Confucian Junzi and the Ethics of Testimony — Epoche B2
Confucian trustworthiness is a condition on telling, not evidence for believing Two questions that look like one Most of what anyone believes came from somebody else's say-so. Philosophy has two distinct questions about this and they are easy to run together. The first is the hearer's question: when is a person justified in believing what she is told? The reductionist answer, associated with Hume [1] , is that the justification must come from something other than testimony—from the hearer's own observation of when speakers of that kind have turned out to be right. The anti-reductionist answer, associated with Thomas Reid [2] and defended at length by C. A. J. Coady [3] , is that testimony is a basic source: the reductive programme cannot even start, because the evidence a hearer would need in order to check testimony's track record is itself almost entirely testimonial. The second is the speaker's question: what does a person undertake when she tells someone something? This is a question about a kind of act, not about a kind of evidence. This essay argues that the Confucian material on xin , trustworthiness, answers the second question and not the first, and that the difference matters. On the reading defended here, xin in the Analects is a condition on telling working at all [4] , not a mark from which a hearer estimates how likely a report is to be true. Placing it in contemporary epistemology as a further indicator of a speaker's truth-conduciveness gets it exactly backwards. The claim can be denied, and the objection that most nearly refutes it is set out below. The complaint that has no target A complaint often made about contemporary work on testimony is that it treats speakers as instruments and ignores their moral character. Whatever its merits elsewhere, it does not describe this literature, and the point has to be made before any comparison is worth attempting. Coady's book turns on a presumption of truthfulness which he treats as constitutive of the practice of assertion rather than as an empirical generalisation. Jennifer Lackey's Learning from Words makes the speaker's contribution one of two indispensable components [5] , against views on which the hearer's side does all the work. And Richard Moran's account [6] , which has shaped the last twenty years of the subject, is built entirely on the moral relation: when I tell you something, I freely assume responsibility for its truth, and my having done so gives you a reason to believe of a kind that no piece of evidence about my reliability could give. On the assurance view, being told is not like reading an instrument. So the moral dimension is not an omission in the field; it is the field's main battleground. Any comparison with Confucius has to begin from that, or it will be arguing with nobody. What the Analects says about xin Turn to the text. At Analects 2.22 Confucius says he does not know what a person without xin is good for, and glosses this with an image drawn from carriages: a large cart without its crossbar pin, a small cart without its collar-bar, cannot be made to go. The image repays attention, because it is not an image of evidence. A crossbar pin is not a sign from which one predicts that the cart will move. It is the part that couples the shaft to the yoke. Without it the cart does not move unreliably; it does not move. Confucius is not saying that untrustworthy people are a poor bet. He is saying that the mechanism by which speech does anything is disengaged. The rest of the text is consistent with this reading. At 1.4 Zengzi's daily self-examination asks whether he has been faithful in his dealings and trustworthy with friends: xin appears in a list of things one owes, alongside loyalty and diligence in what one has been taught. At 1.7 Zixia counts a man learned who is trustworthy in his word to friends, even if others say he has not studied. In every case xin is a property of a person considered as one who acts and speaks among others. It is never introduced as something a hearer inspects. Two complications the tradition supplies itself Anyone tempted to convert the junzi into a maximally reliable informant should read two further passages, because the tradition blocks the conversion. At Analects 13.20 Confucius grades the scholar-official. Asked repeatedly to describe lower grades, he arrives at the man whose word is invariably kept and whose actions are invariably carried through—and calls him the lowest sort, a stubborn small man. Invariable word-keeping is a floor, not a ceiling. Mencius makes the reason explicit at 4B11 [7] : the great person does not insist that words be kept or actions carried through, but attends to what is right. This is the same structure as Analects 4.10, where the junzi has no fixed for and no fixed against but goes with rightness. Xin is subordinate to yi . The consequence is decisive for how the comparison must run. If xin were an epistemic virtue in the reliabilist sense, subordinating it to anything would be a defect: a reporting mechanism that may be overridden by other considerations is a worse mechanism. That the tradition regards the subordination as an improvement shows that xin was never being recommended as a truth-tracking disposition. It is being recommended as what one owes in speech, and what one owes is answerable to what is right. Where this lands Set the cart image beside Moran's account and the fit is close and specific. Both hold that assertion involves the speaker in a commitment, that the commitment is not a probabilistic signal, and that a hearer who receives an assurance stands in a different relation to the speaker from one who has merely collected evidence about her. Confucian xin supplies what Moran's account is thin on, namely an answer to how such a disposition is formed and sustained: by cultivation, within a community that holds people to their word, and as one virtue among several that are answerable to rightness. What xin does not supply is an a