Social Knowledge and Social Norms Peter J. Graham Professor of Philosophy & Linguistics Associate Dean, Arts & Humanities University of California, Riverside Draft: March 21, 2017 Forthcoming in: Knowledge: From Antiquity to the Present, Bloomsbury Stephen Hetherington and Markos Valaris, editors Abstract: A fundamental source of social knowledge is knowledge from testimony. Testimonial knowledge is justified true testimony-based belief that satisfies an anti-luck condition. Why are testimony-based beliefs justified? There are two main answers: reductionism and antireductionism. Reductionists believe the word of another is as such epistemically neutral; epistemic support for acceptance comes from background beliefs in favor of acceptance from other sources. Anti-reductionists deny this; testimony prima facie justifies acceptance. Behind anti-reductionism is the thought that as social creatures we reliably share accurate information. Why do testimony-based beliefs satisfy the anti-luck condition? Commonsense says it is because the speaker’s knowledge secures the hearer’s knowledge: testimony transfers knowledge, and thereby the anti-luck condition. But cases challenge both the sufficiency and the necessity of the speaker’s knowledge for testimonial knowledge. Assuming instead a safe-basis account of the anti-luck condition, it is possible to explain how testimony transfers epistemic force from the speaker to the hearer to explain how the hearer satisfies the anti-luck condition. Both justified belief and knowledge rest on the reliability of testimony. One explanation for the reliability of testimony in human life is the shared social norm of truth-telling. Key words: testimony, testimonial knowledge, knowledge transmission, testimonial justification, safety theory of knowledge, social norms, trust, John Locke, Thomas Reid, Elizabeth Fricker Social Knowledge and Social Norms 1. Introduction Many philosophers in the Twentieth Century argued that knowledge was social, for belief was social, for thought was impossible without language. A weaker view was also popular: though perhaps thought does not depend on other minds, knowledge does, for knowledge requires the ability to justify one’s belief to others according to social standards of justified belief. By the end of the Twentieth Century, however, both views fell on hard times. The current wisdom holds that neither belief nor knowledge is essentially social. Nevertheless, a good deal (if not most) of what we know we know because we believe what other people tell us. The extent of our knowledge owes a great deal to what we learn from others. Epistemology would be far from complete if it did not thoroughly address testimony. Though discussed during the Modern Period, it was not until the late Twentieth Century that testimony came to the fore.1 In this chapter I focus on testimony as a central source of knowledge.2 2. Justification vs. Knowledge (the Anti-Luck Condition) I begin with the distinction between justification and the anti-Gettier or “anti-luck” condition on propositional knowledge. I will rely on this distinction to frame our discussion. On the traditional analysis of propositional knowledge, a subject’s knowledge that P is comprised of (1) S’s belief that P, (2) the fact that P (so that S’s belief that P is true), and (3) S’s having a justification for believing that P. Gettier (1963) showed there are cases of justified true 2 belief that P that fall short of knowledge. Here is an example from Bertrand Russell: an individual may look at a clock and as a result truly believe, with justification, that the time is 2PM. But what if the clock is stopped and the individual only looked at just the “right” time? Then the individual does not know it is 2PM (despite having a justified true belief); you can’t learn the time by looking at a stopped clock. The “Gettier problem” led most epistemologists to conclude that propositional knowledge includes an additional “anti-luck” condition.3 Let’s then assume that justification is one thing and the anti-luck condition is another, for you can meet one without meeting the other. 3. What is Testimony? This is not a question about the speech act per se—viz. What is testimony as one speech act among others?—but rather a question about the process that leads to so-called “testimony-based beliefs.” The question “What is testimony?” (for our purposes) is the question “What makes a belief a testimony-based belief?” Testimonial knowledge is then testimony-based belief that meets the conditions for knowledge; testimonial knowledge is justified true testimony-based belief that satisfies the fourth, anti-luck condition, just as perceptual knowledge is justified true perceptual-belief that satisfied the fourth, anti-luck condition. Here are three easy examples of testimony-based belief. You want to know the time, so you ask a passerby. “2PM” she says. You believe her. You want to know the best time to visit Shanghai so you do a web search. The top hits tell you the second week of October has the best weather and the fewest tourists. You decide you want to know a little chemistry, so you buy a textbook. When you read about the elements, you form a series of testimony-based beliefs. These examples have three ingredients in common: 3 (1) Someone’s testimony that P is the distal cause. A belief is testimony-based because causally based or sustained (in part) on someone else’s testimony, where “someone’s testimony that P” includes assertions, sayings, tellings, even assertive conversational implicatures. (2) A hearer’s representation as of a speaker’s testimony that P is (part of) the proximal cause. The hearer must represent the speaker’s speech act as a telling or saying that P— the hearer must comprehend the speaker’s testimony as a part of the process of forming a testimony-based belief. No comprehension as of a speaker asserting that P as a normal part of the formation of a testimony-based belief, no testimony-based belief that P. (3) The psychological transition from comprehension to belief involves deference or epistemic dependence to the epistemic authority of the speaker. The hearer is “turning to” the speaker (to the passerby, to informants on the internet, to chemistry as a science) for information (for knowledge), and deferring to, or depending upon, the speaker (the sender) for the facts the hearer wants to know. Testimony-based beliefs are “secondhand” beliefs, beliefs formed through deference to, or dependence on, the word—the testimony—of another.4 Given our distinction between justification and the anti-luck condition, I divide the epistemology of testimony into two questions: (i) What is testimonial justification? When and why are testimony-based beliefs justified? 4 (ii) What is testimonial knowledge? When a hearer forms a justified true testimony-based belief, when and why does the hearer satisfy the anti-luck condition?5 But before turning to those two questions, we should first decide whether testimonial knowledge is even possible. 4. The Possibility of Knowledge Transmission John Locke is frequently quoted as claiming that it is not: For, I think, we may as rationally hope to see with other men’s eyes as to know by other men’s understandings. So much as we ourselves consider and comprehend of truth and reason, so much we possess of real and true knowledge. The floating of other men’s opinions in our brains makes us not one jot the more knowing, thought they happen to be true…. In the sciences, every one has so much as he really knows and comprehends. What he believes only, and takes upon trust, are but shreds; which, however well in the whole piece, make no considerable addition to his stock who gathers them. Such borrowed wealth, like fairy money, though it were gold in the hand from which he received it, will be but leaves and dust when it comes to use. (Essay Concerning Human Understanding, Book I, Chapter III, Section 24). People read this as denying the very possibility of testimonial knowledge, of the transmission of knowledge—gold—from one mind to another; all that results from deference and dependence is shreds and dust. 5 This strikes nearly everyone as incredible; we are deeply social creatures after all, living lives of massive mutual interdependence. Surely we learn (come to know) enormous amounts from relying on others. Locke’s British intellectual heir, David Hume, famously remarked that: …there is no species of reasoning more common, more useful, and even necessary to human life, than that which is derived from the testimony of men, and the reports of eyewitnesses and spectators. (Inquiry Concerning Human Understanding, Section 10, Paragraph 5) As opposed to Locke and in line with Hume, commonsense, it seems, asserts that knowledge from one mind to another can be, and often is, transmitted through testimony. C.A.J. Coady agrees: ‘If S knows that P…then S can bring his listeners to know that P by telling them that P’ (1992: 389). Elizabeth Fricker claims: ‘If S’s belief is knowledge, then we may allow that title to H’s belief too’ (1987: 57). And Tyler Burge asserts: ‘If one has acquired one’s belief from others in the normal way, and if the others know the proposition, then one acquires knowledge’ (1993: 477). A cursory review of the literature on this question will result in a score of similar passages. Indeed, the agreement is so widespread on this issue that it must form a piece of our folk epistemology. Though it goes without saying, these passages require qualification: the knowledge is transferred only if the hearer is also justified in forming the corresponding testimony-based belief. If someone knows that P and tells you that P, but you have every reason to believe they are either lying or mistaken, then it is far from obvious, and probably false, that you would come to know that P by believing them. 6 Is the speaker’s knowledge that P also necessary for a hearer’s testimonial knowledge that P? The answer here too seems to be in the affirmative: a hearer can acquire testimonial knowledge that P only if the speaker has knowledge that P to transmit. If the speaker doesn’t have the knowledge to transmit, how could the hearer, in depending on the speaker, acquire knowledge? Magic? Commentators agree. Robert Audi says: ‘My testimony cannot give you testimonially grounded knowledge that P without my knowing that P’ (1997: 410). Tyler Burge writes: “If a recipient depends upon interlocution for knowledge, the recipient’s knowledge depends on the source’s knowledge as well” (1993: 486). Angus Ross asserts: ‘Your telling me that P can only be said to provide me with knowledge if you know that P’ (1986: 62). And Michael Welbourne says: ‘It is necessary, if there is to be a successful process of testimonial transmission, that the speaker have knowledge to communicate’ (1983: 302). Philosophical reflection, in line with commonsense, reveals that (pace Locke) testimony transfers knowledge from one mind to another.6 5. Testimonial Justification: Reductionism vs. Anti-Reductionism This brings us back to our two questions: what is testimonial justification and what is testimonial knowledge? In this section I discuss with the first and then I discuss the second in the two sections following. Why are we justified, when we are, in believing another’s testimony? There are two opposing perspectives on this question. The so-called anti-reductionist believes that we enjoy a default epistemic right to take the assertions and testimonies of others at face value. If someone tells you that P, you have a default defeasible epistemic right or entitlement to believe that P; 7 testimony-based beliefs are default defeasibly justified. Tyler Burge (1993, 2013) calls this the “Acceptance Principle.” A person is entitled to accept a proposition that is presented as true and that is intelligible to him or her, unless there are stronger reasons not to. When Burge advances this principle, he is working within a broadly reliability view of justification (of “warrant”) where entitlement (his word for justification or warrant that does not rely on having a reason or reasons) entails the reliability of the belief-forming competence in normal conditions when functioning normally (Burge 1993, 2013; Graham forthcoming). Put very roughly, his idea is that when you take someone to have told you that P, you enjoy a prima facie defeasible epistemic entitlement to believe what they tell you, for your ability to comprehend others reliably leads to true beliefs in normal conditions when functioning normally; relying on others (in normal conditions) is a good route to truth. Thomas Reid advanced a similar view in the 18th Century. In section 24 of Chapter 6 of his Inquiry into the Human Mind (1764), he argued there were two principles of human social psychology that tallied with one another, the principles of veracity and credulity. The wise and beneficent Author of Nature, who intended that we should be social creatures, and that we should receive the greatest and most important part of our knowledge by the information of others, hath, for these purposes, implanted in our natures two principles that tally… The first…is a propensity to speak truth…so as to convey our real sentiments. This principle has a powerful operation, even in the greatest 8 liars; for where they lie once, they speak truth a hundred times. Truth is always uppermost, and is the natural issue of the mind. It requires no art or training, no inducement or temptation, but only that we yield to a natural impulse. Lying, on the contrary, is doing violence to our nature; and is never practised, even by the worst men, without some temptation. (…) Another original principle implanted in us by the Supreme Being, is a disposition to confide in the veracity of others, and to believe what they tell us…[T]he former…may be called the principle of veracity [and the latter] we shall…call this the principle of credulity. It is unlimited in children, until they meet with instances of deceit and falsehood: and it retains a very considerable degree of strength through life. The result of these paired principles of human psychology is that true information, for the most part, flows reliably from one mind to another. Or so Reid believes. Though Reid attributes this fortunate epistemic situation to the wisdom and good will of the Supreme Being, most contemporary followers of this broadly Reidian view would ground our social psychologies in our evolutionary and cultural history, cultivated in different ways by our varying social norms and traditions. I’ll touch on this issue again. Burge, on the other hand, ambitiously argues that the ground of the reliability of assertive communication does not lie in our social psychologies, but in the very nature of reason. Famously he argued that the Acceptance Principle is a priori true, for the reliability of assertive communication in normal conditions when functioning normally follows from the very nature of the expression of propositional content as a sign of reason, and the very nature of reason as a sign of truth. Regrettably space does not allow for discussion of Burge’s ambitious argument for this surprising conclusion.7 9 Without assuming anti-reductionists are “simple” reliabilists about epistemic justification (generally they’re not), let’s assume that behind the spirit of anti-reductionism is the assumption that our default defeasible entitlement involves the reliability of assertive communication in normal conditions. Not everyone agrees that the Acceptance Principle is true. Those who reject the Principle are often called reductionists. They deny that we can take it for granted, absent defeating evidence, that other people are trustworthy. Reductionists believe that to be justified in believing what another tells you, you need independent, non-essentially testimony-based, positive reasons for believing that the person is sincere and competent on the occasion, either because testimony is generally trustworthy, or they are generally trustworthy, or they are apt to be trustworthy on this occasion. The reductionist believes that another’s report that P as such is epistemically neutral on the question whether P; on its own it provides no support to believe that P. Their main intuitive motivation seems to be this: you just can’t trust other people to be as reliable as you need them to be. There is too much mendacity and honest error, too many sources for error in the chain of communication, to simply risk believing what other people tell you. Indeed, for all you know, most testimony may be misleading. To avoid being duped, either intentionally or accidentally, you need to sort out the good from the bad, and that requires positive, non-testimonial grounds for thinking your source is sincere or reliable or both. Not doing so is a recipe for objectionable gullibility.8 The structure of this debate between the reductionist and the anti-reductionist parallels an older debate over perceptual justification. The classical foundationalist held that beliefs formed through perception about the external world were not prima facie justified (experiences as such were neutral guides to external reality) but instead required for their justification background 10 supporting beliefs from more “basic” or “fundamental” beliefs about the characteristic patterns of our own consciously known inner sensations and perceptual experiences. From beliefs about the patterns of inner experience, we would then infer the characteristics and existence of objects in the external world, presumably via an inference to the best explanation. The justification for beliefs about the external world then “reduced” to the justification for beliefs about our own minds and what we can in turn infer about the external world from those. The classifical foundationalist’s opponent (often called “moderate” foundationalism) insisted instead that perceptual beliefs about the external world are justified “directly” by the perceptual experiences (perceptual representations) that cause them. There was then no need to “reduce” perceptual justification to any other form of justification; we enjoy default entitlement to take perceptual experience at face value. You should now see the parallel and the point of the labels. Reductionists about testimony think testimonial justification reduces to first-hand justification through perception, stored in memory, and extended through reasoning (for testimony as such is epistemically neutral; the fact that someone told you that P is no right on its own to believe that P), just as reductionists about perception think perceptual justification reduces to other sources (for perceptual experience as such is epistemically neutral; the fact that it perceptually appears to you as if P is no right on its own to believe that P). Anti-reductionists about testimony think testimonial justification doesn’t so reduce, but stands on its own two feet, just as antireductionists about perception think perceptual justification doesn’t reduce to other more “basic” sources of justification either.9 What is the main motivation for anti-reductionism? Here is the standard argument. First, most of our testimony-based beliefs are justified. Second, as a matter of fact we lack the 11 reductive, non-testimonial reasons required to reductively justify the extent of our reliance on testimony. The full extent of the justification for our testimony-based beliefs thus does not “reduce” to a non-testimonial basis (see especially Coady 1992). The anti-reductionist sees the reductionist alternative as overly demanding.10 Though many people find this point against the reductionist compelling, it has its detractors.11 Instead of pausing to evaluate these replies, let me rehearse instead its most plausible instance: childhood testimony. Children form justified testimony-based beliefs, even before turning two. From two years on they are deeply dependent for information on what other people tell them, especially on topics they cannot observe for themselves (Harris 2012; Harris and Lane 2014). Infants and young children have a strong and adaptive tendency to rely on testimony (Cole et al. 2012). But children lack some of the conceptual resources to formulate meta-arguments about the reliability of their interlocutors. They also lack sufficient first-hand evidence to justify the premises in those arguments, even if they can formulate them. And lastly, even if they have the evidence and can formulate the arguments, it is unlikely that they rely on that evidence and formulate those arguments; they seem to lack the executive ability. Thus, if children have justified testimony-based beliefs, but they cannot “reduce” that justification to justification from other sources, then testimony-based beliefs do not require positive, nontestimonial reasons for their justification.12 There have been two main responses to this argument in the literature. The first denies that children have justified testimony-based beliefs; since they lack the positive reasons, they lack justified beliefs (e.g. van Cleve 2006). The second denies that children lack the positive reasons—recent developmental psychology has shown us that children are smarter than we think. I simply reject the first reply. As for the second, it would take too much time to review and sort 12 through. But suffice it to say that the recent evidence from developmental psychology, in my view, shows no such thing.13 Fricker (1995) offers a unique and telling reply. She grants the case for children but then goes coherentist for adults. An Acceptance Principle is true for justified testimony-based beliefs when the hearer has little to go on by way of background support (the childhood case), but then coherentism is true when the hearer has a coherent set of beliefs to deploy to justify her reliance on testimony (the adult case). By my lights, this is just to grant anti-reductionism. Firstly, it grants the case where background belief is insufficient. Secondly, it grants the anti-reductionist argument against the distinctively reductionist requirement on background reasons. Thirdly, the existence and relevance of coherent background beliefs is fully compatible with antireductionism (see Graham 2006c). Fourthly, without embracing coherentism, the antireductionist can even explain why so many of those background beliefs are justified and so relevant to new testimony-based beliefs; those background beliefs were themselves justified by testimony, not by further reductive background beliefs (Graham 2006b, Burge 2013). A main concern of Fricker’s, it seems to me, is not the abstract issue of the structure of testimonial justification, but the feared laziness or epistemic irresponsibility that might result if a believer took the Acceptance Principle as a license to ignore relevant background beliefs, or to ignore evidence of untrustworthiness on the speaker’s part. And that concern the anti-reductionist equally shares. The reductionist thinks the Acceptance Principle is too permissive and the antireductionist thinks the alternative is just too demanding. Is there a way to settle this debate?14 Let me offer three points in favor of anti-reductionism. Firstly, as I just remarked, the Acceptance Principle is not a recommendation to ignore counterevidence; it’s not a policy or a license to 13 hand over all your thinking to what other people tell you, willy-nilly. Secondly and relatedly, the Acceptance Principle is fully compatible with hearer’s possessing, developing, and deploying filters and counter-measures. Various forms of monitoring and epistemic vigilance are not only compatible with the Acceptance Principle (compare analogous filters and monitors for perceptual justification), but often recommended by anti-reductionists (Goldberg and Henderson 2006, Henderson 2008, Graham 2010, Sperber et al. 2010). Thirdly, reductionist anxiety is nearly always driven by armchair considerations. Though not objectionable as such, one might naturally wonder what the empirical evidence suggests. Surprisingly, empirical studies suggest that “gullibility” isn’t such a bad thing, especially if our goal is to promote truth and avoid error (Michaelian 2010, Shieber 2014, Ahlstrom-Vij 2016). Some initial computer modeling reaches a similar conclusion (Zollman 2015). I think these three points, individually and collectively, squarely place the burden of proof on the reductionist going forward. This conclusion should not be surprising, once we recognize not only the extent of our reliance on other people for information, but the extent of our nature as social beings. 6. Counterexamples to Transmission I now turn to our question about testimonial knowledge: when and why does a hearer satisfy the anti-luck condition? You might think the answer to this question is easy. Since the speaker knows and thereby satisfies the anti-luck condition, the hearer has only to satisfy the justification condition to come to know. The neat thing about testimonial knowledge is that the speaker (in possessing knowledge) does the work for the hearer. Problem solved. Hence, we don’t need an independent treatment for the hearer, beyond an account of the hearer’s justification. This seems to be the point behind the commonsense view that testimony transfers knowledge. 14 Unfortunately, the issue isn’t so easily resolved. Even if the speaker has knowledge, that may not be enough for the hearer to satisfy the anti-luck condition. Here’s an example of just that possibility: HOSPITAL. A father knows his son is fine today, even though his son suffers serious health problems. The father’s mother (the son’s grandmother) is sick in the hospital. When the father visits, he tells her that her grandson is fine. But if his son were sick (or even dead) he would not tell her to not upset her. Though the father knows his son his well, his mother does not learn from him, for he would easily tell her that her grandson is fine when he is not. Relying on his testimony about her grandson’s well-being, she would easily form a false belief. (Modified from Nozick 1981)15 The hearer forms a justified true testimony-based belief, but the hearer doesn’t acquire knowledge, for in a sense the hearer’s true belief is just luckily true. Though the speaker has knowledge to transmit and the hearer (we assume) has every right to believe the speaker (and so forms a justified belief), something goes wrong. Matters get even worse for the simple commonsense answer, for it may be possible for a hearer’s justified testimony-based belief to satisfy the anti-luck condition, even though the speaker does not have knowledge to transmit, so it cannot be that the speaker’s knowledge that secures the hearer’s knowledge. Here is a variant of a much discussed case: FOSSIL. A devout creationist teaches at a public school where she must teach a section on evolutionary theory. She does not believe a word of it, but is a dedicated and 15 responsible teacher. She develops a near expert understanding based on deep reading of books and articles on evolutionary science. She even develops a deep understanding of fossils that parallels highly skilled scientifically trained expertise. On a field trip she discovers a fossil that proves that ancient humans once lived in this area (itself a surprising discovery no one knew before). Though she does not believe it, when she tells this to her students, they believe her. Because of her commitment to teaching, her exposure to evolutionary science, and her mastery of fossils, she would not say what she did unless it were true. Relying on their teacher, the schoolchildren would not easily be mistaken when coming to believe what she tells them. The children come to know something no one has ever known before. (Graham 2000b, 2006a, 2016b) As I said, variants have been much discussed (see Lackey 1999 and Carter and Nickel 2014).16 Though many people are willing to accept that knowledge transmission does not always succeed (as in HOSPITAL), people find FOSSIL much harder to accept. How can children acquire knowledge from relying on another if the sender doesn’t have knowledge to pass along? That’s like borrowing money from someone who doesn’t have any to lend. How can someone give you something that they themselves don’t have? So how can the speaker ensure that the hearer meets the anti-luck condition for knowledge if the speaker doesn’t have knowledge herself? Digging in their heels against the intuitive force of the example, people have tried three different replies. Firstly, people have argued that even though it looks like the children acquire knowledge, they really don’t, for on closer inspection the teacher has a flaw that prevents her from inducing knowledge in her pupils: she doesn’t always tell them what she believes (when it comes to evolutionary science, she tells them what the science would say, and not what she 16 believes). But this isn’t very persuasive. It would mean they never learn from their teacher (and that you can never learn anything from someone who isn’t completely honest to you about everything). And after all, isn’t it intuitive that because she relies on the science, the children are clearly positioned to acquire knowledge from her (Burge 2013)? Secondly, people have argued that even though it looks like the children form testimonybased beliefs, they are really relying on their background beliefs about the reliability of their schoolteachers and their first-hand, independently acquired knowledge of this teacher’s trackrecord for telling the truth. Their beliefs are more “first-hand” than “second-hand,” and so they are not really deferring to their teacher or epistemically depending on her expertise as opposed to their own, and so they are not really forming testimony-based beliefs, and so their knowledge isn’t testimonial knowledge after all (Audi 2006, Fricker 2006). But this isn’t very persuasive either. The children are young children; they are not nascent critical reasoners sorting out which teacher to trust and which to ignore. They are schoolchildren after all, there to learn from their teachers. If anyone learns from deferring to others, to depending on the expertise of their informants, small children do. Thirdly, people have argued that even though the children acquire knowledge from the teacher and form genuinely testimony-based beliefs, they don’t acquire testimonial knowledge. “Testimonial knowledge,” they say, “is essentially knowledge from another person’s knowledge through testimony. If the speaker doesn’t have knowledge to transmit, it is a priori analytically impossible for the hearer to acquire testimonial knowledge; testimonial knowledge is knowledge secured from someone else’s knowledge. It’s just a part of the very concept, or the very idea, of testimonial knowledge. It is what the phrase ‘testimonial knowledge’ means.” So even though 17 the children acquire knowledge through epistemic dependence on their teacher’s assertion (they form a testimony-based belief that is knowledge), they don’t acquire testimonial knowledge. I’ve never found this reply very persuasive, but I’m happy to grant it. I shall not, that is, dispute about a word. The children then acquire testimonial knowledge* (testimony-based belief that is knowledge) but not testimonial knowledge (knowledge that a priori analytically entails knowledge from a speaker’s knowledge). Though all testimonial knowledge is also testimonial knowledge*, not all testimonial knowledge* is testimonial knowledge. But having just made this linguistic concession, I shall ignore it. For the interesting, worthwhile category of inquiry is not the narrower category of testimonial knowledge, but the broader category of testimonial knowledge*. That, I believe, is the category we should strive to understand. I’ll give a brief argument for this at the end of the next section. We still need an answer to our second question: how and why do hearers satisfy the anti-luck condition when forming justified true testimony-based beliefs? The answer from reflection on commonsense—that the speaker knows and so the speaker, in satisfying the anti-luck condition, takes care of for the hearer—didn’t work. We need to dig a little deeper. 7. A Safe-Basis Account of Testimonial Knowledge Maybe we should choose a proposed anti-luck condition and see where that leads. Since Gettier first published his short article in 1963 there has been no shortage of attempts to articulate the correct “fourth” or “anti-luck” condition on propositional knowledge. The issue is still a matter of controversy, well over fifty years later. Even so, a handful of leading contenders have emerged: the “no-essential falsehoods” account, the “defeasibility” account, the Dretske-Nozick “sensitivity” account (Dretske 1971; Nozick 1981), and the “safety” account (proposed by 18 Luper-Foy (1984) and Sosa (1999) and then taken up by Pritchard (2005), among others). Though the points I am about to make are compatible with both the sensitivity and safety accounts, for reasons that need not detain us, I prefer to work with the safety account over sensitivity. And the other accounts, to my mind, tend to over-intellectualize propositional knowledge; they start at the wrong end of the spectrum. But that’s a debate for another day. In the rest of this section I will explain the safety account and then put it to work in answering our question, to see where it leads. According to the safety account, propositional knowledge that P is (justified, true) belief that P held on a safe basis. When is a true belief that P held on a safe basis? When holding the belief that P on that basis, you would not easily be mistaken. When would you not easily be mistaken? This is usually glossed in terms of possible worlds: a true belief that P is held on a safe basis just in case in all nearby possible worlds, if one forms a belief in one of those possible worlds on the same basis as in the actual world, then one forms a true belief in that world.17 How does the safe-basis account apply to our cases? Take HOSPITAL. The father knows his son his healthy; he believes on a safe-basis (perception, testimony, background knowledge, etc.) so that he would not easily be mistaken. But his mother’s belief that her grandson is healthy (from her son’s testimony) is not formed on a safe basis, for there is a nearby world where she relies on her son’s testimony to the effect that her grandson is fine but forms a false belief. That’s why she doesn’t acquire knowledge from him, even though he knows his son (her grandson) is fine.18 Now take FOSSIL. The schoolteacher relies on a safe-basis (careful observation, a sophisticated understanding of the contents of evolutionary theory and the fossil record, etc.) to reach the conclusion—a conclusion that she “accepts” but because of her personal convictions 19 she does not believe—that ancient humans once lived in this area; she relies on a safe-basis to reach a cognitive state of “acceptance” but not belief (cf. Bratman 1992). “Accepting” that conclusion, she tells the children that ancient humans once lived in the area. Young schoolchildren being young schoolchildren, they believe her. Her “acceptance” results from a safe-basis, and she wouldn’t say what she does unless formed on such a basis, and so her testimony too is itself a safe-basis for the schoolchildren. When they believe her, they too form a belief on a safe-basis. Relying on her testimony, the schoolchildren would not easily be mistaken. That’s why they come to know something, something that (as the example goes) noone has known before.19 The safe-basis account also allows us to argue that ordinary cases (where a speaker transfers knowledge) and unusual cases (like FOSSIL) fall within the same epistemic kind. Imagine an ordinary case where knowledge transfers. Why did that happen? Because the speaker believed on a safe basis, which partly explains why the speaker’s assertion is a safe basis for the hearer, which partly explains why the hearer forms a belief on a safe basis. What happened in the FOSSIL case? The speaker did not believe on a safe basis, but the speaker did “accept” on a safe basis, which then partly explains why she asserts on a safe basis, etc. The underlying “epistemic mechanics” is the same in both so-called testimonial knowledge cases and what I called testimonial knowledge* cases. But if the epistemic mechanics are the same, the epistemic kinds are the same. I conclude that the “broader” category of testimonial knowledge* is the category we want to understand if we want to understand the epistemology of testimony, especially the transmission of knowledge from one mind to another. After all, knowledge per se does not transfer from one mind to another. Suppose I have perceptual knowledge. Then I have knowledge on a safe-basis, where the basis involves 20 perception and perceptual experience. Then I tell you and you learn from me. Did I transfer my perceptual experience to you? Do you now have exactly what I have? No, not at all. You have neither my perceptual knowledge nor my perceptual warrant. But you do have knowledge, testimonial knowledge. Why? Because you formed a belief on a safe basis, because the epistemic force (the safety) of my perceptual belief was transferred through testimony to your belief. Testimony transmits epistemic force—it transmits the safety of one basis (perception) to the safety of another basis (comprehension and belief) through the safety of a medium (assertion, telling). Testimony doesn’t transmit my knowledge to you per se. That’s why the motivation for the so-called narrower category “testimonial knowledge” (as knowledge essentially derived from someone else’s knowledge) puts the wrong foot forward.20 It is not the speaker’s knowledge that secures knowledge for the hearer; it is the epistemic force supporting the speaker’s belief that secures the epistemic force supporting the hearer’s belief. The same story explains unusual cases like FOSSIL. Same epistemic mechanism, same epistemic kind. 8. The Reliability of Testimony and Social Norms But why, you might ask, is testimony reliable enough for justified uptake and knowledge? Consistent with everything else we know about ourselves, what explains the reliability of testimony? There is so much to say about this topic that I can barely begin to scratch the surface. I’ll make three quick remarks and then discuss one answer in a little more detail, just to take a stab at this important issue. Firstly, there’s always the possibility, consistent with Reid, that it is a part of our Godgiven nature. Secondly, there’s the parallel possibility, consistent with the growing literature on 21 the evolution of cooperation, that we’re born to be helpful, especially in communication. Young children are especially helpful at providing information unprompted to adults in apparent need (Tomasello 2009). For a similar reason, parents tell their children the truth out of parental concern. Thirdly, it is obviously frequently the case that, to coordinate with others, we need to tell the truth (I’ll need to tell you the correct time my flight arrives if you are going to pick me up). Another answer, consistent with these three, is that human life is governed by socially shared prescriptions—social norms—for telling the truth and providing useful information. It’s this last answer I’ll discuss in a little more detail. What’s a social norm? According to Cristina Bicchieri, a social norm is not merely a collective pattern of behavior in a group (2006, 2014, 2017). Everyone may want to stay warm when they go out, and so everyone may wear a coat during the winter, so we are all behaving the same way. But that’s not a social norm; that is a social custom, a pattern of behavior in a group where everyone acts the same way because we all have the same needs and we’ve all discovered the same solution. A social norm is not a trend or a fashion either, where people imitate a crowd or follow the leader to go along or to simply rely on what probably works. Nor is a social norm a convention, where people act the same way in a recurrent situation to coordinate their behavior, such as driving on the right-hand side of the road. According to Bicchieri, social norms are patterns of behavior that are collectively approved or disapproved in a group or population, and enforced by informal sanctions, positive and negative. Social norms arise from (1) (second-order) beliefs that we have that other people believe we ought—normatively and not just prudentially—behave a certain way (these are normative social expectations), and (2) a conditional preference to behave a certain way provided we believe that other people believe we ought to believe that way (a preference to meet 22 normative social expectations). The conditional preference exists because of the sanctions, often external but also internal. Positive sanctions include tangible rewards, praise, status, reputation, etc. Negative sanctions for censured behavior include punishments, shaming, ostracism, ridicule, etc. Often just thinking that others would disapprove can stop us in our tracks. Likewise, just thinking others would approve might lead us to conform (Pettit 1990, Graham 2015b).21 Social norms—our normative expectations and conditional preferences to conform—then strongly motivate compliance. “With social norms,” Bicchieri explains, …the normative influence is strong and plays a crucial role in driving compliance. It matters to us that most people in our reference network believe we ought to conform to a certain behavioral pattern. This point must be emphasized…[T]he social pressure to conform, expressed in social expectation that one ought to conform, is a powerful motivator. (Bicchieri 2017: 34-35) And it is the pressure to conform, enforced by sanctions, that really does a good deal of the motivational work. For left to ourselves, we often prefer not to conform. Indeed, many social norms emerge as solutions to social dilemmas, where narrow individual self-interest might lead us to behave in non-cooperative ways, such as prisoner’s dilemmas and public goods games. “With a [social] norm, there is often the temptation to transgress it—this is precisely why norms must be socially enforced” (Bicchieri 2017: 39). On the other hand, sometimes we internalize the norm, so that we positively value the behavior and might conform regardless of social expectations. “Especially with norms that are well established, norm followers tend to value what the norm stands for. An external observer 23 may be induced to think that, since people have a positive attitude toward a norm, they may obey it regardless of what others around them do” (Bicchieri 2017: 40). Even so, personal values—our personal normative attitudes—often fall short of motivating behavior (Eagly and Chaiken 1993, Fishbein 1967), so that social expectations and conditional preferences often play a much stronger role than meets the eye. What if truth-telling is a social norm? What if we believe that other people believe we ought to tell the truth, and what if we also believe there are rewards for conformity and costs for noncompliance? Then we’ll be motivated, in addition to any other motivations that might already be in place, to tell the truth. The social pressure to conform would then be a powerful motivator. We might even positively value telling the truth. Is it? Yes indeed. It’s a textbook case of a social norm: open any textbook discussion of social norms, and tell the truth will surely be on the shortlist of examples. In her contribution to a recent introductory anthology of social science, Bicchieri proclaims that “[e]ach group has its own norms, and some, like reciprocity or truth telling, [are] very general, spanning all groups.” (2014: 208) When providing examples of social norms in his book Social Action, Seumans Miller (2001) lists refrain from violence, remain faithful to one’s spouse, avoid incest, keep promises, tell the truth. Bowles and Gintis (2011) emphasize the social norm that one ought to tell the truth a number of times in their book A Cooperative Species. Philip Pettit (1990) treats truth-telling as the central case in his work on social norms. As I said, it’s a textbook case. Social norms are surely a part of the story—a very complex story—as to why testimony is as reliable as it is, not just in general but in varying domains and in different situations. Though maybe the Supreme Being didn’t implant the principle of veracity in our hearts, it has a powerful operation in our lives even so.22 24 References Adler, J. (2012), ‘Epistemological Problems of Testimony’, in E. Zalta (ed), Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/testimony-episprob/. Alhstrom-Vij, K. (2015), ‘The Social Virtue of Blind Deference’, Philosophy and Phenomenological Research, 91: 545-582. Ahlstrom-Vij, K. (2016), ‘Is There a Problem with Cognitive Outsourcing?’, Philosophical Issues, 26: 7-24. Audi, R. (1997), ‘The Place of Testimony in the Fabric of Justification and Knowledge’, American Philosophical Quarterly, 34: 405-422. Audi, R. (2006), ‘Testimony, Veracity, Credulity’, in J. Lackey and E. Sosa (eds), The Epistemology of Testimony, New York: Oxford University Press. Bachman, Z. and Graham, P. J. (forthcoming), ‘Counterexamples to Transmission’, in M. Fricker, P. Graham, D. Henderson, N. Pederson, and J. Wyatt (eds), The Routledge Handbook of Social Epistemology, Routledge. Bicchieri, C. (2006), The Grammar of Society: The Nature and Dynamics of Social Norms, New York: Cambridge University Press. Bicchieri, C. (2014), ‘Norms, Conventions, and the Power of Expectations’, in N. Cartwright and E. Montuschi (eds), The Philosophy of Social Science: A New Introduction, Oxford: Oxford University Press. Bicchieri, C. (2017), Norms in the Wild: How to Diagnose, Measure, and Change Social Norms, New York: Oxford University Press. Bowles, S. and Gintis, H. (2011), A Cooperative Species, Princeton: Princeton University Press. Bratman, M. (1994), ‘Practical Reasoning and Acceptance in a Context’, Mind, 101: 1-16. Burge, T. (1993), ‘Content Preservation’, The Philosophical Review. Burge, T. (2013), ‘Postscript: Content Preservation’, in T. Burge, Cognition through Understanding: Philosophical Papers Volume 3, Oxford: Oxford University Press. Carter, J. A. and Nickel, P. J. (2014), ‘On Testimony and Transmission’, Episteme, 11: 145-155. Coady, C. A. J. (1992), Testimony: A Philosophical Study, Oxford: Clarendon Press. Cole, C., Harris, P., and Koenig, M. (2012), ‘Entitled to Trust? Philosophical Frameworks and the Evidence from Children’, Analyse & Kritik, 2: 195-216. Dessalles, J.-L. (2007), Why We Talk: The Evolutionary Origins of Language, Oxford: Oxford University Press. Dretske, F. (1971), ‘Conclusive Reasons’, Australasian Journal of Philosophy, 49: 1-22. Eagly, A. H. and Chaiken, S. (1993), The Psychology of Attitudes, Orlando, FL: Harcourt Brace Jovanovich College Publishers. Faulkner, P. (2010), ‘Norms of Trust’, in A. Haddock, A. Millar, and D. Pritchard (eds), Social Epistemology, Oxford: Oxford University Press. Faulkner, P. (2011), Knowledge on Trust, Oxford: Oxford University Press. Fishbein, M. (1967), ‘A Consideration of Beliefs and Their Role in Attitude Measurement’, in M. Fishbein (ed), Readings in Attitude Theory and Measurement, New York: Wiley. 25 Fricker, E. (1994), ‘Against Gullibility’, in A. Chakrabarti and B. K. Matilal (eds), Knowing from Words, Dordrecht: Kluwer Academic Publishers. Fricker, E. (1995), ‘Telling and Trusting: Reductionism and Anti-Reductionism in the Epistemology of Testimony’, Mind 104: 393-411. Fricker, E. (2004), ‘Testimony: Knowing through Being Told’, in I. Niiniluoto, M. Sintonen and J. Wolenski (eds), Handbook of Epistemology, Dordrecht: Kluwer Academic Publishers. Fricker, E. (2006), ‘Second-Hand Knowledge’, Philosophy and Phenomenological Research, 73: 592-618. Fricker, E. (2015), ‘How to Make Invidious Distinctions Amongst Reliable Testifiers’, Episteme, 12: 173-202. Fricker, E. (2016), ‘Unreliable Testimony’, in B. Mclaughlin and H. Kornblith (eds), Alvin Goldman and his Critics, Oxford: Blackwell Publishers. Fricker, E. (forthcoming a), ‘Inference to the Best Explanation and the Receipt of Testimony: Testimonial Reductionism Vindicated’, in T. Poston and K. McCain (eds), Best Explanations: New Essays on Inference to the Best Explanation, Oxford: Oxford University Press. Fricker, E. (forthcoming b), ‘Norms, Constitutive and Social, and Assertion’, American Philosophical Quarterly. Gelfert, A. (2009), ‘Indefensible Middle Ground for Local Reductionism about Testimony’, Ratio 22: 170-190. Gelfert, A. (2015), A Critical Introduction to Testimony, London: Bloomsbury. Gettier, E. (1963), ‘Is Knowledge Justified True Belief?’, Analysis, 23: 121-123. Goldberg, S. (2010), Relying on Others, New York: Oxford University Press. Goldberg, S. (2015), Assertion: The Philosophical Significance of Assertoric Speech, Oxford: Oxford University Press. Goldberg, S. and Henderson, D. (2006), ‘Monitoring and Anti-Reductionism in the Epistemology of Testimony’, Philosophy and Phenomenological Research 72: 600-617. Goldman, A. (1986), Epistemology and Cognition, Cambridge, MA: Harvard University Press. Goldman, A. and Blanchard, T. (2015), ‘Social Epistemology’, in E. Zalta (ed), Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/epistemology-social/. Graham, P. J. (1997), ‘What is Testimony?’, The Philosophical Quarterly, 47: 227-232. Graham, P. J. (2000a), ‘Conveying Information’, Synthese, 123: 365-92. Graham, P. J. (2000b), ‘Transferring Knowledge’, Nous, 34: 132-52. Graham, P. J. (2000c), ‘The Reliability of Testimony’, Philosophy and Phenomenological Research, 61: 695-709. Graham, P. J. (2006a), ‘Can Testimony Generate Knowledge?’, Philosophica, 78: 105-27. Graham, P. J. (2006b), ‘Liberal Fundamentalism and Its Rivals’, in J. Lackey and E. Sosa (eds), The Epistemology of Testimony, New York: Oxford University Press. Graham, P. J. (2006c), ‘Testimonial Justification: Inferential or Non-Inferential?’, The Philosophical Quarterly, 56: 84-95. 26 Graham, P. J. (2008), ‘The Relativist Response to Skepticism’, in J. Greco (ed), The Oxford Handbook of Skepticism, New York: Oxford University Press. Graham, P. J. (2010), ‘Testimonial Entitlement and the Function of Comprehension’, in A. Haddock, A. Millar and D. Pritchard (eds), Social Epistemology, Oxford: Oxford University Press. Graham, P. J. (2012), ‘Testimony, Trust and Social Norms’, Abstracta, 6: 92-116. Graham, P. J. (2013), ‘Review of Faulkner, Knowledge on Trust’, Notre Dame Philosophical Reviews, http://ndpr.nd.edu/news/knowledge-on-trust/. Graham, P. J. (2015a), ‘Testimony as Speech Act, Testimony as Source’, in M. Mi, M. Slote and E. Sosa (eds), Moral and Intellectual Virtues in Western and Chinese Philosophy, New York: Routledge. Graham, P. J. (2015b), ‘Epistemic Normativity and Social Norms’, in J. Greco and D. Henderson (eds), Epistemic Evaluation, Oxford: Oxford University Press. Graham, P. J. (2016a), ‘Formulating Reductionism about Testimonial Warrant and the Challenge from Childhood Testimony’, Synthese, online 16 June 2016. Graham, P. J. (2016b), ‘Testimonial Knowledge: A Unified Account’, Philosophical Issues: A Supplement to Nous, 28: 172-186. Graham, P. J. (forthcoming), ‘Sincerity and the Reliability of Testimony: Burge on the A Priori Basis of Testimonial Warrant’, in E. Michaelson and A. Stokke (eds), Lying: Language, Knowledge and Ethics, Oxford: Oxford University Press. Greco, J. (2012), ‘Recent Work on Testimonial Knowledge’, American Philosophical Quarterly, 49: 15-28. Harris, P. (2012), Trusting What You are Told: How Children Learn from Others, Cambridge, MA: Harvard University Press. Harris, P. and Lane, J. (2014), ‘Infants Understand How Testimony Works’, Topoi, 33: 443-458. Henderson, D. (2008), ‘Testimonial Belief and Epistemic Competence’, Nous, 42: 190-221. Henderson, D., and Graham, P. (forthcoming a), ‘Epistemic Norms and the “Epistemic Game” they Regulate: The Basic Structured Epistemic Costs and Benefits’, American Philosophical Quarterly. Henderson, D., and Graham, P. (forthcoming b), ‘A Refined Account of the “Epistemic Game”: Epistemic Norms, Temptations and Epistemic Cooperation’, American Philosophical Quarterly. Hetherington, S. (2016), Knowledge and the Gettier Problem, Cambridge: Cambridge University Press. Hinchman, E. (2014), ‘Assurance and Warrant’, Philosopher’s Imprint, 14: 1-58. Insole, C. (2000), ‘Seeing off the Local Threat to Irreducible Knowledge by Testimony’, The Philosophical Quarterly, 50: 44-56. Johnson, C. R. (2015), ‘Testimony and the Constitutive Norm of Assertion’, International Journal of Philosophical Studies, 23: 356-375. Kelp, C. (2016), ‘Assertion: A Function First Account’, Nous, online first. 27 Kenyon, T. (2013), ‘The Informational Richness of Testimonial Contexts’, The Philosophical Quarterly, 63: 58-80. Lackey, J. (1999), ‘Testimonial Knowledge and Transmission’, The Philosophical Quarterly, 49: 471-90. Lackey, J. (2011), ‘Testimonial Knowledge’, in S. Bernecker and D. Pritchard (eds), The Routledge Handbook of Epistemology, New York: Routledge. Lyons, J. (1997), ‘Testimony, Induction, and Folk Psychology’, Australasian Journal of Philosophy, 75: 163-178. Luper-Foy, S. (1984), ‘The Epistemic Predicament: Knowledge, Nozickian Tracking, and Skepticism’, Australasian Journal of Philosophy, 62: 26–49. McDowell, J. (1994), ‘Knowledge by Hearsay’, in B. K. Matilal and A. Chakrabarti (eds), Knowing from Words, Dordrecht: Kluwer Academic Publishers. McMyler, B. (2011), Testimony, Trust, and Authority, Oxford: Oxford University Press. Mercier, H. and Sperber, D. (2011), ‘Why do Humans Reason? Arguments for an Argumentative Theory’, Behavioral and Brain Sciences, 34: 57-74. Michaelian, K. (2010), ‘In Defense of Gullibility: The Epistemology of Testimony and the Psychology of Deception Detection’, Synthese 176: 399-427. Miller, S. (2001), Social Action, Cambridge: Cambridge University Press. Moran, R. (2005), ‘Getting Told and Being Believed’, Philosopher’s Imprint, 5: 1-29. Nozick, R. (1981), Philosophical Explanations, Cambridge, MA: Harvard University Press. Owens, D. (2017), ‘Human Testimony’, in his Normativity and Control, Oxford: Oxford University Press. Pritchard, D. (2005), Epistemic Luck, Oxford: Oxford University Press. Ross, A. (1986), ‘Why Do We Believe What We Are Told?’, Ratio, 1: 69-88. Rysiew, P. (2007), ‘Beyond Words: Communication, Truthfulness, and Understanding’, Episteme, 4: 285-304. Schmitt, F. (2010), ‘The Assurance View of Testimony’, in A. Haddock, A. Millar, and D. Pritchard (eds), Social Epistemology, Oxford: Oxford University Press. Shieber, J. (2009), ‘Locke on Testimony: A Reexamination’, History of Philosophy Quarterly, 26: 21-41. Shieber, J. (2014), ‘Against Credibility’, Australasian Journal of Philosophy, 90: 1-18. Shieber, J. (2015), Testimony: A Philosophical Introduction, New York: Routledge. Shogenji, T. (2006), ‘A Defense of Reductionism about Testimonial Justification’, Nous, 40: 331-346. Sosa, E. (1999), ‘How to Defeat Opposition to Moore’, Philosophical Perspectives, 13: 137-48. Sperber, D., Clement, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, G., and Wilson, D. (2010), ‘Epistemic Vigilance’, Mind and Language, 25: 359-393. Sperber, D. (2013), ‘Speakers are Honest Because Hearers are Vigilant’, Episteme, 10: 61-71. Tomasello, M. (2009), Why We Cooperate, Cambridge, MA: The MIT Press. 28 van Cleve, J. (2006), ‘Reid on the Credit of Human Testimony’, in J. Lackey and E. Sosa (eds), The Epistemology of Testimony, New York: Oxford University Press. Welbourne, M. (1985), The Community of Knowledge, Aberdeen: Aberdeen University Press. Wright, S. (2016), ‘The Transmission of Knowledge and Justification’, Synthese, 193: 293-311. Zollman, K. J. S. (2015), ‘Modeling the Social Consequences of Testimonial Norms’, Philosophical Studies, 172: 2371-2383. 29 1 Testimony was discussed by Locke, Hume, Reid and Kant (among others), but then was for the most part marginalized until the late Twentieth Century. Michael Welbourne (1983) was often a lone voice in the 70s, later joined by Elizabeth Fricker, among others, in the 80s. Then Coady’s (1992) book and Burge’s (1993) essay stirred considerable interest in the early 90s, which led to a handful of dissertations completed by the end of the 1990s and then a flowering of interest that continues unabated, including the recent publication of two excellent textbooks (Gelfert 2015, Shieber 2015). 2 I won’t pretend to be as exhaustive or even-handed as possible—I’ll be opinionated for sure— but I’ll do my best in the space available. For other ways of orienting the literature, I recommend Fricker 2004, Lackey 2011, Adler 2012, Greco 2012, Goldman and Blanchard 2015, Gelfert 2015, and Shieber 2015. There are issues I will sideline, including disagreement, epistemic injustice, group knowledge and group testimony, whether the reliability of a process, or even the cognitive process itself, somehow interestingly “extends” to the speaker (Goldberg 2010), among other issues. And the assurance view, regrettably, will receive only the barest mention. For the assurance view see Ross 1986, Moran 2005, McMyler 2011, and Hinchman 2014. Against see Lackey 2008, Schmitt 2010, and Owens 2017. 3 For an up-to-date discussion, see Hetherington 2016. 4 For discussion of testimony as a speech act, see Graham 1997 and Graham 2015a. For more discussion of testimony-based beliefs, see Graham 2015a, 2016. 5 Not every participant to the literature sharply distinguishes these two questions, sometimes resulting in unnecessary confusion. Robert Audi, on the other hand, sharply distinguishes these two, arguing that a hearer can acquire testimonial knowledge without simultaneously acquiring a justified testimony-based belief (Audi 1997). Audi relies on a general framework where knowledge is belief-based on a reliable indicator, and justified belief is belief-based on accessible justifiers that can be cited when justifying the belief. 6 For the record, I think this representation of Locke is far from the truth. For one, in context this passage is targeting an almost blind deference to the so-called “authority of others” on matters of theological and abstract philosophical affairs, exactly the kind of topic where Locke, as a spokesperson of the Enlightenment, believed we can, and should, think for ourselves. For another, Locke also had a very different conception of knowledge than we do today. On Locke’s conception of knowledge, knowledge is the perceived agreement or disagreement among ideas (roughly, but not entirely, what we would consider knowledge of self-evident truths and what can be self-evidently deduced from self-evident truths). Given that conception of knowledge, you cannot just acquire knowledge from relying on someone who knows, for even though they have perceived the relevant agreements among ideas to reach a conclusion, it does not follow that by 30 believing their testimony you will ipso facto perceive the relevant agreement among ideas yourself. Hume probably would have agreed with Locke, for they probably shared this conception of knowledge. So even the quote from Hume is, for the opposite reason, also a little misleading. Furthermore, in line with the quote from Hume, in Book IV of the Essay Locke includes testimony as a “ground of probability,” viz. a ground for reasonable belief, what in many cases we would count as knowledge. In sum, Locke probably did not mean what most people who cite this passage take him to mean (see Shieber 2009 for more discussion). Regardless, this representation of Locke is a useful foil for commonsense, which is why I started with it in the first place. 7 See Graham forthcoming for critical exposition and discussion of Burge’s argumentation. There’s a tendency to equate Burge’s a priori defense of the Principle with the truth of the Principle, so if the Principle isn’t a priori necessary then it isn’t even true (van Cleve 1996; Faulkner 2011). Burge assumes the Principle and then asks for its basis. The Principle, as we’ve just implied, might have more than one basis: God, evolution by natural selection, human psychology, social psychology, social structures, the nature of promising and the institutions of promising, etc., or the very nature of reason itself. Burge opts for the last option. Burge’s account of its basis may fall short without the Principle falling short in any way. 8 Fricker 1994: 144-5. For other arguments, see Fricker 1987, 2004, 2016 and Lackey 2008. Lackey thinks she has a persuasive counter-example to anti-reductionism. Imagine stumbling across an apparent diary written apparently in English where it is clear to you that it was written by an alien from another planet. You possess no positive reasons for what it says, and simultaneously you possess, she says, no defeating reasons not to believe what it says. But intuitively, she thinks, you are not justified in believing what it says. Generalizing, every testimony-based belief requires non-testimonial support. I don’t agree. By my lights, a mature adult will be puzzled by the fact that it was an alien, as so is surely to have defeaters. Perrine argues along these lines: the adult in Lackey’s case “has a wealth of background knowledge that ought to make him skeptical in this case” (2014: 3236; cf. Burge 2013). On the other hand, it the recipient were a young child who could read, but had little thoughts if any about the consequences of the author’s alien status, then we’d have a case of no reasons on either side. But then it seems fine to me to say that child has a prima facie justified testimony-based belief. More on the issue of children’s testimony in the text in a moment. 9 I provide more detailed formulations of some of these issues in Graham 2006 and forthcoming. For testimony aficionados, there are two possible to frame the reductionism/antireductionism debate. One is about justification (that is how I see it), the other is about knowledge (about converting true belief into knowledge). If you see it the latter way, then you see the reductionist as saying that a true testimony-based belief, provided it is reductively justified (usually inductively by a track-record), is then knowledge: to really “reduce” the hearer’s 31 knowledge to perception, memory and induction, everything required for knowledge must “reduce” to the hearer’s track-record. Seen this way the “reductionist” rejects the commonsense idea that the hearer gets what she needs for knowledge from the speaker, and thus rejects testimonial knowledge as involving knowledge transfer of either knowledge or justification from the speaker. This “reductionist” view is then easy to dismiss with Gettier counterexamples. Lackey does this (2008: 142-159). Imagine the speaker is just right by luck, but the hearer has a pretty good track-record of evidence to justify reliance. Then intuitively the hearer too, despite the track-record, has a Gettierized justified true belief. The hearer’s track-record is then not enough to convert a true belief into knowledge. Lackey then argues that a further, anti-Gettier condition is required, involving the reliability of the speaker’s testimony. Paralleling her formulation of reductionism, she formulates “non” reductionism as the view that a true testimony-based belief, provided the hearer has no defeaters, is knowledge. This view also suffers the same fate: Gettier counterexamples are easy to formulate. She concludes that both reductionism (about knowledge) and anti-reductionism (about knowledge) are wrong for the same reason. She then formulates a “hybrid” theory, neither “reductionist” nor “non-reductionist,” for testimonial knowledge. (By these lights, any theory that sees the speaker as playing some role in meeting the anti-luck condition is a “hybrid” theory—which includes nearly everyone working on the topic, including our commonsense that the hearer’s knowledge depends on the speaker’s knowledge.) Putting all of this to one side, Lackey ends up advancing a reductionist account of testimonial justification (she would say the “rationality dimension” of justification) for she adopts a “positive reasons requirement” on testimonial knowledge, where “for each report R, the positive reasons justifying R cannot ultimately be testimonially grounded, where this means that the justificatory or epistemic chain leading up to R cannot “bottom out” in testimony” (2008: 186). Though Lackey spends some time in her book arguing that adults have plenty of positive reasons supporting testimony, she spends little time addressing whether adults satisfy this reductive requirement, which is just the requirement at issue. 10 For other arguments, see Coady 1992 and Rysiew 2007. For some discussion of these arguments, see Fricker 1995 and Graham 2000c. 11 For detractors, see Lyons 1997, van Cleve 2006, Shogenji 2006, Lackey 2008, and Kenyon 2013. Fricker is well-known for the distinction between “local” and “global” reductionism (1995, forthcoming a). She grants that testimonial justification does not “globally” reduce (individuals do not possess a reductive justification for the proposition that testimony is generally reliable, but insists that it “locally” reduces (individuals possess reductive justification for propositions of the form this speaker on this occasion is trustworthy on this topic). Many commentators, the present author included, doubt this move succeeds, for they doubt that individuals possess adequate reductive local justifications, never mind adequate reductive global justifications. See Insole 2000, Gelfert 2009, Graham 2016a. 32 12 Selective trust in special cases some of the time is one thing; reductive trust in every case is another altogether. I discuss this argument in considerable detail in Graham 2016a. 13 See Graham 2016a, section 5, for a review of the literature and references. See also Harris 2012, Harris and Lane 2014, and Cole et al. 2012. 14 Paul Faulkner (2011) claims to occupy a middle-ground. He argues that the anti-reductionist is wrong—we don’t enjoy a prima facie default right to testimony—but the reductionist is wrong too—we don’t need a reductive justification showing that testimony in general, or in the particular “local” case, is reliable. Because of the “problem of communication”—because hearers want the truth but because speakers, as rational self-interested agents, want the freedom to lie when that serves their interests—hearers need a reason in each and every case (even small children) to justifiably believe the speaker’s report; it would be epistemically irrational otherwise to believe testimony without such a reason. But if the reason is not a reductive argument showing the reliability of the speaker, what is it? The reason is the hearer’s affective trust that the speaker will prove trustworthy, where that is the normative expectation that the speaker should prove trustworthy, for the speaker and the hearer are both subject to, and have internalized, the social norm of trustworthiness (of truth telling). (More on social norms in the last section.) The speaker, then aware of the hearer’s normative expectation, will prove trustworthy (will fulfill the expectation). The hearer then possesses a mini-argument rationalizing acceptance of the speaker’s assertion: 1. I normatively expect truth telling; 2. The speaker knows this; 3. The speaker has also internalized the norm, and so will prove trustworthy; 4. So the speaker is telling the truth; 5. So it is rational to believe the speaker’s assertion. Though I agree with Faulkner about the important role social norms play in the overall epistemology of testimony, and I do not agree they tell the whole story: they are one mechanism among others underwriting the reliability of testimony, not the only one. (More on this in the last section.) Also, I think he has conflated two issues. One issue is whether hearers need to represent, in their psychology, a rationalizing argument justifying their reliance. Another issue is whether there is a problem of communication, whether the possibility of deceit shows the antireductionist cannot be right. Once social norms of truth-telling enter the picture, the problem of communication goes away. So, the case against the anti-reductionist dissolves. Why then think everyone, in each and every case, needs a mini-argument to rationalize trust, even small children? Well, if you were an internalist with a tendency to intellectualize (some would say hyper-intellectualize) the epistemology of testimony, then you’d go for that conclusion. But since I don’t, I won’t. Faulkner thinks the problem of cooperation motivates an internalist result, but it is (I believe) his internalist convictions that are doing the work, not the problem of cooperation. The structural problem of communication is one thing, the case for internalism is another. For discussion of Faulkner in more detail, see Graham 2012, 2013. 33 15 Here is another: ASTROLOGIST. Mary sometimes believes that it is raining in her village because she pulls the drapes and looks outside and sometimes because she consults an astrological table with the drapes closed. Today Mary looks outside and sees that it is raining. But when she takes your phone call and tells you that it is raining, you do not come to know that it is, for she would just as easily tell you that it is raining when it is not. Relying on her testimony about the weather, you would easily be mistaken. (Peacocke 1986) For discussion of the issue and additional counterexamples, see Graham 2000a, 2016b, and Lackey 2008. 16 The difference between FOSSIL and Lackey’s SCHOOLTEACHER (1999, 2008) example is that FOSSIL involves testimonial knowledge of a proposition that no-one has ever known before. In Lackey’s example, the schoolteacher is passing on knowledge of evolutionary theory, wellknown throughout the science. FOSSIL seems to illustrate the possibility that testimony can “generate” knowledge, viz. be the very first source of a new piece of knowledge, something that typically only occurs in perception, introspection, a priori understanding, and reasoning from those three sources. Those who think testimony cannot “generate” knowledge allow such cases like Lackey’s (they only insist on knowledge in the chain of sources) hence the need to create a case that avoids the qualification. For further discussion of these and related cases, see Audi 2006, Fricker 2006, 2015, 2016, Graham 2006a, 2016b, Faulkner 2010, Burge 2013, Carter and Nickel 2014, Wright 2016, and Bachman and Graham, forthcoming. 17 “Basis” safety is not to be confused with “belief” safety. A basis is safe just in case it leads you to a true belief in the actual world and doesn’t lead you astray in any nearby possible world. A belief is safe just in case it is true in the actual world and true in all nearby worlds. The standard example to illustrate this difference involves flipping a coin to form the belief that 14+23=37. If you believe because you flipped a coin, then you didn’t believe on a safe basis. In a nearby world the coin flip will come up the other way, and you’ll end up believing a falsehood. But the belief that 14+23=37 (on the other hand) is true in all nearby worlds where you believe it, because it is true in any world where you believe it, because (as a necessary truth) it is true in all worlds! Forming a belief on a coin flip is not a safe basis, but the belief in the mathematical proposition is a safe belief. The safety account of knowledge is a safe basis account, not a safe belief account; there are safe beliefs that fail to measure up to knowledge, but no belief is knowledge without being held on a safe basis. 18 Mary in ASTROLOGY works the same way. By looking out the window through the drapes Mary forms the belief that it is raining outside on a safe-basis. That’s why she knows it is raining. But when she tells you that it is raining, you do not form a belief on a safe-basis. She would easily tell you that it is raining outside when it is not; relying on her testimony about the weather would then easily lead you astray. That’s why you don’t acquire knowledge from her, even though she knows it is raining today. 34 19 For more discussion, detail, defense and further references on the safe-basis account of testimonial knowledge, see Graham 2016b. 20 For additional arguments for a further broadening of the epistemology of testimony, see Graham 2000a, 2015a, 2016b. 21 Among our social practices—our shared patterns of behavior—there are many kinds: customs, descriptive norms, conventions, and social norms. Thus, it is not trivially true that some of our epistemic norms are social norms (or habits, or customs), just because we have a social practice of asking for and giving reasons. For that practice might be a custom, a descriptive norm, a convention, or a social norm… The case must be made. 22 For a longer discussion of some of these issues, see my 2015b. For critical engagement, see Fricker forthcoming a. For more exploration of Bicchieri’s framework for epistemic norms as social norms, see Henderson and Graham forthcoming a, b. Faulkner’s work (2010, 2011) is clearly relevant here too (as discussed previously). The social norms account is an important alternative to both the constitutive norms account of the epistemology of testimony (for example, see Goldberg 2015 and for critical discussion see Johnson 2015) and the assurance view, for each are sensitive to the normative dimension of testimony, but locate it in other kinds of normativity, kinds that do not directly address the motivational basis for the reliability of testimony. 35