Abstract
Reliabilism is an intuitive and attractive view about epistemic justification. However, it has many well-known problems. I offer a novel condition on reliabilist theories of justification. This method coherence condition requires that a method be appropriately tested by appeal to a subject’s other belief-forming methods. Adding this condition to reliabilism provides a solution to epistemic circularity worries, including the bootstrapping problem.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
For an overview of the internalist/externalist debate in epistemology, see Pappas (2014).
Perhaps the first clear statement of reliabilism is in Ramsey’s Truth and Probability (1926). For an overview, see Goldman and Beddor (2015).
Cf. Sosa (2009).
I also intend to remain neutral on the way in which this kind of epistemic norm is related to things like “rationality” or “reasons.” Some epistemologists also distinguish a notion of “justification” that is distinct from warrant or knowledge-level justification, which has to do with the actual reasons subjects give for their beliefs (e.g., Sosa 2009 makes this distinction). I am hopeful that at the end of the day we won’t need this distinction, but for my purposes here I just want to focus on knowledge-level justification. I use this terminology in order to be clear that I am concerned with reliabilism about justification (Goldman 1979), rather than reliabilism directly about knowledge (Armstrong 1973; Dretske 1981). For an overview of that distinction, see Goldman and Beddor (2015).
Barry Stroud offers a related argument against externalism in general, including reliabilist theories (Stroud 1989, p. 15).
Per Vogel, this is taken and modified from Michael Williams.
For an exception to this, see Barnett (2014), who distinguishes between bootstrapping and direct circularity, which he calls “self-verifying” circularity. Cohen (2002) also distinguishes two ways in which circularity problems arise, one via closure from basic beliefs, and the other via bootstrapping, though this latter distinction is different than the one I draw here.
For example, Titelbaum’s (2010) solution involves a ban on “no-lose investigations”: ways of forming beliefs which cannot result in disbelief or disconfirmation. This solution rules out bootstrapping (and perhaps some other things we want to allow in), but does not rule out direct circularity. Al’s crystal ball might tell him it isn’t reliable. Similarly, Weisberg’s “no-feedback” rule (2010), to be discussed below, also fails to rule out direct circularity. Of course, neither of these solutions was aimed at solving direct circularity, so this is less a criticism than a clarificationv.
I am not the first to suggest linking reliabilism with coherence. Ernest Sosa appeals to coherence in his virtue reliabilist account of knowledge and justification (2007). However, Sosa appeals to a traditional kind of coherence.
Note that some reliabilist theories are intended to be more generally applicable than others. That is, some reliabilist theories are only meant to apply for certain kinds of belief-formation. For example, Lyons provides a reliabilist account that only applies to certain kinds of empirical, basic beliefs (2009). I intend to remain neutral on just how general the a reliabilist theory is meant to be, and the MCR schema is compatible with a variety ways of interpreting the theory. I am simply advocating that, if you are a reliabilist about justification (of some kind of knowledge), then you should also adopt the MCC (for that kind of knowledge).
Even without adding an additional defeat condition, MCR might already provide a way out of some of the counter-examples. For instance, it’s not clear that Norman the clairvoyant (BonJour 1980) meets the method coherence condition. Ex hypothesi, though his clairvoyance is reliable, he has never tested its reliability. However, there is not space here to consider how method coherence can help with this and similar counter-examples, and ultimately I suspect that we will still need some form of defeat-exclusion condition.
Note that I am motivating the principle by appeal to instances when we do, consciously, judge a method reliable by testing it against other methods. However, MCC itself does not require awareness or accessibility. While we often are aware of our testing, I do not think this is a requirement of justification.
I am remaining neutral about precisely which probability function we appeal to. It seems plausible that we could appeal to the subject’s own credence function, or the probability function of some (perhaps ideal) evaluator. Fans of evidential probability could even appeal to that sort of probability function.
The details of Fitelson’s measure are somewhat “intricate,” as he puts it, though the basic idea is not. Fitelson defines a measure of evidential support [called “factual support” by Kemeny and Oppenheim (1952)], F. F is a two place function, where F(X, Y) gives the degree to which Y supports X. \(F(X,Y) = 1\) if both propositions are necessary truths, 0 if X is necessary and Y is contingent, and \(-1\) if Y is a necessary falsehood. In all other cases \(F(X,Y) = \frac{Pr(Y|X) - Pr(Y|\lnot X)}{Pr(Y|X) + Pr(Y|\lnot X)} \). Once the F measure of support is defined, this can be used to determine the degree of coherence of any set of evidence E. Fitelson does this by building a set \(P_i\) which, for each \(E_i \in E\), is the powerset of the complement \(E \backslash \{E_i\}\) (which just isolates \(E_i\) and the other members of the set E ). Then, he uses these elements to consider how much each \(E_i\) is supported by the other members of E (and all their possible conjunctions), given by \({\mathbf {S}} = \bigcup \{\{F(E_i,X)|x\in P_i\}|E_i \in E \} \). Finally, the degree of coherence is obtained by simply averaging the support that each member receives from the others, that is the mean of \({\mathbf {S}}\) for E: \(C(E)=_{df} mean({\mathbf {S}})\). For more details, see Fitelson (2003).
This appeal to contextual sensitivity here is similar to the way reliabilist views generally appeal to contextual features to determine the relevant threshold of reliability. Since so may epistemological theories are committed to such features playing a role in setting various thresholds, I don’t think this is a worrisome contextual senstivity, or at least something particularly worrisome for my view.
On some views of mental content, the proposition a belief is held toward will already include a specific time (e.g., if a proposition is a set of possible worlds). On views like this, this added stipulation that the beliefs concern the same time will be redundant, and could be dispensed with. However, I want to remain neutral on questions regarding the nature of propositions and of mental content.
Thanks to an anonymous reviewer for raising this issue.
It does, however, require more than that the subject simply notice if a method is incoherent. Cf. Kent Bach’s “take- for-granted principle” which essentially requires that the subject would become aware if something in a situation was “out of the ordinary” (1985, p. 261), or “No Defeater” conditions (Beddor 2015).
One might be tempted by a counterfactual understanding of “tends,” as a different way of avoiding the problems I have just mentioned. However, such a counterfactual version of the MC requirement fails to address the bootstrapping problem, as it would allow that Roxanne is justified, because if she checked the gas gauge, it would be coherent (since, by hypothesis, it is reliable). The problem is that she does no such testing.
It is worth noting, however, that there is a nearby version of the view that does involve having a limited accessibility requirement. This minimally internalist version of the view would be more stringent than MCR, but still be significantly less stringent than a view which involves full accessbilism. This version of MCR would simply involve the further requirement that the subject have access to the fact that her methods are coherent. That is, a subject must be able to become aware of the fact that she has tested her methods in the past. This version might be amenable to those who seek a more robust conception of justification, while maintaining many of the benefits of reliabilism, and avoiding some difficulties for full-blown accessibilism. Thanks to an anonymous referee for discussion on this point.
If one is still concerned about the use of as method to justify itself, this discussion suggests another solution that makes use of method coherence, although it departs from my schema for justification. One could argue that any time a subject forms beliefs about the reliability of a method, they must do so by the method coherence testing criterion. That is, second order beliefs about whether a method is reliable can only be justified by the process that ensures they are method coherent. This has the benefit of the intuitive appeal of the method coherence procedure, and of making it harder to form justified beliefs about one’s own reliability. However, it has the drawback of having to postulate a special kind of justification for a certain set of beliefs. The justification schema I offer has the benefit of making justification a unified notion.
Some reliabilists are not perturbed by the direct circularity objection in the first place. Some are convinced by Van Cleve (2003) that circularity is both unavoidable and acceptable, as discussed below in §4. The solution presented here to the direct circularity objection may seem less pressing to the reliabilists who are so convinced. However, MCR does also provide a solution to the bootstrapping problem. And it seems nice to have more than one available response to the direct circularity problem. So I think that method coherence should still be attractive to these reliabilists, as well.
The intuitive version of the view gets us this result: being less specific means a proposition rules out fewer possibilities. It will thereby receive and provide less support to other propositions, and will thus be inadequately mutually supportive of the deliverances of other methods. However, MCC will also rule out methods which produce only unspecific predictions on the precisification I give above in terms of Fitelson’s measure of coherence. (At least, this is true as long as the belief-forming methods standardly used by subjects do not also result in beliefs with only unspecific content that is logically equivalent to the predictions offered by the crystal gazer. And I think this is very plausible to assume about ordinary subject’s perceptual belief-forming methods.) This is because Fitelson’s coherence measure appeals to a probabilistic support relation. An unspecific theory, which is compatible with a great deal of evidence, is supported less by evidence E than a more specific theory which predicts E, due to the law of likelihoods (for details on this feature of factual support, see Fitelson 2003; Kemeny and Oppenheim 1952).
The precise version is:
- No Feedback :
-
If (i) \(L_1 - L_n\) are inferred from \(P_1 - P_m\), and (ii) C is inferred from \(L_1 - L_n\) (and possibly some of \(P_1 - P_m\)) by an argument whose justificatory power depends on making C at least x probable, and (iii) \(P_1 - P_m\) do not make C at least x probable without the help of \(L_1 - L_n\), then the argument for C is defeated.
Note that this is a drawback for the reliabilist hoping to employ the No Feedback solution. Weisberg is defending evidentialism, rather than reliabilism. Evidentialism does not grant Roxanne justification at the first step, so the objection does not apply to an evidentialist account of the Roxanne case. However, there are other cases of basic knowledge that many evidentialists do grant, which allow for similar kinds of bootstrapping or “easy knowledge” (see Cohen 2002). No Feedback might suffer from this drawback in some of these cases: it fails to block justification that is too easy at the first step.
Moreover, there is some evidence from the metacognition literature that we receive feedback on how well different methods perform, including how well they perform at the same time (Proust 2013). And this kind of inward tab-keeping on the reliability of methods may very well provide support for method coherence. Though see Nagel (2014) for an opposing view of this research. There is much left to be said on this topic, but I will set it aside for future work.
For more on the generality problem, its application to non-reliabilist theories, and on potential solutions, see Goldman and Beddor (2015).
This could also be made more specific, as in “testimony from John.”
I talk of more and less certainty here to avoid saying “infallible.”If one is not convinced by anti-luminosity arguments that we are fallible about even our own mental states (Williamson (2000)), then this response should look even better. If we have luminous access to our own mental states, we should not expect that our fallible external-world knowledge will be gained in the same way that we infallibly come to know things about our own mental states.
References
Armstrong, D. M. (1973). Belief truth and knowledge. Cambridge: Cambridge University Press.
Bach, K. (1985). A rationale for reliabilism. The Monist, 68, 246–263.
Barnett, D. J. (2014). What’s the matter with epistemic circularity. Philosophical Studies, 171(2), 177–205.
Beddor, B. (2015). Process reliabilism’s troubles with defeat. The Philosophical Quarterly, 65(259), 145–159.
Bergmann, M. (2004). Epistemic circularity: Malignant and benign. Philosophy and Phenomenological Research, 69(3), 709–727.
Bishop, M. A. (2010). Why the generality problem is everybody’s problem. Philosophical Studies, 151, 285–298.
BonJour, L. (1980). Externalist theories of empirical knowledge. In P. A. French, et al. (Eds.), Midwest studies in philosophy 5: Studies in epistemology (pp. 53–73). Minneapolis: University of Minnesota Press.
BonJour, L. (1985). The structure of empirical knowledge. Cambridge, MA: Harvard University Press.
Cleve, Van, & James. (2003). Is Knowledge Easy–Or Impossible? Externalism as the Only Alternative to Skepticism (pp. 45–60). The Skeptics: Contemporary Essays.: Ashgate.
Cleve, Van, & James. (2011). Sosa on epistemic circularity. Philosophical Studies, 2011(153), 19–28.
Cohen, S. (2002). Basic knowledge and the problem of easy knowledge. Philosophy and Phenomenological Research, 65(2), 309–329.
Comesaña (2005). A well-founded solution to the generality problem. Philosophical Studies, 129(1), 27–47.
Conee, E., & Feldman, R. (2001). Internalism defended. American Philosophical Quarterly, 38(1): 1–18.
Douven, I., & Kelp, C. (2013). Proper bootstrapping. Synthese, 190, 171–185.
Dretske, F. (1981). Knowledge and the flow of information. Cambridge: MIT Press.
Earman, J. (1992). Bayes or bust? a critical examination of bayesian confirmation theory. Cambridge: MIT Press.
Elgin, C. Z. (2005). Non-foundationalist epistemology: Holism, coherence, and tenability. In S. Matthias & S. Ernest (Eds.), Contemporary Debates in Epistemology (pp. 156–167). Boston: Blackwell.
Fitelson, B. (2001). Studies in bayesian confirmation theory. Ph.D. Thesis, University of Wisconsin-Madison.
Fitelson, B. (2003). A probabilistic measure of coherence. Analysis, 63, 194–199.
Fumerton, R. (1995). Metaepistemology and skepticism. Lanham: Rowman and Littlefield.
Goldman, A. (1979). What is justified belief? In G. S. Pappas (eds.), Justification and Knowledge. Dordrecht: D. Reidel.
Goldman, A., & Beddor, B. (2015). Reliabilist epistemology. In E. N. Zalta (Eds), The stanford encyclopedia of philosophy (Winter 2015 Edition). http://plato.stanford.edu/archives/win2015/entries/reliabilism/. Accessed 22 March 2017.
Greco, J. (1993). Virtues and vices of virtue epistemology. Canadian Journal of Philosophy., 23, 3.
Kemeny, J. G., & Oppenheim, P. (1952). Degrees of factual support. Philosophy of Science, 19, 307–324.
Klein, P., & Warfield, T. A. (1994). What price coherence? Analysis, 54, 129–132.
Lewis, C. I. (1946). An analysis of knowledge and valuation. LaSalle: Open Court.
Lyons, Jack. (2009). Perception and basic beliefs. Oxford: Oxford University Press.
Nagel, J. (2014). The meanings of metacognition. Philosophy and Phenomenological Research, 89(3), 710–718.
Olsson, E. J. (2005). Against coherence: Truth, probability, and justification. Oxford: Clarendon Press.
Olsson, E. (2014). Coherentist theories of epistemic justification. In E. N. Zalta (Eds.), The stanford encyclopedia of philosophy (Spring 2014 Edition). http://plato.stanford.edu/archives/spr2014/entries/justep-coherence/. Accessed 22 March 2017.
Olsson, E. J., & Schubert, S. (2007). Reliability Conducive Measures of Coherence. Synthese, 157(3), 297–308.
Pappas, G. (2014). Internalist vs. externalist conceptions of epistemic justification. In E. N. Zalta (Eds.), The stanford encyclopedia of philosophy (Fall 2014 Edition). http://plato.stanford.edu/archives/fall2014/entries/justep-intext/. Accessed 22 March 2017.
Poston, T. (2014). Reason and explanation. Palgrave Macmillan UK.
Proust, J. (2013). The philosophy of metacognition: Mental agency and self-awareness. Oxford: Oxford University Press.
Sosa, E. (2007). A virtue epistemology (Vol. 1). Oxford: Clarendon Press.
Sosa, E. (2009). A virtue epistemology (Vol. 2). Oxford: Clarendon Press.
Sosa, E. (2011). Knowing full well. Princeton: Princeton University Press.
Stroud, B. 1989. Understanding human knowledge in general. In M. Clay and K. Lehrer (Eds.), Knowledge and Skepticism. Boulder: Westview Press.
Thagard, P. (2000). Coherence in thought and action. Cambridge: The MIT Press.
Titelbaum, M. (2010). Tell me you love me: Bootstrapping, externalism, and no-lose epistemology. Philosophical Studies, 149(1), 119–134.
Vogel, J. (2000). Reliabilism leveled. The Journal of Philosophy, 97(11), 602–623.
Vogel, J. (2008). Epistemic boostrapping. The Journal of Philosophy, 105(9), 518–539.
Weisberg, J. (2010). Bootstrapping in general. Philosophy and Phenomenological Research, 81(3), 525–548.
Weisberg, J. (2012). The bootstrapping problem. Philosophy Compass, 7(9), 597–610.
Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.
Wimsatt, W. (2007). Re-engineering philosophy for limited beings. Cambridge: Harvard University Press.
Acknowledgements
For a great deal of helpful feedback, I would like to thank Walter Ott, Bill FitzPatrick, Steve Daskal, Ted Parent, Ernest Sosa, Branden Fitelson, Susanna Schellenberg, Andy Egan, Georgi Gardiner, David Black, Lisa Miracchi, Bob Beddor, and Megan Feeney.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Fleisher, W. Method Coherence and Epistemic Circularity. Erkenn 84, 455–480 (2019). https://doi.org/10.1007/s10670-017-9967-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-017-9967-2