An Internalist Rejoinder to Skepticism
Gregg Ten Elshoff
In what follows, I will offer a rejoinder to one popular and influential version of the skeptical argument which avoids (i) begging questions against the skeptical conclusion, (ii) arguing from prudential considerations to a dismissal of skeptical worries, and (iii) arguing for epistemological conclusions from semantic premises. (1)
I. The Skeptical Argument
Let 'd' connote the dream hypothesis as introduced by Descartes in the first meditation. More specifically, to say that d is true for a subject, X, is to say that X is in fact dreaming and that consequently his empirical beliefs are systematically false. (2) Let 'p' be any empirical belief that X has formed on the basis of his ordinary experience of the world. The truth of p, then, logically implies ~d. Consider the following skeptical argument (I):
(1) If I know that p, then I know that ~d.
I represents a straightforward argument for the conclusion that none of my p-beliefs constitute knowledge. The first premise is grounded in the principle that knowledge is closed under logical implication. More specifically, if I know A, and A entails ~B, then I know ~B. This principle has been challenged, however, on the grounds that we are often unaware of the logical implications of things we know. (3) Fortunately for the skeptic, a weaker closure principle will suffice to get the worry under way. A weaker version of the closure principle would be something like the following: If I know A and that A entails ~B, then I know ~B. This weaker closure principle might be incorporated into the skeptical argument as follows (call this argument II):
(4) If I know p and that p entails ~d then I know ~d.
We will take II to be the skeptical argument under consideration in what follows. Our question will be whether or not II represents a successful defeater of my claims to know p. First, however, a brief comment on each of the premises (4-6). As mentioned, (4) is an application of a plausible version of the principle that knowledge is closed under known logical implication. While some have been willing to reject this principle, I find it quite plausible and consequently, suggest as a desiderata of a response to II that it not make this move. (5) is evident once the d-hypothesis is understood since (at least on some formulations of d) ~p is stipulated. (4) What can be said in support of (6)? Descartes suggested that there is no principled way to distinguish between ordinary perceptual states and dream/hallucination states. The idea seems to be that any evidence X might sight for the hypothesis that ~d will be phenomenologically available to X even if d. But if a decision between d and ~d is underdetermined by all available evidence, then one cannot come to have reasons for preferring d over ~d. Thus, (6) I don't know ~d. So goes the skeptical challenge.
II. An Internalist Defeater of Skepticism
My defeater of II will require two assumptions. The first assumption is that JTB - or, if you like, JTB so modified as to appropriate Gettier counter-examples - represents an adequate analysis of knowledge. More specifically, I will assume that a belief, p, constitutes knowledge for X iff (i) X believes p, (ii) X is justified in believing p and is aware of no successful defeaters for p, and (iii) p is true. (5) The second assumption that I will need is that the justification in (ii) is an internalist constraint. Internalists, however, are a disparate group and so I will need to be a bit more precise about this second assumption.
What I have in mind is an internalist notion according to which assessments of the degree to which a subject, X, is justified in believing p will necessarily make reference to the way things are from the subjects point of view. X's subjective, first-person evaluation of p will be the most relevant piece of evidence when trying to determine the degree to which p is justified for X. Notice, however, that this does not entail an internalist position with respect to knowledge. One can be an internalist in the sense here employed and still maintain that there are externalist constraints on knowledge. In fact, JTB has such a constraint 'built-in' - viz. the truth condition. (6)
I will argue that the success of the skeptical defeater under consideration entails a rejection of either JTB or of internalism - a surprising conclusion given that skepticism considerably predates alternative accounts of knowledge and justification.
An Initial Argument (III)
I begin my rejoinder to II with a simple argument for the conclusion that I have prima facie justification that ~d. Consider the following argument, III:
(8) I have prima facie justification for p (where p is any empirical claim for which I seem to have suitable empirical evidence).
(9) If I have prima facie justification for p, then I have prima facie justification for ~d.
(10) I have prima facie justification for ~d.
Now, on the first reading, this may appear to be just another Moorean formulation of the anti-skeptical argument. Upon closer examination, however, it is evident that neither (8) nor (9) is question-begging against the skeptical conclusion, (7). In fact, both (8) and (9) are perfectly consistent with (7) so long as justification is defeasible (and it is). The problem with Moore's argument is that it contains a premise which is the explicit negation of (7) and so begs the question against the skeptic. Not so with III. In fact, the conclusion of III is consistent, not only with (7) but even with the supposition that d. Even the dreamer has prima facie justification for his (seeming) empirical beliefs. Justification is defeasible, however, and the skeptical argument is best viewed as a potential defeater of my prima facie justification that p. Our question will be whether it succeeds.
Perhaps the skeptic will balk at premise (8). He might be reticent to award my p-belief any justificatory status whatsoever. To my mind, this is an unreasonably bold (and implausible) position. This would be tantamount to claiming that no one (not even those who have never considered hypothesis d) has ever had any justification for any p-belief. So far as justification is internal, this is plainly false. People do have prima facie justification for their p-beliefs. People who have not yet considered II take themselves to have very good reasons for (i.e. to be justified in) their p-beliefs. Skepticism is best viewed as an attempt to defeat such justification - not as an attempt to demonstrate that such justification never existed to begin with. Even if the skeptical argument succeeds, it will not follow that I didn't have prima facie justification for p - it will simply follow that whatever justification I once enjoyed has been successfully defeated.
So much for III. It 's function is to remind the reader of the justificatory status of p prior to an analysis of II and to place II within a larger epistemological framework within which it operates as a potential defeater of a subjects prima facie justification. Upon considering III, then, one is armed with prima facie, defeasible justification that ~d. One is not armed, however, either with knowledge that ~d nor with knowledge that p and so no questions have been begged against II.
The Anti-Skeptical defeater defeater (IV)
Our question is whether or not II succeeds as a defeater of X's prima facie justification that p. II will only succeed as a defeater, however, if there is no successful defeater of it. I will argue that there is a successful defeater of II; and that being the case, II fails as a defeater of X's prima facie justification that p. Consider the following argument (IV).
(11) Either d or ~d.
Premise (11) is analytic. Premise (14) lays out what I take to be an uncontroversial constraint on successful defeaters. Premise (15) follows from (11-13). Premise (16) follows from (14) and (15). The crucial premises, then, are (12) and (13). I will discuss them in turn.
Premise (12) falls directly out of JTB. According to JTB, there are three necessary and jointly sufficient conditions for knowledge - one of which is truth. If JTB is to be an analysis of knowledge, then it will follow analytically from any knowledge claim that these three conditions are met. Alternatively, it will follow analytically from any claim that one of these conditions is not met that the subject does not have knowledge. Now consider (12). Here we have a conditional which has as its antecedent a statement to the effect that one of the necessary conditions for knowledge-that-d is not met and has as its consequent that knowledge-that-d does not obtain. But this follows trivially from our analysis of knowledge. The triviality of (12) can be borne out with the following translation:
(12') If d then it is not the case that (1) I believe ~d and (2) I am justified in believing that ~d and (3) ~d is true.
Given JTB, (12') is a literal translation of (12) and (12') is clearly trivial. The upshot is that if d, nothing like II is needed in order to assert (6), since (6) follows analytically from JTB and the assertion of d. This shouldn't bother the skeptic too much, however, since his project is to demonstrate that the subject fails to have knowledge even if ~d. The really crucial premise of IV, then, is (13). What can be said in support of it?
First, remember that our consideration of the skeptical argument, II, and it's potential defeater, IV have been placed within a larger context which includes III, the conclusion of which is (10) I have prima facie justification for ~d. What I will be arguing is that in this context, and given the antecedent of the conditional in (13), X's p-belief satisfies the three necessary and jointly sufficient constraints on knowledge (viz. JTB) and so qualifies as knowledge. But, if X's p-belief qualifies as knowledge, then (6) is false. At this point, we should watch the skeptic for signs of jumping ship. Before he goes, however, perhaps we can get out of him exactly what the worry with (13) is.
Maybe the worry goes something like this: Even if I have a prima facie justified belief that ~d and ~d is true, it doesn't follow that (6) is false. After all, prima facie justified true belief (pJTB) is not sufficient for knowledge. (6) is just the claim that I don't know that ~d. III may suffice to give you prima facie justification for p (and hence for ~d) but this is a far cry from knowledge that ~d. So, in the absence of an argument to the effect that you know ~d, you are unwarranted in your assertion that (6) is false - even given the truth of ~d.
What can be said in response? First, let us grant that there is a sense in which pJTB is insufficient for knowledge. On the other hand, there is a sense in which pJTB can be sufficient for knowledge. pJTB is ambiguous. It could mean strongly (yet defeasibly) prima facie justified true belief or it could mean merely prima facie justified true belief. Premise (8) in III asserts that I have prima facie justification for p. This misleadingly suggests that I have a merely prima facie justified belief (mpJB) that p. What would be more accurate, however, would be to say that I have an at least prima facie justified belief (apJB) that p. If we revise III to rule out the former reading, we get something like the following (call this III'):
(8') I have at least prima facie justification for p (where p is any empirical claim for which I seem to have suitable empirical evidence).
(9') If I have at least prima facie justification for p, then I have at least prima facie justification for ~d.
(10') I have at least prima facie justification for ~d.
Now if the skeptic was willing to accept III then he should be willing to accept III'. The only thing that has changed is that 'prima facie justification' has been replaced throughout with 'at least prima facie justification'. Since it follows from the fact that a subject has prima facie justification that that subject has at least prima facie justification, this should not be problematic. Further, III' prevents one from mistakenly reading 'prima facie justification' as 'merely prima facie justification' while leaving open the possibility that for any specific case, merely prima facie justification may be all the subject has. Let me pause to review a few abbreviations:
pJTB: prima facie justified true belief.
mpJB: merely prima facie justified belief.
apJB: at least prima facie justified belief.
sJB: strong justified belief (where 'strong' denotes whatever degree of justification is sufficient - together with truth - for knowledge).
With these notions before us, we can analyze III' more carefully. It concludes that I have apJB for ~d. This could mean one of at least two things. It could mean that I have mpJB that ~d in which case (6) is true and the skeptical argument goes undefeated. On the other hand, it could mean that I have sJB (however defeasible) that ~d in which case, given the truth of ~d, I know ~d and (6) is false. So the relevant question for assessing the success of III' has to do with the degree of justification which attaches to my belief that ~d. How strong is my justification for ~d? Well, that depends on the strength of my justification for p. According to (9'), the degree of justification I have for p will transfer to my belief that ~d.
So, how strongly justified is my belief that p? Well, one thing is sure. I'm not certain that p. I could be wrong about it. I've been wrong about empirical beliefs before. Whatever justification I enjoy for p is defeasible. However, to the extent that the constraints on justification are internalist constraints, my p-belief's justification is quite strong. (7) Here I am assuming that p is an ordinary perceptual belief about a medium sized object (seemingly) perceived from a medium distance under (seemingly) normal circumstances. My contention is that these kinds of beliefs virtually always meet many (if not all) of the standard internalist tests for strength of justification. (8) Since my p-belief ranks quite high on any of these tests, I can conclude that my belief that p (and hence my belief that ~d) is a sJB.
But is p (and hence ~d) true? After all, continues the skeptic, sJB is not sufficient for knowledge. What we need is true sJB. Now the objector simply needs to be reminded that premise (13) is a conditional premise. So we are evaluating the truth of (6) given ~d. Given the truth of ~d and my sJB for p (and hence for ~d), I can conclude that I know ~d (i.e. I have a strong justified true belief that ~d). If I know ~d, then (6) is false and (13) is vindicated.
One final objection: But you've just begged the question by stipulating the truth of ~d as the antecedent of the conditional premise (13). It is only by so stipulating that you were able to make the move from sJB to knowledge.
Reply: (13) begs no questions against the skeptical conclusion. The reason it does not is because it falls within an argument context which takes the form of a disjunctive syllogism. I can assume the truth of ~d as the antecedent in (13) only because ~d is one of the original disjuncts being discharged in argument III. III begs no questions against the skeptical conclusion because it considers the viability of (6) given both d and it's negation. Since (6) didn't fare well on either score, III succeeds as a defeater of the skeptical defeater, II.
In closing, I will briefly remark on what I take to be the significance of IV. I do not take IV to be a refutation of skepticism. I do take it to be a successful defeater of potential skeptical defeaters for the justification of p-beliefs which depend on some hypothesis d. The success of this defeater-defeater, however, does depend on JTB and an internalist notion of justification. I take both to be viable positions in epistemology generally (although I have not argued for this here) and so I take this skeptical challenge to be successfully rebutted. If the reader is inclined to reject one or both of these assumptions, then he might take this paper as purporting to draw out an interesting commitment (viz. that the skeptic is committed either to a rejection of JTB or internalism or both). Why is this any more interesting than the conclusion that the skeptic is committed to the rejection of the causal theory of reference? (9) Let me briefly suggest two reasons.
First, external world skepticism is an old problem. Alternative theories of knowledge and justification, however, are relatively new. JTB was widely accepted until Gettier, and externalism has only picked up speed in recent decades largely in response to the same problem. Second, there are those who have moved away from internalism in an attempt to deal with the skeptical challenges here discussed. If IV is successful, then this move is unmotivated an ill-advised. It is the internalist who possesses the resources necessary for rebutting the skeptic. Indeed, if the argument of this paper succeeds, then it is the skeptic - and not the realist - who must deviate from traditional theories of knowledge and justification.
(1) To give the reader an idea of the kinds of approaches being avoided, consider G.E. Moore to be representative of (i), Hume to be representative of (ii), and Putnam to be representative of (iii).
(2) Obviously, 'd' as it stands does not entail such systematic failure since a subjects dream-beliefs might be coincidentally true. To rule out such happy coincidences, one might strengthen the skeptical hypothesis, d, with the suggestion that d includes the idea that the only thing(s) in existence are X sleeping or X's mind and an evil demon or X's brain and a brain-sustaining vat.
(3) For a development of this challenge, see Fred Dretske's "Epistemic Operators", The Journal of Philosophy, Vol. 67, No. 24 (Dec. 24, 1970), pp. 1007-1023.
(4) see footnote 2
(5) I don't intend to defend this analysis of knowledge here. I simply maintain that, right or wrong, this is akin to the traditional analysis of knowledge as understood by most epistemologists. For a contemporary defense of a similar analysis, see Lehrer and Paxson, "Knowledge: Undefeated Justified True Belief", in Pojman's The Theory of Knowledge.
(6) For a developed account of internalism along these lines, see Roderick M. Chisholm, Theory of Knowledge, third edition (Prentice-Hall, 1989) especially p.7
(7) Remember, what is being assessed here is the status of my p-belief's justification prior to a consideration of II. What we are trying to assess is the strength of my prima facie justification for p. Until we have a fix on the degree of justification which attaches to p prior to a consideration of potential skeptical defeaters, we will not be able make judgments about what will be required of a successful defeater of p.
(8) What I have in mind here are tests for consistency, coherence, un-give-up-ability, etc. Different internalist approaches will offer different tests for strength of justification. It is not my purpose here to go into them in any detail. What is being claimed is that to the extent that any such test qualifies as an internalist test (as defined above) it will afford a high degree of justification to p-beliefs.
(9) Recall Putnam's semantic argument against skepticism in Reason Truth and History.