Reliabilism and the Limits of Individual Knowledge: An African Perspective — Epoche B2
Reliabilism Is Not Too Individualistic: What Wiredu's Consensus Actually Shows A familiar complaint against reliabilism in epistemology is that it is built for a solitary knower [1] . Its examples are one person's eyes, one person's memory, one person's inference, and its question is always whether that person's faculties are working. Epistemologies that treat knowing as something a community does are then said to expose a blind spot. This essay argues that the complaint, in that form, is mistaken: reliabilism can type a collective practice as a belief-forming process without any change of doctrine, and its founder has done so at book length. But something in the neighbourhood is right. Kwasi Wiredu's account of decision by consensus contains a standard of epistemic appraisal that reliabilism genuinely cannot represent [2] , and identifying it correctly requires giving up the communal-versus-individual framing altogether. What reliabilism claims Two distinct theories of Alvin Goldman's are often run together [3] , and the argument below depends on keeping them apart. In 'A Causal Theory of Knowing' (1967 [4] ) Goldman proposed that S knows that p only if the fact that p is causally connected in an appropriate way with S's believing p [5] . This was an answer to Edmund Gettier's 1963 counterexamples [6] , in which a belief is true and justified but only by accident. The causal requirement was meant to exclude the accident. Goldman later abandoned the theory, in part because it cannot handle knowledge of general propositions or of the future, which have no causal route to the believer. Process reliabilism is the later view, stated in 'What Is Justified Belief?' (1979) and developed in Epistemology and Cognition (1986). A belief is justified when it is produced by a process type that yields a high ratio of true to false beliefs. It is an externalist theory: the believer need not have any access to the process's reliability, or even know which process she used. A child's perceptual beliefs are justified because her visual system works, not because she has views about optics. Three ways a communal epistemology might press The objection from communal knowing can be sharpened in three different directions, and they are not equally strong. First, that the reliable processes are social rather than psychological. This has no force, because reliabilism does not say that a process must be housed inside one skull. A process type is individuated by what it does, and there is no obstacle to counting a council's deliberation, a system of apprenticeship, or scientific peer review as one. Goldman himself made the extension the subject of Knowledge in a Social World (1999), where social practices are assessed by how many true beliefs they produce in a community. A charge that the target has already conceded and developed is not a charge. Second, that the subject of knowledge is the group rather than any member. This is a substantive claim, and it may be true, but it is not an objection to reliabilism. It is a proposal about who or what bears epistemic properties. A reliabilist who accepted it would carry on grading processes exactly as before, and would then apply the grades to a group. Third, that the standard by which a process is appraised is different. This is the one that bites, and it is the one the first two obscure. Reliabilism has exactly one currency: the ratio of truths to falsehoods a process type produces. If a communal tradition appraises its practices by a criterion that cannot be converted into that currency without loss, then the disagreement is real and cannot be settled by widening the notion of a process. Wiredu on consensus Kwasi Wiredu's essay on democracy and consensus, collected in Cultural Universals and Particulars: An African Perspective (1996), reconstructs the deliberative practice of Ashanti councils and defends a general thesis about it. The argument is worth stating carefully, because the label 'consensus' invites two misreadings. It does not mean unanimity of opinion. Wiredu is explicit that people who disagree remain in disagreement. What consensus is reached on is action: what the council will do. Nor does it mean the suppression of dissent for the sake of harmony. Wiredu's claim is the reverse. The procedure is designed so that a decision may not be taken until the concerns of those who objected have been given a place in it — accommodated, in his term, rather than outvoted. His argument against majoritarian voting is that it permanently disenfranchises a minority whose reasons were never taken up, and that a decision procedure which does take them up produces better decisions and better standing for those bound by them. The epistemic content of this is easy to miss. Wiredu is not saying that many heads are better than one, which everybody says. He is saying that a decision has a defect if the reasons against it were set aside without being answered — and that the defect is present even when the decision is correct. Why reliabilism cannot represent it Consider two councils that reach conclusions about disputed matters of fact: where a boundary formerly ran, whether a debt was paid, which of two accounts of an event is right. The first accommodates dissent in Wiredu's sense. The second hears objections and sets them aside. Stipulate that the two get things right equally often. The stipulation is coherent, and it is what the reliabilist must ask us to hold fixed, since the truth ratio is the only variable the theory recognises. Reliabilism must then say the two are epistemically on a par. Anyone who takes Wiredu's point will say they are not: the members of the first council are better placed with respect to what they believe than the members of the second, even on the occasions when both are right. The difference lies in whether the reasons against the belief have been met. That is a property of the deliberation, visible from inside it, and it is invisible to a criterion