1. THE PUZZLE

Lackey presents the puzzle of deficient testimonial knowledge using the following three cases.

The first thing Lackey would have us notice about these cases is that there is something positive to be said for all of the testimonial beliefs in question. Each case includes three people. A first testifier (T1) who has specialized non-testimonial knowledge makes an assertion to a second person (T2) who has good reasons to take T1 as an authority. T2 then asserts without qualification the information received from T1 to a final recipient R. In each case, then, there is a chain leading from T1 to T2 to R. Both T1 and T2 testify on the basis of knowledge that they possess. Testifying from knowledge is a reliable procedure. Though R does not have the robust internal grounds for trusting T2 that T2 has for trusting T1, R has no special reason to doubt what T2 says. Furthermore, there don't appear to be any deviant causal chains in the cases that would lead one to believe that the beliefs have been either Gettiered or brought about by strange and fleeting processes. Reliable, undefeated, un-Gettiered beliefs are good candidates for knowledge, so, plausibly, not only T1 but also R come to have testimonial knowledge.

Despite the fact that there is reason to affirm that the final recipient in each of these chains has testimonial knowledge, Lackey maintains there is something that has gone wrong in each of these cases. Specifically, there is something deficient about the link between T2 and R that does not appear to be deficient about the link between T1 and T2 (Reference Lackey, Henning and Schweikard2013: 36). To draw out this intuition, Lackey has us imagine that R asks T2 a follow-up question only to discover that T2 can offer no further insights in support of or lending detail to T2's assertion. The fictional Lackey cannot elaborate on the paintings at the museum. Sylvester cannot clarify the morality of eating meat. Eliza cannot get into the specifics of Lucas's medical condition. In each case, R should be able to expect competent answers when asking follow-up questions of T2 and would feel somehow cheated were R to discover the limitations of T2's knowledge. In fact, one would expect R to treat the discovery of T2's limited perspective as a defeater for R's own belief.

Lackey drives home her point as follows.

Supposing us to have an adequate grasp of the puzzle, let us turn to the possible solutions Lackey entertains and why she thinks them inadequate. It is worth emphasizing that, as Lackey is introducing a puzzle, the candidates she surveys are adapted from discussions of things other than her puzzle, though often things in the neighbourhood. Moreover, the authors she surveys focus primarily on the recipient of moral or aesthetic testimony rather than on the one doing the asserting.Footnote ¹

The first candidate Lackey introduces concerns Sosa's distinction between animal knowledge and reflective knowledge (Reference Lackey, Henning and Schweikard2013: 40; cf. Sosa Reference Sosa2001, Reference Sosa2009). The thought is that perhaps what is wrong with deficient testimonial knowledge is that T2 in each of the above cases is willing to make assertions based on an item of animal knowledge, a bare reliable belief. Perhaps what is required of T2 qua testifier is the sort of reflective endorsement of the proposition in question that would allow T2 to defend and explicate it.Footnote ² Instead, this Sosa-style response would affirm that T2 has an isolated bit of animal knowledge not connected to a reflective perspective on the proposition asserted. In response, Lackey points out that one could easily describe these cases in a way such that T2 plausibly does have reflective testimonial knowledge without the deficiency of their testimony dissipating. After all, T2 has a rather rich body of information about T1 in the light of which T2 has formed T2's own belief. Sylvester, for instance, forms his belief based on his detailed knowledge of Miriam's track record.

The second candidate Lackey discusses is that what is lacking in T2 qua testifier is not knowledge but rather understanding (Reference Lackey, Henning and Schweikard2013: 40–42). Authors such as Jonathan Kvanvig have defended the idea that understanding should be distinguished as an epistemic good from mere knowledge (cf. Kvanvig Reference Kvanvig2003), so perhaps the intuitive deficiency in these cases arises because T2's assertion is backed by the wrong epistemic good. This stratagem, like the first, fixes upon the isolated nature of T2's testimonial belief as what is deficient with T2's testimony. T2 accepts T1's say so without that belief being undergirded by a grasp of the interconnections of a body of information that explains why the assertion in question is true and yet T2 is willing to make assertions without this understanding. Against this option, Lackey points out that, in the oncology case, Eliza presumably does have a significant degree of understanding of what pancreatic cancer entails and how the outcomes of the sort of tests Lucas has taken would be connected to the diagnosis she has accepted from her colleague. Furthermore, she claims that we can alter the aesthetic and moral cases so that the T2 figure in these cases accepts more propositions on testimony from T1 than the original proposition. The set of propositions that T2 takes on T1's authority can be stipulated to have the explanatory interconnections required for understanding. Lackey thinks that T2's testimony will still be deficient if one changes the cases in this manner, even if less so.

The third candidate Lackey tackles comes from the work of Aaron Meskin on aesthetic testimony (Lackey Reference Lackey, Henning and Schweikard2013: 43–5, interacting with Meskin Reference Meskin2004, Reference Meskin, Kieran and McIver Lopes2006). Meskin claims that aesthetic testimony is deficient because unreliable. The justification for this claim in this context is an elitist position in aesthetics (which, importantly, is tied to a rejection of aesthetic anti-realism). If discernment in aesthetics is difficult to develop and thus rarely possessed, then one would expect aesthetic judgements as a class to be unreliable absent special reason to trust one's informant. Lackey points out, however, that one can generate cases of deficient testimony about taste for low-brow fare that resist an elitist treatment (e.g. the relative tastiness of burgers). She goes on to claim that the deficiency in her aesthetic case is plausibly the same as whatever deficiency pertains to her moral and oncology cases, but it is not plausible that there are similar problems relating to an elitist restriction on expertise in these other cases.

A fourth candidate, due to Robert Hopkins (Hopkins Reference Hopkins2000), is that we expect aesthetic testimony to be justified on evidential grounds and thus one cannot simply be the beneficiary of a reliable extended chain of testimony when it comes to aesthetics (Lackey Reference Lackey, Henning and Schweikard2013: 45–6). Perhaps testimonial knowledge can simply be transmitted on other subjects, but it must be measured against stricter criteria where aesthetic judgements are concerned. One reason that Hopkins puts forward in favour of holding aesthetic testimony to a higher standard is the prevalence of disagreement in aesthetics. In more recent work, Hopkins focuses on holding the using of aesthetic testimony to a higher standard rather than the having of such knowledge (Hopkins Reference Hopkins2011).Footnote ³ Lackey finds Hopkins' imposition of special restrictions on aesthetic testimony theoretically unmotivated. In particular, according to Lackey, he doesn't explain why aesthetic testimony is deficient. He just assumes that it is and imposes his evidential restriction as a response. He does, of course, appeal to disagreement, but there are many kinds of testimony that are surrounded by disagreement, so Lackey thinks that is no reason to think of aesthetic judgements as unique.

Hopkins could respond to Lackey by claiming that deficient testimonial knowledge tracks domains in which disagreement is common in general and by claiming further that aesthetics is one such domain. A rejoinder available to Lackey would be as follows. The oncology case appears to be deficient in whatever way the aesthetic case is, but oncology isn't a domain fraught with disagreement in the manner that aesthetics is sometimes thought to be. Furthermore, not all aesthetic judgements generate significant disagreement; one could presumably generate cases of deficient testimony using uncontroversial aesthetic assertions such as, ‘Seeing Michelangelo's Pieta in person is a powerful experience’. It is completely uncontroversial that the Pieta is a masterpiece, but asserting this proposition without qualification and without ever having seen the sculpture seems to be a case of deficient testimony.

The fifth and final candidate Lackey addresses, due to Alison Hills (Hills Reference Hills2009), again appeals to understanding, but, unlike the second candidate we looked at, Hills focuses on deficiencies in testimonial beliefs rather than deficiencies in testifiers that lack understanding (Lackey Reference Lackey, Henning and Schweikard2013: 46–9). In particular, Hills grants that one can obtain moral knowledge by testimony, but she claims that the problem with moral testimonial beliefs is that they lack understanding. Hills' concept of understanding is very demanding, with six different necessary conditions. Lackey's critique of Hills' account focuses on its being overly demanding. One could, however, reject Hills' account of understanding without rejecting her claim that what is deficient about moral testimonial beliefs is that they lack understanding. One could then generalize Hills' account of moral testimonial beliefs into one that covered all kinds of deficient testimony. The resulting account would be one on which deficient testimony arises when a recipient of testimony lacks understanding.

One could argue against Hills' strategy and its generalization as follows, however. There appear to be many cases of testimonial knowledge where understanding is absent but the deficiency highlighted by Lackey's cases is also not present. Obviously, insofar as understanding is an epistemic good, a belief not accompanied by understanding is missing something valuable, but that is not the same as that belief's being positively deficient. One might instead think of understanding as being supererogatory in a recipient of a testimonial belief.

Suppose, for instance, that Doc from the movie Back to the Future tells Marty that driving over eighty miles per hour will activate the flux capacitor in the time machine. Marty can have knowledge in this case. Doc testifies from his competence as the inventor of the time machine, and Marty justifiably takes Doc to be an authority on his own time machine. Doc has ostensibly picked out the flux capacitor for Marty so that Marty has a minimal sense of what's being told to him that is sufficient to form a belief. What Marty obviously lacks, however, is an understanding of what the flux capacitor does and why driving fast should be required to activate it. In this case, it would be nice if Marty was able to have understanding, but it doesn't get in the way of his having knowledge. It also doesn't lead to the kind of deficiency found in Lackey's cases. In order to get that kind of deficiency, we would need Marty to parrot Doc's assertion to a third party.

One can pattern a moral case on the Back to the Future one. Suppose that God parts the heavens and tells Phil that human beings are destroying coral reefs and that this is morally wrong. Phil need not understand what makes the destruction of coral reefs wrong in order to have moral knowledge in this case. Phil doesn't have to be a divine command theorist in order to suppose that God is in a privileged position relative to the subject matter, and supposing God to be in a privileged position relative to moral truths is a more than adequate basis for coming to believe that destroying coral reefs is wrong. It would be nice for Phil to be able to have Kvanigian understanding of any moral truth that God might share with him, but, surely, it is not required in order for God and Phil to have an epistemically appropriate testimonial interaction. Thus, it does not look like Hills provides us with either a general answer to the deficient testimony puzzle or an account of deficient moral testimony in particular.

2. TESTIMONY AS A TEAM-LIKE ACHIEVEMENT

The solution I will present in the next section could be adopted by different frameworks in epistemology, but it finds its home in a kind of anti-individualistic virtue epistemology (cf. Green Reference Green2012, forthcoming a, forthcoming b). It is within this perspective that the solution is an organic, theoretically motivated extension. In this section, then, I will outline what such a view entails so as to prepare the ground for the solution of the puzzle.

Call the view to be outlined below the extended credit view. A principle central to the extended credit view is as follows.

The italicized portion of CREDIT FOR US points to the respect in which the extended credit view is anti-individualistic. On a standard virtue reliabilist picture like that of Ernest Sosa or John Greco, knowledge is a kind of achievement. It is a kind of success from ability to use Greco's terminology (cf. Greco Reference Greco2010) or a true belief that manifests competence to use Sosa's (cf. Sosa Reference Sosa2009). Though Sosa and Greco certainly do not disparage the contributions that others make to our beliefs, the kind of achievement in view is an individualistic achievement. The traditional virtue epistemologist is concerned with success that is brought about from the agent's ability and with beliefs that manifest the agent's competence. The doings of other agents are effectively background conditions that can create a friendlier or more hostile environment for one to do one's cognitive work in.

In the context of individualistic achievement, sharing credit is at most an honorific exercise devoid of real normative heft. When an Oscar-winning actress gives credit to Spencer Tracy for inspiring her as an artist, the credit attribution does not reflect a role played by Tracy in satisfying the legitimacy conditions on the kind of success an Oscar is. To have that kind of credit, Tracy would have had to have actually been involved in the acting for which an Oscar is a reward. At best, Tracy's influence on the actress is a background condition. What explains why the actress won, however, is the acting competence that she manifested in the film.

Individualistic achievements are not all the achievements there are. Teams of people also achieve things, and a team's achievement can result in the sharing of credit amongst more than one person in a substantive way. To use another Oscar example, suppose that a film wins an Oscar for special effects. The excellence of the special effects in the film may not be a manifestation of one person's competence but the competence of many. Sharing credit in this context is reflective of the distribution of the efforts and abilities that were directly involved in bringing about success. Each of the individuals on the team need not deserve an equal amount of the credit, but there could easily fail to be any one person who deserves anything like primary credit for the excellence of the special effects. It is, rather, the team as a whole that deserves primary credit, and the line between genuine credit and mere background condition is drawn with a mind to whether one truly deserves to be considered as part of the team.

The picture of testimony I want to offer is that it is a team-like achievement. It is an achievement where, in the paradigmatic cases, testifier and recipient cooperate together in an attempt to transfer information from one mind to another. This information includes the target proposition as well as the nature and strength of the attitude appropriately taken to that proposition. A full defence of the extended credit view is beyond the scope of this paper. What is important in this context is that the outline of the view be sufficiently clear for one to see how the solution given in the next section is an organic extension of conceiving of testimony as a team-like achievement.

Let us conclude this sketch of anti-individualistic virtue epistemology by highlighting two points that will be relevant in the next section. The first point is that credit attributions in the context of the achievement of a team are role relative. Achievements have legitimacy conditions. Arguably, it is this fact that gives virtue epistemology its intuitively appealing account of epistemic normativity in the first place. Legitimacy conditions for a team achievement can be applied at the level of the team as a whole as well as on the level of an individual's participation in a team. In a team's success, the credit one deserves is judged according to what the team accomplished through ability and how one's occupation of one's role contributed to the team's success from ability. Moreover, how one's role-relevant behaviour is evaluated is partly a function of how easy or difficult one made it for others to occupy their roles.

Consider as an example K, a supremely talented basketball player. K scores sixty points in his team's losing effort, and one of the reasons that his team loses is that he is a ball hog. Without spreading the shot attempts around, K's teammates cannot get into a rhythm, and K takes too many difficult shots when easier shots are available to his teammates. Consequently, the team loses. What K does with the basketball manifests a great deal of basketball skill. On a purely individualistic level of analysis, K deserves praise. He never misses an easy shot, and he makes many difficult ones. From a team perspective, however, K deserves censure. Why? When one evaluates K's performance from the perspective of the team, one has in mind not just basic basketball skills but a role that K is supposed to play within the economy of the team. Matters would be different if the coach drew up a game plan that called for K to take most of the shots. In that case, the coach would deserve the censure that otherwise should be directed at K. In order to be in a position to give the right evaluation of K, one needs to know how K's actions fit with the role he is supposed to occupy in his team's collective endeavour.

This brings us to our second point. Evaluating intra-team dynamics requires familiarity with the operative social context and its attendant conventions and norms. Suppose our ball hog K calls a play that results in his driving to the hoop and attempting to score. How are we to evaluate this particular play? Ignore K's behaviour over the course of the game, and focus only on the evaluation of the single play. K's role on the team will not completely determine our evaluation of the single play. His role is not inconsistent with his ever driving to the hoop. How we evaluate his behaviour, aside from whether he scored or not, is dependent on further details such as whether K was supposed to call the play, on what play he called, and on how the play developed. Suppose that he failed on all three counts. K called the play when he was supposed to wait for his coach to call the play. He called a play that was designed for him to wait for his teammates to set up and in which passing to one of his teammates was supposed to be the primary option. He did neither. Instead, the play call was a decoy, and he intended to drive to the hoop all along. Finally, the opposing defence positioned itself to force K not to drive. Trying to drive to the hoop when conditions should have caused one to rely more on one's teammates is one more strike against K qua member of a team effort. Even if K's drive to the hoop was a dazzling display of individualistic skill, an awareness of the context and norms in play leads one to a negative evaluation of his behaviour. The context rendered another course of action appropriate.

Having outlined the extended credit view and highlighted the relevance of one's operative social context and the role one plays in that context, the next section will provide a solution to Lackey's puzzle.

3. THE SOLUTION

My suggestion is that what is deficient about deficient testimony is teamwork. Consider the fact that, in all three cases, there is a straightforward way in which T2 could have removed the deficiency in T2's testimony. T2 could have revealed T2's grounds. Suppose that the fictional Lackey had told her friend Vivienne, ‘The paintings at the new art exhibit must be quite beautiful. My neighbour Ken is an art buff, and he's really crazy about them.’ There is nothing deficient about this assertion. Vivienne no longer feels cheated if the fictional Lackey is unable to answer follow-up questions, and even more importantly, Vivienne has no reason to think that the fictional Lackey has problematic grounds for making an assertion about the art exhibit. This latter feature is striking given that the actual Lackey claims that the fictional Lackey from the thought experiment had ‘insufficient evidence’ for her assertion, and the fictional Lackey's grounds have not changed in my alteration of the example. This suggests that the problem in the original case was not insufficient evidence but insufficient transparency regarding what undergirded the assertion in question. In the original case, there is no hedging or qualification that might signal that the testimony is not grounded in the testifier's authority to start a testimonial chain, and there is no explicit reference to such a chain either. Adding these elements erases the problem.

One might object that the following assertion still sounds inadmissible, ‘The paintings at the new art exhibit are beautiful. My neighbour Ken is an art buff, and he's really crazy about them.’ Revealing one's grounds without any hedging or indirectness in the focal assertion still sounds inappropriate. One has to ask, though, what, if any, epistemic difference there is between the first version I gave and this version.

Assertoric inappropriateness is not always a matter of whether one's assertions are epistemically well grounded. Certainly, it is customary to remark that paintings are beautiful only if one has personally viewed them. One cannot jump to the conclusion that this is because only a personal viewing could provide adequate grounds for believing or asserting something about the beauty of a painting. Claiming that the paintings ‘must be beautiful’ commits one to their being beautiful after all. Though the hedging may mitigate Vivienne's wrath if the paintings turn out not to be beautiful, it would still be appropriate for Vivienne to tell a hedging testifier that the assertion made to her was mistaken and that the testifier asserted something false. Vivienne's reprisals upon discovering that the paintings are not beautiful would make no sense if it were not the case that asserting that the paintings must be beautiful committed one to their actually being beautiful.Footnote ⁴

For an alternative example, it is customary to understate one's achievements in polite conversation, but that doesn't prevent people from communicating clearly where they think they're doing well (and to be judged by others as having imparted untruths). The understating of one's achievements is not meant to signal a defect in one's grounds. It has an alternative, non-epistemic function. Likewise, the more natural sounding phrasing of ‘the paintings must be beautiful’ signals an indirectness in the grounds of the testifier, and, in consequence, it communicates some epistemic information. It does not necessarily communicate weakness in the testifier's grounds, however. If the fictional Lackey followed up on her assertion that the paintings must be beautiful by saying, ‘I would bet you $100 that your guests will be blown away by that art exhibit’, then Vivienne shouldn't think that Lackey is being inconsistent after just having asserted that the paintings ‘must be’ beautiful. Rather, Vivienne should just think that Lackey puts quite a lot of stock in Ken's aesthetic expertise. Thus, I conclude that one can account for deficient testimony in terms of transparency despite the oddity of some un-hedged aesthetic assertions.

One can get equivalent results in the moral and oncology cases. Suppose that Sylvester had told his daughter, ‘Eating meat must be wrong. The most moral person I know is absolutely convinced that it is wrong’. Even if Sylvester's child is old enough to reason out moral matters for herself, she should regard this assertion as completely appropriate and not as providing information based on ‘the wrong kind of reason’. Granted, Sylvester's child may not be willing to form a moral belief based on the saintliness of someone she herself may not know, but it is very implausible that, in this altered case, the child will think Sylvester has made a defective assertion. Saintliness is, after all, a morally relevant property.

Eliza, in the oncology case, could have asserted, ‘Your results are in and have been analysed by my colleague. I'm afraid you have cancer.’ In this case, one does not even have to include any indirectness to find a natural sounding assertion. Lackey claims of Eliza's assertion in the original oncology case that it has an ‘inadequate basis’, but is the basis of the assertion the problem? Medical professionals outsource tasks to their colleagues all the time. If Eliza's colleague had been a radiologist giving an expert opinion on an X-ray, then there wouldn't be a problem even if Eliza herself was competent in reading X-rays. It's not clear that there should be anything wrong with a doctor making an assertion about a medical fact on the basis of a competent colleague's assessment.

Given that Eliza is the lead doctor in the case and her colleague does not possess a specialty that Eliza lacks, there is certainly something untoward about Eliza proceeding with treatment without having her own look at the test results. After all, Eliza is legally responsible for the medical care that Lucas receives at her hands. One would expect her to go the extra mile in personally verifying Lucas' diagnosis. There is nothing wrong with the alternative assertion, however. If Eliza had followed up her assertion by saying, ‘I want you to schedule your treatment with the front desk, and, if my analysis of the test results concurs with my colleague, we will begin as soon as possible’, Lucas should not be surprised, and I don't see why he should, at that moment, find any of Eliza's actions inappropriate.

If what is wrong with the assertion of T2 in each of these T1–T2–R chains is a lack of transparency rather than the nature of T2's grounds, what explains what is deficient about this lack of transparency? After all, T2 testifies from knowledge, and, if there is no in principle reason for refraining from testifying on T2's grounds, then why should T2 have to make those grounds explicit in these cases?

Recall that evaluations of individuals who are part of a team are role relative. If one looks at the assertion of T2 in isolation, then what is wrong with the assertions in the original cases will be obscured. The assertions received by R in each chain are reliable after all. The problem is that R is likely to misconstrue the role that T2 is playing in each of these chains because the fact that T2 is acting as a conduit of the good-making properties of the cognition of others is hidden. Because T2 is hiding the role that T2 is playing in the chain, T2 puts R at a two-fold disadvantage. First, R is not given a choice about whether he or she is willing to trust T1. In each of these cases, R knows T2, and they may be more comfortable trusting chains of testimony beginning with T2 than they are in trusting an unknown source. If R were uncomfortable with simply taking T2's word on the matter testified about, R has the option of following up with T2 and requesting additional grounds or explanations. R has no similar access to T1, and no awareness of what it is that R is missing. When T2's grounds are made transparent, then R is empowered to choose for him or herself whether R wants to trust T1. T2's lack of transparency impoverishes R's epistemic agency. That is one deficiency.

Second, in failing to acknowledge their role in the testimonial chain, T2 invites R to map the domain of T2's epistemic authority errantly. Lackey herself points out that the unqualified assertions made by the T2 figures in her cases could easily lead R to ask follow-up questions about their assertions to which R would expect an answer. Why? T2's unqualified assertion conveys not only the information that the proposition asserted is true but conveys a sense that T2 has information or abilities that T2 actually lacks. When deciding whether or not to take part in a collective endeavour, it is important to have a sense of what the relevant capabilities of one's prospective teammates are. The situation is analogous to picking someone for one's sports team based on his talking a good game when, in fact, he can't back it up with his play on the field. By failing to make the limitations of T2's domain relevant knowledge explicit, T2 has invited R to misjudge how R should relate to T2 as a teammate. This has two relevant consequences.

The first consequence is that, when R gets information that indicates that T2 is not capable of filling the role that is actually filled by T1, R will mistakenly come to believe that her team is inadequate, and her testimonial belief will acquire a misleading defeater. Thus, if Vivienne discovers that the fictional Lackey has no clue as to what is on display at the Art Institute, Vivienne should take her testimonial belief to be defeated, and she will thereby lose the benefit of Ken's actual expertise. This vulnerability to defeat is a second deficiency.

This second deficiency can be mitigated by further features of the context or of one's relationship to a testifier. Perhaps a testifier will have had such a good track record that one can appropriately trust her even when she fails the follow-up test, for example. Such cases do not show that vulnerability to defeat is not a real feature of these cases or not a deficiency, however. What such cases show is that the deficiency in question can be defeated.Footnote ⁵

The second consequence of failing to map T2's epistemic authority accurately is that R is in danger of trusting T2 to fill roles that T2 is not qualified to fill in future cooperative endeavours. Suppose that, in the moral case, Sylvester's child asks Sylvester whether she can order fish when they are out at a restaurant. Sylvester says, ‘Sure, honey, that's fine’. Sylvester's child is put in danger of supposing that eating fish is not in tension with the immorality of eating meat generally. Sylvester, however, is not in a good position to tell whether eating fish should be ruled out or not. Miriam, Sylvester's source, might have been qualified to play this role for Sylvester's daughter, but, because Sylvester has not been forthcoming about the role he was playing in the prior testimonial chain, his daughter might well treat him as someone qualified to play a T1 role for this further proposition. The disposition that R acquires to depend on T2 to fill roles in future endeavours for which T2 is not necessarily well suited is a third deficiency that comes about due to T2's lack of transparency.

One would expect that the dangers that attend treating a conduit of information as more than a conduit could easily lead to social norms that require one to make one's position in a chain of testimony explicit in a variety of circumstances. If a testifier flaunted these norms, then they would be doing something wrong from a team-related perspective even if, from an individualistic perspective, there were positive things to be said about their assertion such as that it was reliable and well-grounded. The testifier might even know that there was little danger of the three deficiencies I have outlined actually having a deleterious effect on the recipient and still be subject to an operative transparency norm.

Such social norms would be unsurprising in a variety of scenarios. Here are four sample scenarios in which one might expect social norms concerned with making one's intermediate position in a chain explicit.

• • When chains with very different structures could easily look the same to someone at the end of the chain.

• • When intermediate links in testimonial chains could easily fail to be discriminating.

• • When the desirable qualities of a chain originator are rare or contested.

• • In high stakes cases in general.

Let us take one final look at Lackey's cases in light of this list and ask of each case whether one might expect there to be an operative social norm in that sort of context which would require the T2 figure in the relevant chain to make their role explicit to R. In the first case, the stakes are moderately high. Vivienne is stressed out about where to take her out-of-town guests. The fictional Lackey's words are offered as a recommended course of action to remove her friend's anxiety. Aesthetics has its uncontested exemplars, but not every exhibit valued by an art buff will be pleasing by the lights of someone trying to entertain guests who may not themselves by artistically savvy. Aesthetics is a domain in which, arguably, identifying expertise can be difficult even if Lackey herself is in a good position to judge Ken's merits. Moreover, the fictional Lackey's assertion is not easy to distinguish from one in which she is relating an experience that she has had of the artistic works in question. Vivienne might well care which kind of testimonial chain is present, however. For all these reasons, one would expect there to be an operative social norm requiring Lackey to make her role in the testimonial chain explicit.

In the second case, the moral nature of the case makes it a high stakes case. If eating meat is wrong, then that means that Sylvester and his daughter have likely been guilty of doing something wrong, contributing to needless suffering, and doing it every single day. Moreover, giving up meat might be a very irksome lifestyle change not to be committed to without good reason. Moral expertise is contested and certainly the ethics of eating meat is contested. Plausibly, the moral testimony that gets passed on in society varies tremendously in its quality and the standards of those who pass it on. Moreover, the number of people who should be believed on moral matters solely in virtue of their demonstrated virtuosity à la Miriam is comparatively small.

The third case fairs better on some of these parameters. Rigid social norms that control the quality of medical care standardize information-sharing in a way such that Lucas can reasonably trust the quality of the information given him. Moreover, we are told that he picked out Eliza specifically because of her good track record. Eliza is a discriminating intermediary. She would have been unlikely to trust a colleague that was not worthy of trust. Where the requirement for transparency gains traction in this case is in the high stakes involved. Lucas has cancer. The ways of treating it are extremely uncomfortable, and any mistakes that are made in Lucas' treatment have significant consequences not only for Lucas but also for Eliza. Consequently, the high stakes involved should lead toward greater standards regarding the transparency of where the critical information so important to all parties is coming from. Compare Lucas' case to a parallel one in which a patient has a trivial condition. The intuition that there is something deficient about the case would be greatly diluted or simply absent if one were told that Lucas has, not pancreatic cancer, but an ear infection.

We are now in a position to form a hypothesis as to why one could very naturally mistake the deficiency with which we have been concerned for a problem unique to aesthetic or moral testimony. Notice that the aesthetic and moral cases above satisfied several different criteria that could easily lead to a social norm that requires transparency in a testifier, especially if that testifier is a mere conduit of information and does not possess relevant non-testimonial expertise or experiences. The social norm is over-determined in these cases. In contrast, in the oncology case, it is only the high stakes involved that push one in the direction of a transparency norm. If one finds it plausible that this contrast is not an artefact of the details of these cases but is reflective of aesthetic and moral testimony generally, one is given a possible explanation as to why one might associate the deficiency in play with aesthetic and moral testimony. The transparency norm is not unique to these domains, but it is over-determined in these domains. It is psychologically easy to confuse those two properties. If a norm is over-determined in a domain, then that provides a reason why the norm and the domain would be associated in people's minds. It would be very easy to slide from the fact that a norm is over-determined in a domain into the idea that the norm was uniquely applicable to that domain or was harder to satisfy in that domain than others. Neither of these further theses are entailed by a norm's being over-determined, however.Footnote ⁶

4. AN OBJECTION

One might object that, whatever deficiency T2's assertion has in these cases, it cannot be epistemic for all the reasons that might lead one to think that R comes to have testimonial knowledge. If T2 makes a reliable assertion, an assertion from knowledge, then surely it meets whatever epistemic standards one could possibly have. This objection gains force if one agrees with my claim that it is not the grounds of T2's assertion that makes them deficient. For Lackey, having appropriate grounds is the problem, and, since the adequacy of the grounds of one's belief is clearly an epistemic matter, then deficiency in T2's assertion is an epistemic one as well. The problem with T2's assertion, then, must be a pragmatic one. Perhaps making a flat-out assertion on secondhand knowledge, for example, invites the recipient to form false beliefs in propositions other than the proposition asserted, and, therefore, the testifier has asserted something true that has practical disutility for the recipient because of the way it tempts the recipient into falsehoods that should not be acted on. Likewise, although the testifier has asserted something true, they've done it in a way that makes it easy to lose the practical value of that true belief because one could easily lose confidence in it.Footnote ⁷

In response, it is important to remember that the epistemic and the pragmatic overlap. There being a pragmatic dimension to the deficiency in Lackey's cases is not a reason to think that the deficiency is not also epistemic. Whether the objection succeeds is a function of whether the deficiency in these cases is purely or merely pragmatic. An example of something that has mere pragmatic value is the belief that one is attractive. The belief that one is attractive makes one more likely to make a good impression on potential mates whether it is true or not. Believing that one is bad at math has pragmatic disvalue whether it is true or not. In contrast, believing that one can use one's martial arts training to disarm a mugger without harm coming to oneself or one's loved ones is a belief that has a great deal of pragmatic value if it is true and a great deal of pragmatic disvalue if it is false. Pragmatic value in this last case should not be detached from the epistemic properties of the belief in question.

In light of the fact that the epistemic and the pragmatic overlap, consider what a testifier does when he asserts that something is true. By asserting p, a testifier takes some epistemic responsibility for the p-beliefs that others form on the basis of his act of assertion. True, he might not care about this epistemic responsibility, or, as in the case of a liar, he might attach value to violating that responsibility. Nonetheless, in asserting that p is so, a testifier offers p to the recipient as something that should be believed and thus comes to bear some epistemic responsibility for the recipient's belief should the recipient accept the testifier's word.

The intersection of epistemic responsibility and assertion is why there is a literature debating what the norm of assertion is. If there was no epistemic responsibility incurred by asserting something, then it would not make sense to claim that assertion has a norm that should be honoured. If it's right that assertion implies epistemic responsibility, then one has to ask why it would be that epistemic responsibility should attach itself only to the proposition asserted and not over other propositions that a recipient of testimony could be reasonably expected to infer from one's assertion. I see no reason, if assertion has a norm, to think that conversational implicature shouldn't have a norm as well.Footnote ⁸ If, however, the norm of assertion is just that in virtue of which we judge whether a testifier has fulfilled an epistemic responsibility that he bears qua testifier, then there being a parallel norm of conversational implicature would be a matter of epistemic responsibility as well. Epistemic responsibilities, however, are not merely or purely pragmatic.

Furthermore, each of the three deficiencies that I identify in Lackey's cases are epistemic, even if they have pragmatic import as well. Taking away one's opportunity to decide whether one trusts the ultimate source of information one receives is a blow to one's epistemic agency. Vulnerability to defeat is an epistemic limitation of a belief. A vulnerability to forming future beliefs through unreliable methods is an epistemic problem. After all, if a memory belief lacks a sign of what faculty produced it, we would not hesitate to label this feature of the memorial belief an epistemic property, even if it also had pragmatic import. The same goes for other non-testimonial beliefs that are vulnerable to defeat or that might lead to unsafe methods of future belief formation. Thus, I take the objection discussed in this section to be off the mark.

In conclusion, in this paper, I have used a perspective on which testimony is a team-like activity to develop an account of what deficient testimonial knowledge consists in. Deficient testimony is a matter of deficient teamwork. It occurs when a testifier acting solely or primarily as a conduit of information fails to make his or her role in the activity of the team as a whole transparent when that is what good teamwork would have demanded. This lack of transparency impoverishes a recipient's epistemic agency, makes a recipient's testimonial belief vulnerable to defeat, and promotes dispositions in the recipient that could lead to further poorly formed testimonial beliefs. One would expect there to be social norms designed to mitigate this deficiency in a variety of scenario types, and it is eminently plausible that such a social norm would be operative in many of the contexts in which aesthetic and moral testimony is given. This could explain why it is often thought that moral and aesthetic testimony have special restrictions that apply to them despite the fact that Lackey seems right to argue that deficient testimonial knowledge is a generalizable phenomenon.