Abstract
Coherentists on epistemic justification claim that all justification is inferential, and that beliefs, when justified, get their justification together (not in isolation) as members of a coherent belief system. Some recent work in formal epistemology shows that “individual credibility” is needed for “witness agreement” to increase the probability of truth and generate a high probability of truth. It can seem that, from this result in formal epistemology, it follows that coherentist justification is not truth-conducive, that it is not the case that, under the requisite conditions, coherentist justification increases the probability of truth and generates a high probability of truth. I argue that this does not follow.
Similar content being viewed by others
Notes
Coherentists deny that there can be noninferential justification. But coherentists need not, and should not, deny that there can be noninferential belief-formation, that is, belief-formation not consisting in an inference from beliefs. See BonJour (1985, sec. 6.1).
On foundationalism, a belief can be justified even if it is not evidentially supported by beliefs. On some varieties of foundationalism, a belief can be justified in virtue of being evidentially supported by a perceptual experience. Paul Moser (1989) has a sophisticated view along these lines.
For general discussion of the elements of coherence, see BonJour (1985, Chap. 5). For discussion of probabilistic conceptions of coherence, see Olsson (2005a, pp. 95–102), Siebel (2005), and Douven and Meijs (2007). For a nonprobabilistic account of coherence, see Lehrer (2000, Chap. 6). For discussion of forms of coherentism requiring, for justification, more than coherence, e.g., reliability, see Roche (2006). It might be that coherentists should hold that what matters for justification is not the coherence of the subject’s belief system as a whole, but the coherence of a certain proper subset (or “module”) of that system. See Lycan (1996) and Olsson (1997).
Just what the requisite conditions are is a main issue of this paper.
A related, but distinct, question is whether, ceteris paribus, greater coherentist justification makes for a greater probability of truth. For discussion, see Klein and Warfield (1994, 1996), Merricks (1995), Cross (1999), Shogenji (1999, 2005b, 2007), Bovens and Olsson (2000, 2002), Olsson (2001, 2002, 2005a, b), Bovens and Hartmann (2003, 2005, 2006), Huemer (2007), Meijs and Douven (2007), and Schupbach (2008).
I shall assume, as seems plausible, that epistemic justification is truth-conducive. This is a generous assumption. If epistemic justification is not truth-conducive (in the sense in question), and it is not incumbent on coherentists to show that coherentist justification is truth-conducive, then the formal epistemological result explained in §II, below, poses no threat to coherentism. Even if, contrary to what I shall argue, the result in question showed that coherentist justification is not truth-conducive, it might still be that coherentism is the correct theory of epistemic justification. For discussion, in addition to that in this paper, of how to understand the “truth connection,” i.e. the connection between justification and truth, see e.g., Lehrer and Cohen (1983), Cohen (1984), Conee (2004), and Kvanvig (2007).
Admittedly, the case as it stands is underdescribed.
This would depend on whether the conditions in question are the requisite conditions.
Or at least can be read as doing so. But see §VI, below, where I suggest an alternative reading of BonJour’s position.
See also Bovens and Olsson (2000), and Olsson and Shogenji (2004). Two comments are in order. First, strictly speaking, the thesis defended by Huemer, Olsson, and Shogenji has an independence clause, stating that the various witness reports are probabilistically independent of each other (conditional on the truth or falsity of the hypothesis reported). This clause is quite important. See Huemer (2007). See also Olsson (2005a, pp. 58–60) and Shogenji (2005a, p. 321). But for the purposes of this paper, the independence clause may be ignored. Second, though Olsson and Shogenji side with Lewis in arguing for (A), Olsson and Shogenji disagree with Lewis on certain other related issues. See Olsson and Shogenji (2004), and Olsson (2005a).
A witness can have no individual credibility with respect to some propositions (e.g., propositions in quantum mechanics) and yet have some individual credibility with respect to other propositions (e.g., propositions about the weather). Individual credibility need not be construed as attaching to witnesses (relative to propositions). It may instead be construed as attaching to witness reports; Lewis and BonJour talk in this fashion. Nothing of substance hinges on which construal is used. I shall continue to speak of individual credibility as attaching to witnesses.
See also Huemer (1997, pp. 470–471).
(C) relies on several assumptions. One assumption is that a witness is a truth-teller, a liar, or a “randomizer.” A randomizer is a witness who testifies randomly. Suppose, in a particular case, there are n suspects, s 1, . . . , s n , and they are all equally likely to be the criminal. Suppose a certain witness w is a randomizer, and is set to incriminate one of the suspects. Then, regardless of which of the suspects is guilty, the probability of w’s incriminating s i , for any i, is 1/n.
It is being assumed that if there are n suspects, there are n-1 ways to lie.
(B) is simply a specification of how to understand the notion of a witness’s having no individual credibility. The substantive claims are (A) and (C).
(E), like (B), is not a substantive claim. (E) merely specifies how to understand the notion of a cognizer’s having no individual credibility.
I noted above that (C) relies on the assumption that a witness is a truth-teller, a liar, or a randomizer. Accordingly, (F) should be understood as relying on the assumption that a cognizer’s processes are reliable, unreliable, or random. Suppose, in a particular case, there are n possible belief contents, p 1, . . . , p n , and they are all equally likely to be correct. Suppose a certain process r is random, and is set to produce a belief with one of the contents. Then, regardless of which of the contents is correct, the probability of r’s “picking” p i , for any i, is 1/n.
I take it that if (D), (E), and (F) do not lead to (G), then neither do (A), (B), and (C).
Recall, from §I, that I am assuming that epistemic justification is truth-conducive, and so unless doxastic coherence (coherentist justification) is truth-conducive, coherentism is not the correct theory of epistemic justification.
(H) says: Doxastic coherence is truth-conducive only if under conditions of no individual credibility, doxastic coherence both increases the probability of truth and generates a high probability of truth. (D) says: Under conditions of no individual credibility, doxastic coherence neither increases the probability of truth nor generates a high probability of truth. (D) thus implies that: It is not the case that under conditions of no individual credibility, doxastic coherence both increases the probability of truth and generates a high probability of truth. Hence, if (D) is correct, the consequent of (H) is false. Hence, if (H) is correct, and (D) is correct, it follows that the antecedent of (H) is false, in which case (G) is correct.
I am assuming that when a cognizer forms, say, a perceptual belief, there are two or more ways for her to get things wrong.
It might be argued, in favor of (I), that for doxastic coherence to be truth-conducive, it needs to be the case that under any conditions in which there can be doxastic coherence, even conditions in which the cognizer is much more likely to have unreliable processes than reliable processes, doxastic coherence increases the probability of truth and generates a high probability of truth. I consider this proposal in §V.
I am simplifying a bit. I take it to be plausible that, with respect to the issue of reliability, the requisite conditions are conditions of ignorance as to the reliability of the cognizer’s processes. It might be, though, that the requisite conditions involve more than just ignorance as to reliability.
(K) does not say, or imply, that doxastic coherence is truth-conducive only if the only conditions in which doxastic coherence increases the probability of truth and generates a high probability of truth are conditions in which the cognizer is just as likely to have unreliable processes as reliable processes.
(K) says that: Doxastic coherence is truth-conducive if and only if under conditions in which the cognizer is just as likely to have unreliable processes as reliable processes, doxastic coherence increases the probability of truth and generates a high probability of truth. (K) thus implies that: It is false that doxastic coherence is truth-conducive only if under conditions in which the cognizer is more likely to have unreliable processes than reliable processes, doxastic coherence increases the probability of truth and generates a high probability of truth. (This inference fails if the following conditional holds: If doxastic coherence increases the probability of truth and generates a high probability of truth under conditions in which the cognizer is just as likely to have unreliable processes as reliable processes, then doxastic coherence increases the probability of truth and generates a high probability of truth under conditions in which the cognizer is more likely to have unreliable processes than reliable processes. But this conditional seems false.).
This is the requirement that a belief system “contain laws attributing a high degree of reliability to a reasonable variety of cognitively spontaneous beliefs (including in particular those kinds of introspective beliefs which are required for the recognition of other cognitively spontaneous beliefs)” (1985, p. 141).
BonJour’s argument makes no appeal to witness agreement. In fact, in Chapter 8, the chapter in which he tries to give a metajustification for his coherentism, BonJour discusses witness agreement not at all.
For BonJour, one’s grasp of one’s belief system is not at issue. One may simply take it for granted, when engaged in epistemological investigation, that one’s grasp of one’s belief system is by and large correct. See BonJour’s discussion of the “Doxastic Presumption” (1985, pp. 103–106). What is at issue, for BonJour, is whether one’s external-world beliefs are correct. BonJour’s metajustification is meant to be of help in defending one’s external-world beliefs against the external-world skeptic.
Reliability should here be understood so that reliability comes in degrees, and a process can be reliable (though not fully reliable) even if it does not invariably produce true beliefs.
I am glossing over an important but, for my purposes, tangential distinction between justification and “well-foundedness.”
See e.g., Goldman (1979).
(L*), like (L), leads to (I) and (H).
Unreliability should here be understood so that unreliability comes in degrees, and a process can be unreliable (though not fully unreliable) even if it does not invariably produce false beliefs.
A better test, it seems, is whether evidentialist justification increases the probability of truth and generates a high probability of truth under conditions of ignorance as to the reliability of the cognizer’s processes. (Again, I am simplifying. The suggestion is that, with respect to the issue of reliability, the requisite conditions are conditions of ignorance as to the reliability of the cognizer’s processes).
(C) should be understood so that conditions in which a witness is just as likely to be a truth-teller as a liar are conditions in which she has some individual credibility—her testifying that p increases the probability of p. See Olsson (2005a, p. 71, pp. 218–219). (F) should thus be understood similarly.
Though, strictly speaking, BonJour speaks of individual credibility as attaching to witness reports, not to witnesses.
James van Cleve (2005) reads BonJour in this manner. See also Shogenji (2005a, pp. 314–315). Cf. Olsson (2005a, p. 67, n. 4). If this reading of BonJour is correct, and if Lewis understands “no individual credibility” as in (B), then BonJour’s and Lewis’s claims are compatible. It can be true that, in some cases, witness agreement increases the probability of truth and generates a high probability of truth even though, for each witness w, Pr(p | w said that p) is less than or equal to .5, indeed, even though Pr(p | w said that p) is much less than .5, and yet also be true that witness agreement neither increases the probability of truth nor generates a high probability of truth when, for each witness w, Pr(p | w said that p) is equal to Pr(p). There might be reason, however, for doubting that Lewis understands “no individual credibility” as in (B). See van Cleve (2005, pp. 170–171).
Or should hold, to be consistent.
Hereafter, unless otherwise noted, “no individual credibility” is to be understood as in (E).
I take it that if (1) and (2) are true, coherentists as such hold that the requisite conditions (the conditions under which doxastic coherence should be expected to increase the probability of truth and generate a high probability of truth) are conditions of no noninferential justification and thus of no individual credibility. Hence (3).
Or at least we can suppose. Surely there can be cases in which there is a low prior probability that the President is in New York City.
See e.g., Davidson (2000).
References
BonJour, L. (1985). The structure of empirical knowledge. Cambridge: Harvard University Press.
BonJour, L. (1999). The dialectic of foundationalism and coherentism. In J. Greco & E. Sosa (Eds.), The Blackwell guide to epistemology (pp. 117–142). Oxford: Blackwell.
BonJour, L. & Sosa, E. (2003). Epistemic justification. Malden: Blackwell.
Bovens, L. & Hartmann, S. (2003). Solving the riddle of coherence. Mind, 112, 601–633.
Bovens, L. & Hartmann, S. (2005). Why there cannot be a single probabilistic measure of coherence. Erkenntnis, 63, 361–374.
Bovens, L. & Hartmann, S. (2006). An impossibility result for coherence rankings. Philosophical Studies, 128, 77–91.
Bovens, L. & Olsson, E. (2000). Coherentism, reliability and Bayesian networks. Mind, 109, 685–719.
Bovens, L. & Olsson, E. (2002). Believing more, risking less: On coherence, truth and non-trivial extensions. Erkenntnis, 57, 137–150.
Cohen, S. (1984). Justification and truth. Philosophical Studies, 46, 279–295.
Conee, E. (2004). The truth connection. In E. Conee & R. Feldman, Evidentialism (pp. 242–254). Oxford: Clarendon.
Cross, C. (1999). Coherence and truth conducive justification. Analysis, 59, 186–193.
Davidson, D. (2000). A coherence theory of truth and knowledge. In E. Sosa & J. Kim (Eds.), Epistemology: An anthology (pp. 154–163). Malden: Blackwell.
Douven, I. & Meijs, W. (2007). Measuring coherence. Synthese, 156, 405–425.
Feldman, R. & Conee, E. (2004). Evidentialism. In E. Conee & R. Feldman, Evidentialism (pp. 83–101). Oxford: Clarendon.
Goldman, A. (1979). What is justified belief? In G. Pappas (Ed.), Justification and knowledge (pp. 1–23). Dordrecht: D. Reidel.
Huemer, M. (1997). Probability and coherence justification. Southern Journal of Philosophy, 35, 463–472.
Huemer, M. (2007). Weak Bayesian coherentism. Synthese, 157, 337–346.
Klein, P. & Warfield, T. (1994). What price coherence? Analysis, 43, 129–132.
Klein, P. & Warfield, T. (1996). No help for the coherentist. Analysis, 56, 118–121.
Kvanvig, J. (2007). Coherentist theories of epistemic justification. Available via The Stanford Encyclopedia of Philosophy, at http://plato.stanford.edu/entries/justep-coherence/. Cited 20 September 2009.
Lehrer, K. (2000). Theory of knowledge. Boulder: Westview.
Lehrer, K. & Cohen, S. (1983). Justification, truth, and coherence. Synthese, 55, 191–207.
Lewis, C. I. (1946). An analysis of knowledge & valuation. La Salle: Open Court.
Lycan, W. (1996). Plantinga and coherentisms. In J. Kvanvig (Ed.), Warrant in contemporary epistemology: Essays in honor of Plantinga’s theory of knowledge (pp. 3–23). Lanham: Rowman & Littlefield.
Meijs, W. & Douven, I. (2007). On the alleged impossibility of coherence. Synthese, 157, 347–360.
Merricks, T. (1995). On behalf of the coherentist. Analysis, 55, 306–309.
Moser, P. (1989). Knowledge and evidence. Cambridge: Cambridge University Press.
Olsson, E. (1997). Coherence and the modularity of mind. Australasian Journal of Philosophy, 75, 404–411.
Olsson, E. (2001). Why coherence is not truth-conducive. Analysis, 61, 236–241.
Olsson, E. (2002). What is the problem of coherence and truth? Journal of Philosophy, 99, 246–272.
Olsson, E. (2005a). Against coherence. Oxford: Clarendon.
Olsson, E. (2005b). The impossibility of coherence. Erkenntnis, 63, 387–412.
Olsson, E. & Shogenji, T. (2004). Can we trust our memories? C. I. Lewis’s coherence argument. Synthese, 142, 21–41.
Roche, W. (2006). Can a coherentist be an externalist? Croatian Journal of Philosophy, 5, 269–280.
Schupbach, J. (2008). On the alleged impossibility of Bayesian coherentism. Philosophical Studies, 141, 323–331.
Shogenji, T. (1999). Is coherence truth conducive? Analysis, 59, 338–345.
Shogenji, T. (2005a). Justification by coherence from scratch. Philosophical Studies, 125, 305–325.
Shogenji, T. (2005b). The role of coherence of evidence in the non-dynamic model of confirmation. Erkenntnis, 63, 317–333.
Shogenji, T. (2007). Why does coherence appear truth-conducive? Synthese, 157, 361–372.
Siebel, M. (2005). Against probabilistic measures of coherence. Erkenntnis, 63, 335–360.
van Cleve, J. (2005). Why coherence is not enough: A defense of moderate foundationalism. In M. Steup & E. Sosa (Eds.), Contemporary debates in epistemology (pp. 168–180). Malden: Blackwell.
Acknowledgments
I wish to thank Gregg Franzwa, Nicholaos Jones, William Melanson, Michael Roche, and Joshua Smith for very helpful questions and comments on earlier drafts of this paper. I also wish to thank audiences at the 1st Annual USA Philosophy Workshop: Perspectives on Coherentism, the 42nd Meeting of the North Texas Philosophical Association, and the 105th Annual Meeting of the American Philosophical Association Central Division, where I presented earlier drafts of this paper. The completion of this work was financially supported by the Junior Faculty Summer Research Program at Texas Christian University.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Roche, W.A. Coherentism, Truth, and Witness Agreement. Acta Anal 25, 243–257 (2010). https://doi.org/10.1007/s12136-009-0065-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12136-009-0065-1