Simplicity and the Sceptic: Comparing and Contrasting W.V. Quine and Kripke's Wittgenstein

At first blush, this dispute may not appear to offer much philosophical excitement, not only because it has the look and feel of an irresolvable case of Quine says/KW says but also because the import of its resolution is less than obvious. However, the dispute is not irresolvable and the philosophical payoff of its resolution (in Quine's favor) is rather significant, viz., it allows us to reject KW's case against private language. In particular, I'll show that KW's main argument against using SCs to help us decide between competing meaning hypotheses leans on an unsupported premise that is not only at odds with Quine's account, as well as the rest of KW's account, but which simply cannot be accepted by anyone who allows that there is a sceptical solution to KW's sceptical paradox. Since KW is irrevocably committed to there being a sceptical solution to his sceptical problem, I conclude that he argues in bad faith and, with Quine's assistance, conclude that SCs can be used to justify choosing between competing meaning hypotheses. So much for the irresolvability of the dispute.

As for the significance of this result, recall that KW's sceptic purports to establish the arbitrariness of our rule following and the nonsensicalness of our meaning one function rather than another, by showing that there are no "meaning facts", i.e., facts that would, e.g., establish that someone means/meant plus rather than quus. (K, p. 13). Those of us who wish to resist the sceptic's conclusion allegedly have two solution options, viz., a "straight solution", which would involve producing, discovering or otherwise inventing meaning facts, or a "sceptical solution", i.e., a way of blocking the sceptic's conclusion while yet acknowledging the nonexistence of meaning facts. Notoriously, KW not only opts for a sceptical solution, he also contends that it is our only option.

KW's sceptical solution recommends a shift from a truth conditional account of language to one based on justification or assertion conditions. Allegedly, such a shift allows us to legitimate our usual rule following and meaning talk (effectively blocking the sceptic's "insane and intolerable" conclusion that rule following and meaning talk is arbitrary and nonsensical) while yet accepting the nonexistence of meaning facts.4 For my purposes, the most important feature of KW's sceptical solution is that it allegedly has the impossibility of private language as a corollary (K, p. 60 and p. 68). More specifically, KW contends that the assertion conditions for attributing correct or incorrect rule following to an individual considered in isolation are "simply that [his/her] own authority is unconditionally to be accepted", a situation which is obviously at odds with our usual concept of following a rule. (K, pp. 88-89). According to KW, it is only in a community setting that there can be substantive assertion conditions for the attribution of correct or incorrect rule following. Such is KW's case for the impossibility of private language.

Donald Davidson is, as far as I can tell, the only commentator to have appreciated that KW's conclusion about the impossibility of private language is unwarranted.5 In particular, Davidson rightly notes that KW has shown, at best, that his own assertion conditions-based sceptical solution is incompatible with private language. But, suggests Davidson, there very well could be, for all KW shows, sceptical solutions which are compatible with private language. Of course, it's one thing to note that KW has not ruled out the possibility of solutions which are compatible with private language and it's another to actually find such a solution. Simplicity considerations are, I contend, capable of providing just such a solution, i.e., SCs provide a sceptical solution to the paradox which does not, a la KW's own solution, yield the impossibility of private language as a corollary.

Simply put, by allowing us to justify a choice between competing meaning hypotheses, SCs enable us to justify rejecting the quus hypothesis in favor of the plus hypothesis, which in turn is enough to solve KW's sceptical problem, albeit sceptically.6 Most importantly, since an SCs-based sceptical solution is obviously unaffected by the number of people whose talk or rule following is being assessed, this solution will not be, a la KW's own, inapplicable to a person considered in isolation. Of course, for those benighted enough to want a straight solution to KW's sceptical problem (something which is not to be confused with wanting a way to ignore or otherwise reject the sceptical problem), such a result is cold comfort. But for the rest of us, i.e., those of us willing to grant the nonexistence of meaning facts, the ability of SCs to justify the plus hypothesis over the quus hypothesis and so sceptically solve the sceptical paradox, is extremely important, for it allows us to reject KW's case against private language.

The remainder of my paper will be devoted to making the case for SCs providing a sceptical solution that is compatible with private language. I begin by showing that Quine allows SCs to justify choosing between competing meaning hypotheses and then turn to an examination of KW's argument against our being able to use SCs to justify such choices. I find that there is nothing in KW's argument to threaten Quine's position. More specifically, KW will be seen to have provided an unsuccessful argument against SCs providing a sceptical solution to the paradox. Quine's authority, bolstered by KW's failure, leads to the conclusion that not only can SCs provide, ceteris paribus, a sceptical solution to the paradox but they provide a solution which, unlike KW's own solution, poses no threat at all to the possibility of private language.

II. Quine on the use of SCs to choose between competing translation manuals

Quine's views on SCs are most succinctly and thoroughly stated in his "Reply to Gibson" in The Philosophy of W.V. Quine.7 In this reply, Quine probably startled a few people by allowing that there is no bar to using SCs to choose between competing manuals of translation. He obviously startled Follesdal, who, Gibson tells us, argues that for Quine, SCs play a different role in physics from that which they can play in translation. To his credit, Gibson sees Follesdal as "assigning simplicity dubiously different roles in physics and linguistics", and Quine, of course, concurs. He probably also startled Rorty, who, Gibson tells us, believes that if Quine allows that SCs can be used to choose between competing manuals then he ought also allow that there is a fact of the matter to translation. He may have even surprised Gibson a bit, since Gibson seems unaware that Quine allows for the possibility that SCs permit us to reject all but one of our competing manuals of translation.

Without further ado, here are Quine's remarks on the role of SCs in helping us choose between competing manuals of translation:

There are indeed further virtues that we can seek and agree on,

in devising a manual of translation, besides conformity to verbal

dispositions. We can seek simplicity, and we can try to maximize

truth in the natives' assertions. If these and kindred canons of

procedure suffice to weed out all conflicting codifications of the

native's verbal dispositions, which is doubtful, I would still say

that they confer no fact of the matter. This is . . . because . . . the

factuality is limited to the verbal dispositions themselves, however

elegantly or clumsily codified. (Q, p. 155).8

For my purposes, the two most important points here are, first, that Quine allows that SCs can be used to justify choosing a translation manual over a "competing manual" (i.e., one that is satisfied by the same speech dispositions as the first one), even to the point of eliminating all competing manuals, and second, that Quine insists that even if SCs allow us to find a single, simplest manual, we still get no fact of the matter about translation. More particularly, it should be clear that insofar as Quine allows SCs to justify choosing one competing translation manual over another, he finds no difficulty with understanding what the manuals state, despite the nonexistence of a fact to decide between them. Though I will say a bit more about Quine's view below, this is enough to bring out a contrast with KW.

III. Kripke on the use of SCs to choose between meaning hypotheses

On pp. 38-9 of his book, Kripke (presumably in the name of KW, so I'll use 'KW' henceforth) savages the very idea that SCs can be of any use against rule scepticism, providing a preemptive strike against anyone who would dare suggest that they can help solve the book's sceptical challenge. Indeed, KW goes so far as to say that the suggestion "that the hypothesis that I meant plus is to be preferred [to the hypothesis that I meant quus] as the simplest hypothesis . . . must be based either on a misunderstanding of the sceptical problem, or of the role of simplicity considerations, or both." (K, p. 38). That KW sees the attempt to solve the sceptical problem via SCs as wrong-headed is obvious. No less obvious, at least to veterans of Kripke's book, is the need for KW to establish that SCs cannot be used to justify a choice between the plus and quus hypotheses. I will show here that this need is not met and that KW's argument against the use of SCs shows, at best, that they cannot provide a straight solution to the sceptical challenge.

KW seems to have two complaints against the use of SCs, both of which are contained in his opening salvo against them:

Recall that the sceptical problem was not merely epistemic.

The sceptic argues that there is no fact as to what I meant,

whether plus or quus. Now simplicity considerations can help

us decide between competing hypotheses, but they obviously

can never tell us what the competing hypotheses are. If we

do not understand what two hypotheses state, what does

it mean to say that one is 'more probable' because it's

'simpler'? If the two competing hypotheses are not genuine

hypotheses, not assertions of genuine matters of fact, no

'simplicity' considerations will make them so. (K, p. 38).

Here KW is claiming:

(i) the nonfactual character of the plus and quus hypotheses (assuming the correctness of the sceptic's case against meaning facts, of course)9 prevents us from understanding what the hypotheses are, or what they state, rendering SCs unusable,

and,

(ii) the plus and quus hypotheses are not "genuine hypotheses, not assertions of genuine matters of fact", and SCs are powerless to turn them into "genuine hypotheses".

KW's first complaint against SCs appears to be in direct conflict with Quine's view. Most prominently, KW appears to hold that SCs can't be used to justify a choice between competing meaning hypotheses because, for want of a fact to show which hypothesis we meant, we allegedly don't know what the hypotheses state. Since I think Quine would be sympathetic to KW's contention that if we don't know what two meaning hypotheses state, it's nonsense to speak of one of the hypotheses as simpler than the other (indeed, we should all be sympathetic to this), it's clear that Quine denies that the nonexistence of "meaning facts" prevents us from understanding what meaning hypotheses state. In short, Quine rejects KW's assumption that our understanding of meaning hypotheses is threatened by the nonexistence of a fact to decide between them.

So the conflict between KW and Quine on the use of SCs appears to be due to their having very different views about whether our understanding of meaning hypotheses requires the existence of meaning facts that decide between them. However, the potentially irresolvable debate on who is correct on this matter can be avoided by appreciating that not only is there much in KW's text to suggest that he does not accept (i) as true, but (i) can also be seen to be incompatible with the very possibility of a sceptical solution. It is noteworthy that neither before nor after KW's attempt to dismiss the use of SCs is there any suggestion from him that the lack of meaning facts leaves us unable to understand what the plus and quus hypotheses state. On the contrary, KW allows (indeed, must allow) that we understand our meaning hypotheses well enough to appreciate that my past history (both "public" and "private") is "equally compatible" with both hypotheses. (K, p. 15). He also allows, at several points, that his sceptic in no way casts doubt on our understanding of what is required of someone who means plus or quus. (See, e.g., K, pp. 12-13). Indeed, if Hilary Putnam is to be believed, it's even the case that at least one of the sceptic's meaning hypotheses is understandable enough to admit of empirical refutation.10

Needless to say, all of this makes it hard to see how KW can grant the nonexistence of meaning facts and accept (i) above. Since KW is obviously committed to the nonexistence of meaning facts, he cannot, I contend, really accept (i). Rather, (i) appears to be little more than an ad hoc device for rejecting the use of SCs to solve the sceptical challenge. Furthermore, (i) appears to be incompatible with the very possibility of a sceptical solution to the sceptical challenge. For such a solution begins by granting the nonexistence of meaning facts. But given (i), this leads to the conclusion that we don't know what our meaning hypotheses state. It is clear then that KW cannot really accept (i) above.11

Simply put then, not only is Quine's view the correct view but it is also KW's view as well. But the point admits of further clarification and support by appeal to an important distinction of Quine's. The distinction in question is that between the nonfactualness of a choice between competing meaning hypotheses and the nonfactualness of the meaning hypotheses themselves. Quine makes this distinction explicit in his "Reply to Roth", as follows:

Farther along [Roth] ascribes to me a thesis (F) to the effect that

"there is no warrant . . . for attributing a fact of the matter to semantic

theories." On the contrary, the conformity of a translation manual to

speech dispositions is decidedly a matter of fact. It is only the choice

between certain rival manuals that lacks factuality. (Q, pp. 459-60).

Much ink could have been spared if Kripke and his commentators kept this distinction in mind. For the distinction allows us to see that the sceptic's case against meaning facts is narrower than advertised or commonly understood. In particular, KW's sceptic simply does not establish the nonexistence of any and all facts of meaning. Rather, following Quine's lead, we can see that KW's sceptic shows, at best, that the choice between the plus and quus hypotheses is nonfactual. By my lights, a fact that would justify such a choice is what KW means, or ought to mean, by a "superlative fact". Of course, I am urging then that the lack of such a superlative fact poses no threat to our understanding of what our meaning hypotheses state, for either Quine or KW. As such, KW's first argument against the use of SCs is unsuccessful. For given that we can understand our meaning hypotheses despite the nonexistence of superlative facts, nothing stands in the way of our using SCs to provide a justification for choosing the plus hypothesis over the quus hypothesis.

In contrast to KW's first complaint against SCs, his second complaint, viz., "that the plus and quus hypotheses are not "genuine hypotheses, not assertions of genuine matters of fact", and SCs cannot make them so", doesn't seem to conflict with Quine's view at all. For this complaint does not rule out using SCs to justify choosing between competing meaning hypotheses (or translation manuals). Instead, the complaint says, that a "genuine hypothesis" just is an hypothesis that asserts a genuine matter of fact, and second, that the plus and quus hypotheses do not state genuine matters of fact, and third, even if SCs lead us to choose, say, the plus hypothesis over the quus hypothesis, we still can't call the plus hypothesis a "genuine hypothesis", can't say that it asserts a genuine matter of fact. But since KW requires of "genuine hypotheses" that there be a fact to decide between them (K, pp. 38-9), what he says is both perfectly correct, as well as in line with Quine's view on the matter.

For given KW's understanding of "genuine hypothesis", it's clear that Quine's manuals of translation are not "genuine hypotheses", and the same goes for the analytical hypotheses that allow us to create a manual in the first place. But genuine or not, our meaning hypotheses (and Quine's analytical hypotheses) are understandable and so there's no good reason, for either Quine or KW, to think that SCs cannot be used to choose between them. Quine simply insists that the use of SCs to decide between competing hypotheses does not yield a fact of the matter about which hypothesis is correct. Since such a fact is what is needed to turn Quine's manuals and analytical hypotheses into "genuine hypotheses", it's clear that Quine agrees with KW that even if SCs allow us to discard all but one manual, one meaning hypothesis, this can't turn the "nongenuine" meaning hypothesis into a genuine hypothesis in KW's sense. Just as clear, however, is the fact that none of this poses any difficulty for those who advocate using SCs to choose between competing meaning hypotheses.

But this raises the following question: Since the fact that SCs are unable to turn nongenuine hypotheses into genuine ones is compatible with using SCs to justify choosing one hypothesis over another, why does KW bother reminding us that SCs won't produce meaning facts? The simple answer is that by reminding us that SCs cannot produce meaning facts, (or, in KW's terms, that SCs can't turn nongenuine hypotheses into genuine hypotheses), KW has at least shown that SCs cannot provide a straight solution to the sceptical paradox. Recall that a straight solution is one that provides a fact that shows which of plus or quus is meant, and which thereby justifies our going on one way rather than another, which in turn prevents our language from being a series of "unjustified leaps in the dark". Since it is obvious that SCs are not in the fact-producing business, regardless of whether they allow us to justify choosing one meaning hypothesis as "the simplest hypothesis", it is clear that KW has at least established that SCs can't provide a straight solution to the sceptical paradox. It's also clear that Quine agrees that SCs are unable to produce a fact of the matter about which of two competing meaning hypotheses is the "right" one. As such, Quine would allow that SCs cannot provide a straight solution to KW's sceptical paradox, as should we all. Still, this does nothing to show that SCs cannot provide a sceptical solution to the paradox.

IV. Conclusion

Be the question of determining the relative simplicity of KW's competing meaning hypotheses as it may12, it has been shown that KW's case against using SCs to solve the sceptical problem is only partially successful, for it shows, at best, that SCs cannot provide a straight solution to the sceptical problem. Since a sceptical solution is, according to KW, all that one can hope to provide against the sceptical problem, his failure to show that SCs cannot provide a sceptical solution to the paradox is a serious shortcoming. It is so primarily because, unlike KW's own sceptical solution, an SCs-based sceptical solution does not yield the impossibility of private language as a corollary. For the use of SCs can be used to rule out competing meaning hypotheses, whatever the number of people considered. Though I have no desire to silence those still intent on finding fault with the details of KW's sceptical solution or his case against private language, let alone claim to have rendered previous such critiques otiose, I might be allowed to have presented a critique of KW's enterprise that is preferable because simpler.

ENDNOTES

1 Saul Kripke, Wittgenstein on Rules and Private Language (Cambridge, Massachusetts: Harvard University Press, 1982). Hereafter referred to by K. This paper presupposes some familiarity with the main points of Kripke's book, especially the nature of the sceptical paradox (or "problem" or "challenge", as Kripke sometimes calls it), our inability to provide a "straight solution" to it, and the so-called sceptical solution. Kripke discusses Quine's view on pp. 14-15 and again on pp. 55-57.

2 By my lights, Kripke fails to appreciate important similarities between KW and Quine, although compare the penultimate sentence on p. 57 where, somewhat typically for this book, Kripke appears to undo a couple of pages of previous discussion in a single sentence. Also, Kripke never addresses the $64 question, viz., Why is it, given the obvious similarities between KW's and Quine's respective cases against "meaning facts", that only KW suggests that we get a language threatening paradox from the nonexistence of meaning facts, let alone a "sceptical solution" to it, which forces us to admit the impossibility of private language?

3 In particular, KW contends that the problem with our so using SCs is not that simplicity is relative or hard to define. The problem, as we'll see in more detail below, is alleged to be due to the fact that the sceptical problem is not "merely epistemic"(K, p. 38). Also, because the roles, natures and upshots of KW's meaning hypotheses and Quine's translation manuals are similar, I will, for ease of exposition, use 'meaning hypotheses' as a generic term covering KW's singular meaning hypotheses, e.g., claims like "Jones means plus by '+'", "Jones means quus by '+'", as well as Quinean manuals of translation. Quine's manuals of translation are, after all, a more elaborate and complete set of the kinds of hypotheses scrutinized by KW. It's worth noting though that KW recognizes that the sceptic's bent hypotheses (to use Blackburn's phrase), e.g., "Jones means quus by '+'", cannot be rendered likely without being surrounded with many other "bent hypotheses" as well. (See K, pp. 16-7 especially). The point is that although KW focuses his attention on the plus and quus hypotheses, a complete justification of any particular meaning hypothesis requires invoking a rudimentary manual of translation. Finally, the big difference between Quine's competing manuals and KW's competing meaning hypotheses is that the competing manuals are alleged to predict, and be satisfied by, the same speech dispositions. Obviously, this is not so for Kripke's plus and quus hypotheses, for the former hypothesis predicts that Jones will say that 68 + 57 is 125 while the latter predicts that Jones will say that 68 + 57 is 5.

4 It's worth noting that in laying out the sceptical challenge, Kripke never mentions the truth conditions/assertion conditions distinction. Of course, to do so would have taken much of the drama out of the sceptic's challenge. For KW's sceptical solution, so-called, serves to downplay the success of the sceptic. For the sceptic is allowed only to have shown the nonexistence of "superlative facts", or to have threatened only a misleading philosophical misconstrual of our language, i.e., the truth conditional view of Wittgenstein's Tractatus. (See K, p.65 and p. 69). Furthermore, the extent to which this view has really been shown to be a misconstrual is rather limited. For we should keep in mind that the sceptical paradox not only appeals to an artificial function par excellence, , but it also leans on a simplifying proviso, viz., that 68 + 57 is a computation that I've never performed before. (K, p. 8). It is just one more paradox of KW's account that although he stresses, in his sceptical solution, that we need to look at how we actually use meaning attributions (K, p. 86), his case against the truth conditions account is based on a philosopher's fiction.

5 Davidson's remarks appear in his "The Second Person", in Midwest Studies in Philosophy, Vol. XVII, (Notre Dame: Notre Dame Press, 1992), pp. 261-2.

6 Obviously, there are differences between this way of sceptically solving the paradox and KW's way of solving it. An SCs-based solution provides a way of choosing the plus hypothesis over the quus hypothesis without appealing to meaning facts. KW's solution, on the other hand, seems to do no such thing. In particular, KW's solution focuses almost exclusively on showing how we are able to say that someone means plus, effectively ignoring the original sceptical problem, viz., whether we are able to find a legitimization of the plus hypotheses that does not at the same time legitimate any of its bent brethren. In short, KW ignores the fact that whatever assertion conditions there are that legitimate the assertion of the plus hypothesis can be matched by assertion conditions which legitimate the assertion of the quus hypothesis. As such, KW's sceptical solution seems to have succeeded in changing the problem rather than solving it.

7 W.V.Quine, "Reply to Roger F. Gibson, Jr." in The Philosophy of W.V. Quine, (LaSalle, IL: Open Court, 1986), p. 155.

8 For any readers who may have been wondering what is meant by 'simplicity' or 'simplicity considerations' in either Kripke or Quine, or to what extent the two mean the same thing by 'simplicity', the following remarks will, I hope, suffice. To begin with, any talk of simplicity should keep in mind Quine and Ullian's claim that "[s]implicity is harder to define when we turn away from curves and equations." (The Web of Belief, (New York: Random House, 1978), p. 71). Second, I don't think a precise account of simplicity for meaning hypotheses is either possible or required. By my lights, Quine's use of "kindred canons of procedure" in the passage from p. 155 suggests that as far as he is concerned there is no reason to be fussy about distinguishing simplicity from other considerations "beyond" speaker dispositions which could be used to choose between meaning hypotheses. For a list of such procedures, see Quine's "Indeterminacy of Translation Again", in The Journal of Philosophy, Vol. LXXXIV, No. 1, Jan. 87. As far as KW's argument is concerned, I don't see that it would be affected if his use of 'simplicity' were taken to include not only what Quine calls simplicity but also his "kindred notions of procedure". For I think KW sees himself as providing a general case against using any "nonfactual" considerations to help us justify a choice between the plus and quus hypotheses. (See again footnote 6 above).

9 Some, e.g., Crispin Wright, (in his "Kripke's Account of the Argument Against Private Language" in The Journal of Philosophy 81, 1984, p. 773, fn. 5) take Kripke's assumption that the sceptic is correct about the nonexistence of "meaning facts" (i.e., facts that show which of plus or quus someone meant) to be question-begging. I do not, although I allow that it could be. I willingly grant the nonexistence of a fact to justify a choice between the plus and quus hypotheses. What I deny is KW's contention that without such facts we are unable to understand what our meaning hypotheses state. If KW is read so as to say that it is only within an assertion conditions account of language that we can understand our meaning hypotheses despite the nonexistence of meaning facts, then the assumption that the sceptic is correct is question begging. Since I don't think this is a proper reading of KW but merely a vain attempt to save his argument against using SCs to solve the sceptical paradox, (see below, as well as fn. 11), I don't see his assumption that the sceptic is right to be question begging.

10 Putnam's remarks appear in his paper, "On Wittgenstein's Philosophy of Mathematics", in Aristotelian Society, Supplementary Volume, 70, 1996, pp. 243-264.

11 Of course, KW might hold that (i) above is to be understood as part and parcel of a truth conditional view of language. As such, (i) would commit KW to holding that only a truth conditional view of language is incompatible with there being a sceptical solution to the paradox rather than that there can be no sceptical solution to the paradox at all. Even ignoring the fact that this reading of (i) is incompatible with the remainder of KW's text, the simplest response to it is twofold.
First, this defense ends up granting precisely what KW doesn't wish to grant, viz., that meaning facts are not needed to make sense of our meaning hypotheses. And even if it's allowed that an assertion conditions account of language is the only way to give meaning to our meaning attributions, KW still owes us an argument to show that SCs cannot be included in the assertion conditions for meaning hypotheses. Personally, I see nothing to prevent SCs from being part of the assertion conditions for our saying that someone means plus rather than quus. Second, and relatedly, it's still one thing to allow that there can be no sceptical solution to the paradox from within a truth conditional account of language but quite another to allow that the only way to give understanding to our meaning hypotheses, in the face of the nonexistence of meaning facts, is to adopt an assertion-conditions account of language. (Recall Davidson's point noted above). Unless and until KW can show the latter, he must be seen, on the reading considered here, as having begged the question against the use of SCs, as well as any other sceptical solutions based on something other than a shift to assertion conditions.

12 Recall that Kripke claims that his complaint against the use of SCs to solve the sceptical problem is NOT based on the "usual" problems with appeals to simplicity, e.g., that it is hard to define or that it is relative. (See K, p. 38). By my lights, this suggests that Kripke recognizes the futility of arguing that, although SCs are applicable in the plus/quus case, the plus hypothesis is not really simpler than the quus hypothesis.

Top of Page | WRPL Ultimate HomePage