Abstract
Matthew Frise claims that reliabilist theories of justification have a temporality problem—the problem of providing a principled account of the temporal parameters of a process’s performance that determine whether that process is reliable at a given time. Frise considers a representative sample of principled temporal parameters and argues that there are serious problems with all of them. He concludes that the prospects for solving the temporality problem are bleak. Importantly, Frise argues that the temporality problem constitutes a new reason to reject reliabilism. On this point, I argue that Frise is mistaken. There are serious interpretive difficulties with Frise’s argument. In this essay, I show that there are principled and reasonable temporal parameters for the reliabilist to adopt that successfully undermine the interpretations of Frise’s argument that only invoke plausible premises. There are interpretations of Frise’s argument that leave reliabilism without a clear parameter solution. However, I argue that these interpretations invoke controversial premises that are at best unmotivated, and at worst they merely re-raise older disputes about reliabilism. In any event, the temporality problem fails to constitute a new reason to reject reliabilism.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Frise considers the temporality problem to be just one part of what he calls the reliability problem: “[T]he problem of supplying a suitable account of reliability” (924). Interestingly enough, a discussion of the temporality problem consumes almost his entire paper on the reliability problem. We can assume that he takes the temporality problem to constitute at least a significant part of the reliability problem.
See (926–928) for Frise’s presentation of three other parameters.
Enduring reliability is static because the truth ratio mentioned in the consequent must be high for any time t, not just some finite range of times or some particular time.
As an anonymous referee insightfully points out, EW and TF are rather odd in how, given their current formulation, they’re principles describing how a given belief b might have justification at any time—including times at which the belief b hasn’t even been formed yet. The referee, correctly, points out that these principles misrepresent the notion that reliabilists (and other epistemologists, for that matter) actually care about analyzing: whether a given belief b has justification at times when b exists/is being held. As the referee notes, Frise could accommodate this concern by adding the further necessary condition in EW and TF to specify that “the belief that p exists” at the given time in question. Furthermore, it does not appear that this modification would affect the cogency of Frise’s arguments in any way.
As I’ll note later, Frise does not offer a lengthy discussion of reliabilism’s ability to handle the unnatural deduction problem when combined with the TF trickle-down theory. This is because he thinks that “[t]here are indefinitely many ways to map justification on to beliefs formed by dynamically reliable processes. I won’t state them all. I will employ the first view [EW], since the other [trickle-down theories] in the end aid [reliabilism] no better” (928).
In his essay, Frise also presents a version of the argument that begins with the premise, Guesses are unjustified. This version of the argument proceeds in a structurally analogous manner, so I’ve omitted this version for brevity’s sake.
Frise identifies the process type perception with “taking perception at face value” (931).
For the version of the argument that proceeds from the premise, guesses are unjustified, Frise presents a hypothetical scenario called Blindluck. In this scenario, it turns out that for a large part of the history of the evolution of the human cognitive system, “guesses,” (i.e., beliefs formed without the basis of consciously accessible information and phenomenology) constituted the majority of beliefs being formed, and that they were for the most part correct (because they allowed us to survive as a species). Hence, for a very large temporal window, guessing had a high truth ratio (931–932).
Given the nature of enduring reliability, this conclusion follows regardless of whether EW or TF is the operative trick-down theory.
One might think that the unnatural deduction problem for AWR merely constitutes a good reason for the reliabilist to opt for a modalized interpretation of reliabilism’s temporal parameter, like NWR. However, Frise argues that the unnatural deduction problem plagues these interpretations of reliabilism as well.
On all defended Modal Reliabilist views, features of the actual world partly determine whether processes used in it are reliable and justifying… And, our ignorance about the actual past and future extends to the pasts and futures of worlds neighboring ours. Within the special domain, either there is too much diversity or too much similarity to our world for us to suppose the domain contains sufficiently few…Experience Machine-type scenarios. If we had reason to believe any Modal Reliabilist theory on any of the accounts of reliability, we could make unnatural deduction about not just our world, but also about other…worlds in the neighborhood—while nonetheless being fairly ignorant about which worlds these are! (941)
So, according to Frise, the unnatural deduction problem is actually worse for NWR. Not only could we implausibly come to know obscure facts about our world, but we can also gain a sort of “insight” into innumerable neighboring worlds.
As I discuss in Sect. 7, this conjunction also includes a theory of process type relevance.
For clarity’s sake, I’m building into these two readings that the agent stores her perceptually formed beliefs in memory in a justification-enabling fashion, and that the agent never acquires defeaters for her perceptually formed beliefs held in storage.
The reliabilist can defend proximate reliability from the version of the unnatural deduction argument that proceeds from the premise guessing is unjustified or guessing never justifies in an analogous fashion. However, there might be an even simpler way to show that guessing never justifies fails to lead to an unnatural deduction.
Frise describes a guessed belief as one that “seems to the subject formed arbitrarily, seems formed without enough relevant information, without even a hunch or a memory of a relevant good track record” (930, 931). As Frise notes, some reliabilists view reliability as merely a necessary condition on justification. Some epistemologists hold that a belief must be based on the right sorts of phenomenological or doxastic grounds in order to have justification. For example, Robert Audi claims that the only justified non-inferential beliefs involve some sort of intuitive, perceptual, introspective, or memorial phenomenology (2011: 6, 7). Guesses, according to Frise’s description, lack these sorts of grounds. Hence, Reliabilists who hold to the right grounds necessary condition as well can account for the claim guesses never justify without even invoking the unreliability of any sort of process—thus blocking an unnatural deduction.
Much thanks to an anonymous referee for raising this reasonable defense of the unnatural deduction problem.
See Goldman (1979) for a reliabilist-friendly formulation of this principle of inferential doxastic justification.
This could include basing a belief in PJa on expert testimony. Presumably, testimonial belief-formation is not a form of armchair reasoning.
Frise claims that it is “intuitive” that both guessing never justifies and that perception will always justify (935). But the PJb rendering of this sentence is an eternal truth claim—one that characterizes all times in the future—and it’s not clear why we should think that we could “intuit” a claim like this.
For example, Alston (1988) argues that the deontological conception is inconsistent with plausible principles of doxastic involuntarism. Goldman (2012: 147–149) argues that there are many kinds of justified belief (e.g., introspective justification, memorial justification, arithmetical inference-based justification) for which best-explanationism cannot account.
As opposed to the deontological conception, Alston argues that we should prefer a conception according to which being “justified in believing that p is to believe that p in such a way as to be in a strong position thereby to attain the truth and avoid error. It is to believe that p in a ‘truth conducive’ way” (1986: 193). Similar to Alston, as opposed to best explanationism, Goldman suggests that we adopt a “veritistic” conception of justification (2012: 147). According to veritism, the fundamental, or, cardinal epistemic value is true belief (1999).
For example, see Goldman’s classic arguments in his (1979). Elsewhere, Goldman argues that “The obvious strategy for veritistic unitarianism is to defend a reliabilist theory of justification, or at least some form of truth-linked justification theory…Beliefs are regarded as justified when they are produced by these very truth-conducive processes” (2002: 53, 54).
Reliabilists will be apt to point out that, insofar as we have good reason to believe reliabilism, we also have good reason to believe that justification for PJb is very hard to come by. According to reliabilism, process types like perception generate justified belief only so long as they are sufficiently reliable. In order to justifiedly believe something as strong as PJb, the reliabilist will demand that we use some reliable belief-forming process that is sensitive to the truth-conductivity of perception throughout all future points in time. But it’s implausible that we have ready access to any sort of belief-forming process like this.
Consider Conee and Feldman’s account of evidentialism, according to which believing p is justified for S iff believing p fits S’s evidence, where some p fits a body of evidence E iff p is the best explanation of E (2008: 97, 98). Reasonably, there comes a point at which, after observing N number of green emeralds, S’s evidential base Eg is determinately best explained by q (all emeralds are green). But this means that the trivial feature of observing one green emerald (the difference between N and N-1) makes a non-trivial difference pertaining to the justification of q for S according to Conee and Feldman’s evidentialism.
For brevity, I’ve left off a relevant qualification of Frise’s claim. More precisely, he grants that there can be some cases in which the truth of a belief constitutes a content-sensitive explanation for its justification (936). For example, consider an occurrent demonstrative belief with content that (successfully) refers to an occurrent experience, like, I am having this experience. A belief with this content might very well be justified (at least partially) in virtue of the successful reference of the thought, and hence, in virtue of the truth of its content. Importantly, Frise takes reliabilism to be committed to the thesis that the truth of a belief can constitute a content-neutral explanation of its justification, and it’s this commitment that’s, allegedly, so counter intuitive. For ease of presentation, I’ve left the “content neutral” qualifier out of D1, D2, and D3. However, following Frise, I am assuming a content-neutral notion of explanation in my formulations of these principles.
As Alvin Goldman notes, many theories of justification—both internalist and externalist—are keen to posit some sort of connection between truth and justification (2002). However, reliabilism uniquely analyzes this truth connection in terms of the kinds of truth ratios mentioned in TR.
Contrast the TR account of the truth connection with Earl Conee’s evidentialist account: “Epistemic justification of a proposition is evidence of its truth. The relation of evidential support is the truth connection” (1992: 667). Conee clarifies that the relation of evidential support “is not obviously within the ontology of any current or prospective science” (668). One might have thought that evidential support could be analyzed in terms of naturalistic concepts, but Conee admits that “no such reduction seems to be in the offing for the relation of giving evidence” (668). This is in sharp contrast to how most reliabilists view the truth ratios mentioned in TR—namely as invoking naturalist-friendly notions like event and causation. For further discussion of the relationship between naturalism and reliabilism, see Goldman (1994).
Frise also hints at a related but different sort of difference maker problem that specifically plagues AWR:
Something so trivial as the truth-value of a single belief should not make a nontrivial difference in its formation process’s truth ratio, such that all beliefs formed by that process share one justificatory status rather than another. It appears arbitrary that some processes are reliable in the actual world and are consequently justifying in all worlds. A trivial difference has made all the difference. (936, emphasis mine)
The implication articulated in the italicized portion only follows on AWR, which rigidifies the reliability and justification-conferring properties of process types. This implication does not obtain on SWR and NWR. Hence, insofar as one finds the italicized implication too problematic, one can take this consideration as merely a reason to reject AWR rather than reliabilism per se.
Frise presents versions of this problem for proximate reliability combined with either EW or TF (937). I’ll simply discuss the latter formulation, seeing as how proximate reliability combined with TF was shown to present a plausible solution to the unnatural deduction problem in Sect. 3.
See Conee and Feldman (1998) for an extended discussion of the generality problem.
This approach to the generality problem might be attractive to one who thinks that the concept of justification fundamentally plays the role of allowing us to flag reliable informants in our community (on various matters). See Sosa (1991: 275) for a discussion of this “reliable informant” approach to the concepts of justification and knowledge.
Thanks to an anonymous referee for highlighting the relevance of modal reliabilism to this version of the instability problem.
For an influential defense of a mental accessibility requirement on justificatory factors, see Ginet (1975: 34–36). For arguments against such a requirement, see Alston (1986: 203–215). There, Alston shows that the extant arguments in favor of a mental accessibility requirement on justificatory factors either presuppose the controversial deontological conception of justification, or invoke principles that make the requirements for justification far too demanding (e.g., requiring that a belief have second-order justification just in order to have first-order justification). Also see Bergmann (2006: 13–19) for a similar argument against accessibility or “awareness” requirements on justificatory factors.
In his presentation of the instability problem, Frise presents another version of this argument that seems to invoke a premise like PJb:
And it is plausible that a given perceptual belief is determinately justified, and that a given guess is determinately unjustified, regardless of when formed. We could therefore deduce that perception has performed well enough recently, that its less recent performance isn’t bad, and that the same cannot be said for guessing. These conclusions may be correct, but this way of establishing them is illegitimate (938, 939, emphasis mine).
Here, Frise seems to merely re-raise the unnatural deduction problem for dynamic parameters like proximate reliability—to which I have responded in Sect. 3.
References
Alston, William P. (1985). Concepts of epistemic justification. The Monist, 68(1), 57–89.
Alston, William P. (1986). Internalism and externalism in epistemology. Philosophical Topics., 14(1), 179–221.
Alston, William P. (1988). The deontological conception of epistemic justification. Philosophical Perspectives, 2, 257–299.
Audi, Robert. (2011). Epistemology: A contemporary introduction to the theory of knowledge (3rd ed.). New York: Routledge.
Bergmann, Michael. (2006). Justification without awareness. Oxford: OUP.
Conee, Earl. (1992). The truth connection. Philosophy and Phenomenological Research, 52(3), 657–669.
Conee, Earl, & Feldman, Richard. (1998). The generality problem for reliabilism. Philosophical Studies, 89(1), 1–29.
Conee, E., & Feldman, R. (2008). Evidence. In Q. Smith (Ed.), Epistemology: New essays (pp. 83–104). Oxford: OUP.
Frise, Matthew. (2018). The reliability problem for reliabilism. Philosophical Studies, 175(4), 923–945.
Ginet, Carl. (1975). Knowledge, perception, and memory. Dordrecht: D. Reidel.
Goldman, A. I. (1979). What is justified belief? In G. S. Pappas (Ed.), Justification and knowledge (pp. 1–23). Dordrecht: D. Reidel Publishing.
Goldman, Alvin I. (1994). Naturalistic epistemology and reliabilism. Midwest Studies in Philosophy, 19(1), 301–320.
Goldman, Alvin I. (1999). Knowledge in a social world. Oxford: OUP.
Goldman A. I. (2002). The unity of the epistemic virtues. In Pathways to knowledge (pp. 51–72). Oxford: OUP.
Goldman A. I. (2012). Toward a synthesis of reliabilism and evidentialism? or: evidentialism’s troubles, reliabilism’s rescue package. In Reliabilism and contemporary epistemology (pp. 123–150). Oxford: OUP.
McCain, Kevin. (2014). Evidentialism and epistemic justification. New York: Routledge.
Sosa, Ernest. (1991). Knowledge in perspective. Cambridge: CUP.
Funding
Funding was provided by University of Notre Dame.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Tolly, J. Does reliabilism have a temporality problem?. Philos Stud 176, 2203–2220 (2019). https://doi.org/10.1007/s11098-018-1122-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-018-1122-7