By DOUG SHAVER
June 9, 2009
Reliabilism is one of several alternatives to the traditional analysis of knowledge as justified true belief (JTB) that were proposed in response to objections raised against JTB. Various objections were raised in turn against those alternatives, including reliabilism, and as a result JTB appears to continue as the predominant theory of knowledge. This paper will summarize the arguments for and against reliabilism and suggest an accommodation between it and JTB.
Gettier is usually credited with the initial discovery of problems with JTB.1 His article offered two counterexamples demonstrating the possibility of a subject's having a justified true belief in some proposition that, according to our intuition, he does not in fact know.2 One reason for the failure of knowledge to obtain in those cases was a disconnect between the subject's reason for believing that particular fact that made his belief true, and so various externalist theories, including reliabilism, were proposed in the hope of remedying that defect. Another motivation was that externalism seemed to avoid the problems of infinite regress to which internalist theories were susceptible.3
Reliabilism holds that justification consists in a belief's having been, in BonJour's phrasing, "produced or caused in a way or via a process that makes it objectively likely that the belief is true."4 Canonical examples of such processes include sensory perception, valid reasoning, and memory. Thus, if I see a tree in my back yard, then, assuming there actually is a tree there, I know there is a tree in my back yard. The reason I know is that it is nearly always that case that whatever I see actually exists and actually is in the place where I see it. It is the fact of this reliability that constitutes my justification for believing that I see the tree in that place. I need not, however, be aware of this fact in order to derive justification from it, and this is what makes reliabilism an externalist theory of knowledge.
Feldman summarizes the case against reliabilism with three objections. One, the "brain in a vat" objection, purports to demonstrate the possibility of a reliable process that produces mostly false beliefs. Another, the "accidental or unknown reliability" objection, discusses hypothetical situations in which (a) a subject's beliefs are true, (b) they arise from a process that actually is reliable, but (c) the subject knows nothing about or has some reason to doubt the reliability of the process. (The paradigmatic example is a naive clairvoyant.) Reliabilist theory says the subject in these cases has a justified belief, but this is contrary to our intuition. The third objection is the generality problem. This is the difficulty of identifying, in any given instance of belief-formation, the specific characteristics of the process producing or causing that belief that make it reliable, so that we could formulate a rule for distinguishing reliable from unreliable processes by recognizing the presence or absence of those characteristics.5
With these objections in view, Feldman seems to have a cogent argument for his conclusion that "reliabilists have difficult problems to solve."6 If the brain in a vat is using a reliable process, then reliability is clearly not a sufficient condition of knowledge, but to argue against its reliability is to raise the generality problem. Reliability is stipulated for the naive clairvoyant, but it again seems insufficient in light of our intuition that, given the present state of scientific knowledge, everyone should distrust clairvoyance. It seems improbable that a belief can count as knowledge for someone who has good reason to reject that belief. And, even if reliability could be proved sufficient, the generality problem seems to undermine its theoretical utility. Any particular process that induces a belief in someone can be considered an instance or token of an indefinitely large number of types of processes, and those types will have various track records of reliability ranging from poor to excellent. There is not yet an agreed-upon way to ascertain which type any particular instance ought to be associated with for determining its reliability, and until there is, reliability theory cannot be usefully applied to a judgment of whether a particular person S, using a particular process P to acquire belief in a particular proposition q, knows q.
These problems reinforce a more general argument against any externalist theory of knowledge. BonJour encapsulates it as follows.
Do we have any good reasons for thinking that our beliefs about the world are true (or at least approximately true)? And if so, what specific form or forms do these reasons take? I suggest . . . that the issue posed by these questions is the most central one in epistemology. . . .
The first thing to notice about this issue is that it is essentially a first-person issue, one that is in fact better captured in a first-person singular formulation than in the first-person plural formulation just employed. The basic question (which each person must in the end ask for himself or herself) is whether I have good reasons for thinking that my beliefs are true (and, if so, what form those reasons take).7
As BonJour goes on to observe, "the reasons in question are supposed to be reasons that I have" in the sense that they are cognitively accessible. Whatever the reliability of my doxastic sources, if I do not know whether they are reliable, then I do not know whether my beliefs are justified, and if that is my situation, then it is not apparent why I should not revert to pure skepticism. Even if someone else can tell me from their externalist perspective that I am using reliable sources, I will have no reason to take their word for it.
Pure reliabilism, then, seems not to work as a theory of knowledge, but it does not follow that the reliability of our doxastic sources is irrelevant to an internalist theory such as foundationalism. The quest for justification of basic beliefs appears really to be an effort to prove Feldman's Two Theses.9 There is a list of propositions that we are uncontroversially said to know (Thesis 1). Then we extract a definition of "know" from its use in relevant contexts. That definition tells us: (a) We believe those propositions; (b) We believe them for good reasons, and those reasons involve the sources, which include sensory perception, listed in Thesis 2; and (c) The propositions are true as a matter of fact.
But how do we confirm whether (c) obtains? We cannot. We have to regard (c) as an axiom.If we do not — if (c) cannot be confirmed — then we have no examples of knowledge, and epistemologists are investigating a subject without content. This is not to bar the question "But how do we know those things?" It is just to note rather that in order to ask that question, we would have to define knowledge in a way that does not presuppose those propositions to be instances of it. In most epistology debates about how we might justify believing that what we see is really there, it seems to be simply stipulated that something, as a matter of fact, is there — a red triangle, a chicken, a fake sheep, whatever. So, notwithstanding frequent thought experiments about what would be our epistemic status in the Matrix or a demon world, epistemologists routinely take it for granted that the world we all think we are aware of is real and is more or less the way it looks, sounds, feels, smells, and tastes, at least most of the time under ordinary circumstances. We may call this supposition, that the world's existence and gross characteristics actually are as they ordinarily appear to be, the Axiom of Apparent Reality.
And so if we assume this axiom, then we necessarily assume the general reliability of our perceptions, since according to the axiom the world actually is approximately the way it looks to us. Our perceptions are therefore reliable by definition, and this gives us a Reliability Hypothesis: Our perceptions usually produce true beliefs. If the Axiom of Apparent Reliability is false, then most if not all of what we think we know is false (and so is the Reliability Hypothesis), and in that case the extent of our actual knowledge is meager or nonexistent. This would be particularly so of our foundational beliefs, and in that case the quest for justification would appear to be misguided, since what we should want to do with false beliefs is discover their falsity, not justify them.
As everyone arguably is entitled to their epistemic systems pending a credible challenge,10 so might we all be blindly entitled to believe our senses pending a confrontation with compelling evidence that they could be deceiving us. Most of us cannot believe otherwise than that our senses are generally reliable under most circumstances. We also, although aware of hypothetical possibilities that this belief could be mistaken, are aware of the apparent nonexistence of any evidence supporting such a possibility. Until a credible challenge transpires, Feldman's Two Theses seem secure, and they presuppose the Axiom of Apparent Reality, which incorporates the Reliability Hypothesis. To at least that extent, some kind of reliabilism has a place in foundationalist epistemology.
(The original version of this essay was written for a class in epistemology taught by Dr. Matthew Davidson at California State University, San Bernardino.)
__________
Notes
1. Richard Feldman, Epistemology, 25.
2. Edmund L. Gettier, "Is Justified True Belief Knowledge?"
3. Laurence BonJour and Ernest Sosa, Epistemic Justification, 24.
4. BonJour and Sosa, 25.
5. Feldman, 94-99.
6. Feldman, 99.
7. BonJour and Sosa, 174.
8. Ibid.
9. Feldman, 3-4. He labels the theses SV1 and SV2.
10. Paul Boghossian, Fear of Knowledge, 99.
Bibliography
Boghossian, Paul. Fear of Knowledge: Against Relativism and Constructivism. New York: Oxford University Press, 2006.
BonJour, Laurence, and Ernest Sosa. Epistemic Justification: Internalism Vs. Externalism, Foundations Vs. Virtues. Malden, MA: Blackwell, 2003.
Feldman, Richard. Epistemology. Upper Saddle River, NJ: Prentice-Hall, 2003.
Gettier, Edmund L. "Is Justified True Belief Knowledge?" Analysis 23 (1963): 121-23. http://www.ditext.com/gettier/gettier.html. (Accessed May 28, 2009.)
(This page last updated on January 5, 2022.)