Evidence, experience, and externalism

Abstract

The Sellarsian dilemma is a famous argument that attempts to show that nondoxastic experiential states cannot confer justification on basic beliefs. The usual conclusion of the Sellarsian dilemma is a coherentist epistemology, and the usual response to the dilemma is to find it quite unconvincing. By distinguishing between two importantly different justification relations (evidential and nonevidential), I hope to show that the Sellarsian dilemma, or something like it, does offer a powerful argument against standard nondoxastic foundationalist theories. But this reconceived version of the argument does not support coherentism. Instead, I use it to argue for a strongly externalist epistemology.

Notes

1 ‘Standard’ versions of foundationalism might not be traditional versions. The traditional view may have been that basic beliefs are self-justifying, but few or no foundationalists hold that today. Instead, most hold that the basic beliefs (whether beliefs about external objects or about one's own current mental states) are justified by nondoxastic appearance states. The beliefs would still count as epistemologically basic, a basic belief being one whose justification does not rely on support from other beliefs.

2 I lack the space here to argue against doxastic theories, which hold that only beliefs can serve as justifiers. However, the widespread exodus from such theories in recent years suggests that the famous isolation argument, coupled with the observation that any belief can be held for bad reasons, have convinced most epistemologists that doxasticism is a lost cause. It is worth pointing out in this connection that the recent renaissance of old-fashioned foundationalism [DePaul 2001 has been a rebirth of nondoxastic foundationalism, which allows experiential states a crucial epistemological role.

3 For discussions of the basing relation and the controversies surrounding it, compare Korcz 1997; 2000, Pollock 1986, Kvanvig 2003b.

4 Thus, the controversy surrounding the assumption that opens this paragraph is likely more terminological than substantive. Evidentialism as Feldman and Conee see it 1985; Feldman 2003 is perhaps primarily a theory of what I am calling justifiability rather than justifiedness. (Others [e.g., Haack 1993 use ‘evidentialism’ differently.) Feldman and Conee use ‘justification’ to refer to a belief's fitting all the evidence S possesses and ‘well-foundedness’ for a belief's being based on appropriate evidence. What they call justified but not well-founded, I call justifiable but not justified. The dispute here is largely, perhaps entirely, terminological. I define evidential justifiers as justifying grounds, rather than, say, potential justifying grounds, because my focus is an justifiedness rather than justifiability. I believe that the major theses of this paper could be translated into Feldman and Conee's language, though I won't pursue this.

5 One might hold, as a matter of substantive doctrine, that a justified belief must be grounded in an at least tacit awareness of all the factors relevant to that belief's justification; though I think such a view is mistaken, I am not denying it here but merely clarifying the conceptual distinction between evidential and nonevidential justifiers.

6 Though it may be uncontroversial that the deductive relation nonevidentially justifies the belief, epistemologists with different theoretical commitments will certainly disagree about why this is true. That is, we will disagree about what the deeper nonevidential justifiers are: does deductive inference justify because it is reliable? because it increases overall coherence? because it exemplifies intellectual virtue? etc.

7 Standard versions of coherentism [e.g., Lehrer 1990; BonJour 1985 endorse (a). One might count as a ‘nondoxastic coherentist’ either by embracing a more standard coherentism about only some beliefs (and a nondoxastic foundationalism about the others), or by allowing experiences a justificatory role, but only in the presence of other beliefs [e.g., Kvanvig 2003a; Gupta 2006. Most, though perhaps not all, of the present discussion of nondoxastic foundationalism will apply to nondoxastic coherentism as well.

8 Conee and Feldman 2005, for example, hold a doctrine of ‘strong supervenience’, a.k.a. ‘cognitive essentialism’[Pollock 1986, according to which evidential relations hold necessarily. This is presumably compatible with their mentalism: the view that only mental states can serve as evidence. Yet the fact that my evidence is good evidence is not itself a mental state; nor is the fact that I possess said evidence. Thus, even on this strongly evidentialist and internalist view, there are things that are relevant to justification but do not serve as evidence; i.e., there are nonevidential J-factors.

9 Versions of this argument have been endorsed by Sosa 1980, van Cleve 1985, and Steup 1996, 2000.

10 This sort of point generalizes to all so-called doxastic theories, but I will restrict my attention to coherentism here to simplify matters.

11 It is unclear that supervenience implies even this much. Premise 4 of SA is only obviously true if we read ‘confer justification on’ as ‘evidentially justify’.

12 Standard characterizations of doxasticism tend to run together several distinct theses. See Pollock 1986: 19], Pollock & Cruz 1999: 22], where one of the two formulations of doxasticism includes the view that no two doxastically identical agents can differ epistemically. Not only does this overlook the evidential/nonevidential justifier distinction, but it leaves no room for the notion of basing/grounding at all. Surely two agents can share all their beliefs even though some of their beliefs have different bases (especially given Pollock and Cruz's endorsement of a causal theory of the basing relation). Pollock and Cruz also seem to equate ‘nondoxastic foundationalism’ with a view I call external object foundationalism: that version of foundationalism according to which some beliefs about ordinary physical objects are basic (though Pollock 2001 rectifies this). Steup 2000: 76 – 8] seems to make this last mistake as well, though in a different terminology. Most traditional foundationalists, however [e.g., Chisholm 1977; 1989; BonJour 2001; 2002; and Fumerton 1995; 2001, are nondoxastic foundationalists who nonetheless reject external object foundationalism.

13 Other adherents to experientialism likely include Alston 2002, Audi 1998, BonJour 1999; 2001; 2002, Brewer 1999, Chisholm 1966; 1977; 1989, Fumerton 1985; 2001, Haack 1993, Huemer 2001, Pollock 1986, Pryor 2000, Quinton 1966, Reynolds 1991, and Steup 1996; 2000.

14 Although doxasticism is usually taken to imply internalism, one could conceivably maintain that what determines whether a given belief counts as evidence for some other belief is some factor external to the agent. Such a view would parallel Alston's externalist yet evidentialist view, mentioned above.

15 By following the contemporary literature in calling the ensuing argument ‘the Sellarsian dilemma’ I do not thereby intend to make any claims about whether Sellars himself endorsed any part of this. The Sellarsian dilemma has appeared in many forms, perhaps the clearest and most influential of which is in BonJour 1985, whom I follow roughly here, and who claims to find it or something like it in Sellars 1956.

16 The version of the argument treated here begins with the question of whether experiential states are propositional, though other variants ask whether they are conceptual, others whether they are cognitive. These differences will not affect the main points of this paper.

17 I assume, but do not argue, the unacceptability of doxasticism. See note 2, above.

18 Recall that experiential states are being invoked here to explain the justification of basic beliefs. Thus they must evidentially justify these beliefs in the absence of any other beliefs. It is this that makes their nonpropositional status troublesome; otherwise, they might serve as evidence in the way that, say, a smoking gun—which obviously lacks propositional content—does.

19 Thanks to correspondence with James Montmarquet for this phrase.

20 At the very least, the state must have a content, and this content must be in some sense commensurate with the content of the justificandum belief in order for the relation to be an evidential one. Perhaps those variants of the Sellarsian dilemma that proceed in terms of cognitive or conceptual content rather than propositionality fare better in this regard, though I won't try to sort out those differences here.

21 This point has also been noticed, I think independently, by Aaron Champene.

22 This point is brought out vividly, though inadvertently, by Steup 2000, who argues that experiential states, though propositional, are not thereby in need of or susceptible to epistemic justification, just as other nondoxastic propositional attitudes, like desires, are not in need of or susceptible to epistemic justification. This analogy does him more harm than good, since it is obvious that desires cannot serve to evidentially justify beliefs.

23 Additional necessary conditions might well hold: the agent may need to believe or be justified in believing that the premises support the conclusion, etc.

24 An objection along these lines was offered by an anonymous referee for this journal.

25 The previous objection might be formulated by employing this broader sense of ‘evidence’. Again, however, if experiences only play the epistemological role that certain distal states of affairs play, then they lose the special status distinctive of experientialism.

26 Thanks to Aaron Champene, Alvin Goldman, James Montmarquet, Patrick Rysiew, Thomas Senor, and two anonymous referees for this journal for discussions of this material or comments on earlier drafts of this paper. A longer version of the argument presented here can be found in Lyons (forthcoming).