:: Article

on testimony

Jennifer Lackey interviewed by Richard Marshall.

[Photo by Rose Lenehan]
Jennifer Lackey is the rootin’ tootin’ jive falutin’ philosopher of testimony and social epistemology. She broods on summative and non-summative approaches to social epistemology, on how much this breaks with traditional epistemology, on collective responsibility, on the epistemology of testimony and the difference between a Reidian and Humean approach, on learning from words, on whether memory is a generative epistemic source, on whether infants produce a counter-example to reductionism in the epistemology of testimology, on whether speakers or listeners have to be reliable and on disagreement between peers. All in all, this is a burning jive along the wild epistemic highway. Wham Bam Thank U Ma’am!

3:AM: What made you become a philosopher? Was it something you always wanted to be or did it creep up on you?

Jennifer Lackey: From the time I was a very young child, I had philosophical thoughts, but didn’t conceive of them this way. I remember being as young as three and struggling with the concept of death or with finding meaning in things that seemed to have no significance beyond my own pleasure, such as my prized sticker collection. These thoughts became more developed when I was in elementary school and began reading literature with philosophical content, such as short stories and novels by Dostoyevsky and Tolstoy. I began reading philosophy in a humanities class in high school, though it still wasn’t clear to me that I wanted to do this for a living. When I went to college, I was a declared psychology major, and took both psychology and philosophy classes my first semester. This led to an epiphany of sorts, as I immediately discovered that the issues I thought I would be grappling with in psychology turned out to instead be philosophical ones. From that point onward, I knew that philosophy was for me, and cultivated this in a myriad of ways before graduate school, from taking just about every course that the department offered to working through my moral and political views through social activism.

3:AM: You’re a leading figure in social or collective epistemology. There are various positions one can take under that term, so there’s what’s called a summative and a non-summative approach isn’t there? What’s the difference between these two positions?

JL: The distinction between summative and non-summative approaches in the collective epistemology literature applies to a broad range of phenomena, including group belief, group justification, group knowledge, and group testimony. According to summativism, a group’s state can be understood in the sense that all or some members of the group are, or would be, in that state. So, for instance, on a summative account of group belief, a group’s believing a proposition amounts to all or some members of the group believing that proposition. On such a view of group testimony, a group’s testifying to a proposition means that all or some members of the group would testify to that proposition were the relevant opportunity to arise. For summativists, then, collective states can be understood entirely in terms of the states of individuals.

In contrast, non-summativists claim that a group’s state cannot be understood in this way. Instead, the group itself is the bearer of the state, where this is something over and above, or otherwise distinct from, the states of the individual members. Such views in general are supported by divergence arguments, which purport to establish that phenomena at the group level can diverge from what is happening at the individual level among the group’s members. For instance, the divergence arguments with respect to group belief hold that a group can be properly said to believe a proposition even if not a single one of its members believes it. This might happen, for example, when a department agrees to put forward a candidate as the best applicant for admission to its Ph.D. program, despite the fact that not a single one of its members actually believes this is correct; instead, they all think that this is the candidate who is most likely to be approved by the administration. In such a case, non-summative accounts of group belief take this to be a good reason to accept that the department itself, rather than any particular individual(s), believes this proposition. For non-summativists, then, a new, collective epistemology is needed to understand the states of collective entities.

3:AM: So where are you on this?

JL: There are important differences between my view of group belief and, say, my view of group testimony, but, broadly speaking, much of my work in collective epistemology aims to develop reductionist views of collective phenomena. The central idea is that collective states can be understood entirely in terms of the states of individuals and the relations between them. This does not mean, however, that the states of groups with a particular content are merely the summation of individual states with the same content. For instance, I argue that a group can testify that p, even when not a single member of the group testifies, or would testify, that p. This might happen when group members each enter data into a system that is then aggregated and conveyed by a spokesperson. The individual members would not be able to testify that p in such a case because they aren’t even aware of this aggregated conclusion. Nevertheless, the only thing we need to use, in explaining and evaluating the testimony of the group, is an epistemology of individual testimony. When it comes to belief, justification, and knowledge at the collective level, I include a summativist constraint, which requires that at least some of the relevant members possess the state in question, e.g., belief that p, knowledge that p. But this does not exhaust collective epistemology, as there are relations between group members and epistemic norms that apply to individuals qua group members that go beyond what is found in individual epistemology. For instance, given the structure of a certain group and the power relations among its members, there might be distinctive epistemic norms requiring that individuals qua group members share information, where these norms do not apply to the same members qua individual agents.

3:AM: How much of this approach would be acceptable to traditional epistemology and how much is a radical break?

JL: One of the goals of my project in collective epistemology is to determine how far we can get understanding and explaining group phenomena using only the resources of traditional, individual epistemology. And my conclusion is that we can get pretty far, but not all the way. For instance, our best theories of individual belief will, in many ways, ground the account of group belief that I offer. But there are questions that arise only at the group level, such as which members of a group need to believe that p in order for the group to believe that p. The same is true of group justification and knowledge: understanding the justification of individual beliefs will lay the foundation for the view of group justification that I defend, but when evidence is available to a group as a whole and what renders a group collectively rational or irrational will require answers unique to collective epistemology.

3:AM: Social epistemology has been influential on things like legal doctrines of collective knowledge hasn’t it? Where do you stand on this kind of idea, that even though individuals in a firm, say, don’t have certain knowledge, the firm as a collective does and so should be treated as a weird sort of epistemic agent?

JL: I agree that there might be instances in which we want to hold a collective responsible for actions grounded in knowledge that no single individual member possesses. But I do not think that it is necessary to attribute knowledge to the group that no member has in order to do so. Just as we can hold an individual responsible for what she should have known, so, too, can we hold a group responsible for what it should have known. This is highly relevant to the collective knowledge doctrine in the law, since one of the main reasons for its existence is to discourage the deliberate compartmentalization of knowledge by groups in an effort to avoid culpability. For instance, suppose that a corporation sets up its structure so that one set of employees is privy to proposition p1, another set is privy to a different proposition, p2, and a still different set is privy to p3. The corporation does this precisely to avoid having knowledge of the conjunction attributed to it, which is central to the satisfaction of the knowledge requirement needed to establish mens rea. According to the collective knowledge doctrine, the knowledge of the individual employees can be added or aggregated and then properly attributed to the bank itself. Thus, even if no single individual knows p1 & p2 &p3, such a doctrine maintains that the bank itself does. On my view, however, if a group deliberately avoids having knowledge that p, it can still be held accountable for what it should have known, even if it gets away with not knowing it. Moreover, what a group should have known will be grounded in what the individual members know, plus the individual and collective epistemic norms they are bound by.

3:AM: Presumably then you have situations where groups know stuff that might be disbelieved by every individual in it? So then we’d have knowledge that is held by things (banks for example) that we don’t normally think of as being literally capable of mental thoughts. Are you happy with that?

JL: While it is not uncommon for philosophical and legal views to maintain that groups can be in states, such as that of believing or knowing, that no individual member is in, the summativist constraint that I include on group belief, justification, and knowledge denies this. Instead, I argue that a group can accept or have a public view that p, where no member of the group believes or knows that p. But this is not at all surprising, as this can happen at the individual level as well. A judge, for instance, might accept or have a professional view that a defendant is innocent, despite the fact that she believes that the defendant is guilty.

3:AM: You’re a leading thinker in the related epistemological topic of testimony. You contrast Reid with Hume to establish the importance of this area of social epistemology. Their difference is about reductionism or non-reductionism isn’t it? Can you start by explaining what the philosophy of testimony involves?

JL: One of the central questions in the epistemology of testimony is how, exactly, hearers acquire justified beliefs from the testimony of speakers. Answers to this question have traditionally fallen into one of two camps: non-reductionism or reductionism. According to non-reductionists—whose historical roots are typically traced to the work of Thomas Reid—testimony is a basic source of justification, on an epistemic par with sense perception, memory, inference, and the like. Given this, non-reductionists maintain that, so as long as there are no relevant undefeated defeaters, hearers can be justified in accepting what they are told merely on the basis of the testimony of speakers.

In contrast to non-reductionism, reductionists—whose historical roots are standardly traced to the work of David Hume—maintain that, in addition to the absence of undefeated defeaters, hearers must also possess non-testimonally based positive reasons in order to be justified in accepting the testimony of speakers. These reasons are typically the result of induction: for instance, hearers observe a general conformity between reports and the corresponding facts and, with the assistance of memory and reason, they inductively infer that certain speakers, contexts, or types of reports are reliable sources of information. In this way, the justification of testimony is reduced to the justification for sense perception, memory, and inductive inference.

There are two different versions of reductionism. According to global reductionism, the justification of testimony as a source of belief reduces to the justification of sense perception, memory, and inductive inference. Thus, in order to be justified in accepting the testimony of speakers, hearers must possess non-testimonially based positive reasons for believing that testimony is generally reliable. According to local reductionism, which is the more widely accepted of the two versions, the justification of each instance of testimony reduces to the justification of instances of sense perception, memory, and inductive inference. So, in order to be justified in accepting the testimony of speakers, hearers must have non-testimonially based positive reasons for accepting the particular report in question.

3:AM: Are you more with Reid than Hume?

JL: My view incorporates aspects that are both Reidian and Humean in that I argue that the justification of testimonial beliefs is neither globally nor locally reducible to justification involving more basic epistemic sources, but I also require that hearers have positive reasons for justifiedly accepting the testimony of speakers.

I reject reductionism because the possession of epistemically good positive reasons by a hearer is not sufficient for accepting a speaker’s testimony with justification. This is made clear by cases where the reasons one has for trusting a source are nested within the reasons that one has for trusting another source. For instance, suppose that A has excellent epistemic reasons for believing B to be a highly reliable source of information on a wide range of topics and that B tells A that C is a highly trustworthy person, especially when it comes to information regarding wild birds. Because of this, A believes C when she reports that albatrosses, not condors, have the largest wingspan among wild birds. Suppose further that it turns out that while B is an epistemically excellent source of information, B was incorrect on this particular occasion: C is, in fact, a highly incompetent and insincere speaker, especially on the topic of wild birds. Moreover, though C is correct in the report about albatrosses, C came to hold this belief merely on the basis of wishful thinking. Here, it seems clear that A does not justifiedly believe that albatrosses have the largest wingspan among wild birds on the basis of C’s report. For even though B’s testimony provides A with excellent positive reasons for accepting the report in question, C is not only a generally unreliable speaker, she is also reporting a belief which, though true, fails to be reliably produced. Because of this, the testimony that C offers to A fails to be appropriately truth-conducive, thereby preventing it from leading to justified belief for A. What this case reveals is that the possession of positive reasons on behalf of a speaker’s report, even objectively excellent ones, does not necessarily put one in contact with testimony that is truth-conducive and, thus, the justification of testimonial beliefs cannot simply be reduced to the justification provided via sense perception, memory, and inductive inference. In this sense, then, my view is Reidian.

I also argue, however, that accepting testimony without any epistemically relevant positive reasons is incompatible with epistemic justification. I first show that there are at least three classes of inductively based positive reasons that are available to epistemic agents for distinguishing between reliable and unreliable testimony. The first class includes criteria for individuating epistemically reliable contexts and contextual features. For instance, one may take a less critical attitude in the context of an astronomy lecture or a National Geographic report than one does in the context of an astrology lecture or a National Enquirer report. The second broad class of positive reasons includes criteria for distinguishing between different kinds of reports. So, for example, a hearer may quite reasonably take an uncritical stance when a speaker is reporting the time of day, her name, what she had for dinner, and so on, but may take a more critical stance when receiving a speaker’s testimony about political matters, the achievements of her children, alien encounters, UFO sightings, and the like. The third class includes criteria for identifying epistemically reliable speakers. For instance, one may have accumulated inductive evidence for believing that accountants tend to be reliable sources of information about taxes, while politicians in the middle of their campaigns tend to be unreliable sources of information about the characters of their political opponents.

Once we are aware of all of the different kinds of reasons that can guide our acceptance of testimony, we can test the plausibility of non-reductionism by envisaging a scenario in which a recipient of testimony does not have any of them. So, suppose that an average human being is taking a walk and sees an alien drop a book that appears to be written in English and looks like what we on Earth would call a diary. After reading the first sentence of the book, the human forms the corresponding belief that tigers have eaten some of the inhabitants of the author’s planet. It turns out that the book is a diary, the alien does communicate in English, and it is both true and reliably written in the diary that tigers have eaten some of the inhabitants of the planet in question. But since the book in question is written by an alien, the human truly has no epistemically relevant positive reasons: she has no commonsense alien-psychological theory, she has no beliefs about the general reliability of aliens as testifiers, she has no beliefs about the reliability of the author of this book, she has no beliefs about how ‘diaries’ function in this alien society, and so on. Does the human justifiedly believe that tigers have eaten some of the inhabitants of the planet in question on the basis of the alien’s diary?

I argue no, because it seems plainly irrational epistemically for the human to form the belief in question on the basis of the alien’s testimony. For, it may very well be accepted practice in alien society to be insincere and deceptive when testifying to others. Or, the language that the aliens use, though superficially indistinguishable from English, may really be Twenglish, where Twenglish uses the ‘negation’ sign for affirming a proposition. Or, ‘diaries’ in the alien society may be what we on Earth regard as science fiction, and so on. For all the human knows when she reads the book, each of these scenarios is just as likely as the possibility that this alien is a reliable testifier who speaks English. But, in the absence of any way to discriminate among these possibilities, it seems clear that the appropriate epistemic response is to withhold belief. Testimonial justification, therefore, requires that a hearer have epistemically relevant positive reasons for accepting a speaker’s report. In this sense, then, my view is Humean.

3:AM: Traditionally testimony is seen as merely transmissive but doesn’t generate new knowledge itself. You disagree don’t you?

JL: The traditional view is that testimony is a merely transmissive, rather than a generative, epistemic source, where this means that testimony can only transmit epistemic properties from one person to another and is thus incapable of generating new epistemic properties in its own right. Given this, hearers can acquire knowledge on the basis of testimony only if the speakers themselves possess the knowledge in question. The widespread acceptance of this transmission view of testimony is undoubtedly connected to its enormous intuitive appeal. For testimonial chains of speakers and hearers are often pictured as being much like bucket brigades: each person in a bucket brigade must have a bucket of water in order to pass it to the next person and, moreover, there must be at least one person who ultimately acquires the water from another source. Similarly, each speaker in the chain of transmitting knowledge must have the knowledge in question in order to pass it to the next person, and moreover, there must be at least one speaker in the chain who ultimately acquires the knowledge from another source, such as through sense perception or reason.

In my book Learning from Words: Testimony as a Source of Knowledge, I show that there are multiple ways in which hearers can acquire testimonial knowledge from speakers who do not themselves possess the knowledge in question, thereby undermining the transmission view. In particular, I argue that there are scenarios revealing that unreliable knowers may nonetheless be reliable testifiers and, accordingly, that the testimony of speakers who do not possess knowledge of a given fact can nonetheless be the source of knowledge that hearers acquire of this very fact. One such scenario involves an elementary school teacher who is a devout creationist, rejecting all of the evidence supporting evolutionary theory and thus failing to either believe or know that Modern day Homo sapiens evolved from Homo erectus. Nevertheless, she researches evolutionary theory to prepare for her biology lessons and reliably conveys this fact about Homo sapiens to her students, thereby imparting knowledge to them that she fails to possess herself. Another such scenario involves a student in the grips of reading Descartes’s First Meditation who has skeptical doubts about reality that are so strong that he can no longer be accurately described as knowing that there is a café around the corner; that is, the student has an undefeated defeater for the justification that underwrites his belief about the café. But he is still able to reliably convey the whereabouts of the café to a passerby—without communicating his skeptical doubts—resulting in the passerby acquiring knowledge that the student does not have himself. Contrary to what has been regarded as received wisdom in epistemology, then, I show that testimony is not merely a transmissive source but instead can generate new knowledge in its own right.

While the views dominant in the literature on testimony focus on the internal states of speakers—such as states of knowing and believing—I advance a theory that instead focuses on the linguistic or communicative items in testimonial exchanges—such as statements and other acts of communication. This view explains how knowledge can be acquired through the testimony of speakers, despite the fact that such speakers fail to possess the knowledge in question. The upshot of my view is that, strictly speaking, we do not learn from one another’s states of knowing—we learn from one another’s words.

3:AM: Interestingly, traditional epistemology considers memory as non-generative as well but you have written arguing that it too can be generative. Can you explain this?

JL: The traditional view is that memory, like testimony, is not a generative epistemic source. While testimony is said to be capable of only transmitting epistemic properties from one person to another, memory is thought to be capable of only preserving such properties from one time to another. So, for instance, just as it is said that a hearer cannot know that p on the basis of testimony unless the speaker from whom it was acquired herself knows that p, the thought underlying this picture of memorial knowledge is that a subject cannot know that p on the basis of memory unless she non-memorially knew that p at an earlier time.

I argue against this view by showing that a subject can memorially know that p even though she did not non-memorially know this at an earlier time. One way that this can happen is through a lack of belief at the earlier time. For instance, suppose that during his morning commute, a driver takes in more pieces of information than he actually processes at that time. When a friend later asks the driver whether construction has begun on I55, he calls to mind passing I55 on his drive to work yesterday, and correctly remembers seeing construction work being done on this freeway. Prior to the recollection of the visual image triggered by this question, however, it would not be correct to say that he believed or knew that there is construction on I55, which is evidenced by the fact that he wouldn’t have made even minor attempts to avoid it.

In place of the preservation view of memory, I argue that S knows that p on the basis of memory at T2 only if: (i) the information that p was reliably registered in S’s cognitive system at an earlier time, T1, (ii) the information that p was registered in S’s cognitive system at T1 via a source other than memory, (iii) the information that p is reliably retrieved by S at T2, and (iv) S does not have any undefeated defeaters for the belief that p at T2.

3:AM: There’s an objection to your position on non-reductionism involving infants isn’t there? You don’t think it holds though do you? Can you explain this issue?

JL: One of the central problems afflicting reductionism in the epistemology of testimony is the apparent fact that infants and small children are not cognitively capable of having the inductively based positive reasons required by this view. The argument goes like this:

1. According to reductionism, a hearer, B, is testimonially justified in believing that p on the basis of A’s report that p only if B has non-testimonially based positive reasons for accepting A’s report that p.
2. However, infants and young children lack the cognitive capacity for acquiring and possessing non-testimonially based positive reasons.
3. Therefore, infants and young children are incapable of satisfying the reductionist’s requirement.
4. Infants and young children do have testimonial justification for at least some of their beliefs.
5. Therefore, reductionism is false.

Since non-reductionism does not impose a requirement of this sort, it is thought to avoid this problem and is therefore taken to have a significant advantage over reductionism. I argue, however, that if this objection undermines reductionism, then a variant of it similarly undermines non-reductionism. I first distinguish two ways that a condition might be satisfied:

Trivial Satisfaction: If X does not φ merely because X does not have the capacity to φ, then X has trivially satisfied the no- φing condition.
Substantive Satisfaction: If X has the capacity to φ and does not φ, then X has substantively satisfied the no- φing condition.

I then argue that the non-reductionist’s no-defeater condition has to be understood as requiring substantive satisfaction, or it fails to have any epistemic significance. But if infants and young children are not capable of having positive reasons, then they are also incapable of having negative reasons, which constitute many of the defeaters relevant to defeating knowledge. Thus, I argue that if reductionism has a problem with infants and young children, then non-reductionism has the following version of this problem:

1*. According to non-reductionism, a hearer, B, is testimonially justified in believing that p on the basis of A’s report that p only if B has substantively satisfied the no-undefeated-defeater condition.
2*. However, infants and young children lack the cognitive capacity to substantively satisfy the no-undefeated-defeater condition.
3*. Therefore, infants and young children are incapable of satisfying the non-reductionist’s requirement.
4*. Infants and young children do have testimonial justification for at least some of their beliefs.
5*. Therefore, non-reductionism is false.

I conclude that considerations about the cognitive capacities of infants and small children do not effectively discriminate between these two competing theories of testimonial justification.

3:AM: Does your position shift reliability onto the reliability of the speaker rather than the believer?

JL: I defend what I call a dualist view of the epistemology of testimony that emphasizes the work that needs to be done by both the speaker and the hearer in a testimonial exchange. Reductionists, on the one hand, focus entirely on the hearer in a testimonial exchange. For in order for testimonial justification to be reduced to the justification of perception, memory, and inference, all of the epistemic work needs to be shouldered by the hearer since it is precisely her positive reasons that are supposed to provide the reductive base. But no matter how excellent a hearer’s positive reasons are on behalf of an instance of testimony, a speaker may still be offering a report that is thoroughly unreliable. Because of this, an adequate account of testimonial justification must include a condition requiring that the testimony in question is reliable or otherwise properly truth-conducive. Non-reductionists, on the other hand, focus on the work that needs to be done by the speaker in a testimonial exchange by requiring knowledge, but neglect the positive contribution that a hearer needs to make. Specifically, they mistakenly assume that the hearer merely has to satisfy the no-defeater condition. However, no matter how reliable a speaker’s testimony is, this cannot by itself make it rational for a hearer to accept her report. For this, the hearer needs to have some epistemically relevant positive reasons on behalf of the testimony in question. Thus, I claim that the positive epistemic work of testimonial beliefs can be shouldered neither exclusively by the hearer nor by the speaker. While I emphasize the necessity of the reliability of the speaker’s testimony, the hearer also has to be a properly functioning recipient of testimony, where having relevant positive reasons is one component of this.

3:AM: Do you also think that testimony justifies us asserting things that are true rather than things we know? Does this allow us to worry less about Gettier? In fact, doesn’t this reduce the issue of knowledge, an issue that has been the centre piece of orthodox epistemological work? Testimony seems a very radical change of direction doesn’t it?

JL: My view holds that hearers can acquire knowledge from the testimony of speakers, even when the speakers lack the knowledge in question. But I require that the testimony be reliably produced or otherwise truth conducive, so the mere truth of the assertion is not enough for a hearer to acquire testimonial knowledge. So while I do argue that we need to shift our attention away from what speakers believe or know toward what they say, more than the mere truth of what they assert is relevant.

Nevertheless, in saying that speaker-knowledge is not necessary for hearer-knowledge, there is a sense in which I am minimizing the relevance of knowledge. Moreover, I argue against the widely accepted Knowledge Norm of Assertion and defend instead a norm where what is required for epistemically proper assertion is that it is reasonable for the speaker to believe the asserted proposition. This means that on my view, assertions grounded in reasonable belief, even if Gettiered, will still be epistemically proper, and this is because the asserter is not subject to any sort of criticism in such cases. Here, then, we are allowed to worry less about Gettier.

But hearers can still fail to acquire knowledge via testimony because of Gettier worries. Suppose, for instance, that one reliable speaker is surrounded by many unreliable ones, and through nothing but good luck you happen to ask the reliable one for directions. You might end up with a justified true belief that fails to be knowledge for reasons very similar to those operative in barn façade cases.

3:AM: Linked to this are issues related to disagreement amongst peers. Can you explain why, and why numbers matter?

JL: In the epistemology of disagreement, there is widespread consensus that when you are disagreeing with multiple peers, you do not need to revise your belief in light of more than one of them if there is dependence among them. So, for instance, if I believe that tomatoes are a fruit and two people tell me that they are not, I should count this as only one instance of disagreement if one of their beliefs is dependent on the other’s. But I argue that there is no interpretation of this thesis that turns out to be true. When the dependence in question is only partial in a case of peer disagreement—as happens when, for instance, I partly depend on your testimony but also partly depend on firsthand evidence for believing that p—additional evidence relevant to the belief in question can be possessed by the dependent party, thereby necessitating doxastic revision that goes beyond that required by the original source’s belief. When the dependence is complete but both agents are autonomous—as happens when a normally functioning adult depends on another for information—additional epistemic force can be brought to the disagreement in various forms, such as through the hearer monitoring the incoming information for defeaters, possessing beliefs about the reliability and trustworthiness of the source, and bearing the responsibility of offering a flat out assertion. Finally, when the dependence in question is complete but non-autonomous—as happens when a child merely parrots what her parent says—then the parties to the disagreement are not epistemic peers, and are thus not subsumed by the thesis in the first place. Thus, the amount of doxastic revision required in the face of peer disagreement does not track how independent your peers’ beliefs are from one another.

So what does such belief revision track? According to the justificationist account I have developed, the epistemic power of an instance of peer disagreement, or lack thereof, depends on the degree of justified confidence with which the belief in question is held. So, for instance, if a belief is held with a very high degree of justified confidence, then no doxastic revision may be required in the face of peer disagreement, while substantial doxastic revision may be required if such a belief is held with a very low degree of justified confidence. What I argue is that the emphasis on independence ought to be replaced with a focus on the overall justificatory status of the beliefs involved in peer disagreement, thereby providing additional support for a broadly justificationist approach to the epistemology of disagreement.

3:AM: And finally, for the readers here at 3:AM looking for further testimonies of your philosophical world, are there five books (other than your own which we’ll all be dashing away to read straight after reading this) you could recommend?

JL: The book that ignited my interest in testimony is C.A.J. Coady’s Testimony: A Philosophical Study (Oxford: Clarendon Press, 1992). While I disagree with much of what Coady argues in this book, it was the first book-length treatment of the topic of testimony, and it is really a fabulous introduction to the key issues in this area.

Alvin Goldman is arguably the father of social epistemology, and his Knowledge in a Social World (Oxford: Oxford University Press, 1999) contains articles addressing many of the central topics in this area of philosophy, showing both the breadth and depth of social epistemology.

Sanford Goldberg’s Relying on Others: An Essay in Epistemology (Oxford: Oxford University Press, 2010) develops the extent to which we are epistemically dependent on others, further cementing the importance of testimony in epistemology.

My current work in collective epistemology has benefited enormously from Margaret Gilbert’s On Social Facts (Princeton: Princeton University Press, 1989), which is impressively comprehensive in its consideration of issues in this area. It is an absolute must-read for anyone interested in groups and related phenomena.

The collection of essays edited by Richard Feldman and Ted A. Warfield, Disagreement (Oxford: Oxford University Press, 2010) inspired much of the recent interest in the epistemology of disagreement, including my own, and contains many of the articles at the center of the debates in this area.

Richard Marshall is still biding his time.

First published in 3:AM Magazine: Friday, May 31st, 2013.