Abstract
In this paper, I present a solution to the problem that the need to accommodate the phenomenon of epistemic defeat poses for reliabilism. Defeaters are supposed to remove justification for previously justified beliefs. According to standard process reliabilism, the justification of a belief depends on the reliability of a process that is already completed when a defeater for that belief is obtained. It is hard to see, then, how a defeater can affect reliabilist justification, if that justification, from the perspective of the defeater, lies wholly in the past. What is more, undercutting defeaters plausibly defeat by generating incompatibilities with higher-order propositions (Sturgeon in Philos Stud 169(1):105–118, 2014; Melis in Philos Stud 170(3):433–442, 2014). This is hard to integrate with reliabilism which explicitly denies higher-order conditions for justification. I argue that both of these phenomena can be accommodated by it via what I call the “Replacement Principle”: when a defeater is obtained, it generates either a logical incompatibility (in cases of rebutting defeat) or a doxastic incompatibility (in cases of undercutting defeat). The resulting tension is registered and initiates a reasoning process that compares the epistemic support for the defeater with that for the defeated belief. This process may replace the original reliable formation process as a doxastic base for the defeated belief. If the defeater is not given its due weight in the reasoning process, that process was unreliable and if the defeated belief is retained as a result, it will be unjustified. This is intuitively the right result. Since reasoning is reliable, only if it is rational, the account manages to capture the intuition that disregarding a defeater would be irrational without having to make rationality a condition for justification. Replacement is not only plausible on causal pictures of doxastic basing, but also required in order to allow for the possibility of belief reevaluation in general and is therefore not an ad hoc addition to reliabilism, but rather part of its inherent explanatory resources.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In order to stay neutral on the details here, I will use the term “information”, meaning propositional input to the subject’s doxastic system that can be used to base doxastic attitudes on (on reliabilism, as part of a belief-formation or -re-evaluation processes). Defeaters are then understood as a certain kind of information one has obtained to account for the idea that doxastic defeaters need to have found entry into one’s doxastic system and cannot just be external facts. Importantly, information is not taken to have normative components, like evidence. It can, but need not, be used as a doxastic base.
I here subscribe to what can be called a “causal” account of defeat, which requires a defeated belief to be justified beforehand. It is possible to conceive of defeaters as counter-reasons instead, which may also target unjustified beliefs (for details on both, see Loughrist 2015). The account I will present can be adapted to such a view by dropping the justification-condition without loss, since conflicts with unjustified beliefs plausibly also trigger re-evaluation processes.
In light of the phenomenon of partial defeat as highlighted by Thune (2010), Bergmann’s definition may need weakening. Specifically, a partial defeater need not be “an epistemically appropriate basis for” the belief that one’s belief is false. It should be enough to claim that a rebutting defeater epistemically supports the belief in the negation to some extent. In order to keep things as clear as possible, I will mostly ignore partial defeat for the purposes of this paper, making only some comments on the issue in footnotes. Therefore, I will use Bergmann’s account as a working definition.
The comment made on Bergmann’s view of rebutting defeat concerning partial defeat applies here as well. Luckily, assuming that undercutting defeaters turn out to be higher-order rebutting defeaters, as I will hold in this paper, a similar solution can be implemented.
Lasonen-Aarnio suggests this during a discussion of safety-conditions on knowledge and defeat. What I want to do here is flesh out this idea in a way that hopefully solves a number of problems for the reliabilist.
If one understands such a replacing re-evaluation process as a sustaining process, it need not be ongoing. In fact, as I will point out, it would make more sense to conceive of it as a reinstating process, as it is plausible that a belief that is being re-evaluated is sometimes temporarily suspended.
While I, like Plantinga (1994), take normally functioning subjects to be such that they notice and urge to resolve tension in their doxastic systems, since it causes them discomfort, I don't assume that it is part of their normal function that they react rationally to defeaters.
One might wonder how this works if, as is often the case, we forgot, what the target-belief was originally supported by. Two scenarios are possible in such cases: for the most part, at least indirect access to the original belief-formation is available to us. I might not remember exactly why I believe that Austerlitz is in Austria, but I do remember that I have learnt it at school. Since I know that learning things at school is reliable, I can weigh this against a given rebutting defeater. If not even indirect access is available and all I have left is a confident belief, there is nothing I can weigh against the defeater when re-evaluating and in such cases it is quite plausible that I should simply give into the defeater.
As is common in epistemology, I take (doxastic) justification for a belief to consist in the belief being based on some adequate justifier.
One may suspect that, instead of replacing the formation-process as the belief’s doxastic base, a re-evaluation is instead added to it. Without going into too much detail, there is reason to prefer the replacement-option. While there is counterfactual dependence between the formation-process and a belief’s epistemic status after it has been re-evaluated, it is unlikely to be sufficient for causal dependence. For while it is true that, if the belief had not been formed, it would not have been justified or unjustified as a result of re-evaluation now, it is much less plausible that epistemically significant variances in the formation process are also counterfactually connected to the belief in this way (cf. Lewis 2000). For example, it is far from obvious that, had the formation process been slightly less reliable, the belief, after re-evaluation, would also have been justified to a lesser degree, given that normal human subjects cannot pick up on the small difference in reliability during re-evaluative reasoning. This suggests that there is no (relevant) causal influence present. For reasons like this, I do not presume too much when I stick to the replacement-option for the purposes of this paper.
Generally, what exactly the effect of the hooligans' actions is, depends on how we individuate effects and how modally fragile effects are (Lewis 2000). If one feels that the hooligans-case is wrongly described in that their spray-painting has as its effect a different coat of yellow and therefore replaces the original effect, one can say that the target-belief is replaced by an indistinguishable, but different belief, which is then based on the re-evaluation process. This might be somewhat more plausible, if we assume the re-evaluated belief to be suspended during the relevant process and then get reinstated.
It may be preferable to assume that the target-belief is temporarily suspended while it is being re-evaluated and is then, as a result of the process, either dropped or reinstated. However, since not much hangs on this and because talk of holding on to a defeated belief is more common in the debate, I will continue to use terms like “kept” or “held on to”.
Couldn’t the subject just weigh the defeater against the justifier, just like in the case of rebutting defeat? As mentioned in the first section, this is implausible because undercutting defeat behaves differently from rebutting defeat when additional support is introduced. In Red-False, further first-order support, such as a second, clear look at the widgets, does not appear to counterbalance the information that the widgets are illuminated by a red light. In contrast, introducing higher-order support that questions the veracity of the defeating proposition does intuitively does.
This can also be modeled as an automatic inferential process, in the course of which Sally “jumps to the conclusion” that is the suspension of judgment. This kind of smooth movement from one propositional attitude to the next is akin to Kent Bach's notion of “default reasoning” (Bach 1984: p. 254), where the subject automatically goes through the inference step by step with the underlying assumption that each step follows from the previous, without consciously reflecting on this.
If there is no independent support on either side, whether defeat happens depends on a more general stance on the possibility of unjustified defeaters. I believe that there can be no unjustified defeaters, but even if there could be, in such a case, higher-order suspension of judgment is the result, which is enough to ensure defeat.
This is a position I find attractive for a variety of reasons I cannot go into here. However, it must be noted that, once one accepts normative conditions on justification to account for passively ignored defeaters, there is some pressure to give up the original project of explaining justification in terms of only descriptive conditions. After all, normative factors may also allow one to account for cases of properly acknowledged defeaters in simpler ways than suggested in this paper. So, a reliabilist that wants to hold onto the general project and minimize the degree of normativity will need to give additional reasons to limit the admitted normative conditions to the issue of passively ignored defeaters. It seems that the prospect for principled reasons of this sort are somewhat dim, which should motivate the reliabilist to make a more concerted attempt to reject the possibility of passively ignored defeaters. The parallels between them and cases of far apart beliefs should provide a decent starting point.
References
Bach, K. (1984). Default reasoning: Jumping to conclusion and knowing when to think twice. Pacific Philosophical Quarterly, 65(1), 37–58.
Baker-Hytch, M., & Benton, M. A. (2015). Defeatism defeated. Philosophical Perspectives,29, 40–66.
Beddor, B. (2014). Process reliabilism’s troubles with defeat. The Philosophical Quarterly,65(259), 145–159.
Bergmann, M. (2005). Defeaters and higher-level requirements. The Philosophical Quarterly,55(220), 419–436.
Bergmann, M. (2006). Justification without awareness: A defense of epistemic externalism. Oxford: Oxford University Press.
Boghossian, P., & Williamson, T. (2003). Blind reasoning. Proceedings of the Aristotelian Society, Supplementary Volumes,77, 225–293.
Conee, E., & Feldman, R. (1998). The generality problem for reliabilism. Philosophical Studies,89(1), 1–29.
Frances, B. (2008). Live skeptical hypotheses. In J. Greco (Ed.), Oxford handbook of skepticism (pp. 225–245). Oxford: Oxford University Press.
Goldman, A. (2012). What is justified belief? In A. Goldman (Ed.), Reliabilism and contemporary epistemology: Essays (pp. 29–48). Oxford: Oxford University Press.
Grundmann, T. (2009). Reliabilism and the problem of defeaters. Grazer Philosophische Studien,79(1), 65–76.
Grundmann, T. (2011). Defeasibility theory. In S. Bernecker & D. Pritchard (Eds.), The Routledge companion to epistemology (pp. 156–166). London: Routledge.
Janvid, M. (2017). Defeater goes external. Philosophia,45(2), 701–715.
Korcz, K. A. (2015). The epistemic basing relation. In: E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. http://plato.stanford.edu/archives/fall2015/entries/basing-epistemic/. Accessed July 2016.
Lackey, J. (2008). Learning from words: Testimony as a source of knowledge. Oxford: Oxford University Press.
Lasonen-Aarnio, M. (2010). Unreasonable knowledge. Philosophical Perspectives,24, 1–21.
Lewis, D. (2000). Causation as influence. Journal of Philosophy,97(4), 197–282.
Loughrist, T. D. (2015). Reasons against belief: A theory of epistemic defeat. Proquest LLC, UMI Dissertation Publishing.
Melis, G. (2014). Understanding undermining defeat. Philosophical Studies,170(3), 433–442.
Plantinga, A. (1994). Naturalism defeated. Resource document. Calvin College virtual library. https://www.calvin.edu/academic/philosophy/virtual_library/articles/plantinga_alvin/naturalism_defeated.pdf. Accessed January 2016.
Pollock, J. (1974). Knowledge and justification. Princeton: Princeton University Press.
Priest, G. (2008). An introduction to non-classical logic (2nd ed.). Cambridge: Cambridge University Press.
Pryor, J. (2004). What’s wrong with Moore’s argument? Philosophical Issues,14(1), 349–378.
Sturgeon, S. (2014). Pollock on defeasible reasoning. Philosophical Studies,169(1), 105–118.
Swain, M. (1965). Justification, reasons and reliability. Synthese,64(1), 69–92.
Thune, M. (2010). “Partial defeaters” and the epistemology of disagreement. Philosophical Quarterly,60(239), 355–372.
Acknowledgements
Research presented in this paper was funded by Deutsche Forschungsgemeinschaft (D73285Z) and conducted as part of the research project “Disagreement in Philosophy” at the University of Cologne. It has further profited from critical discussions at the Second Cologne-Leuven Epistemology Meeting 2016 in Leuven and at the European Epistemology Network Meeting 2016 in Paris. Thanks also go to Thomas Grundmann, Jens Kipper, Joachim Horvath, Dominik Balg, Steffen Koch, Anja Wieben-James and John Henry James for invaluable feedback.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Constantin, J. Replacement and reasoning: a reliabilist account of epistemic defeat. Synthese 197, 3437–3457 (2020). https://doi.org/10.1007/s11229-018-01895-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11229-018-01895-y