Does anti-reductionism in the epistemology of testimony imply interest relativism about knowledge attributions ?

Received on: 30 jun. 2021. Approved on: 10 set. 2021. Published on: 27 dez. 2021. Abstract: Anti-reductionism in the epistemology of testimony is the thesis that testimonial knowledge is not reducible to knowledge of some other familiar kind, such as inductive knowledge. Interest relativism about knowledge attributions is the thesis that the standards for knowledge attributions are relative to practical contexts. This paper argues that anti-reductionism implies interest relativism. The notion of “implies” here is a fairly strong one: anti-reductionism, together with plausible assumptions, entails interest relativism. A second thesis of the paper is that anti-reductionism in the epistemology of testimony creates significant pressure toward attributor contextualism (a version of interest relativism). Even if anti-reductionism does not strictly entail attributor contextualism, the most powerful motivations for anti-reductionism also motivate attributor contextualism over alternative positions.

Abstract: Anti-reductionism in the epistemology of testimony is the thesis that testimonial knowledge is not reducible to knowledge of some other familiar kind, such as inductive knowledge. Interest relativism about knowledge attributions is the thesis that the standards for knowledge attributions are relative to practical contexts. This paper argues that anti-reductionism implies interest relativism. The notion of "implies" here is a fairly strong one: anti-reductionism, together with plausible assumptions, entails interest relativism. A second thesis of the paper is that anti-reductionism in the epistemology of testimony creates significant pressure toward attributor contextualism (a version of interest relativism). Even if anti-reductionism does not strictly entail attributor contextualism, the most powerful motivations for anti-reductionism also motivate attributor contextualism over alternative positions.

Reductionism and Anti-reductionism in the epistemology of testimony
The label "anti-reductionism" is applied to various positions in the epistemology of testimony. 3 But a standard and well-motivated meaning of the term is the following: At least some testimonial knowledge is not reducible to knowledge of some other familiar kind, such as inductive knowledge. 4 One way to better understand the commitments of anti-reductionism, so understood, is to consider the position that it is denying. Namely, "reductionism" in the epistemology of testimony is the thesis that testimonial knowledge can be reduced to knowledge of some other kind. The relevant sense of "reduced" here is that testimonial knowledge is simply a species of some other kind of knowledge.
For example, standard Humean reductionism holds that testimonial knowledge is merely a species of inductive knowledge. 5 If reductionism is true, then there is nothing epistemically special about testimonial knowledge-the epistemology of testimony is just the epistemology of induction.
On the contrary, anti-reductionism holds that testimonial knowledge is special, requiring its own theoretical treatment.
Accordingly, we may define reductionism as follows: Reductionism: Testimonial knowledge is merely a species of (and is in that sense reducible to) some other kind of knowledge, for example inductive knowledge.
Anti-reductionism, then, is the denial of Reductionism: Anti-Reductionism: At least some testimonial knowledge is not merely a species of some other kind of knowledge. For example, not all testimonial knowledge is inductive knowledge.
For present purposes, it is important to note that anti-reductionism, so understood, entails that the standards for testimonial knowledge in general are not the same as the standards for inductive knowledge in general. At least some testimonial knowledge satisfies different standards from those associated with inductive knowledge.
The next thing to note is that, whereas Reductionism is defined with a universal quantifier, Anti-Reductionism is (and should be) defined with an existential quantifier. That is because Anti-Reductionism is simply the denial of Reductionism.
Of course, Anti-Reductionism is consistent with the position that all testimonial knowledge is distinct from inductive knowledge, but in fact this would be an implausibly strong position. For it seems clear that, at least in some cases, a hearer knows on the basis of testimony only because she has good inductive evidence that what the speaker testifies is true. Here are two such cases.

Police Investigator. A police investigator is
interviewing a suspect in a criminal case and the suspect testifies that she is unaware that the crime has occurred. The investigator follows up with further questions and, eventually, judges that the suspect is telling the truth.

Job Applicant.
A personnel director is interviewing a job applicant and the applicant testifies that he has extensive experience in sales.
The personnel director follows up with further questions and, eventually, judges that the job applicant is telling the truth.
Let us further suppose that, in some such cases, the hearer may come to know that the speaker is telling the truth. But if so, that is only because the hearer has adequate evidence of the relevant sort. That evidence might include extensive experience with similar kinds of interviews, or relevant background knowledge regarding the speaker, or some other kind of relevant evidence.
But in such cases, it looks like we are dealing with the same kinds of evidence and evidential standards that are relevant for inductive knowledge generally. If the hearer knows, it is because she has good inductive evidence regarding the point in question.
Contrast these cases with a third: Good Friends. Two good friends are talking and the first tells the second that he is going for a job interview later in the week. He also says that he feels good about his chances, because he has extensive experience in sales. The second friend believes that the first is telling the truth.
The present point is that any plausible Anti--Reductionism will treat this third case differently from the first two. Depending on the version of Anti-Reductionism at issue, the difference will be that the third case involves a special kind of trust, or assurance, or address, or speech act, or norms of cooperation, etc. 6 But whatever the view, the essential point will be that some testimonial knowledge requires some such special treatment, not that all testimonial knowledge does.
Again, the essential point of Anti-Reductionism is that not all testimonial knowledge is just more inductive knowledge. It is perfectly consistent with Anti-Reductionism that some testimonial knowledge is. 7 Accordingly, we may define "plausible anti-reductionism" as follows: Plausible Anti-Reductionism: Some testimonial knowledge is a species of inductive knowledge (and therefore shares similar standards) and some is not.
Anti-Reductionism in the epistemology of testimony is best understood as Plausible Anti-Reductionism.

Interest relativism and knowledge attributions.
We said that Interest Relativism about knowledge attributions is the thesis that the standards  Greco (2020). 7 This is explicit in Stephan Wright, "In Defence of Transmission," Episteme 12, 1 (2015): 13-28. Paul Faulkner makes essentially the same point with different terminology, by distinguishing between "knowledge from testimony" and "testimonial knowledge." See Faulkner, op. cit. See also Greco (2020). 8 More explicitly, Attributor Contextualism holds that two speakers can each utter the same sentence "S knows that p at time t," whereby the first says something true and the second says something false, while keeping the values of 'S', 'p', and 't' constant. Without this additional index to same time, Attributor Contextualism would be trivial. Obviously, the same person can know that p at one time and not know that p at another. Below I will follow the usual practice of leaving indexes to same time implicit. For some early defenses of attributor contextualism, see Stewart Cohen, "Knowledge, Context, and Social Standards," Synthese 73 (1987) ). 9 If we interpret "standards" and "practical context" broadly, any plausible version of Attributor Contextualism will be a standard version Attributor Contextualism. However, plausible versions of Attributor Contextualism need not be motivated by high stakes/ low stakes cases, as there are other reasons why standards might vary with practical contexts. This becomes apparent in Section 3.
for knowledge attributions are relative to practical contexts. We can better understand the commitments of Interest Relativism by considering its two most prominent versions: Attributor Contextualism and Subject Sensitive Invariantism.
Attributor Contextualism is a meta-linguistic thesis about the semantics of "knows," and hence the semantics of sentences containing "knows" and its cognates. According to Attributor Contextualism, the truth-values of sentences of the form 'S knows that p' (and the like) are variable across conversational context. That is, two speakers, engaging in different conversations, can each utter the same sentence "S knows that p," whereby the first says something true and the second says something false. Importantly, this can be the case while keeping the values of 'S' and 'p' constant. 8 On standard versions of Attributor Contextualism, the truth-values of such sentences are variable across conversational contexts because the standards for "knows" are relative to practical contexts. Standards Contextualism is often motivated by the consideration of pairs of cases, the first of which involves a "high stakes" practical context and the second of which involves a "low stakes" practical context. The idea is that "knows" takes on higher standards in higher stakes contexts. 9 We may further clarify Attributor Contextualism by making a distinction between a conversational context and a practical context. A conversational context is a situation in which some conversation is taking place. A practical context is a situation in which various practical interests, needs and concerns are in play. So as not to conflate the two senses of "context," going forward we will adopt the terminology of conversational contexts and practical environments. The central idea of Attributor Contextualism, then, is that different conversational contexts pick out different practical environments, which in turn determine different standards for "knows." Accordingly, we may define Attributor Contextualism as follows:  With these several positions now clearly in view, we may now turn to an argument that Anti-Reductionism entails Interest Relativism.

Anti-Reductionism implies Interest Relativism
The argument begins by considering an implication of Anti-Reductionism. Namely, if Anti-Reductionism is true, then different hearers are subject to different standards for testimonial knowledge. For recall, on that view, some testimonial knowledge is merely a species of inductive knowledge and some is not. But that entails that some testimonial knowledge is subject to the standards of inductive knowledge and some is not.
This does not yet imply Interest Relativism, however. That is because Interest Relativism is the view that the standards for knowledge attributions are relative to practical environments, and Anti-Reductionism, at least as defined above, is silent on that point. But consider: on any plausible anti-reductionist view, the reason that the standards for testimonial knowledge are sometimes those of inductive knowledge and sometimes not, is due to differences in the hearer's practical environment.
That is, sometimes features of the hearer's practical environment make the standards for inductive knowledge appropriate, and sometimes they make anti-reductionist standards appropriate.
That will be true on any plausible anti-reductionist view, and it is illustrated by the cases from Section 1. Thus, it is the practical environment described in Police Investigator that makes the standards of inductive knowledge appropriate.
Likewise for the practical environment described in Job Applicant. Accordingly, it was argued, any plausible anti-reductionist view should acknowledge that some testimonial knowledge is subject to the standards of inductive knowledge more generally. However, if anti-reductionism is true, then not all testimonial knowledge is like this. In the practical environment described by Good Friends, for example, the quality of evidence appropriate to inductive knowledge is not required.
Different anti-reductionist views will vary on the details, but all will agree on the basic point.
That is, given the practical features constituting the hearer's (and speaker's) environment, inclu-ding such things as the hearer's role within that environment, and the relationship between the speaker and the hearer, anti-reductionist standards for testimonial knowledge are appropriate.
Depending on the version of anti-reductionism at issue, the difference will be that this third case involves a special kind of trust, or perhaps some appropriate form of cooperation. But independently of these details, it will be some such features of the hearer's (and speaker's) practical environment that make the difference. For example, suppose that the conversational context is a discussion among good friends. Let us suppose that a third friend, Frederica, asks Henry whether Sean has experience in sales, and Henry answers, "Yes, I know that he does-he told me so. " In this case, let us suppose, the practical environment involves relevant relations of trust and cooperation, making anti-reductionist standards for testimonial knowledge relevant. Relative to this practical environment, the knowledge attribution comes out true. Now consider the same knowledge attribution, made in a different conversational context-a discussion between Henry and his boss, Beatrice.
Beatrice asks Henry whether Sean has experience in sales, and Henry answers, "Yes, I know that he does-he told me so. " Beatrice might take pause. What about alternative positions? First, it would seem that Standard Invariantism is inconsistent with the case as Anti-Reductionism understands it. That is because Standard Invariantism denies that the standards for knowledge are relative to practical interests, and so it must deny that the standards for testimonial knowledge are different relative to the two practical environments described in the case. Accordingly, Standard Invariantism must deny that Henry knows relative to the standards for testimonial knowledge associated with one of those practical environments, and does not know relative to the standards associated with the other. This is all in keeping with the results of Section 3, which included that Standard Invariantism denies Interest Relativism.
For somewhat different reasons, Subject Sensitive Invariantism also seems inconsistent with the case as Anti-Reductionism must understand it. As we saw above, Subject Sensitive Invariantism is a form of Interest Relativism, and so accepts that the standards for knowledge are relative to practical interests. But according to the view, it is always the subject's practical interests that govern. Accordingly, it is always the subject's practical environment that determines which standards are relevant. The problem with Job Applicant and Friends, however, is that in this case the subject (Henry) occupies more than one practical environment. Subject Sensitive Invariantism, therefore, does not have the resources for saying which of the two practical environments described in the case is the relevant one for determining the standards for knowledge. 13 Here is Grimm: it seems that our judgments about knowledge are sensitive not just to the practical concerns of the subject and not just to our own concerns as evaluators; in addition, it seems, they are sensitive to the concerns of certain third parties as well.
It therefore seems that the best way-perhaps the only way-to take all of these judgments at face value is to claim that our judgments about knowledge are sensitive to the highest of these stakes In other words, to accept something along the lines of the "rising tides" account an account on which rising costs either for the subject or for the evaluator or for certain third parties can raise the thresholds relevant to knowledge. 15 Accordingly, we can define the Rising Tides view as follows: Relative to the standards for testimonial knowledge associated with their friendship, Henry does know that Sean has extensive experience in sales.
Relative to the standards for testimonial knowledge associated with Henry's role as a personnel director, Henry does not know.
Put differently, Anti-Reductionism entails that a certain kind of case is possible-one in which a hearer occupies multiple practical environments, and where commitments 1-3 apply.
We may now see that both Attributor Con- 15  Contrast this with motivation for the Rising Tides view. Grimm motivates this view by emphasizing that practical interests can involve higher or lower stakes, and by arguing that the standards for knowledge ought to reflect this. The basic idea is that higher stakes demand higher standards. But given that different persons are faced with different practical interests, whose stakes matter? Grimm's thought is that knowledge should be of high enough quality to serve anyone who might need it, and so the highest standards should govern: "when we judge that a belief amounts to knowledge we are judging that the belief is reliable enough not only given our first-person interests and concerns, but given the interests and concerns of third parties who might come to depend upon the belief as well." 17 Hence, "our judgments about knowledge are sensitive to the highest of these stakes, whether the stakes are those of the subject, or of the evaluator, or of some third party." 18 Grimm's view ties higher standards to higher stakes, but this particular point does not seem essential to his core motivation for the view. Rather, the driving ideas are a) that different practical interests require different standards for knowledge, b) that knowledge should be of high enough 16 Other ways of articulating the "cooperation insight," consistent with this one, are to stress the importance of trust, and to characterize testimonial exchanges in terms of joint intention and agency. See Greco (2020). 17 Grimm,op. cit.,p. 131. Grimm glosses the "might" claim this way: "in our judgments about knowledge the "might" here is tracking the notion of what we may call a "real possibility"-that is, a possibility that we think "might actually" happen, as opposed to a possibility that we think "merely might" happen." (132) Thus, "while we do not think that to count as a knower someone's position needs to be strong enough to respect the "merely possible" elevated stakes of others, we do seem to think that his position needs to be strong enough to respect the elevated costs of those who "might actually" appeal to the belief." (134) 18 Grimm,op. cit.,p. 126. 19 What about other competitor views? Are there other competitors to Attributor Contextualism that can accommodate these conside-quality to serve all who might need it, and c) that the highest standards will achieve that purpose.
That is, the highest standards will yield knowledge of high enough quality to serve the practical needs of all who might need it. In that sense, a rising tide lifts all boats.
If we apply Grimm's reasoning here to Job Applicant and Friend, we get the result that the stricter standards associated with Henry's role as personnel director apply, and so Henry does not know that Sean has extensive experience in sales. Importantly, this is so relative to any conversational context, including the conversation among friends. So, when Frederica asks Henry whether Sean has experience in sales, and Henry answers, "Yes, I know that he does," this self-attribution of knowledge by Henry is false.
My own intuition is that Rising Tides gives the wrong result here. But that is not what is presently at issue. Rather, the present question is whether Rising Tides can accommodate the most important motivation for Attributor Contextualism-that some testimonial exchanges take place in contexts of cooperation, and are therefore governed by anti-reductionist norms and standards that reflect this. It seems clear that the answer is no.
On the contrary, the Rising Tides view dictates that the stricter standards associated with Henry's role as personnel director trump the anti-reductionist standards associated with his friendship, and so the stricter standards apply even relative to that context. This is in contrast to Attributor Contextualism, which easily accommodates the idea that the anti-reductionist standards ought to govern in the one conversational context, and the stricter standards ought to govern in the other. 19

Conclusions
In conclusion, I have argued that Anti-Reductionism in the epistemology of testimony implies Interest Relativism about knowledge attributions.
More specifically, the former position, when wedded to two plausible assumptions, entails the latter. The first assumption is that not all testimonial knowledge is subject to anti-reductionist standards. The second is that, whether testimonial knowledge is subject to anti-reductionist standards depends on features of the practical environment in which the testimonial exchange takes place. I have also argued that, although Anti-Reductionism does not strictly imply Attributor Contextualism, the most compelling motivation for the former view-that anti-reductionist norms ought to govern testimonial exchanges in cooperative contexts-also motivates the latter over to Interest Relativism and has good reasons for adopting Attributor Contextualism. I find a modus ponens inference attractive here, but others might opt for a modus tollens. 20