve some of his ends is a consideration that typically generates an instrumental reason for him to perform the act in question, it does not seem to him invariably do so, for he can conceive of cases where it does not. When, for instance, he imagines some sadistic desire arising in him—a desire to hurt Mary just for fun—his faculty of reason seems to represent him to have no reason whatsoever to act on that desire, rather than some instrumental reason that is being eclipsed by stronger moral reasons not to do so. Normally, if he desires to do something, his desire to do it generates an instrumental reason for him to do it. But in this case, the moral disfavoring—so the negative moral normative valence that his sadistic desire possesses—seems to be operating to prevent what might otherwise have generated an instrumental favoring of performing a given act from doing so. So, a moral reason seems to be preventing an instrumental reason from being created.

Finally, and most importantly for my purposes here, Ted believes the same is true for epistemic reasons. That is, Ted believes that, typically, the fact a proposition is true or likely true will generate an epistemic reason for him to believe it. But he does not think it will invariably do so, any more than he thinks the fact an act will serve some of his ends invariably generates an instrumental reason to perform it, or the fact an act will make someone suffer invariably generates a moral reason not to do it. Whether the fact a proposition is true generates an epistemic reason for him to believe it depends, he thinks, on what other features are present, including what other normative valences may be operating in the situation. Thus, Ted believes that a fact about a proposition's truth does not necessarily generate a normative reason to believe the proposition. Of course, when or if it does so, the favoring in question will qualify as an epistemic normative reason. But the fact a proposition is true does not, thinks Ted, always and everywhere generate a favoring of believing it.

At this point one might object that the kind of normative holism that Ted believes in is quite radical and controversial. For Ted believes that among the kinds of consideration that can, in principle, affect what normative valence another feature has in a given context includes the normative valences of other features present.

In reply, it may be true that Ted's normative holism is quite radical and controversial, but it seems no less believable for that. And I described a case in which it seems believable that this happens: it seems believable that a sadistic desire to hurt another for fun will not typically generate any instrumental reason for its bearer to perform that act precisely due to the moral valences of the desire and act in question—that is, the negative moral valence of that feature is preventing it from generating any instrumental reason. And I describe another below. Consider, too, the following analogy with tastiness, about which Ted is also a holist of this kind. It seems entirely believable—and highly plausible—that whether a flavor tastes pleasant or unpleasant or registers as neither (so, whether a flavor operates with a positive or negative or neutral taste valence) varies according to context. And clearly one way in which the taste valence of a flavor might be altered is by what other flavors are present. For many people, the flavor of garlic, for instance, typically has a positive taste valence when accompanying the flavor of mushroom. However, it has a negative taste valence when accompanying the flavor of strawberry. But other flavors are not the only considerations that can alter a given flavor's taste valence and it would be rather odd to insist otherwise. It seems every bit as true and believable that the taste valence of a flavor can be affected by other features of a foodstuff, such as its temperature or texture. To use myself as an example: I find the flavor of tomato unpleasant if the substance bearing it is liquid and cold, yet I find it pleasant if the substance bearing it is a cold solid, such as a tomato itself, and also pleasant if the substance bearing it is liquid and hot, such as a hot tomato soup. And even prior to carrying out such experiments, I would consider it peculiarly dogmatic to have as one's working assumption that a flavor's taste valence can be affected by other flavors and nothing else. What Ted and I believe about flavor valences is what Ted believes about the behavior of normative valences too. That is, he believes the moral valence of a feature can vary from context to context; and sometimes this can be due to the moral valence of other features present, but sometimes it can vary due to the instrumental or epistemic valence of other features present. This seems a much more plausible kind of holism than one that—contrary to the whole spirit of holism—insists moral valences can be affected only by moral valences and no other kind. But this is not the place to argue the matter, for as mentioned above, all I need for my case to go through is for Ted's view to be believable.

True Propositions That We Have No Reason to Believe

I said above that Ted is a normative holist who thinks that the truth of a proposition does not always and everywhere generate a normative reason for us to believe it. Below I offer five concrete examples of such cases.

Believing What You Believe There Is No Reason to Believe

Thus far, nothing I have said directly challenges the truth of B4. However, below I seek to show how variations of the above kinds of case can be used to generate counterexamples to B4. They do not all have the same probative force—at least not in my view—but each has at least some and their cumulative effect provides a significant challenge to B4's credibility.

Take the first example above. Ted believes that the proposition ‘normative error theory is true’ is capable of being true, yet if or when it is true it will not generate any normative reason for him to believe it. Well, imagine that he wakes one morning to find that he is thinking, and thinking very confidently, that normative nihilism is true. What non-question-begging grounds are there for thinking that his thought that normative error theory is true fails to qualify as a belief?

True, he does not believe there is any normative reason to believe this proposition. But by hypothesis he believes that this is entirely compatible with its being true. So, we cannot on these grounds conclude that he lacks confidence in or understanding of the truth of the proposition. Does his believing he has no normative reason for believing the proposition entail that he believes there are no facts about its truth? No, for as has already been noted, facts are not themselves normative reasons to believe the proposition. If normative error theory is true, facts can still exist, it is just that they will not be generating any normative reason to believe in the truth of the propositions corresponding to them. And note that Ted understands this: Ted is not one of those who conflates favoring relations with considerations that typically feature as a relatum within that relation. So, he can believe there are facts that make the proposition ‘normative error theory is true’ true, consistent with believing that he has no normative reason to believe it. And there seems no ground for thinking that Ted does not properly understand the proposition's content either, for he understands that its truth entails that he has no normative reason to believe it, for this is something he already believed about this proposition. That is, prior to having the thought ‘normative nihilism is true’, Ted believed that propositions of this kind are ones that, if or when true, are ones that do not generate any normative reason to believe them. If he did not evince a lack of understanding earlier, there seem no non-question-begging grounds for thinking he does now. (Others have also proposed that the belief in normative nihilism may itself be a counterexample to B4—for instance, see Olson [Reference Olson2014: 171–72]—however, their cases are a little different to mine, as I explain below.)

Now consider a variation of the second example. The proposition ‘It is raining, but nobody believes it is raining’ is capable of being true, but it is not capable of being true and believed to be true at the same time. It was on precisely this basis that Ted believes he does not have any reason to believe it, when or if it is true. But imagine first that Ted has temporarily forgotten the basis upon which he believes there is no reason to believe the proposition ‘it is raining, but nobody believes it is raining’ if or when it is true. That is, he believes that the proposition is capable of being true, and he also believes that he has no reason to believe it, he just does not currently remember why. And now imagine that Ted finds himself thinking ‘it is raining, but nobody believes it is raining’. Given what I have said about Ted, can Ted be said to believe that it is raining and nobody believes it is raining?

I think the answer is no. But it is instructive to consider why. What is the most plausible explanation of why Ted fails to believe the proposition? Surely, if asked to explain, we would cite his failure to satisfy B3 in respect of it, rather than his failure to satisfy B4. That is, we would say that Ted does not, on this occasion, really understand the proposition, for if he did then he would realize that believing it makes it false. The fact he also believes he has no reason to believe the proposition seems, by comparison, to be by the by. That is not what one would expect if B4 were true, or at least not what one would expect if B4's truth were as clear and distinct as B3's. For in the case as I have described it, it is as clear—if not clearer—that Ted is in violation of B4 as it is that he is in violation of B3 (after all, I stipulated that he does not believe he has any reason to believe the proposition in question). So, other things being equal, we would expect his violation of B4 to be either the more prima facie plausible explanation of his failure to believe, or at least as prima facie plausible an explanation as his violation of B3 is. Yet it does not appear to be, and as such, I think this kind of case, even though it is one in which Ted fails to believe something he believes he has no reason to believe, provides us with at least some reason to think that B4 may not be true. Certainly, it gives us reason to think B4 has less prima facie plausibility than B3.

Now consider the third example. Ted believes that he does not have reason to acquire a vast number of true beliefs given that he lacks reliable control over acquiring them. But now imagine that he awakes one morning to find that he has acquired a vast number of true beliefs. He still believes he has no normative reason to believe all of these propositions. For Ted believes that the mere fact one has acquired something is not sufficient to establish that he acquired them on the basis of a reliable control mechanism, and thus not sufficient to establish that one had the ability—and so normative reason—to acquire them (see Austin Reference Austin1961: 153–80; Kane Reference Kane1996: 52–54). To illustrate: acquiring a million dollars by winning a lottery is not something that Ted believes he has reliable control over doing (and thus not something he believes he has normative reason to do). And that belief is not altered by his actually winning a million dollars on the lottery. Similarly then, the mere fact he has acquired, by pure luck, a vast number of true beliefs, including some of incredible complexity, does not commit him to concluding that he had the ability to do so. Thus, just as he would not conclude, from his winning the lottery, that he had normative reason to win the lottery, so too he does not conclude from his acquiring a vast number of true beliefs that he had normative reason to acquire them all.

If what I have said is correct, then Ted is now in a situation in which he believes many things that he believes at the same time he has no normative reason to believe. And there seems no non-question-begging grounds for denying he truly believes them, for the basis upon which he disbelieves he has reason to believe them has to do with what he believes himself capable of doing, and so does not indicate any lack of confidence in, or understanding of, the propositions in question.

One might object that though Ted may continue to believe that he had no normative reason to acquire all these true beliefs, the fact he now has acquired them will nevertheless mean he believes he has normative reason to continue believing them. As such, he will not truly be in the situation of believing propositions, he believes himself to have no reason to believe, and so this is not really a counterexample to B4.

However, this seems false too. For just as it is beyond Ted's practical ability to acquire all these beliefs, he also believes—equally plausibly—that it is beyond his practical ability to maintain them all. Accordingly, Ted believes he does not have normative reason to continue to believe them all (which is consistent, note, with believing he does not have normative reason to divest himself of them either).

Now consider a variation of the fourth example. Ted believes that he has no normative reason to acquire trivial true beliefs, such as beliefs about the color and length of hairs on David Cameron's left arm. But now imagine that Ted starts to think that David Cameron has eight thousand hairs on his left arm, perhaps because he overheard Cameron's doctor say so. What non-question-begging reason is there for thinking that he does not qualify as really believing it? Yes, he does not believe there is any normative reason to believe it (for it remains as trivial as ever). But how does that give us reason to think he does not believe it given that the basis upon which he believes he has no reason to believe it is the proposition's triviality. He is not evincing any lack of understanding, for the proposition in question is trivial. And similarly, I do not see any non-question-begging reason to think that Ted lacks confidence in its truth. The burden of proof, it seems to me, is squarely on the B4 defender to explain why, exactly, Ted cannot be said to believe the proposition in question given that it seems entirely believable that such propositions are trivial, and entirely believable that we do not have reason to believe trivial propositions, and entirely believable that someone might nevertheless acquire such a belief and still be able to recognize it as trivial.

Finally, consider a case involving examples of the fifth kind. We know from the previous section that Ted believes that (typically anyway) if a couple is in a committed loving relationship with one another, then both parties have moral and instrumental reason to be certain that their love for the other will last a lifetime and no reason to think otherwise. The fact that the statistics show that there is a significant chance that their love will not last a lifetime is a fact that, in their case, is being prevented from generating any epistemic reason for them to be less than certain about the robustness of their love. Reason may favor others thinking there is a significant chance the loving couple's love will not last the distance, but Reason does not favor the loving couple themselves thinking this. But now imagine that Ted is in a loving relationship with Eva. As such, he believes he has moral and instrumental reason to believe his love for her will last a lifetime and believes as well that he lacks any normative reason to think otherwise. Yet—and to his shame—he has nevertheless formed the belief that his love for Eva stands a significant chance of not lasting a lifetime, and he has formed this belief on the basis of the statistical evidence. Well, now he believes a proposition that he at the same time believes he has no normative reason to believe. And as with the other cases, there seems no non-question-begging grounds for denying that he confidently and with understanding believes the proposition in question, for he has acquired it on the basis of the evidence. I mean I believe that Ted's love for Eva stands a significant chance of not lasting a lifetime, and I believe it on the basis of the statistical evidence, which is the same basis upon which Ted believes it. Now there is surely no ground for claiming that I do not truly believe this. Ted believes it, too, and he believes it on the same grounds that I do. It is just that he believes as well that he has no reason to believe it. But note that the basis upon which he believes that he has no normative reason to believe it is his belief that loving relationships of the kind that he stands in to Eva are such as to generate moral and instrumental reasons that block evidence from generating any epistemic reasons in such contexts. So, his belief that he has no normative reason to believe what he believes does not, in his case, betoken any lack of understanding or confidence on Ted's part in the truth of the content of his belief. Once more then, we seem to have a way in which Ted has acquired a belief while at the same time believing that he positively has no normative reason to believe it.

Apart from my second example, all of these counterexamples make the same point. Ted believes the truth of a proposition does not always and everywhere generate a normative reason for him to believe it. And in the cases I offer, Ted believes the propositions in question fail to generate a normative reason for him to believe them for reasons unrelated to their truth. Thus, when or if he acquires the belief in the propositions in question, there seems no ground for denying that he understands their content and is confident in their truth. Thus, in each case bar the second, there seems no non-question-begging ground for denying he truly believes them.

Note that my counterexamples differ from ones that others have offered against B4 but that might be thought to bear a surface similarity to some of mine.Footnote ¹ For example, my first counterexample appeals to the proposition that normative nihilism is true. Others have also appealed to this proposition, notably Jonas Olson (Reference Olson2014: 171–72; see also Lillehammer and Möller Reference Lillehammer and Möller2015: 457–58). But their cases focus on how a person might acquire such a belief, rather than on the background beliefs of the putative believer. As a result, their cases seem either to beg the question or else to rest on an appeal to the fact that a person can believe a proposition without believing that there is any normative reason to believe it (which does not challenge B4, for as Streumer points out in response to such challenges, believing a proposition without believing there is a reason to believe it is distinct from believing a proposition while believing there is positively no reason to believe it) (Reference Streumer and Machuca2017b: 79–80). In my case, by contrast, the emphasis is not on how Ted acquires the belief in normative nihilism, but rather on the background beliefs he already possesses—in particular his belief in normative holism—and whose believability does not seem to be in doubt.

Alexander Hyun and Eric Sampson (Reference Hyun and Sampson2014) attempt a counterexample to B4 that bears passing resemblance to some of mine. They describe a case in which a person, Susan, becomes convinced on philosophical grounds that she has no reason to believe her senses are reliable (Reference Hyun and Sampson2014: 634–35). Nevertheless, Susan finds that she still believes that her senses are reliable. However, in Susan's case but not Ted's there seems no non-question-begging reason to think that Susan really does believe her senses are reliable at the same time as believing there is no reason to think this. Far from it, for Susan thinks there is reason to believe that her senses are unreliable. If she nevertheless finds that she still believes that her senses are reliable, then all she has really discovered is that her belief in the reliability of her senses is unresponsive to what she believes she has reason to believe. One might think that this is nevertheless sufficient for this to be a counterexample to B4, given that Susan believes—albeit compulsively—that her senses are reliable at the same time as believing she has no reason to think they are. But, as Streumer points out, that would be a pyrrhic victory given that all it would show is that normative nihilism can, at best, be believed compulsively (Reference Streumer and Machuca2017b: 78).

None of this applies to my kinds of case. For Ted's background beliefs are about the nature of normative reasons and their relation to truth. As such, when Ted acquires a belief in the truth of a proposition of a kind that he already believes is of a sort that does not—or does not in this context—generate a normative reason for him to believe it, there seems to be no non-question-begging reason to think either that he does not believe it, or to think that he believes it compulsively.

Objections

One might object that normative holism is implausible. But in addition to seeming false on its face, this objection is beside the point, for as I have been at pains to point out, my case requires not that normative holism be true, or even plausible, but that it be believable. It is not, after all, the truth—or plausibility—of normative error theory that is at question, but its believability. And it is hard to deny that normative holism is believable.

Nevertheless, I suppose some might make such a claim. But on what non-question-begging basis could one make it? There seems nothing prima facie unbelievable about it. And it would clearly be question-begging to insist it is unbelievable on no better basis than the putative truth of B4, for it is B4's truth that is at issue.

One might object that epistemic normativity is different from other kinds, and that while one can believe that, say, the moral normativity of a feature might vary from context to context—the fact an act causes suffering may operate with a negative normative valence in one context, but positive in another, and neither in yet another—facts and evidence are somehow different and have an invariable epistemic valence.

I agree that epistemic normative reasons are distinct from other kinds, but what marks them out as distinct does not provide any grounds for the above objection. For what is distinctive about epistemic reasons is that they are favoring relations in which relatum b has to be a belief, and relatum c has to related to the truth or likely truth of the belief in question. Yet clearly nothing in that distinction gives us any ground for thinking that facts have invariable epistemic normative valences. If a fact or consideration indicative of a fact generates a normative reason to believe something, and to believe it due to its being true or likely true, then the favoring relation in question qualifies as an epistemic one, but it does not follow from this that facts and considerations indicative of them invariably generate such favorings. So, nothing in what is distinctive about epistemic normative reasons provides grounds for thinking that holism about normativity will be false in respect of them.

One might object that the fact a belief is true is a reason to believe it—or the fact that the evidence indicates its truth is a reason to believe it. Ted is therefore simply confused to think that evidence can indicate a proposition's true and yet at the same time not provide any reason to believe the proposition in question.

But facts are not normative reasons. Normative reasons are favoring relations, or at least that is what Ted—in common with many others—believes about them. As such, and as already mentioned above, although we may sometimes say things such as ‘the fact it is true is a reason to believe it’, this can plausibly be taken to be elliptical for ‘the fact it is true is what gives rise to there being a favoring of believing it’ (see Bedke Reference Bedke2010: 48). It should also be mentioned again that normative error theory is not the theory that there are no facts. It is the theory that there are no normative reasons. That is, the normative nihilist does not deny the existence of any of the elements within the relation constitutive of a normative reason; rather they just deny the existence of the favoring relation itself. And it is the believability of normative error theory, so construed, that is at issue.

One might object that it is not possible for Ted to arrive at his beliefs on the same basis as others without thereby believing that the considerations in question provide him with reason to acquire the said beliefs.

In reply, such a claim is question begging in this context. As virtually everyone will admit, one can base a belief on a consideration without having to make a normative judgment that there is reason to believe the proposition in question. Children and nonhuman animals form beliefs on the basis of perception ‘although they presumably lack the relevant normative thoughts’ (Olson Reference Olson2014: 171). Clearly then, one can believe something without having to believe that there is a reason to believe it (and so believing something on the basis of the same consideration as someone else does not essentially involve believing there is a normative reason to believe it). Granted, this leaves open whether a person can base a belief on a consideration at the same time as believing that the consideration in question positively does not generate a normative reason for them to hold it (Streumer Reference Streumer and Machuca2017b: 145). But that is precisely what my cases are designed to settle.

One might object that if Ted recognizes that a consideration might provide others with epistemic reason to believe a proposition, then he cannot at the same time fail to believe the proposition himself. For as I have already noted above, an epistemic reason is a reason to believe a proposition due to the truth or indicated truth of the proposition in question. Thus in acknowledging that someone—anyone—might be provided with an epistemic reason to believe that there are eight thousand hairs on David Cameron's left arm, or that there is a significant chance his love for Eva will not last a lifetime, Ted must eo ipso believe these propositions himself.

True, but it is beside the point, as it does nothing to undermine my case against B4. For in my examples Ted does believe the propositions in question (with the exception of the second example, but that counterexample operates differently). That is, he does believe that there is a significant chance his love for Eva will not last a lifetime and he does believe that there are eight thousand golden hairs on David Cameron's left arm, and so on. The important point is that he believes them while believing he has no normative reason to believe them. Nothing the objector has said calls this into question.

Perhaps someone might object that my cases are not true counterexamples to B4 because Ted still believes that there are, or could be, normative reasons to believe the propositions in question (for Ted is not a normative nihilist). But this objection misses the point. For if Ted can, on the basis of his normative holism, believe a proposition at the same time as believing that there is no normative reason to believe it, then what grounds do we have for supposing that he would be unable to pull off this feat if the basis upon which he believed it was normative error theory instead? None, I think.