1. Introduction and overview
We form judgements by relying on the opinion of experts about all sorts of things. This fact has not gone unnoticed in epistemology. Epistemologists have been addressing the significance of the phenomenon of expert deference by taking up questions such as: What does it take to be an expert about a given matter (see Goldman [Reference Goldman2001])? Under what conditions is (expert and nonexpert) testimony a genuine source of knowledge (see Lackey and Sosa [Reference Lackey and Sosa2006])? Does the very idea of acquiring knowledge by relying on others’ opinions speak in favor of a specific account of knowledge rather than another (see Pritchard [Reference Pritchard2012]; Sosa [Reference Sosa2007])?
A common trait of these questions is that they focus on the first-order, normative epistemological significance of expert deference. Indeed, expert deference has been traditionally brought to bear on issues such as how to form our beliefs in an epistemically correct way, what knowledge is, what a good epistemic agent is, and so on. Instead of picking up one of these familiar threads in first-order epistemology, the more general aim of this article is to broaden the philosophical significance of expert deference by drawing attention to the as-yet overlooked metaepistemological implications of such a phenomenon.
In broad strokes, the strategy is to focus on the phenomenon of making epistemic judgements—that is, judgements such as “John does not know that Francis has left,” “Mary is irrational to believe that it’s raining while disbelieving that it’s not raining,” and so on, by relying exclusively on the opinion of somebody we are disposed to take to be an expert about the epistemic matter at issue. It will become apparent that, in a set of central cases where deference is distinctively about epistemic matters, one is doing something suspicious in making one’s epistemic judgements by exclusively relying on the opinion of the alleged expert.
I take the suspiciousness of expert deference about epistemic matters to be an instance of the more general phenomenon of suspicious expert deference about normative matters, which can be roughly yet intuitively contrasted with the unsuspiciousness of expert deference in squarely nonnormative regions of thought, such as biology, regional geography, physics, and the like. However, while much attention has been devoted to suspicious expert deference in morality (see e.g., Enoch [Reference Enoch2014]; Hills [Reference Hills2009]; McGrath [Reference McGrath2011]) and aesthetics (Driver [Reference Driver2006]), the question of the status of epistemic expert deference has been so far neglected.Footnote 1 This is an unfortunate lacuna in the study of epistemic normativity. Given the uncontroversial fact that most of what we know is testimonially acquired, deference about epistemic matters has important theoretical and practical implications. Suppose, for instance, that you defer to somebody you take to be an expert about whether you know that there’s a package of pasta in the cupboard. Surely, this episode of deference would be relevant theoretically, for it would allow you to evaluate yourself qua epistemic agent. Such an episode of deference would also be relevant practically: several epistemologists (see e.g., Hawthorne and Stanley [Reference Hawthorne and Stanley2008]; Williamson [Reference Williamson2000]) have argued in favor of a close relationship between knowledge and reasons for action such that, when one’s choice is p-dependent, one can treat p (say, that there’s a package of pasta in the cupboard) as a reason for acting (say, invite your friends over for a pasta dinner) only if one knows that p. Once again, deferring to an expert about such an issue allows you to find out whether you have reason to act in such-and-such ways.
Examining the phenomenon of expert deference about the epistemic by focusing on which cases of epistemic expert deference are suspicious and which are not is going to be relevant to a host of issues. The more general aim of this article is to carry out such an examination.Footnote 2
A moment of reflection shows that focusing on suspicious epistemic expert deference is also, and perhaps surprisingly, significant for understanding the nature of epistemology. On the face of it, metaepistemological views denying the objectivity of epistemology have a readily available explanation of this phenomenon: since there are no (in principle accessible) objective epistemic facts, there is no room for some agents to be better positioned epistemically vis-à-vis such facts than others. This is what makes deference to alleged experts about epistemic matters suspicious. Since, on pain of self-defeat, the objectivist about epistemology cannot deny the existence of objective epistemic facts, she cannot accept the explanation just sketched, or so it seems. This raises the question: How can the epistemic objectivist explain the asymmetry between suspicious deference in the epistemic case and unsuspicious deference in the geography or physics cases? Answering this question is the more specific aim of this article.
The article is organized as follows. In section 2, I offer a more precise characterisation of epistemic expert deference and clarify which instances of deference are relevant. In section 3, I present and criticize some fairly natural objectivist-friendly answers. In section 4, I offer an alternative explanation of the phenomenon. The core of my explanation is that an agent A’s deference to an alleged expert agent B is suspicious since A fails to acknowledge B is, as a matter of fact, A’s epistemic peer, and it is not epistemically rational to completely defer to the opinion of one’s epistemic peer. I argue that such an explanation fares better than the ones criticized in section 3, and I show that it is compatible with some, but not all, varieties of epistemic objectivism. In section 5, I offer some thoughts on the metaepistemological significance of my explanation.
2. Characterizing the phenomenon
To begin with, let me clarify what I mean by epistemic expert deference. We face an instance of epistemic expert deference just in case:
i. An agent A makes a judgement of the form “x is E.”Footnote 3 Let “x” range over concepts picking out whatever is subject to epistemic assessment, such as individuals, groups, mental states, propositions, and so on. Let “E” range over concepts picking out properties that are relevant to epistemic evaluation, such as justification, rationality, evidence, knowledge, and so on.
ii. A’s grounds for the judgement are wholly constituted by the fact that another agent B says that x is E. Footnote 4
iii. B is such that A sincerely and wholeheartedly takes her to be better positioned epistemically than A herself vis-à-vis the question whether x is E.
To forestall misunderstandings: epistemic expert deference neither requires disagreement between A and B nor does it require that the epistemic matter under scrutiny be controversial. This means that, on the one hand, A could defer to B about a question A has never thought about; on the other hand, A could defer to B about uncontroversial questions, such as whether or failing to believe that p amounts to knowing that p.
Having said this, the next step is to get clear about the scope of the phenomenon of suspicious epistemic expert deference. As registered above, we are rather unfamiliar with discussions of deference about epistemic matters. This unfamiliarity, I suspect, might lead us to conclude that epistemic expert deference is seldom, if at all, suspicious on the grounds that we can easily come up with examples where A makes an epistemic judgement by exclusively relying on B’s opinion and there is nothing suspicious about it. An analogous point, however, has already been noticed in the literature on moral deference.Footnote 5 That is to say, deferring to somebody who is better positioned than we are about normative matters is not always suspicious. Hence, my next task is to bring out which cases of epistemic expert deference are suspicious and which are not.
I submit that unsuspicious cases of expert deference about epistemic matters fall into two broad categories. The first consists of cases in which A and B agree on epistemic principles, A defers to B, and B is an expert on some relevant empirical or conceptual question. The second consists of cases in which A and B agree on epistemic principles, A defers to B, and A recognizes that B can better apply those principles. To illustrate the first category, consider the following case:
(SWANS)
Mary believes that all swans are white. John wonders whether Mary ought to believe so while accepting that one ought to believe true propositions only. However, John has no clue as to whether all swans are white. For this reason, he asks Ann, whom he takes to know a lot about swans. She tells John that, in 1667, Willem de Vlamingh discovered the existence of several specimens of black swans (Cygnus atratus) in Australia. So, John judges that Mary ought not to believe that all swans are white on the basis of Ann’s opinion.
Even though John makes the epistemic judgement that Mary ought not to believe that all swans are white on the basis of Ann’s opinion, it seems that his doing so is perfectly legitimate. The reason, I contend, is that Ann is not really acting as an epistemic expert. Rather, she is an expert about an empirical question, namely whether all swans are white, and John is harnessing this empirical expertise to make an epistemic judgement.
Consider now a different case belonging to the same category:
(INFERENCE)
John and Ann agree that it is epistemically correct to form a belief in p on the basis of a logically valid inference whose premises are justified. John, however, does not know whether he would violate this general principle if he believed q on the basis of his additional beliefs that p and that p implies q for he lacks the concept of material implication. Hence, John defers to Ann, who possesses the concept: she tells John that it is epistemically correct to infer q from p and p implies q, and he judges accordingly.
Is John doing something suspicious? It seems that he is not, and the reason is that Ann is not providing him with epistemic expertise—i.e., expertise about an epistemic matter—but rather with conceptual expertise. Ann possesses the concept of material implication and she can therefore draw an epistemic conclusion about the case at stake, whereas John doesn’t possess the concept and is therefore unable to draw any epistemic conclusion about the case.
These examples make it apparent that the cases of epistemic expert deference that are suspicious are not cases in which A’s deference to B’s expertise concerns an empirical or a conceptual question. Let us turn now to give an example of cases falling within the second category of nonsuspicious cases of epistemic expert deference:
(COFFEE)
John and Ann agree that in order for one to correctly form one’s belief that p on testimonial grounds, one should have beliefs about the reliability of the testifier. John asks his colleagues, “Has the dean decided whether we’ll have a new coffee machine this year?” George replies, “Yes, she did. I’ve run into her this morning and she said that we won’t have a new coffee machine.” John dislikes George very much, though. For this reason, he defers to Ann, who is rather indifferent about George, as to whether he can reasonably believe that George is a reliable testifier with respect to this matter. Ann tells John that George is reliable, John takes her word for it and judges to be correct to form the belief that the dean has decided that they won’t have a new coffee machine on the basis of what George says.
Is John’s deference to Ann suspicious? It does not seem so. What explains this verdict is the fact that Ann is not really acting as an epistemic expert. Rather, she’s better positioned than John in applying the general epistemological principle about testimony they both accept since she is less biased than he is. To generalize a lesson from this example: cases of suspicious epistemic expert dependence are not cases in which the alleged expert is better cognitively equipped (in the sense of not undergoing any general cognitive impairment, such as being biased, being dishonest, having a terrible headache, and so on, which would prevent the full exercise of one’s judgemental abilities) than the alleged nonexpert.
Summing up. (SWANS), (INFERENCE), and (COFFEE) are such that deference is made unsuspicious by some asymmetry either at the level of possession of empirical and conceptual facts, or at the level of one’s cognitive equipment. One might contend that many everyday cases of epistemic, as well as moral, deference fall under the two categories exemplified by these examples just discussed. Yet, it is equally plausible to have (and conceive of) cases of epistemic deference in which A defers to B and they differ neither in cognitive equipment nor in terms of conceptual repertoire and empirical facts available to assess the question at issue. Take, for instance:
(PEOPLE)
Ann, John, and Joanie are the only three people in the room, and John has to assess whether Joanie’s belief that there are five people in the room amounts to knowledge. John asks Ann, whom he takes be an expert about epistemic matters. Ann says that Joanie’s belief does not amount to knowledge, and John judges accordingly.
We can easily take (PEOPLE) to be a case in which John and Ann are equally cognitively equipped—that is, their cognitive systems are working normally, they have not taken any drugs, they have no particular bias towards Joanie, and so on. Furthermore, since both Ann and John are in the room, they both have access to the same evidence as to the number of people in the room. Finally, we can make sure that John and Ann share the same relevant conceptual repertoire by taking them to be part of the same cultural and socioeconomic group. If we read (PEOPLE) this way, the impression that John is doing something suspicious becomes much more salient. It indeed seems perfectly legitimate to react to John’s deference by asking: “Why is he asking Ann?”
Notice that expert deference in presumably objective and nonnormative areas of discourse does not give rise to suspiciousness. Consider:
(GEOGRAPHY)
Samuel wants to know where Saguenay is and takes Christine to be an expert about the geographical location of cities. Samuel therefore asks Christine, who tells him that Saguenay is in Quebec. Hence, Samuel judges that Saguenay is in Quebec by deferring to Christine’s opinion.
Samuel’s deference to Christine—as opposed to John’s deference to Ann in (PEOPLE)—is unsuspicious, and it would certainly be too harsh to criticize Samuel for not knowing where Saguenay is, even in spite of the lack of controversiality of this issue. This speaks in favor of the existence of an asymmetry between deference about epistemic matters and deference about geographical ones.
The cases presented so far motivate the following claim:
Distinctively Epistemic
The cases of epistemic expert deference which are suspicious are such that:
(I) A and B do not differ because B is better cognitively equipped than A.
(II) A and B do not differ because B is better positioned than A vis-à-vis empirical and/or conceptual facts bearing on whether x is E.
One might certainly worry that Distinctively Epistemic is too restrictive in that it rules out too many cases of genuine expert deference. To assuage this worry, let me offer a more general defence of Distinctively Epistemic.
The first condition imposed by Distinctively Epistemic has it that any difference in intelligence, conscientiousness, freedom from bias, and so on between A and B makes deference about epistemic matters not distinctively epistemic. This is so since being intelligent, free from bias, and thoughtful are character traits which make you better at making any kind of judgement and not only at judging epistemic matters. So, while it is true that if B is better cognitively equipped than A, then B counts as better positioned than A vis-à-vis the matter at hand, and while it is true that the matter at hand is epistemic, this expertise is not distinctively about epistemic matters. Now, if A defers to B about a given question, B is better cognitively equipped than A, and this—contra condition (I) of Distinctively Epistemic—counts as a case of epistemic expert deference in which the deference indeed is distinctively epistemic, then we have that cases such as (PEOPLE) and (GEOGRAPHY) should be regarded as on a par in terms of their suspiciousness—suppose that Christine acts as an expert since she’s more careful and conscientious in assessing geographical facts than Samuel. And yet, this prediction would fly in the face of the powerful asymmetry datum that Samuel is not doing anything wrong in (GEOGRAPHY) and doesn’t deserve criticism, whereas John is clearly doing something fishy in (PEOPLE) and does deserve some criticism. This explains why condition (I) is warranted.
A similar line of reasoning can be deployed to defend condition (II) of Distinctively Epistemic. For instance, if one lacks the concept of, say, knowledge and defers to, say, an epistemologist who presumably possesses it, one’s deference is not distinctively epistemic: possessing the adequate conceptual repertoire to judge a given question is something that makes us better than those who do not possess it with respect to any question, and not just with respect to epistemic ones.
These observations point to the same fact: we need to take the phenomenon of suspicious expert deference to be distinctively about epistemic matters. Since Distinctively Epistemic tells us which cases of deference are distinctively epistemic and which are not, we can use it to pin down the scope of the phenomenon at stake.
Since the main features and scope of the phenomenon of suspicious epistemic expert deference have been now clarified, we can turn to consider various possible explanations of it. A simple—yet far from trivial!—explanation of the suspiciousness of epistemic expert deference is that it is not the case that, in (PEOPLE), Ann is better positioned epistemically than John vis-à-vis the question of whether Joanie knows because since there are no objective epistemic facts, there are no epistemic experts. By contrast, expert deference about geographical matters is not suspicious because there are objective geographical facts about which one can be an expert.
This line of explanation can be developed within an antiobjectivist metaepistemology. It will suffice for the purposes of this section to characterise epistemic antiobjectivism as the negation of one of the main tenets of epistemic objectivism. Let epistemic objectivism be the family of metaepistemological views subscribing to the following three tenets:
(EO1) Epistemic judgements are truth-apt.
(EO2) Certain epistemic judgements are nontrivially true in virtue of certain epistemic facts.
(EO3) Epistemic facts exist independently of our cultures, evaluative perspectives, historical traditions, conceptual schemes, attitudes, linguistic practices, and so on.
In section 4.2, I will discuss different ways of articulating (EO1) through (EO3). However, the point that is of import here is that all epistemic antiobjectivist views deny (EO3), though for different reasons. By doing so, epistemic antiobjectivists can account for the idea that epistemic experts do not exist since there is no domain of distinctively epistemic facts (irrespective of whether this distinctiveness derives from their being identical to a certain kind of natural facts or for being irreducibly epistemic facts) with which such experts are better in touch than novices. This warrants the contention that deferring about epistemic matters to alleged experts is suspicious because there are no experts about such matters.
Clearly, epistemic objectivism cannot reject the existence of epistemic experts by giving up (EO3), for that would be self-defeating. Thus, epistemic objectivists have to take up the challenge of providing a different explanation of the phenomenon.
One might, of course, think “Not so fast,” for it seems that an antiobjectivist can clearly maintain that while there are no epistemic facts to be expert about, one can be better positioned than another vis-à-vis what kind of epistemic judgements are licensed by a given community, culture, epistemic system, conceptual scheme, and the like. On closer inspection, though, the kind of antiobjectivist friendly notion of expertise just sketched is not distinctively epistemic. According to this notion of expertise, one is an expert since one is well-positioned vis-à-vis facts about what follows from the adoption of the epistemic standards of a given community (culture, epistemic system, or conceptual scheme). Facts about what habits, standards, and practices are endorsed in certain cultures or communities are sociological or historical, as opposed to epistemic. The same holds, mutatis mutandis, for conceptual schemes. Hence, the notion of expertise available to the antiobjectivist is purely descriptive, as opposed to epistemic or normative.
So, while suspicious epistemic expert deference is a phenomenon that needs to be explained by objectivists and antiobjectivists alike, there is a presumption in favor of taking it to be a more serious problem for objectivism than for antiobjectivism. This article takes this presumption at face value and offers an investigation of the problem from an objectivist viewpoint. In the next section, I will consider what I take to be some natural options the objectivist can go for and argue that they are all unsatisfactory though for different reasons. This will warrant investigation of a new option.
3. Some unsatisfactory objectivist-friendly explanations
The first option available to the epistemic objectivist is to deny the existence of the challenge by resisting the intuition that, in cases such as (PEOPLE), John is doing something distinctively suspicious. More generally, the objectivist could deny that epistemic deference—and, more generally, deference in normative domains—is suspicious, while deference in purely nonnormative domains—such as geography—is not. She could substantiate this point by emphasising that there are cases in which deference is suspicious even in the latter case, e.g., suppose one defers about whether Rome is on Neptune.
This strategy strikes me as unsatisfactory though. For one thing, we might clearly acknowledge the existence of some suspicious cases of geographical deference and the possibility of some asymmetry between the moral case and the epistemic case while at the same time maintaining that there is something to the appearance of suspiciousness in the epistemic case which needs some explaining. It might well be the case that the suspiciousness of expert deference is, to some extent, ubiquitous across normative and nonnormative domains of discourse, but this doesn’t make its explanation in the epistemic domain less urgent. For another, one can reasonably take the phenomenon of suspicious deference about epistemic matters to be presented as an explanatory challenge arising from the antiobjectivist camp. I believe that one had better concede the existence of an apparent phenomenon and show that one’s own theory (i.e., objectivism) can make sense of it without any explanatory disadvantage vis-à-vis the account offered by the opposite theory (i.e., antiobjectivism) rather than refusing to take up the challenge by sheltering behind a stalemate of intuitions. Thus, I take this first explanation as ruled out.
A second option would be to deny the existence of epistemic experts by claiming that even though objective epistemic facts exist, they are much harder to know than facts obtaining in other objective yet nonnormative areas of discourse, such as geography. As far as I can see, this explanation can gain support from two often cited problems with the epistemic realm: first, the pervasiveness of disagreement amongst epistemologists about what one ought or is permitted to believe; and second, the absence of a track record of success or progress in epistemology that is comparable to that of geography or other nonnormative and allegedly objective regions of discourse. On reflection, however, this strategy turns out to be highly contentious. First, we might question the alleged data of the pervasiveness of disagreement and lack of progress in epistemology by pointing both to the many epistemic issues on which we agree and to the fact that geographers also disagree about many issues. Second, accepting the absence of epistemic experts by appealing to our massive inability of accessing epistemic facts amounts to conceding quite a bit of ground to skepticism about epistemology itself. Third, the claim that epistemic facts are hard to know is itself an epistemic claim. Now, do we know or are we warranted to believe it? If epistemic facts are hard to know, then it seems that we should not believe it. So, the claim that epistemic facts are hard to know borders on self-defeat.Footnote 6
A third way of explaining the suspiciousness of epistemic expert deference has it that making a judgement about an epistemic matter solely on the basis of expert deference prevents us from attaining understanding of that matter.Footnote 7 The basic strategy is this: we begin with the observation that when A judges that x is E on the basis of B’s opinion, A might well know that x is E without understanding why it is so; we then turn to defend the idea that understanding epistemic matters is somehow significant; we use that to explain the suspiciousness of epistemic expert deference in terms of its inability to provide the deferring agent with an understanding of the target epistemic matter. To develop this strategy into what we may call an “understanding-based explanation” of suspicious epistemic expert deference, we should first get clearer about what is required of the notion of understanding in order for it to play a role in the explanation of the suspiciousness of epistemic expert deference.
It appears fair to say that when A judges that x is E on the basis of B’s opinion, A makes this judgement without responding to the epistemically relevant features that determine whether x is E. To give an example, suppose that A judges that S ought not to believe p on the basis of B’s expert opinion. When A does so, A is not responding to the epistemically relevant features that make it incorrect for S to believe p such as, say, the falsity of p, the lack of sufficient evidence in its favor, the mistaken evaluation of the probative force of evidence, and so on. To put the same point in a different way, when A makes an epistemic judgement by deferring to B, A has not figured things out by herself. This shows that in order for the suspiciousness of epistemic expert deference to be explained by the fact that expert deference frustrates understanding, we should take understanding to have the following feature: if A judges that x is E and understands why x is E, then A makes that judgement since she is appreciating and responding to the epistemically relevant features of the target case. To put it differently, understanding requires figuring things out by oneself.Footnote 8 I take this to be an important constraint that a characterization of understanding must respect in order for it to be part of the understanding-based explanation.
This said, we should now make the best of the idea that understanding plays a somewhat important role in epistemic discourse for, in (PEOPLE), John’s failure of understanding why Joanie doesn’t know that there are five people in the room is what explains why there’s something amiss with his forming the belief that Joanie doesn’t know this proposition on the basis of Ann’s testimony.
A potentially illuminating articulation of this idea—inspired by McGrath (Reference McGrath2011)—consists in maintaining that epistemic discourse has a twofold aim: it aims both at getting epistemic facts right and at understanding them. That is to say, when we judge that x is E, we aim at correctly representing x as being E as well as understanding why x is E. By contrast, when we make a geographic judgement such as “Saguenay is in Quebec,” we just aim at getting the geographical location of Saguenay right without understanding why Saguenay is in Quebec. This enables us to single out the asymmetry between epistemic and geographical discourse that is needed to bolster the contention that the lack of understanding resulting from epistemic deference is more problematic than the lack of understanding resulting from geographical deference. This, in turn, enables us to explain the striking suspiciousness of the former as opposed to the (comparative) unsuspiciousness of the latter.
The aiming-at-φ claim has it that the aim of a given practice is such that it individuates the type of practice it is. Think, for instance, of the claim that belief is the type of cognition which aims at truth (see e.g., Shah and Velleman [Reference Shah and Velleman2005]), actions the type of events in which agents aim at self-constitution (Korsgaard [Reference Korsgaard2009]). When applied to epistemic discourse, the aiming-at-φ claim amounts to this: one’s judgement of the form “x is E” is an epistemic judgement only if one aims at getting epistemic things right and understanding them. On reflection, this licenses the following reasoning:
(1) One makes an epistemic judgement only if one aims at the distinctive aim of epistemic discourse (aiming-at-φ assumption).
(2) One is engaged in epistemic discourse only if one aims at getting epistemic facts right and understanding them (principle about what identifies epistemic discourse).
(3) If one makes one’s judgement “x is E” on the basis of expert deference, one does not aim at understanding (understanding-based explanation).
(4) If one does not aim at understanding, then one is not engaged in epistemic discourse (from 2).
(5) If one is not engaged in epistemic discourse, then one does not make an epistemic judgement (from 1 by contraposition).
(6) If one forms one’s judgement “x is E” on the basis of expert deference, then one does not make an epistemic judgement (from 3, 4, and 5).
The conclusion of this reasoning is highly contentious. Epistemic agents, epistemologists included, make judgements about epistemic matters on the basis of expert deference, and they do take such judgements to be epistemic in kind. If (6) were correct, though, it would follow that they are blind to the subject matter of their judgements: they take them to be epistemic judgements, but they are not. As familiar from the literature on epistemic contextualism, positing such systematic error and blindness is a highly contentious move whose careful examination would lead us astray. I shall instead make the following observation: fleshing out the understanding-based explanation by adopting the aiming-at-understand claim carries the highly contentious commitment to the idea that we are mistaken about the fact that when we made judgements featuring epistemic concepts on the basis of expert deference, these judgements are epistemic in type. Such contentiousness, to my mind, warrants dissatisfaction with the understanding-based explanation.
One might observe that premise (2) of the previous reasoning features a principle about what identifies epistemic discourse which is too strong. I agree with this observation.Footnote 9 However, I believe that the supporter of the understanding-based explanation must be committed to such a strong principle in order for her explanation to cover all possible cases of expert deference about distinctively epistemic matters. Let me explain.Footnote 10
Suppose we accept the idea that one need not pursue understanding on every occasion. Rather, what matters is to pursue that aim to a certain extent. Thus, there is a suitably defined threshold such that if one has formed enough of one's epistemic judgements by figuring things out by oneself, one thereby counts as having pursued the aim of understanding.
Now, suppose that A has met the required threshold and therefore counts as having pursued the aim of understanding. Suppose, moreover, that A defers in a distinctively epistemic way to B as to whether x is E. This instance of deference seems suspicious, but this suspiciousness cannot be explained by claiming that A has failed to pursue the aim of understanding. So, the resulting understanding-based explanation would not be able to cover those cases of epistemic expert deference which are suspicious and in which A counts as having pursued the aim of understanding. This amounts to a significant loss of explanatory power which makes this version of understanding-based explanation ultimately unappealing.
Let us take stock. I have rejected three possible and somewhat natural ways in which epistemic objectivism could make sense of the suspiciousness of cases like (PEOPLE). However, all is not lost for the epistemic objectivist. In the remainder of the article, I argue that there is a different and arguably less flatfooted explanation of the phenomenon which can be accepted by some varieties of nonskeptical epistemic objectivism.
4. Epistemic parity and objectivity: The parity-based explanation
The explanation I will defend goes as follows: cases of distinctively epistemic deference are suspicious since the two individuals fail to acknowledge that they are epistemic peers and it is not epistemically rational to ground one’s judgement solely on the basis of one’s peer’s opinion. Call this the parity-based explanation. Footnote 11
I take this to be a not-so-obvious and surprising objectivist-friendly explanation of the phenomenon at stake, so let us proceed carefully. In section 4.a, I present the parity-based explanation. In section 4.b, I outline some varieties of epistemic objectivism, and discuss which of them can accept this explanation.
4.a The parity-based explanation
The parity-based explanation contends that cases of distinctively epistemic expert deference are suspicious since A and B fail to acknowledge that they are, as a matter of fact, epistemic peers and complete deference to one’s peer’s judgement is not epistemically rational. By contrast, cases of geographical expert deference are not suspicious since the two individuals do stand in a novice-expert relation. To articulate this explanation in greater detail, let me first offer a definition of epistemic parity:
Epistemic Parity
A and B are epistemic peers with respect to a given question Q just in case: (evidential equality) They possess the same evidence which bears on Q; (cognitive equipment equality) They are equally cognitively equipped in terms of thoughtfulness, sincerity, freedom from bias, and so on.
This formulation captures the driving thought behind notationally different definitions of epistemic parity proposed in Christensen (Reference Christensen2007, 188), Kelly (Reference Kelly, Gendler and Hawthorne2005, 173–74), Lackey (Reference Lackey, Haddock, Millar and Pritchard2010, 302), and many other authors working in the epistemology of disagreement.Footnote 12
Armed with a better grip on epistemic parity, we can ask: Why should we endorse the claim that cases of suspicious expert deference are, as a matter of fact, cases of peer deference? This claim is clearly less of a natural explanation than, say, the understanding-based explanation. So, let me motivate this explanation.
The rationale of the parity-based explanation lies in the claim that I call Distinctively Epistemic formulated at the end of section 2. Distinctively Epistemic has it that the relevant cases of suspicious epistemic deference are such that the agents do not differ with respect to their cognitive equipment and possession of empirical and conceptual facts that bear on the target matter. This is tantamount to maintaining that cases of distinctively epistemic deference are such that the agents meet both the cognitive equipment equality condition and the evidential equality condition of epistemic parity where the latter is met under the assumption that the evidence bearing on epistemic questions is constituted by empirical and/or conceptual facts only. Call this the “evidential equality assumption.”
I hasten to acknowledge that while the satisfaction of the cognitive equipment equality condition straightforwardly follows from Distinctively Epistemic, the evidential equality assumption is much more controversial in that it is far from established that the evidence that bears on epistemic questions is constituted by empirical and conceptual facts only. To address this delicate issue, I will proceed as follows. In the remainder of this section, I will articulate the parity-based explanation in its finer details by uncritically accepting the evidential equality assumption. I will then turn to argue for the plausibility of this assumption.
The parity-based explanation has two components: first, it claims that cases of distinctively epistemic expert deference are suspicious because the parties involved fail to acknowledge that they do not stand in a novice-expert relation but they stand in an epistemic parity relation; second, it maintains that making a judgement solely on the basis of the opinion of an epistemic peer is not epistemically rational. So, the parity-based explanation has to be tested against instances of distinctively epistemic deference which come in two broad varieties:
(1V) Prior to deferring to an expert about whether x is E, A believes or suspends judgement about whether x is E. Then, A asks B if x is E, and B judges that x is not E.
(2V) A has never thought about whether x is E prior to deferring to B.
The first variety comprises cases in which A and B disagree about whether x is E and, after such a discovery, A gets to adopt B’s opinion. Thus, in order for the parity-based explanation to succeed with regard to (1V), we have to establish whether it is rational for one to respond to a disagreement with one’s epistemic peer by completely giving up one’s initial attitude and adopting one’s peer’s.
A quick survey of the literature on the epistemology of disagreement teaches us that giving up completely one’s opinion and ending up embracing one’s peer’s opinion is not put on the table as a plausible rational response to peer disagreement. For one thing, it is an open question whether or not peers are rationally required to revise or retain their original opinions in the face of their disagreement. For another, even conceding that revision is rationally mandated, such revision will never require of one to completely embrace the opinion of one’s peer, and vice versa. Indeed, so-called “conciliationism” about the rational resolution of peer disagreement (see Christensen [Reference Christensen2007] and Elga [Reference Elga2007] for some early statements of the view) maintain that one has to partially revise one’s own initial doxastic attitude, for instance by splitting evenly one’s original degrees of confidence and those of one’s peer.
In light of this, we can safely claim that the parity-based explanation succeeds in accounting for the suspiciousness of cases of distinctively epistemic deference where agents disagree about the matter at hand. These cases are suspicious since the two agents fail to acknowledge that they are epistemic peers, and it is not epistemically rational to give up completely one’s initial attitude to entertain the attitude held by one’s peer.
Let us turn now to (2V) and take up cases of suspicious distinctively epistemic deference in which A has never thought about the target question prior to asking B about it. It must be kept in mind that, according to the parity-based explanation, these are cases where A and B are epistemic peers. Thus, if A has never thought about the question, this means that she has not begun thinking about how to address the question, is not aware of what the body of evidence bearing on the question is, has not collected and assessed evidence, and so on. Hence, A is in a bad epistemic condition vis-à-vis answering the question at hand. If A is in such a bad epistemic condition vis-à-vis the target question and B is her epistemic peer, B is a fortiori in the same (bad) epistemic condition as A. Thus, if A deferred to B in such cases, she would be forming her judgement solely on the basis of the opinion of somebody who has no clue about the matter at hand. This strikes me as ultimately irrational. So, these cases of distinctively epistemic deference are suspicious since it is not rational to make a judgement solely on the basis of the opinion of your peer when your peer has no clue how to answer the target question. The parity-based explanation captures this plausible verdict.
An objector might ask: “What if, in (PEOPLE), John has some (ultimately misleading) evidence that he is not Ann’s peer and that she is much more knowledgeable and reliable than him on epistemic matters?” Call this scenario (PEOPLE*). (PEOPLE*) might be regarded as a counterexample to the parity-based explanation for the following reason. John’s (misleading) evidence about Ann’s superiority gives him a reason to believe that that Ann is an epistemic superior. Assuming that (PEOPLE*) is as suspicious as (PEOPLE), we couldn’t therefore explain the suspiciousness of John’s deference in (PEOPLE*) by claiming that it’s irrational for him to rely on Ann since John does have reason to regard Ann as an epistemic superior.
To handle cases such as (PEOPLE*), I’ll avail myself of the well-tried distinction between pro tanto and all-things-considered reasons. The distinction, in a nutshell, is that while there might be a pro tanto epistemic reason for ϕ and a pro tanto epistemic reason for not-ϕ, it is not the case that there is all-things-considered reason for both ϕ and not-ϕ. Surely the misleading evidence possessed by John in (PEOPLE*) gives him a pro tanto reason to believe that Ann is not his epistemic peer. Yet, the fact that Ann is John’s epistemic peer provides John with a pro tanto reason to disbelieve that Ann is not his epistemic peer. Now, John’s reason to disbelieve that Ann is not his epistemic peer has a better epistemic pedigree than John’s reason to believe this proposition in that the former is not misleading, whereas the latter is. As a consequence, John has most reason to treat Ann as his epistemic peer. Thus, the suspiciousness of John’s deference in (PEOPLE*) can still be accounted for by the irrationality of deferring to an epistemic peer.
In light of the discussion pursued so far, we can highlight the advantages of the parity-based explanation vis-à-vis the other accounts examined in section 3. First, the parity-based explanation concedes the intuition that there is something suspicious about alleged expert deference in the epistemic domain. Second, it does not explain the suspiciousness by saying that epistemic facts are very hard to access, thereby avoiding opening the door to a potentially worrisome and self-defeating skepticism about epistemic discourse. Thirdly, it does not commit us to any claim about the aim(s) of epistemic discourse which might give rise to a very contentious method of type-individuating epistemic judgements. Given these advantages, it’s worth trying to defend the most controversial bit of the parity-based explanation, namely the evidential equality assumption to the effect that agents involved in cases of distinctively epistemic deference are equals with respect to the evidence that bears on the target question.
The first point I want to make is the following: antiobjectivists have to concede this assumption to objectivists. It is indeed important to bear in mind that insofar as the phenomenon of suspicious epistemic expert deference poses a problem for a given account of the nature of epistemic discourse, we must allow that account to use whatever resources it has at its disposal to address that problem. Let us now suppose that the antiobjectivist criticizes the parity-based explanation as follows: “Such explanation fails because even if two agents possess the same empirical and conceptual facts, in order for them to be epistemic peers they have to share the same evaluative perspective, cultural context, preferences, or what have you.” This line of reasoning begs the question against the epistemic objectivist because it presupposes the falsity of (EO3), thereby presupposing the falsity of objectivism. This shows that the claim to the effect that agents in cases of distinctively epistemic deference are epistemic peers must be conceded by the antiobjectivist.
This provides the first part of my defence of the evidential equality assumption. The second part of such defence consists in arguing that such assumption is indeed compatible with some varieties of epistemic objectivism.
4.b Varieties of objectivism and the parity-based explanation
In section 2, I have offered a general characterization of epistemic objectivism by claiming that any view subscribing to (EO1) through (EO3) is objectivist in kind. It is time now to explore the various ways in which one can be an epistemic objectivist.Footnote 13
I begin with Reductive Epistemic Realism (RER). (RER) satisfies (EO1) through (EO3) by reducing epistemic facts and properties to nonepistemic facts and properties and by taking epistemic judgements to be about and made true by such nonepistemic facts and properties. Depending on how one conceives of nonepistemic facts, there are two versions of (RER), what I shall call Synthetic Reductive Epistemic Realism (SRER) and Conceptual Reductive Epistemic Realism (CRER).
(SRER) maintains that epistemic judgements are about and made true by empirical facts—that is, facts that can be discovered by using the methods employed in scientific inquiry. Jenkins (Reference Jenkins2007) explicitly defends (SRER) about epistemic ought. Jenkins maintains (264) that a judgement such as “S epistemically ought to believe p” is made true by the same states of affairs that make true a judgement such as “p is probably true, in an obvious and salient way, given S’s state of information.” Jenkins stresses that this is a synthetic, as opposed to conceptual, reduction. Thus, while the concept “ought” and the concept “being probably true in an obvious and salient way given S’s state of information” are different, they refer to the same natural property.
Unlike (SRER), (CRER) maintains that epistemic judgements are about and made true by empirical and conceptual facts, where the latter can be discovered via conceptual analysis.Footnote 14 To illustrate an instance of (CRER) about knowledge, take Sosa’s virtue-reliabilist account of what he dubs “animal knowledge.” On Sosa’s account, a judgement such as “A animal-knows p” is about and made true by the fact that A has an apt belief that p, where aptness is constituted by accuracy (viz the belief is true) and adroitness (i.e., the belief is the product of the agent’s reliable cognitive abilities) instantiating a “because of” relation. This account of animal knowledge holds in virtue of empirically explainable facts about which agents’ dispositions tend to produce true beliefs as well as facts about our concept of knowledge uncovered by intuitions elicited by thought experiments, such as Gettier’s.
I will turn now to present Nonreductive Epistemic Realism (NER). (NER) vindicates (EO1) through (EO3) by maintaining that epistemic judgements are ultimately about and made true by irreducibly epistemic facts. An instance of (NER) about knowledge is “knowledge-first” epistemology (see Williamson [Reference Williamson2000]).Footnote 15 Knowledge-first epistemology has it that knowledge is to be regarded as an irreducibly epistemic property—that is, a property which cannot be in turn decomposed in or reduced to a combination of other allegedly more fundamental epistemic and nonepistemic properties.Footnote 16
In light of this, I submit that (SRER) and (CRER) can accept the parity-based explanation. (SRER) has it that what determines whether a given epistemic judgement is true is an empirical fact. So, the evidence bearing on epistemic questions is constituted by empirical facts only. Since Distinctively Epistemic has it that the relevant cases of suspicious deference are such that the two parties are equal with respect to their possession of empirical facts, it follows that (SRER) can endorse the parity-based explanation. (CRER) has it that what determines whether a given epistemic judgement is true is whether this judgement is made true by empirical and conceptual facts. Since Distinctively Epistemic has it that the cases giving rise to the feeling of suspiciousness are such that A and B are equals with respect to possession of empirical and conceptual facts, it follows that (CRER) can endorse the parity-based explanation.
The foregoing assumes that if two agents are equals with respect to the possession of empirical and conceptual facts, then they possess exactly the same evidence. Yet, one might want to object to this conditional.Footnote 17 To see how, suppose that (a simplification of) Jenkins’s version of (SRER) is true and consider a scenario in which:
(a) A and B have to establish whether one ought to believe that p.
(b) It is true that what one ought to believe reduces to what makes a given proposition probably true relative to an agent’s state of information.
(c) A and B possess the same empirical evidence about what makes a given proposition probably true relative to an agent’s state of information.
(d) However, only B knows that the evidence bearing on this question is also evidence about epistemic matters—that is, evidence that enables us to settle the question of what agents ought to believe.
If (a) through (d) are in place, it seems that equal possession of empirical facts does not entail equal possession of epistemic facts. Therefore, we should not take A and B to be epistemic peers since, despite their being with respect to empirical facts, B is the only one who appreciates the epistemic import of such empirical facts for the question of what agents ought to believe.
I believe, however, that these considerations do not undermine the parity-based explanation. Consider the following example:
(PROBABLE)
John defers to Ann’s expert opinion about whether Lucas ought to believe that it’s raining in a scenario where (a) through (d) are met.
Is (PROBABLE) a case of suspicious epistemic expert deference? I don’t think so. The reason why is that Ann is not acting as an epistemic expert but as a metaepistemic one. In (PROBABLE), Ann is better positioned than John with respect to the fact that what an agent ought to believe reduces to what makes a proposition probably true given the agent’s state of information. This is a metaepistemological fact—viz a fact about what epistemic facts are. Thus, the fact that (PROBABLE) is not a case of epistemic deference which is distinctively about epistemic matters is explained by the idea that Ann’s expertise bears on the nature of epistemic thought and talk as opposed to the positive or negative epistemic status of a given proposition.
Having clarified this, let us turn now to (NER). As far as I can see, (NER) cannot accept the parity-based explanation. To see why, we should pause on the commitments that this view has about the nature of the evidence that bears on epistemic questions. (NER) maintains that in order for us to answer epistemic questions, we must access irreducibly epistemic facts—that is, facts which are neither about our concepts nor about the empirical world. To illustrate, suppose that A has to establish whether John is justified in believing that p. To answer this question, A should establish whether John knows that p, where knowledge is an irreducibly epistemic property. It is therefore possible that A and B are peers with respect to their cognitive equipment and possession of empirical and conceptual facts but still differ with respect to their access to the irreducible epistemic property. Thus, (NER) cannot endorse the evidential equality assumption, for there is more to evidential equality about epistemic matters than empirical and conceptual equality. Therefore, (NER) cannot accept the parity-based explanation.
5. Conclusion
The parity-based explanation of the phenomenon of suspicious expert deference about epistemic matters is very appealing for the following three reasons. First, even though nonreductive epistemic realists cannot accept it, all the other epistemic objectivist accounts of the nature of epistemic discourse can. So, this explanation enjoys a great deal of generality. Second, to avoid begging the question against such views, epistemic antiobjectivists will have a hard time showing where the parity-based explanation fails in that they have to concede to reductive realists about epistemic discourse that cases of epistemic deference which are distinctively about epistemic matters are cases of peer deference. Third, the parity-based explanation does not suffer from the problems affecting the other objectivist-friendly explanations reviewed in section 3.
In light of this, I have accomplished the more specific aim of this article: offering an objectivist-friendly explanation of the suspiciousness of epistemic expert deference.
I would like to conclude by explaining how the article also accomplishes its more general aim—that is, showing that the phenomenon of expert deference can be brought to bear on an account of the nature of epistemology. As I see it, the arguments of this article have a bearing on two important metaepistemological issues. First, I have focused on what the aims of epistemic thought and talk are by taking up the hypothesis that epistemology aims not only at getting epistemic matters right, but also at understanding them. This claim might well be right but no support for it is going to come via an inference to the best explanation of the following form: given that the phenomenon of suspicious epistemic deference is best explained by the claim that epistemic discourse aims at understanding, (it is probable) that epistemic discourse aims at understanding. This is so since I have argued explaining this phenomenon via the claim that epistemic thought and talk aim at understanding leads to a highly contentious way of type-individuating epistemic judgements.
Second, the fact that nonreductive epistemic realism cannot accept the parity-based explanation tells us that there is a pro tanto reason to take reductive versions of realism to be better metaepistemological views than nonreductive realism: if, as argued here, the parity-based explanation is the best one, reductive realisms can explain the suspiciousness of epistemic expert deference in a satisfactory way, whereas nonreductive realism cannot.
Acknowledgments.
I wish to thank Nathan Ballantyne, Fernando Broncano-Berrocal, Annalisa Coliva, Hichem Naar, Dylan Popowicz, and two anonymous referees for this journal for their helpful comments on various parts of this material.
Michele Palmira is a postdoctoral fellow at the University of Barcelona and a senior member of BIAP/LOGOS research group. His research interests include disagreement, epistemic rationality, and inquiry. His work has appeared in journals such as Proceedings of the Aristotelian Society, Philosophical Quarterly, Synthese, and American Philosophical Quarterly.
