2.1 Zagzebski's account of the authority of belief

Zagzebski's account of epistemic authority is based on two fundamental theses she derives from Joseph Raz (Reference Raz1986): the Preemption Thesis and the Justification Thesis 1 for the Authority of Belief.Footnote ⁷ Both are meant to provide justification for a seemingly counterintuitive action, such as believing on authority. The former is “a thesis about what it means to take someone as an epistemic authority” (Zagzebski Reference Zagzebski2014: 176) and can be defined as follows:

Thus, according to the Preemption Thesis, believing on authority means that if I consider someone to be an epistemic authority for me, then I am not only in the position to defer to her; rather, I am epistemically compelled to adopt her reasons for belief as the decisive ones to settle the matter. I am not required to dismiss or completely disregard my own reasons for believing p as long as I let the authority's reasons be those for why now I believe p.Footnote ⁸ However, although the Preemption Thesis accounts for the epistemic attitude I am rationally supposed to adopt, it says nothing about what justifies my adopting EA's reasons for belief.

Zagzebski's latter thesis explains “the normal way to show that [E]A is an authority for S” (Reference Zagzebski2014: 177) as follows:

According to JAB1, what allows me to consider someone an EA is my reflective judgment that she is more conscientious than I am, which in turn makes it rational for me to believe what she believes if I want to increase my chances of getting the truth.

Let us consider an ordinary example:

According to Zagzebski, once I reflectively judge through JAB1 that I am more likely to form a true belief if I trust the ornithologist and believe that p than if I try to figure out what to believe myself, I have a compelling reason to consider her an EA and to adopt her belief, as demanded by the Preemption Thesis.

2.2 Jäger and Zagzebski's dispute on the Preemption Thesis

As mentioned above, there is debate over whether believing on authority requires preempting. Jäger, among others, raises several concerns against the Preemption Thesis: here I focus on what he calls “the problem of unhinging proper bases” (see 2016: 173-ss.). This problem does not directly pertain to cases of initial disagreement between S and EA, such as the one considered in the previous example.Footnote ⁹ Rather, it concerns cases of partial or full agreement between S and EA that, according to him, Zagzebski's account is not in a position to accommodate.Footnote ¹⁰

In order to illustrate Jäger's first objection, consider the following example:

The Preemption Thesis, Jäger notices, requires that I stop considering my visual experience as the reason for which I believe p and instead preempt; but in such a situation where I partially (or fully) agree with EA, (i) it might not be psychologically possible for me to disregard my evidence and (ii) I need not do so. Clearly, the fact that EA shares my belief is additional supporting evidence for my true belief, which is (partially) justified by my visual experience. Grounding my belief in both my own perceptual evidence and the authority's judgment seems epistemically better off than simply supporting such a belief with uncertain perceptual reasons. Nonetheless, there is no reason why I should set aside my own evidence, since it revealed itself as accurate.

Zagzebski's reply to Jäger's objection aims at deflating the relevance of the subject's own reasons compared with the authority's. She argues that there is nothing I can see about the bird that the ornithologist has not already noticed. Thus, it would be illusory to think that adding my own reasons to the ornithologist's would provide a better justification for believing that p (see Zagzebski Reference Zagzebski2016: 190). Furthermore, Jäger's remark that it might not be psychologically possible to preempt in cases of agreement like this should not affect Zagzebski's account, which is meant to provide a rational way to justify believing on authority, rather than determining whether or not it is possible to preempt in such and such a situation. What matters to her is that preemption assures the best track record in terms of getting the truth.

Without entering the debate too deeply and straying too far from my purposes, there is a compelling reason why Zagzebski's reply does not make her argument safe from Jäger's objection. Consider the following modified version of the Red Cardinal Agreement Case, where the authority is not a socially recognized expert, e.g. an ornithologist, but just someone who is better epistemically placed than I am.

If we assume that there are circumstances, as in the Red Cardinal Sister Agreement Case, in which it is not necessary for one to be an EA to already be aware of possible reasons in support of p, then preempting might not be the most rational thing to do for S. Let us grant that deferring to my sister, i.e. the epistemic authority here, is rational, since she is better epistemically placed than I am. Nevertheless, despite my nearsightedness, when I asked for her help, I already had some reasons in support of my belief that were not part of her justification for p – namely, auditory evidence concerning the red cardinal's song. These perceptual reasons enforce both my justification for p and my confidence that p is the case. Therefore, preempting might still lead to the best track record in getting the truth, yet considering my sister's testimony as a further evidential reason for my belief that p would give me a stronger justification for p than she has. Hence, contra Zagzebski, it would be irrational for me to stop also basing my belief on my auditory evidence and just preempt.

2.3 Jäger's account of Socratic authority

Jäger's second objection to Zagzebski's account of epistemic authority emerges in cases of initial disagreement between S and EA within important domains like philosophy, scientific research, politics, and religion, or in those ordinary situations where gaining understanding is a more fundamental goal than merely having true beliefs. In such domains, the Preemption Thesis fails to capture what epistemic authority actually is, because the layperson S who disagrees with EA would not be willing to simply switch beliefs, adopt EA's, and improve his track record of true beliefs. What S wants is to enhance his understanding of some subject matter, that is, “grasping systematic connections among elements of a complex whole, or gaining insight into certain relations between items within a larger body of information” (Jäger Reference Jäger2016: 180).

Let us clarify the objection with an example: suppose I falsely believe that the recipe for tuna tartare requires that I cook the tuna in the oven for half an hour, but a chef, i.e. the epistemic authority for me here, tells me that I ought not to cook the tuna in the oven when preparing a tuna tartare. According to Zagzebski's account, once I conscientiously judge that the chef is more likely to get the truth than I am, I should just preempt my own reasons and adopt her belief. However, preempting would not give me any further understanding of the tuna tartare recipe, which is what I want to learn. I would expect an epistemic authority to improve my understanding by presenting further reasons in support of her belief. Merely knowing that I need not cook the tuna in the oven would not be of much help: acquiring more understanding would entail that I also learn, for instance, that tuna tartare is made with raw tuna, sliced into little strips or cubes, and served with different sauces or salads depending on the cook's preferences. This kind of expectation from an epistemic authority is something Zagzebski's theory does not explain, as she does not take a stand on whether her account of the authority of belief can also apply to understanding.Footnote ¹²

In contrast, Jäger argues that we should consider authorities in the epistemic realm those who are concerned with fostering S's understanding rather than merely providing him with true beliefs. This is what he calls a Socratic authority:

A comprehensive account of epistemic authority has to make room for Socratic authorities. The example of Socrates shows that the notion of epistemic authority not only serves as an efficient “litmus paper” for true beliefs, on Zagzebski's view; it also ought to account for the epistemic guidance that authorities offer.Footnote ¹⁴ I will be accounting for the authority of understanding in the next section. Here, it is worth adding a final remark: it is wise to separate the negative part of Jäger's project, that is undermining the Preemption Thesis, and its positive proposal that we define authority in the epistemic realm in terms of Socratic authority. For admitting SA as a fundamental level of epistemic authority, as in fact I do, does not commit us to accepting Jäger's argument against the Preemption Thesis. Therefore, in what follows, I suggest that there are good reasons for both saving the Preemption Thesis and offering a theory of epistemic authority that is able to account for the authority of belief as well as for the authority of understanding.

3.1 The (unexpressed) potential of Zagzebski's account

As anticipated, this section is devoted to defend the following claim:

Now, the notion of “understanding” is open to different interpretations. Here, following a recent but well-established tradition, I consider “understanding” in its “objectual” dimension, that is, as related to subject matters, theories, and bodies of information requiring explanation, rather than mere propositions (“propositional understanding”, Kvanvig Reference Kvanvig, Haddock, Millar and Pritchard2009: 96).Footnote ¹⁶ Moreover, I take understanding to be an epistemic success that varies along three main dimensions: breadth, depth, and significance (see Elgin Reference Elgin2007). According to this perspective, a subject A has broader understanding of a subject matter x than S if A is able to imbed some true belief pertaining to x into a more comprehensive set of beliefs concerning x. Instead, A has a deeper understanding of x if her network of beliefs related to x is more tightly intertwined than S's, that is “it contains more propositions, and/or more non-trivial inferential connections between propositions” (p. 36). Finally, A's understanding of x is more significant than S's if A grasps the relevance of the beliefs included in x better than S does; in other words, if A is in a better position to appreciate the significance of the truth, not merely to recognize that it is a truth (see p. 36).Footnote ^17, Footnote ¹⁸

Consider the following example:

Such an example of broader understanding, a very common one within education, clearly shows that the teacher's considerations, which were in fact absent from S's initial understanding of the issue, would improve his understanding of the reasons why the US attacked Iraq. Indeed, they allow him to figure out why starting that war was potentially profitable from an economic and financial point of view. Thus, the History Professor Case clarifies why S can consider the professor an epistemic authority even though they share a true belief – namely, that “George W. Bush sent US troops to Iraq to depose Saddam Hussein” – and they (partially) agree on the causes of the Iraq War.

For what makes the professor an authority for the student is not her having more true beliefs than him, but her possessing a broader understanding and sharing it with him. This is why she counts as an authority of understanding, rather than an authority of belief. Based on this example, we can introduce a thesis for the authority of understanding that parallels Zagzebski's JAB1:

This thesis expands Zagzebski's account, since it suggests another way to establish that EA is an authority for S, i.e. showing that S's chances of acquiring a better understanding of something increase if S trusts EA, instead of sticking to his own view.

3.2 Weakening but extending the Preemption Thesis

Claim 1 only concerns the way in which we can justify broadening the notion of epistemic authority so as to include the authority of understanding. Whether or not the Preemption Thesis works for the authority of understanding is a different matter, which I explore with the following consideration:

Claim 2 can be split into two different questions: (i) whether it is possible to maintain that the Preemption Thesis explains what epistemic authority is, and (ii) whether this thesis can apply to both the authority of belief and the authority of understanding. Jäger concedes that in several ordinary circumstances preempting is still the most rational thing to do.Footnote ²⁰ Suppose I go to the doctor and ask for some cure for my cough, but she tells me that, instead, I need to cure my acid reflux, which is the real cause of the cough. In such a case, once I conscientiously judge that my doctor is an authority for me in this domain, it is rational to give up my belief and the epistemic reasons supporting it and preemptively adopt the doctor's belief.

However, as discussed in sections 2.2 and 2.3, Jäger holds that preempting is irrational both when S has epistemic reasons regarding whether p is the case and in those domains where understanding is a more important epistemic goal than acquiring true beliefs. I disagree with him on both claims. Concerning the former, I will argue that it is rational for S to let EA's reasons preempt his own both when S disagrees with EA and when they agree but EA's reasons include and exceed all the reasons S has for believing p. As for the latter, I will demonstrate that a refined version of the Preemption Thesis is compatible with the notion of the authority of understanding. Thus, I endorse positive answers (partially, at least) to both (i) and (ii).

Consider (i). I suggest that the best way out of the Zagzebski–Jäger debate is to conditionalize the Preemption Thesis on particular, yet not-too-narrow, circumstances:

Weakening Zagzebski's Preemption Thesis in this particular way restricts its applicability to specific situations, thereby achieving two important objectives: first, the thesis accounts for those ordinary cases in which preempting is the most rational option for S; and second, it allows for situations in which preempting may not be rational. We can distinguish (at least) two types of situations in which S must rationally let EA's reasons for belief preempt his own:

1. (a) Cases of disagreement between S and EA.Footnote ²¹

2. (b) Cases of agreement between S and EA, where S is entitled to suppose that his reasons for believing p are included among EA's reasons.

This analysis of PTW-B accommodates instances of (a), such as the doctor case and the Red Cardinal Disagreement Case, and instances of (b) such as the Red Cardinal Agreement Case. However, condition (b) does not settle the Red Cardinal Sister Agreement Case, where I am entitled to suppose that my sister is not sensitive to birds’ sounds, and hence that aggregating her reasons with mine is more rational than just preempting.

Someone might object to (a), arguing that not every case of disagreement requires S to preempt his own reasons: indeed, there might well be situations in which EA gets something wrong and S should have held on to his original belief. It is important to notice though that such circumstances would call EA's status of epistemic authority into question, rather than S's rational commitment to preempting. Indeed, when you disagree with an EA on something, if you have conscientiously judged that she is an authority – that she is more likely to get the truth than you are – it is still rational for you to believe that you are more likely to be wrong than she is, and hence to adopt her belief.Footnote ²² However, if you find out that she is wrong (and you are correct), then you should start wondering whether it is still rational to consider her an epistemic authority, that is, whether adopting her belief maximizes your chances of getting the truth.Footnote ²³

Arnon Keren (Reference Keren2014b) offers an argument against the functioning of Zagzebski's Preemption Thesis, which challenges condition (a). He recalls that Zagzebski's justification of the rationality of preempting in cases of disagreement is grounded in Raz's argument for practical authority. As Raz clearly shows, once S acknowledges that he is more likely to be wrong than the authority in some domain, endorsing the authority's judgment makes S's rate of mistakes decline and equal that of the authority. In contrast, when S disagrees with the authority and sticks to his own judgment, his rate of mistakes remains greater than that of the authority. From these considerations, Raz concludes that “only by allowing the authority's judgment to preempt mine altogether will I succeed in improving my performance and bringing it to the level of the authority” (Reference Raz1986: 68).

Keren points out that the Razian argument is meant to explain preemption in cases of practical authority, where S has just two options – namely, doing an action a and not doing a. In contrast, in the epistemic case – relevant to Zagzebski – S has at least three options, namely “believing p”, “believing not-p”, and “suspending judgment”, and two epistemic goals, namely “forming a true belief” and “avoiding a false belief”. Given these elements, Keren argues that S's suspending judgment “in at least some proportion of the cases” of disagreement between S and EA – instead of preempting – lowers S's probability of error not only below that of S's independent judgment, but also below that of EA (see Keren Reference Keren2014b: 74). If S has better scores than EA in avoiding falsehood by withholding judgment, then it is not the case that S will be more likely to form a true belief and avoid a false one if S lets EA's judgment preempt his own. Thus, we should conclude with Keren that Zagzebski's account of the rationality of preempting in cases of disagreement fails and hence that PTW-B should not apply to condition (a).

However, Keren's argument is susceptible to criticism. Zagzebski (Reference Zagzebski2014) acknowledges the importance of considering “withholding judgment” an available strategy for S, yet she replies that Keren's objection loses its strength once we show that if S has three options, EA should have three options too. In other words, if we concede that EA can suggest that S suspend judgment then preempting still assures the best track record in satisfying both epistemic goals, for it is rational for S to withhold judgment whenever EA says to do so.

Taking a different line of argument, I want to show that, despite its initial plausibility, Keren's strategy leads to unwelcome and counterintuitive results. Let us assume that in cases of disagreement between S and EA, EA is more likely to be right than wrong.Footnote ²⁴ If so, then we can easily demonstrate that by withholding judgment S is more likely to fail to acquire true beliefs than to avoid false ones. Thus, it is more likely that suspending judgment will lead S to not reach the former goal than to reach the latter.

Although Keren is merely concerned with proving that suspension of belief lowers S's probability of error, he should not underestimate the (potential) loss of true beliefs that withholding judgment involves. For, assuming that forming true beliefs and avoiding false ones are equally relevant epistemic goals, as appears to be the case in Zagzebski's view, S's overall score with respect to the two goals considered together is lower when he suspends judgment than it is when he preemptively adopts EA's belief. In other words, withholding judgment leads S to obtain worse overall scores than he does by preempting. These results strike me as good reasons to avoid suspension of belief, especially because both Keren and Zagzebski postulate that the epistemic goals are equally relevant. Therefore, despite the technical soundness of his response to Zagzebski, Keren's strategy still looks implausible due to his underestimation of the costs of withholding judgment.

Now, let us examine (ii). Despite the examples supporting Jäger's argument against the Preemption Thesis, i.e. Red Cardinal Agreement Case and Red Cardinal Sister Agreement Case, I think Jäger is wrong to claim that preempting is in itself incompatible with the authority of understanding. Consider the following example:

In such a case, the child is in a position to disregard his initial understanding of the fish respiratory system and to adopt his grandmother's, for he realizes that he is more likely to gain a better understanding of this subject if he trusts his grandmother than if he sticks to his own judgment, as suggested by JAU1.

However, to demonstrate that it is rational for S to let EA's understanding preempt his own, I first need to resist the objection that “understanding can't simply be given to another in the way knowledge can” (Gordon Reference Gordon, Grimm, Baumberger and Ammon2017: 298), that is, via testimony. Philosophers like Emma C. Gordon, Allison Hills, and Linda Zagzebski have already defended what Kenneth Boyd recently called the Indirectness Thesis, according to which “[t]estimony cannot be a direct source of understanding: at best, it can be an indirect source of understanding by laying the groundwork for potential understanding” (Gordon Reference Gordon, Grimm, Baumberger and Ammon2017: 105).Footnote ²⁶

Following Boyd (Reference Boyd2017), we can conceive the notion of understanding as consisting of two main components: the informational component and the grasping component. The former amounts to possessing the apt information we need for understanding a subject matter x; hence it can easily be transmitted via testimony, as the teacher does when explaining the causes of the Iraq War to the student and as the grandmother does when she tells the child about how fish breathe. The latter component involves both the apt exercise of some intellectual ability on S's part and “the relationships between the thing understood and either the reasons that make that thing true or propositions that are related to it” (Boyd Reference Boyd2017: 114). Given that testimony does not convey epistemic virtues from the speaker to the hearer, it seems that one cannot acquire this grasping component through testimony, as stated by the Indirectness Thesis.

Nonetheless, Boyd claims that this thesis only applies to a specific kind of understanding, i.e. difficult understanding – that is, to cases requiring long-term effort and the exercise of many abilities, such as understanding the tiki-taka strategy in football or Navies-Stokes equations in fluid mechanics. Other kinds of understanding, i.e. easy understanding and easy _s understanding, have more relaxed demands. Anyone possessing a general background familiarity with the subject matter at stake can achieve easy understanding simply “by making sense of” the interlocutor's utterances, as the student and the child can do in the aforementioned examples (see Boyd Reference Boyd2017: 119). On the other hand, easy_s understanding works as easy understanding for a particular range of subjects who have specific background familiarity with a more complex issue and who have developed the necessary abilities to grasp the information they receive. For example, understanding the Gettier problem is a difficult challenge for a young philosophy student, yet it is definitely easier for a philosophy professor.

To demonstrate that one can achieve these kinds of understanding via testimony, Boyd argues that satisfying their grasping component does not demand too much effort from the subject, since making sense of the testimony itself suffices to allow one to grasp the relationships between bodies of information. Thus, even though the grasping component, technically speaking, does not transmit from one person to another it is still true that the mere fact that one conveys her understanding via testimony suffices to let the hearer acquire that understanding. For making sense of the speaker's utterances entails fulfilling both the informational and the grasping component at once. Therefore, we can conclude that, pace Gordon, easy understanding and easy _s understanding can be given to another as knowledge can, that is, via testimony.

Furthermore, easy understanding and easy _s understanding allow for preemption. Preemptive transmission of understanding between EA and S requires S to disregard whatever understanding of a subject matter x he possesses and adopt EA's understanding of x without considering her testimony as evidence to weigh against his own. Boyd's analysis explains why preempting understanding is possible, as acquiring easy understanding and easy_s understanding by making sense of another's words does not require the cognitive effort and the exercise of intellectual abilities that weighing evidence demands.

Some might want to resist the idea that it is possible to preempt understanding by arguing that understanding does not satisfy Jäger's account of preemption in terms of proper basing.Footnote ²⁷ According to Jäger, preemption demands that “the authority's belief should typically constitute a reason for which one adopts” her belief (Reference Jäger2016: 175). In the case of understanding, holding that the authority's understanding of x constitutes a reason for which one adopts her understanding of x seems (at best) unclear. For adopting another's understanding always requires that I grasp the inner relationships between the various parts of her testimony as well as the relationship between the testimony itself and my background knowledge concerning x.

However, Boyd's distinction between the two components of understanding suggests a way to accommodate this worry and to argue that Jäger's proper basing account explains preemption in the case of understanding as well. We should not be surprised that proper basing does not apply to the grasping component, as we have already shown that this component cannot be transmitted from one person to another. In contrast, this account does apply to the informational component: the epistemic reason for which S accepts the information conveyed by the authority via testimony – and therefore satisfy the informational component – is exactly the fact that it is EA's testimony of her understanding. Thus, we can account for preempting understanding in terms of proper basing, as the communicable component of understanding satisfies Jäger's requirements.

Before moving to the analysis of the circumstances in which preempting EA's understanding is rational for S, I think it is worth mentioning one more point about the role of the authority of understanding. Letting another's understanding preempt my own is possible only if the interlocutor is able to convey her understanding (or some portion of her understanding) in such a way that I can grasp the relevant relationships between bodies of information simply by making sense of her utterances – as both easy understanding and easy _s understanding demand. Such a condition casts light on a specific ability that authorities of understanding need to display: namely, sensitivity to S's epistemic resources. I will say more about the authority's virtues in section 4.

Now it is possible to introduce a version of PTW and clarify how preempting works in cases of the authority of understanding as follows:

The two main circumstances in which PTW-B justifies S's preemptively adopting EA's beliefs apply to PTW-U as well. When S disagrees with EA about x, that is, condition (a), preempting may well be the most rational option for him, as in the Grandmother Fish Case, where the grandmother clearly has a broader, deeper, and more significant understanding than the child. Moreover, we can imagine that the grandmother has developed the necessary sensitivity to her grandson's epistemic resources such that she is able to convey to him an easy understanding of how fish breathe.

Someone might object that there are cases of disagreement in which preempting is not rational, since EA is wrong and S is right. As in PTB-W, similar cases threaten EA's status of authority, not the rationality of preempting. Preempting understanding is rational insofar as EA's understanding of x fares better than S's understanding of x on at least one of the three dimensions of understanding – i.e. breadth, depth, and significance – without faring worse on any of them.

This rule applies to condition (b) as well: when EA's understanding of some subject matter x is just broader, deeper, or more significant than S's, and hence the informational component of EA's understanding of x already includes all the information about x that S possesses, it is rational for S to preempt. This is what happens in the History Professor Case, where the teacher's broader understanding of the Iraq War includes the student's explanation for the US military intervention in Iraq but places it within a more comprehensive story of the causes of conflict. Therefore, both JAU1 and PTW-U apply to the History Professor Case, for it is rational for the student to disregard his previous understanding and preemptively adopt the teacher's.

However, as condition (b) of PTW-B is not in a position to account for the Red Cardinal Sister Agreement Case, there are also cases where S's understanding is not completely included within EA's. Consider the following example:

In situations like the Philosophy Professor Case, even if it were possible to preempt, it would still not be rational. Indeed, since acquiring U¹ via preemption entails that my own understanding of x (U) “stops operating” (Jäger Reference Jäger2016: 175), by doing so I would be acquiring an understanding of the relationships between the traditional views and hybrid theories from the professor without grounding this new information on the more detailed story of x that I already possess. Suppose JAU1 still applies to the Philosophy Professor Case insofar as I acknowledge that the professor has a broader understanding U¹ than I have and I judge that it is more likely that I gain a better understanding of x if I believe what the professor believes. In this case, it would still be wise not to preempt, because basing the new information about hybrid theories of testimony provided by the professor on my deeper and more significant understanding of x would allow me to establish more robust explanatory relationships between traditional views and hybrid theories than the ones I would acquire if I let the professor's understanding preempt mine. Therefore, PTW-U does not apply to the Philosophy Professor Case.

To summarize, Claim 2 allowed me to defend a weak form of the Preemption Thesis, that is, to conditionalize the application of this thesis upon specific circumstances in which preempting is still the most rational thing to do for S, both in cases of the authority of belief and in cases of the authority of understanding.

3.3 The limits of Jäger's account

A final point about Jäger's account of epistemic authority can be found in Claim 3:

The argument in support of Claim 3 is quite straightforward. What Jäger has in mind when he offers his account of epistemic authority is Socrates’ maieutic ability. This ability is such that Socrates does not tell anyone that such and such is the case: he merely asks questions that guide interlocutors to understand things for themselves. Hence, Socrates’ questions provide people with reasons to expand their understanding in a very indirect way. On the one hand, Socrates’ method rightly places him and those who adopt the maieutic method among the authorities of understanding. But on the other hand, his higher ratio of true to false beliefs than the interlocutor plus his ability to recognize false beliefs within the corpus of the interlocutor's beliefs should make him an authority of belief as well.Footnote ²⁸ The fact that SA has this twofold authoritative role for the layperson justifies considering it a special case of authority. Indeed, Jäger's view illustrates a specific, viz. indirect, way in which someone can fulfil what I take to be the general scope of epistemic authority, that is, responding to S's various epistemic needs.

Nevertheless, it would be a mistake to define epistemic authority merely in terms of Socratic authority, for two main reasons: first, Jäger's account rules out authorities of understanding that lack maieutic ability and therefore guide laypeople in a more direct way, e.g. the grandmother who straightforwardly presents the child with a better explanatory story of how fish breathe, rather than leading him to work out that his understanding is wrong through critical questions. Second, it is insensitive to any distinction between authorities of belief and mere experts, neither of which is considered an epistemic authority on his view. Thus, Jäger's account severely restricts the opportunities for a layperson to defer to an epistemic authority.

Some might believe that this consequence of Jäger's view is unproblematic for the agenda of social epistemology. For even if experts and authorities who lack maieutic ability do not fulfil his account of authority, we can still acquire testimonial knowledge from them. Notice, though, that this reply is flawed. Many epistemologists concede that someone's testimony provides the hearer with weighing reasons. Furthermore, Jäger holds that this view may well apply to the interaction between the authority and the layperson. However, the weighing-model of epistemic authority differs from testimonial exchanges among epistemic peers in an important respect. According to this model, it is precisely the fact that I acknowledge my interlocutor's epistemic authority that explains why EA's testimony that p provides me with a stronger reason to believe that p than the one I would acquire from an epistemic peer's testimony that p.

Thus, if we follow Jäger and deny experts and authorities who lack maieutic ability the status of epistemic authorities, it is doubtful whether an expert's testimony that p provides us with the strong reason to believe p that we expect to possess whenever we deal with such epistemically superior subjects. This is bad news for social epistemologists aiming at explaining the value of our dependence on epistemic authorities, because Jäger's view fails to appreciate an important epistemic advantage to deferring to authorities, rather than to epistemic peers. Thus, unless Jäger offers us better reasons for endorsing this restrictive view, we can conclude that Socratic authority cannot represent a plausible account of epistemic authority; rather, it constitutes a special kind of authority.

We can draw further evidence for this conclusion from considerations about the Preemption Thesis. Jäger does not account for any authority whose exercise entails issuing preemptive reasons to the interlocutor. But if the reasons I offered in support of PTW-B and PTW-U are compelling, then Jäger owes us an argument for rejecting the idea that in specific circumstances epistemic authorities can provide the interlocutor with preemptive reasons. Furthermore, no version of the Preemption Thesis can work for Socratic authorities. To understand why, consider the following case, involving the grandmother and her grandson again:

This example shows how the interaction between a SA and a layperson might work. Indeed, the grandmother does not use her authority to present the child with a better understanding; rather she challenges him with critical questions so as to lead him to reassess his own judgment. It seems evident that Socratic authority leaves no room for preempting: if SA cares about a layperson's intellectual development and limits herself to putting S in a better position to improve his understanding by himself, then she does not give any true and clear-cut testimony of her understanding that S might preemptively adopt. Therefore, we can conclude that SA should be merely conceived as a special subset of epistemic authority including both the authority of belief and the authority of understanding.

As a final remark, notice that there may be cases of Socratic authority that violate Zagzebski's assumption that an epistemically conscientious S needs to acknowledge the interlocutor's authority over him. In other words, it is not necessary in the case of Socratic authority that S possesses a reason to believe that SA is more likely to have a true belief or a better understanding than S is. This violation is legitimate because the maieutic method does not require that S defer to SA, neither by preempting nor by weighing the reasons provided via testimony. Rather, S enhances his understanding simply by reflecting on the interlocutor's questions.

The discussion of the three claims introduced in this section has set the ground for the broader theory of epistemic authority I will develop in section 4. Claim 1 provided reasons for broadening Zagzebski's account so as to elicit S's appeal to an authority to expand his understanding of some subject matter. Claim 2 explained why it is wise to maintain a weak version of the Preemption Thesis (i.e. PTW-B) and to admit that it also applies to the authority of understanding (i.e. PTW-U). Finally, Claim 3 made clear that a comprehensive account of epistemic authority ought to define Jäger's Socratic authority merely as a special case of authority of both belief and understanding.