1. Preemptive reasons and epistemic authorities

In order to understand the PRV, we need to get a handle on two things. First, what is a preemptive reason, and how does it differ from evidence? Second, how should the notion of an epistemic authority be understood? Let us start with the former.

According to the PRV, if one acquires evidence for p, then in the absence of any preemptive reasons, that evidence must always be added to, and thus weighed against, all of the other evidence that one has. For instance, if I have some evidence that the plant we are looking at is poison oak, and if Sally, a student who is learning about plant identification, tells me that it is actually poison ivy, then I must decide what to believe by considering my total body of evidence, i.e., the evidence provided by Sally’s testimony and whatever other evidence I might happen to have.

However, as Linda Zagzebski puts it, when one acquires a preemptive reason for believing that p, that reason

More specifically, Arnon Keren says that receiving testimony from an epistemic authority provides you with a preemptive reason in the sense that the authority

And Jan Constantin and Thomas Grundmann make this same point by suggesting that when an epistemic authority provides a subject, S, with a preemptive reason to believe that p, all of S’s other relevant evidence, E, is preempted in the following sense:

So if I have some evidence that the plant we are looking at is poison oak, and if Beth, a known epistemic authority on all things botany, tells me that it is actually poison ivy, then her testimony preempts, or screens off, all of my other relevant evidence such that, rationally speaking, I should not rely on it when determining what to believe. Instead, I should just adopt whatever attitude Beth has.Footnote ²

This, then, is how the difference between evidence and preemptive reasons should be understood. Let us now turn to the question of what makes someone an epistemic authority.

According to the PRV, epistemic authorities are those you justifiably take to be experts about the domain in question, where an expert is someone who, in addition to possessing loads of evidence about their domain of expertise, has methods that allow them to use this evidence to form true beliefs and accurate credences in a highly reliable way (Constantin and Grundmann (Reference Constantin and Grundmann2020, pg. 4114) and Goldman (2001, pp. 91–92)).Footnote ³

In order to become an expert, then, you must cross two thresholds. First, you must cross the evidence threshold in the sense that, at some point in your education, you acquire enough information that your total body of evidence counts as being “substantial.” Second, you must cross the reliability threshold in the sense that, at some point in your education, you are able to use your evidence to form attitudes in a way that counts as being “highly reliable.”Footnote ⁴

Moreover, while someone can be an expert regardless of their relationship to you, one can only be an epistemic authority relative to someone else – i.e., someone is an epistemic authority for you only if you have good reason to think that they are your epistemic superior.Footnote ⁵ That is, when it comes to propositions within their domain of expertise, an epistemic authority is someone you justifiably take to be an expert who is much more likely than you are to form true beliefs and accurate credences about the proposition in question. For instance, if Beth and Ben are both expert botanists and thus epistemic peers with respect to plants, then when it comes to plant identification, they are both epistemic authorities for me but not for one another.

Finally, notice that an epistemic authority is someone whom you justifiably believe to be both an expert and your epistemic superior with respect to the domain in question. Thus, if you have no clue that Beth is an expert botanist and is thus much more likely than you are to correctly identify the plant in question, then she is not an epistemic authority for you. As a result, her testimony can only provide you with evidence.Footnote ⁶

Consider, then,

Two clarificatory remarks before moving on. First, for now, I am going to focus on Constantin and Grundmann’s (Reference Constantin and Grundmann2020) notion of an epistemic authority. This is because I take their account to be the most plausible on offer. But in Section 3, I will consider some alternative views of what it takes to be an epistemic authority (e.g., Zagzebski’s (Reference Zagzebski2012) and Keren’s (Reference Keren2014)) and explore how it might help proponents of the PRV avoid the problem that I am going to raise below.

Second, in order to have sufficiently good reason to think that someone is an epistemic authority, you must justifiably believe that they have considered all of the evidence that you possess about the relevant proposition.Footnote ⁷ For instance, if you know that the plant in question was cleverly painted to look like poison ivy when it is really poison oak, and if you also know that Beth the botanist is not privy to this evidence, then even if she tells you that the plant is poison ivy, her testimony does not give you a preemptive reason to believe what she says. And this is because you should no longer think that Beth is more likely than you are to identify the plant correctly.

Now that we have seen how the PRV should be understood, I am going to develop a new problem for the view.Footnote ⁸

2. The Problem of Opinionated Apprentices

Even if you know that someone is not an expert (and thus not an authority for you), you can still be well aware that they possess a good amount of training in their field. Consequently, even if someone is just shy of becoming an expert, you can still justifiably take them to be your epistemic superior because they not only have loads of evidence about the domain in question (albeit slightly less evidence than the experts have) but also because they are able to use this evidence to form beliefs in a very reliable way (albeit in a way that is slightly less reliable than the way in which the experts form their beliefs).

For instance, suppose that Fiona goes to the hospital because her foot hurts. And suppose that she meets with her physician, Phyllis, and one of Phyllis’s student interns, Irene. Insofar as Fiona knows very little about medicine, Phyllis is clearly an epistemic authority for Fiona with respect to her health. And while Irene is not an epistemic authority for Fiona (namely because she has not yet risen to the level of a medical expert), she is still Fiona’s epistemic superior who can provide her with loads of useful medical advice.

More specifically, recall that by their own lights, proponents of the PRV maintain that expertise is a threshold notion such that, in order to become an expert, you must cross the evidence threshold and the reliability threshold. Thus, while Irene is a very talented medical student, she is not yet an expert for the following two reasons. First, let us stipulate that while Irene has a lot more information about medicine than people like Fiona, she has not quite crossed the evidence threshold because her total body of evidence falls just short of being “substantial.” Second, let us also stipulate that while Irene is better than most at using her evidence to form true beliefs and accurate credences, she has not quite crossed the reliability threshold, since, at this point in her training, her diagnostic skills fall just short of being “highly reliable.”

But even so, Irene still has lots of medical training, and with a bit more schooling, she too will be an expert doctor. In contrast to someone who is an epistemic authority, then, let us say that people like Irene are epistemic apprentices in the following sense

There are four things to consider here. First, just as expertise is a threshold notion, so too is the notion of being an epistemic apprentice. This is because in order for one to be “close to becoming” an expert, one must cross two thresholds. First, one must cross an evidence threshold such that one has quite a bit of evidence about the domain in question, i.e., they will have way more evidence than the total novice, but a bit less evidence than the experts. Second, one must cross a reliability threshold such that one can use this evidence to form beliefs about the relevant propositions in a reliable way, i.e., they will be way more reliable than the complete layman, but not quite as reliable as the experts. Thus, in the course of one’s training, one will start out as a novice. Then, when one crosses the first set of evidence and reliability thresholds, they will be an epistemic apprentice. And finally, when one crosses the second set of evidence and reliability thresholds, they will be an expert who is capable of serving as an epistemic authority.

Second, like an expert and unlike an epistemic authority, being an epistemic apprentice does not entail that you are epistemically superior to a particular individual. This is because, depending on your own credentials, you can be an epistemic apprentice while also being someone’s epistemic superior, peer, or inferior.

Third, by their own lights, proponents of the PRV maintain that receiving testimony from an epistemic apprentice does not provide you with a preemptive reason to believe what you are told. Rather, because epistemic authorities are the only ones that can provide you with preemptive reasons, receiving testimony from an epistemic apprentice just provides you with regular evidence.

Fourth, epistemic apprentices are not hard to find. Advanced graduate students, professional interns, and other well-trained individuals who are in the early stages of their careers fit this bill. And while epistemic apprentices make up an important but often overlooked group of epistemic agents, I am now going to argue that adequately accounting for the epistemic significance of their testimony leads to a new problem for the PRV.

Consider these two cases,

A few clarifications. First, let us stipulate that Fiona knows that Irene is an epistemic apprentice, that Phyllis is an epistemic authority, and that both are her epistemic superiors with respect to medicine.

Second, because she is not an authority for Fiona, the PRV says that in CONSISTENT DIAGNOSES, the testimony that Fiona receives from Irene provides her with regular evidence in favor of her having Condition A.

Third, because Phyllis is an epistemic authority for Fiona with respect to the diagnosis in question, the PRV says that Phyllis’ testimony provides Fiona with a preemptive reason for believing that she has Condition A in both CONSISTENT DIAGNOSES and SINGLE DIAGNOSIS.

Fourth, let us stipulate the following two things about Phyllis: she knows Irene is an epistemic apprentice and thus her epistemic inferior with respect to Fiona’s ailment, and in CONSISTENT DIAGNOSES, Phyllis hears Irene testify to Fiona that she has Condition A before giving her own diagnosis.

Here is the key question: rationally speaking, should Fiona be at least slightly more confident that she has Condition A in CONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS? And here is what strikes me as the intuitively correct answer: yes. In SINGLE DIAGNOSIS, Phyllis’ testimony does justify Fiona in being very confident that she has Condition A. However, in CONSISTENT DIAGNOSES, both testifiers examined the same evidence and independently concluded that Fiona has Condition A. The fact that there is a consensus here makes it rational for Fiona to be a little bit more confident that this is indeed the case. (Notice: I am not claiming which credences Fiona should actually have. Rather, I am only making the weaker claim that she should be at least a tiny bit more confident that she has Condition A in CONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS.)Footnote ⁹

But the PRV cannot account for this. For in CONSISTENT DIAGNOSES, Fiona first receives testimony from Irene that she has Condition A. And because she is not an epistemic authority, Irene’s testimony only provides Fiona with evidence that this is the case. However, because Phyllis is an epistemic authority, her testimony provides Fiona with a preemptive reason for believing that she has Condition A. Thus, because Phyllis’ testimony preempts the evidence that Fiona acquired from Irene, in CONSISTENT DIAGNOSES, the PRV says that the only non-preempted reason that Fiona has with respect to her medical condition is Phyllis’ testimony that she has Condition A.

So far, this may not be a problem for the PRV.Footnote ¹⁰ But now consider SINGLE DIAGNOSIS, in which Fiona only hears from Phyllis. In this case, the PRV also says that the only non-preempted reason that Fiona has with respect to her medical condition is Phyllis’ testimony that she has Condition A, since this is the only testimony that Fiona received in general.

The key is that in both cases the PRV says the same thing, namely that the only non-preempted reason Fiona has with respect to her medical condition is Phyllis’ testimony that she has condition A; that is, because the testimony from Irene drops out of the picture completely, Phyllis’ say-so is all that Fiona has to go on in both cases. Thus, the PRV gives symmetric verdicts; it says that, however, confident Fiona should be that she has Condition A in SINGLE DIAGNOSIS, she should be equally confident that she has Condition A in CONSISTENT DIAGNOSES.

The PRV thus faces what I will call The Problem of Opinionated Apprentices: when one receives testimony from both authorities and apprentices, the PRV cannot account for situations in which the apprentices’ testimony should rationally impact how confident one should be about the proposition in question.

3. First objection: apprentices are epistemic authorities too

Perhaps one could insist that in CONSISTENT DIAGNOSES, Irene should really be considered an epistemic authority who also provides Fiona with a preemptive reason for belief. This is because our characterization of an epistemic authority should be revised to include both experts and apprentices as follows:

The upshot is that when it comes to CONSISTENT DIAGNOSES and SINGLE DIAGNOSIS, Fiona should be more confident that she has Condition A in the former case than in the latter. This is because in CONSISTENT DIAGNOSES, she has two preemptive reasons that support this conclusion, whereas in SINGLE DIAGNOSIS, she only has one. Thus, insofar as preemptive reasons can be aggregated to justify a higher degree of confidence in this way, the PRV can secure the verdict right after all.Footnote ¹¹

I agree that by relaxing the requirements on becoming an epistemic authority, proponents of the PRV can capture the right verdicts in CONSISTENT DIAGNOSES and SINGLE DIAGNOSIS. However, to see why the same underlying problem still remains, consider:

The important thing to notice here is that because Fiona knows that she and Irene are both epistemic apprentices and thus epistemic peers, Irene cannot be an epistemic authority* for Fiona. This is because in order for Irene to serve as an epistemic authority*, Fiona must have good reason to think that Irene is her epistemic superior. Thus, even if Irene is an epistemic apprentice, she cannot be an epistemic authority* who provides Fiona with preemptive reasons for belief.Footnote ¹²

Once again, here is the key question: rationally speaking, should Fiona be at least slightly more confident that she has Condition A in CONSISTENT DIAGNOSES* than in SINGLE DIAGNOSIS? And for the same reasons given above, here is what strikes me as the correct answer: yes. The problem is that the PRV cannot account for this. Because Irene is not an epistemic authority* for Fiona, once Fiona hears from Phyllis, all of her other evidence drops out of the picture; that is, the evidence provided by Irene’s testimony, as well as any other evidence that Fiona gathered herself, becomes rationally unusable. In both CONSISTENT DIAGNOSES* and SINGLE DIAGNOSIS, then, the only thing that Fiona has to go on is Phyllis’s say-so. Thus, because the PRV says that Fiona should be equally confident that she has Condition A in both cases, the Problem of Opinionated Apprentices remains.

Proponents of the PRV might respond by saying that the problem here is not with the PRV per se, but with the notion of Epistemic Authority* in particular. Instead of endorsing Epistemic Authority*, perhaps the solution is to follow Zagzebski’s (Reference Zagzebski2012) influential account of what it takes to be an epistemic authority as follows:

The key difference is that being an expert or an apprentice is no longer necessary for

providing one with a preemptive reason for belief. All that is needed is that you are justifiably taken to be someone’s epistemic superior.

But even if this is how we should understand what it takes to be an epistemic authority, the worry still remains. To see why, recall that Fiona knows that Irene is her epistemic peer. Thus, Irene cannot be an epistemic authority** for Fiona, since Fiona does not have sufficiently good reason to believe that Irene is her epistemic superior. Thus, in CONSISTENT DIAGNOSES* and SINGLE DIAGNOSIS, once Fiona hears from Phyllis, all of her other relevant evidence drops out of the picture. Consequently, because the only thing that Fiona has to go on is Phyllis’s say-so, the PRV says that Fiona should be equally confident that she has Condition A in CONSISTENT DIAGNOSES* and SINGLE DIAGNOSIS. But intuitively, Fiona should be more confident that she has Condition A in the former case than in the latter.

The upshot here is that the Problem of Opinionated Apprentices cannot be avoided by tinkering with how exactly epistemic authority should be understood.Footnote ¹⁴

4. Second objection: the asymmetric Preemptive Reasons View

Katherine Dormandy (Reference Dormandy2018) has highlighted some considerations that help motivate what I will call the Asymmetric PRV (APRV).Footnote ¹⁵ According to the APRV, epistemic authorities are the only ones that can provide you with a preemptive reason for belief. However, if an epistemic authority tells you that p, then while their testimony does preempt all of your evidence that supports not-p, it does not preempt any of your evidence that supports p. Otherwise put, the authority’s testimony only preempts your evidence when that evidence does not support what the authority said.

The APRV secures the right verdicts in SINGLE DIAGNOSIS and CONSISTENT DIAGNOSES. It says that in SINGLE DIAGNOSIS, Fiona should be confident that she has Condition A because she can rely on the preemptive reason that she acquired from Phyllis’s testimony. And in CONSISTENT DIAGNOSES, Fiona should be even more confident that she has Condition A, since, in addition to the preemptive reason she acquired from Phyllis, she can also rely on the evidence that she acquired from the testimony of Irene.Footnote ¹⁶

But even if preemptive reasons only function asymmetrically, the underlying problem still remains. To begin to see why, consider

In this case, it is because Phyllis justifiably views Irene as her epistemic inferior that she sticks to her guns and tells Fiona that she has Condition A, i.e., even after hearing what Irene thinks, in INCONSISTENT DIAGNOSES, Phyllis continues to endorse her original assessment.

Now, contrast SINGLE DIAGNOSIS, in which Fiona just hears from Phyllis, with INCONSISTENT DIAGNOSES, in which Phyllis tells Fiona that she has Condition A, but Irene tells Fiona that she has Condition B. Here is the key question: should Fiona be at least slightly less confident that she has Condition A in INCONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS? And here is what I take to be the intuitively correct answer: Yes.Footnote ¹⁷ For in SINGLE DIAGNOSIS, the fact that Fiona hears from a single authority justifies her in being highly confident that she has Condition A. But in INCONSISTENT DIAGNOSES, there is a known apprentice who, after evaluating the same evidence, disagrees with Phyllis’s verdict. And the fact that a well-trained and fairly reliable medical intern arrived at a different verdict suggests that Fiona’s confidence that she has Condition A should be lower than it is when there is no disagreement whatsoever. (Note again that I am not claiming which credences Fiona should actually have. Rather, I am only making the weaker claim that she should be at least a tiny bit less confident that she has Condition A INCONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS.)

But the APRV cannot account for this. For according to the APRV, in SINGLE DIAGNOSIS, the only non-preempted reason Fiona can rationally rely on is Phyllis’s testimony that she has Condition A. Moreover, in INCONSISTENT DIAGNOSES, Phyllis’s testimony preempts the evidence that Fiona acquired from Irene, since Irene’s testimony was incompatible with what Phyllis said. In INCONSISTENT DIAGNOSES, then, the only non-preempted reason that Fiona can rationally rely on is Phyllis’s testimony that she has Condition A. Consequently, the APRV says that Fiona should be equally confident that she has Condition A in both cases. Thus, since Irene’s testimony should not be completely disregarded in this way, the Problem of Opinionated Apprentices remains.Footnote ¹⁸

5. Third objection: downgrading epistemic authorities

There are times when someone you initially took to be an authority should no longer be regarded as such, at least for the time being. For instance, suppose Phyllis tells Fiona that in order to cure her foot pain, she should take 40,000 doses of Tylenol every hour for the next week. As Thomas Grundmann (Reference Grundmann2021) has argued, while Phyllis can usually serve as an authority for Fiona with respect to her foot condition, in this case, Phyllis’s testimony does not provide Fiona with a preemptive reason for taking a gargantuan dose of pain relievers. For insofar as Fiona knows that this is an outrageous thing to say, she has good reason to think that Phyllis is unable to give a reliable diagnosis at this time. Whether the best explanation of what is going on here is that Phyllis is massively sleep deprived, intoxicated, or just in the mood to bullshit, there is good reason to think that her ability to prescribe the right treatment is wildly off base. And if Fiona should not regard Phyllis as being remotely reliable at the present moment, then for the time being, she should not take Phyllis to be an expert with respect to her foot condition – recall that you should only take someone to be an expert if you should also take them to be able to form true beliefs and accurate credences in a way that surpasses the reliability threshold for expertise. Thus, because Fiona should only regard Phyllis as an authority if she should also regard her as an expert, and because Fiona should not take Phyllis to be an expert at the present moment, she should not view her as being an epistemic authority either. Consequently, Phyllis’s testimony does not provide Fiona with a preemptive reason to down a mega dose of Tylenol.

Just as Fiona can downgrade Phyllis’s status as an authority when it comes to outrageous drug prescriptions, perhaps proponents of the APRV can say something similar about INCONSISTENT DIAGNOSES. For recall that in this case, Phyllis disagrees with Irene about whether Fiona has Condition A. Perhaps when Fiona realizes this, she should think that Phyllis’s reliability at diagnosing foot pain has plummeted below the threshold for medical expertise. And once Fiona realizes that Phyllis has temporarily lost her status as an expert, she should also downgrade Phyllis’s status as an epistemic authority. Thus, because Phyllis is currently unable to serve as an epistemic authority for Fiona, her testimony does not provide Fiona with a preemptive reason that cancels out the evidence that was provided by Irene’s testimony.

The purported upshot is that once Fiona downgrades Phyllis’ testimony in this way, the APRV can secure the right result. For in SINGLE DIAGNOSIS, Fiona should be very confident that she has Condition A, since Phyllis’s testimony provides her with a preemptive reason for thinking that this is the case. And in INCONSISTENT DIAGNOSES, Fiona should be less confident that she has Condition A, since she must weigh the evidence provided by Phyllis’s testimony in favor of Condition A against the evidence provided by Irene’s conflicting testimony in favor of Condition B.

The problem with this response, though, is that, unlike cases involving outrageous recommendations, the same explanation as to why Fiona can downgrade Phyllis’s status as an authority is unavailable. To see why, notice the mechanism by which Fiona is able to legitimately downgrade Phyllis’ status as an expert when it comes to the Tylenol: Fiona can stop treating Phyllis as an expert, and thus as an authority, because she has evidence that Phyllis is currently unable to reliably prescribe the right treatment plan, i.e., given that Phyllis said something totally outrageous, Fiona should think that Phyllis’ current reliability at giving the right prescriptions has dropped below the threshold for expertise.

But the same mechanism for downgrading Phyllis’ status as an expert is not available in INCONSISTENT DIAGNOSES. This is because Fiona does not possess evidence that justifies her in thinking that Phyllis is positively unreliable when it comes to diagnosing her foot condition, just that she is slightly less reliable than was initially thought. And thinking that Phyllis is slightly less reliable than she initially thought does not mean that Fiona should think that Phyllis’s reliability has dropped so low that it now fails to meet the threshold for expertise. Rather, Fiona should still view Phyllis as being an expert, albeit one who is now a bit more likely to have made a mistake.

Of course, if Phyllis and Irene were epistemic peers, then upon receiving their conflicting verdicts, perhaps Fiona should think that there is only a 50% chance that Phyllis is the one who got it right. And surely if Fiona should only think there is a 50% chance that Phyllis got it right, then she should not view her as being reliable enough to meet the threshold for expertise on this occasion. However, Phyllis and Irene are not epistemic peers. As Fiona is well aware, Phyllis is Irene’s epistemic superior. Thus, Fiona should not think that whenever the two disagree, there is only a 50% chance that Phyllis diagnosed the foot condition correctly; rather, she should think that the odds of Phyllis being right are still quite high, just not as high as they were before. And because expertise is a threshold notion such that some experts are more reliable than others, Fiona should acknowledge that when Phyllis and Irene disagree, Phyllis still meets the reliability threshold for expertise, even if her reliability has dipped a bit.

But if Fiona should still view Phyllis as an expert despite what Irene says, then she has no good reason to downgrade Phyllis’s status as an epistemic authority. Thus, according to the APRV, Phyllis’s testimony in INCONSISTENT DIAGNOSES really does provide Fiona with a preemptive reason for belief.

The upshot so far is that while there is a mechanism that allows Fiona to downgrade Phyllis’s status as an authority whenever Phyllis says something truly outrageous that same mechanism cannot explain why Fiona can downgrade Phyllis’s status simply because Irene offers a conflicting opinion. Before moving on, I want to consider one more explanation that proponents of the APRV might offer for why Fiona really should stop treating Phyllis as an authority in INCONSISTENT DIAGNOSES.

Suppose that in order for Fiona to rationally downgrade Phyllis’s status, she does not need to have a good reason to think that Phyllis’s reliability is impaired in a serious way, just that there is a possibility that Phyllis got it wrong on this occasion. When it comes to INCONSISTENT DIAGNOSES, then, Irene’s conflicting verdict should make Fiona think that it is at least possible that Phyllis is the one who made a mistake; that is, while Fiona should not think that the likelihood of Phyllis being wrong is all that high, she should still recognize that there is at least a chance that Phyllis screwed up. Thus, if you can rationally downgrade someone as an authority whenever you justifiably take there to be even a small chance that they made a mistake, then with respect to INCONSISTENT DIAGNOSES, proponents of the APRV could explain why Fiona can rationally downgrade Phyllis’ status this time around.

The problem with this response, though, is that it makes downgrading someone as an authority too easy. To see why, let us stipulate that while Fiona knows that Phyllis is a great doctor, she also knows that Phyllis is not completely infallible when it comes to diagnosing medical conditions. But if Fiona knows that Phyllis is not an infallible physician, then she always has a good reason for thinking that there is at least some possibility that Phyllis has botched a given diagnosis. Thus, Fiona would always have a good reason to downgrade Phyllis as an authority, which is not something that proponents of the APRV should accept. To the contrary, they should think that in many cases (e.g., SINGLE DIAGNOSIS), Fiona really should treat Phyllis as an authority, even if she knows that Phyllis’ diagnostic abilities are not perfectly reliable. Thus, by their own lights, proponents of the APRV should think that it takes a lot more for Fiona to downgrade Phyllis’s status than merely recognizing that she is fallible.

The main takeaway here is that in INCONSISTENT DIAGNOSES, the burden is on proponents of the APRV to explain how Fiona can rationally downgrade Phyllis’s status simply because one apprentice thinks otherwise. And as we have just seen, the explanation cannot be that Phyllis has said something so outrageous that we should stop regarding her as an expert for the time being, nor can it make the process of downgrading authorities too easy. Until such an explanation is forthcoming, the Problem of Opinionated Apprentices remains.

6. Authorities as advisors

So far, I have argued that the Problem of Opinionated Apprentices poses a new worry for the PRV and the APRV. But how serious of a worry is this, really? Perhaps one could acknowledge that while the PRV and the APRV are saddled with a new problem here, the problem is not unique to these two views; that is, perhaps one could argue that because the Problem of Opinionated Apprentices poses a general concern for any view of authoritative testimony whatsoever, it does not amount to a unique strike against the PRV and the APRV in particular.

I am now going to argue that the Problem of Opinionated Apprentices does not cause trouble for every view of authoritative testimony, because it can be adequately handled by the AAV.

While there are various versions of the AAV that differ in their details, they all maintain the following: testimony from an authority does not provide you with a preemptive reason, but with a piece of evidence that must be weighed with all of the other evidence that you possess (Dougherty (Reference Dougherty2014), Jäger (Reference Jäger2015), Lackey (Reference Lackey, Benton, Hawthorne and Rabinowitz2018), and Bokros (Reference Bokros2021)). And because the AAV treats all testimony as evidence, it can explain the difference between authoritative testimony and testimony from everyone else in terms of the quality of evidence that each speaker provides. More specifically, relying on the testimony of an epistemic authority is better than relying on the testimony of an epistemic apprentice, because the former provides you with better evidence for the truth of the proposition in question. And for the same reason, relying on the testimony of an epistemic apprentice is better than relying on the testimony of a layman.

For our purposes, the key insight here is that because everybody’s testimony provides you with the same kind of reason for belief, the AAV does not say that it is necessarily the case that the evidence provided by an epistemic apprentice is going to drop out of the picture whenever an authority weighs in on the issue at hand. And it is for this reason that the AAV has the resources to avoid the Problem of Opinionated Apprentices.

For instance, the AAV secures the correct result that Fiona should be more confident that she has Condition A in CONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS.Footnote ¹⁹ This is because in CONSISTENT DIAGNOSES, Fiona has two pieces of evidence that support this hypothesis, i.e., one great piece of evidence that came from Phyllis’s testimony, and another pretty good piece of evidence that came from the testimony of Irene. However, in SINGLE DIAGNOSIS, Fiona just has the one great piece of evidence that came from Phyllis’s say-so. Thus, insofar as Fiona should increase her confidence that she has Condition A when she receives the additional evidence from Irene, the AAV says that Fiona should be more confident that she has Condition A in CONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS, since this is what her total body of evidence supports.

And for the same reason, the AAV secures the correct result that Fiona should be less confident that she has Condition A in INCONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS. This is because in SINGLE DIAGNOSIS, she has one great piece of evidence from Phyllis’s testimony that she has Condition A, whereas in INCONSISTENT DIAGNOSES, she also has a pretty good piece of evidence from Irene’s testimony that this is not the case. Thus, insofar as Fiona should reduce her confidence that she has Condition A after Irene tells her otherwise, the AAV says that Fiona should be less confident that she has Condition A in INCONSISTENT DIAGNOSES than in SINGLE DIAGNOSIS, since this is what her total body of evidence supports.

The main takeaway here is that the AAV avoids the Problem of Opinionated Apprentices, and that it does so because it treats all testimony as evidence. The Problem of Opinionated Apprentices, then, is not a concern for all views of authoritative testimony in general, but a worry for the PRV and APRV in particular.