International Journal of Philosophical Studies
ISSN: 0967-2559 (Print) 1466-4542 (Online) Journal homepage: http://www.tandfonline.com/loi/riph20
The Metaphysics of Meaning: Hopkins on Wittgenstein Steven Gross To cite this article: Steven Gross (2015) The Metaphysics of Meaning: Hopkins on Wittgenstein, International Journal of Philosophical Studies, 23:4, 518-538, DOI: 10.1080/09672559.2014.999249 To link to this article: http://dx.doi.org/10.1080/09672559.2014.999249
Published online: 28 Jan 2015.
Submit your article to this journal
Article views: 242
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=riph20 Download by: [Johns Hopkins University]
Date: 17 September 2015, At: 12:11
International Journal of Philosophical Studies, 2015 Vol. 23, No. 4, 518–538, http://dx.doi.org/10.1080/09672559.2014.999249
The Metaphysics of Meaning: Hopkins on Wittgenstein Steven Gross Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
Abstract Jim Hopkins (2012) defends a ‘straight’ (non-skeptical) response to Wittgenstein’s rule-following considerations, a response he ascribes to Wittgenstein himself. According to this response, what makes it the case that A means that P is that it is possible for another to (correctly) interpret A as meaning that P. Hopkins thus advances a form of interpretivist judgment-dependence about meaning. I argue that this response, as well as a variant, does not succeed. Keywords: meaning; Wittgenstein; response-dependence; rule-following; interpretation; Davidson
I. Introduction Readers of Wittgenstein disagree about whether his discussion of rule-following is meant to raise and answer genuine questions or rather to dispel the illusion that genuine questions arise here. Among those taking the former tack, there is disagreement about the nature of the questions raised – whether they are epistemic, constitutive (or in some other sense metaphysical), or both. Further, there is disagreement about the thrust of Wittgenstein’s proposed answers whatever the questions – in particular, about whether his responses are skeptical or nonskeptical. Jim Hopkins’ ‘Rules, Privacy, and Physicalism’ (Hopkins 2012)1 stakes out the following position on these matters: the rule-following considerations are indeed meant to raise genuine questions – both epistemic and constitutive (cf. pp. 116–117) – to which Wittgenstein provides a straight (non-skeptical) response.2 These questions include:
‘How do I know I am following a particular rule … correctly?’ (p. 117) ‘How is such correctness constituted?’ (p. 117) ‘Do we really think and mean as we take ourselves to?’ (p. 118) ‘If meaning or content is real, what makes it so?’ (p. 118) How does one solve the ‘problem of representation’ (or, the problem of intentionality)? (p. 110) ‘What makes it the case that my understanding of my own concepts … is correct …’? (p. 121)
© 2015 Taylor & Francis
THE METAPHYSICS OF MEANING
What is ‘[t]he reality … of meaning … constituted by’? (cf. p. 121) What ‘justifies’ our ‘exercise of first-person authority’? (cf. p. 130, fn. 15) Though the questions range generally over thought and language use, we can let questions concerning some speaker’s meaning that P (in uttering some words on some occasion) stand in for the rest.3 Thus among the questions Wittgenstein addresses, according to Hopkins, are the following:
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
epistemic e.i) How do I know what I mean? What justifies, or entitles, my belief that I mean that P? e.ii) How do I know whether I am applying correctly the concepts or words that I (know that I) mean? e.iii) What renders my belief that I mean that P authoritative? constitutive (metaphysical) c) What makes it the case that A means that P? (and thus also, if things are so, that so meaning is correct; and, if A believes that she means that, that this belief is correct?) On Hopkins’ view, Wittgenstein not only provides straight answers to these questions, he provides answers that are ‘relatively straightforward’ (p. 118): … what makes it the case that my understanding of my own concepts (rules, etc.) in my own case is correct, as opposed to merely seeming so to me, is that I admit of understanding by others (via my participation in ‘the common behaviour of mankind’ which, as Wittgenstein says in §206, is ‘the system of reference by which we interpret an unknown language’). This entails that there are regularities in my verbal and nonverbal bahaviour which others might interpret, so that in this they might understand me as thinking and acting in accord with the same concepts (rules) as I take myself to, and as doing so correctly. So the fully objective correctness which I presuppose but cannot justify in my own case, can nonetheless be justified on my behalf by the interpretive understanding of others. (p. 121) In sum, what entitles A’s belief that she means that P, renders her belief authoritative, and makes it the case that she means that P is that it is possible for another to (correctly) interpret A as meaning that P.4 This possibility in turn ‘rests on a system of empirical regularities, which hold over linguistic behaviour, non-linguistic behaviour, and the environment’ (p. 129, adapting §207). 519
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
Hopkins has developed his views in a powerful series of papers (1994a, 1994b, 2004), of which ‘Rules, Privacy, and Physicalism’ (2012) is the latest. If he is correct, much Wittgenstein scholarship is misguided and a central problem in the philosophy of mind and language has yielded to a solution. In what follows, I critically discuss Hopkins’ claims, focusing on the constitutive questions, not the epistemological. (While some would collapse this distinction when it comes to meaning, my discussion can be read as exploring one way of putting pressure on such views.) My main concern is not whether Hopkins has got Wittgenstein right, but rather the position considered in and of itself. I believe it faces serious hurdles. If so, and absent some alternative, the ‘problem of representation’ retains its force (unless Wittgenstein as others read him is right that no genuine problem has in fact been raised). II. Clarifying the Constitutive Claim: Necessity, Sufficiency, and Basicality Let’s begin by clarifying the constitutive claim: (C) What makes it the case that A means that P is that it is possible for another to (correctly) interpret A as meaning that P. What is it for X to constitute or make it the case that Y, on Hopkins’ understanding? Immediately after the long passage quoted above, Hopkins (p. 121) writes: Such justification does not require that others should actually understand me or share my concepts, but only that this should be possible. For insofar as it is possible that another should succeed in understanding me as correctly thinking as I take myself to do, then in fact I do think that way, and correctly. The first sentence states a necessary condition, the second a sufficient condition. Applied to constitution, we have: (N) If A means that P, then it is possible for another to (correctly) interpret A as meaning that P. (S) If it is possible for another to (correctly) interpret A as meaning that P, then A means that P. Moreover, Hopkins talks of what’s ‘basic’ (pp. 113 and 116) and ‘explanatory’ (p. 113), and of what provides ‘foundations’ (p. 125), and he (pp. 121 and 123)
520
THE METAPHYSICS OF MEANING
reads Wittgenstein as addressing head on ‘what this going-by-the-sign really consists in’ (§198, my emphasis). This suggests:
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
(B) That it is possible for another to (correctly) interpret A as meaning that P is explanatorily more basic than that A means that P: in some sense, the latter is so because of the former. I shall therefore assume that Hopkins intends (C) at least to entail (N), (S), and (B). It is true that (N) on its own yields the impossibility of a private language – a topic that occupies the final sections of Hopkins’ paper – since, on Hopkins’ view, ‘for a language to be private in Wittgenstein’s sense is just for it to be uninterpretable’ (p. 133, Hopkins’ emphasis – I briefly return to this in endnote 10). But if we identify (C) with just (N), we must allow that any necessary condition of something is what makes it the case. This might be fine for some notions of partial constitution, but it does not exhaust what Hopkins claims for constitution in the cited passages. Similarly for the conjunction of (N) and (S): a biconditional doesn’t capture the asymmetry Hopkins’ language suggests. In any event, Hopkins (2004) contains additional textual support for our explication. Hopkins’ Wittgenstein is thus most naturally read as advancing a form of interpretivist judgment-dependence about meaning. But, as we’ll see, there are standard worries that such views must address. III. Further Clarification, As Constrained by Plausibility (N), (S), and (B) require further clarification: What is meant by ‘interpret’? Who might be the ‘other’ who’s doing the interpreting? What modality is at issue? What is the relative scope of the possibility operator and ‘another’? What of the parenthetical correctness requirement: should it be included or not? – And can we answer these questions in a way that renders (C) plausible?5 On perhaps the least demanding construal of interpretability, one interprets A to mean that P just in case one believes, or judges, that A means that P. (N) would seem to follow directly – so long as, if A means that P, then it’s possible for others to have beliefs about A and to entertain the thought that P (suppose we concede that it is). But (S) so construed is obviously false. Utterances are rarely so meaningful as to include everything that others might take them to mean. What A means does not include, for example, what is ascribed by recipients of false testimony about what A means. Nor does it include the ascriptions accurately entered on possible worlds where A meant something else. If believing does not make it so, all the more for possible believings. As we’ve seen, Hopkins sometimes speaks of understanding rather than interpretation.6 ‘Understand’ can be used as a success verb. In this spirit, one 521
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
might suggest that one interprets A as meaning that P just in case one correctly believes that A means that P. This suggestion is bolstered by Hopkins’ talk – quoted above – of another’s ‘succeed[ing] in understanding me as correctly thinking as I take myself to do,’ which is the basis of our formulations’ parenthetical correctness requirements. Now, even this, at least on one construal, wouldn’t save (S). For, again, even if A does not (actually) mean that P, it is (metaphysically) possible to believe correctly that A means that P: there is a possible world in which A does mean that P and one correctly believes that A means that P. What’s needed rather is that it be possible that A be relevantly similar to how she actually is and someone else correctly believe that A means that P. (Henceforth, I’ll suppress the italicized clause but remind us of it by using ‘A@’ to refer to A in the scope of the possibility operator.) This would indeed restore (S), albeit trivially. But it would run afoul of (B), for the possibility of interpretation is supposed to constitute there being a fact to be correct or incorrect about. If correctness is built in, then the possibility of interpretation cannot be explanatorily more basic. For, ‘it is possible for another to correctly interpret A as meaning that P’ is plausibly glossed as: it is possible for another to interpret A as meaning that P and it’s the case that A means that P. But something can’t be more basic than itself.7 How then might we construe interpretation so that it is possible but not already characterized in terms of what speakers in fact mean? Talk of interpretation suggests that the judgments concerning what others mean are arrived at in a certain way – via a process of interpretation. And Hopkins’ reference to radical interpretation (p. 129) suggests one way of specifying what that process is: one interprets A as meaning that P just in case one arrives at the belief that A means that P on the basis of radical interpretation – that is, by basing one’s intentional ascriptions on non-intentionally characterized evidence concerning the speaker’s verbal and non-verbal behavior and her environment.8 But what if A is shy, uncooperative, or a mumbler, or the interpreter lazy, inattentive, or just bad at interpreting? Might A then be uninterpretable and yet mean something, contra (N)? Furthermore, radical interpreters can make mistakes (cf., e.g., p. 119); so, wouldn’t (S) still over-generate meanings? Addressing these prima facie problems is a delicate matter. (N) and (S) are each other’s inverse and thus pull in opposite interpretive directions. If we defend (N) by offering a weak reading of what interpretability requires, we risk exacerbating the problem for (S) – and vice versa if the demands are more stringent. Can we find a reading that satisfactorily deals with both at once? Hopkins writes: ‘We can acknowledge that order renders language interpretable without holding that we can always find such order when it is there.’ (p. 133, fn. 17, my italics). This suggests that we distinguish modalities: while it may not be really possible, say, for some interpreter to interpret A@, it may be metaphysically possible. A lazy interpreter might have more get-up-and-go in another possible world where coffee is stronger than is nomologically 522
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
THE METAPHYSICS OF MEANING
possible. Perhaps there’s room to wonder whether there might be interpreters essentially blocked from interpreting some speaker. But we needn’t pause over this. For, in any event, if it’s really possible for radical interpreters to make mistakes, then all the more so is it metaphysically possible. This suggestion, taken on its own, helps with (N), but exacerbates the problem with (S). Similarly if we read weakly who the ‘other’ is who does the interpreting. The existence or possibility of inept interpreters (even ones essentially incapable of interpreting certain speakers) poses no problem for (N) if (N) requires, not that all (possible) radical interpreters are such that it is possible for them to interpret A@ to mean that P, but only that it’s possible that there exist some. But, again, though this helps with (N), it exacerbates the problem with (S): if the possibility of one mistaken interpretation suffices for A to mean that Q, then (S) is implausible indeed. IV. The Constitution Claim and Optimal Radical Interpreters Our construal of interpretability must remain sufficiently weak to render (N) plausible, but block the possibility of incorrect interpretations, since they would undermine (S) – albeit not by building correctness into our characterization. A natural thought is to screen off the effects of inattentiveness, mumbling, and the like by restricting our attention to optimal interpreters in optimal conditions. (See Hopkins 1999a, pp. 8, 19, 21, and 50, for mentions of best interpretations.) Now, both Wittgenstein and Hopkins refer repeatedly to the ability we humans have to understand one another: interpretability, recall, is supposed to rest on the ‘common behaviour of mankind’ (§206). So, let’s suppose that optimal interpreters are human beings – not idealized rational agents with super-human powers – who are as attentive, skilled, etc. as humans can be in ways that matter to interpretation and who have access to any possibly relevant evidence that a human could have.9 Is there reason to think that (C) is plausible so construed? a. assessing (N) First, consider (N). Might A mean that P, but it not be possible for optimal interpreters in optimal conditions to judge that? Davidson (1994, p. 121) thinks so: I have never argued that every language is radically interpretable; I have not even argued that every language can be understood by someone other than its employer, since it would be possible to have a private code no one else could break.10 His view is rather that ‘anyone with any language must have a non-private language.’ (Davidson 1994, p. 121). But even sticking to a shared language, there 523
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
are prima facie counter-examples to (N). What precludes A’s uttering ‘That’s a nice dress’ on some occasion and no one’s being able to tell – even an optimal interpreter – whether A was being sarcastic? Suppose A intends her tone of voice to clearly indicate her sarcastic attitude, but she coughs as she speaks or simply does not succeed in striking the intended tone – and nothing else seems to tell either way. Suppose further no one then asks her to clarify what she meant (or perhaps she’s struck by lightning before she can answer). Does it follow that she does not mean that it’s not a nice dress?11 It might be replied that this does not take seriously the requirement that conditions be optimal. The optimal interpreter, after all, would have indeed judged that A meant that the dress was not nice had A produced the tone. Conditions in which she produces the tone thus seem better, so far as interpretability goes, than how things actually were. But if optimal conditions are construed in a manner that allows changes in a speaker’s actual conditions, we run afoul of the ‘conditional fallacy’ (cf. Shope 1978). Being in the better conditions might affect A’s attitudes and what she means: if it was her cough-inducing cold that put A in a sarcasm-inducing mood, then, had she not coughed, she might not have meant that it’s not a nice dress. The realization of optimal conditions, if they are not actual, must not causally affect the speaker in a relevant way. Otherwise put, there are cases where conditions necessary for interpretation would violate our suppressed clause that A be relevantly similar to how she actually is. Indeed, ‘conditional fallacy’ worries might arise more broadly. For just as altering the interpretee might affect her attitudes, so might introducing an optimal interpreter. The presence of an optimal interpreter (or, indeed any interpreter, if none had been present) might not only induce greater shyness, it might relevantly affect the specific attitudes to be interpreted. Byrne (1998) suggests that we handle this by allowing interpreters to possess any relevant evidence without actually being present to the subject or even in indirect causal contact; the interpreter and the subject might even occupy different possible worlds. There is a potential problem, however, if there are causal conditions the interpreter must satisfy in order to entertain thoughts about A and to grasp the content to be ascribed. Could there be a case in which the introduction of any possible optimal interpreter would induce a relevant change in A, no matter how indirect or discrete the satisfaction of the causal conditions? If so, then, assuming that A is not actually interpreted to mean that P, it would not be possible for A@ to be so interpreted. Perhaps it’s nomologically impossible for a speaker to be so sensitive: there will always be some way to gather evidence that does not relevantly affect her. But suppose it’s not. We might then try dealing with this worry by further reconstruing interpretability. Perhaps (N) requires only, for A to mean that P, that it is possible for optimal interpreters in possession of all relevant evidence to judge of a person evidentially equivalent to A that that person means that P (where A and B are evidentially equivalent just in case all the relevant facts about behavior and environment true 524
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
THE METAPHYSICS OF MEANING
about one are mutatis mutandis true about the other). Delicate questions would remain, however: for example, if what A means is that this is the actual world, can an optimal interpreter on another possible world interpret A’s evidential equivalent as meaning that? Whether ‘conditional fallacy’ worries arise more generally thus turns on a variety of issues. For our purposes, it suffices that our original challenge to (N) still stands. Even if we can distinguish better and worse interpreters (and conditions for interpreters, understood in terms of evidence possessed) without running into the ‘conditional fallacy’, it remains the case that altering the interpretee is problematic.12 But perhaps an optimal interpreter can (correctly) interpret the ‘nice dress’ utterance without adverting to evidence the interpretee might have produced. According to this alternative reply, we seem to have a counter-example to (N) only because of a narrow view of the evidence already available. In particular, as Hopkins emphasizes (p. 140), the evidence may include, not just what is casually observable, but also facts about the interpretee’s neurophysiological states and their relation to the environment.13 If, say, ‘regular connections’ among neurophysiological states, behavior, and the environment support an interpretation that includes the hypothesis that ceteris paribus a tokening of neurophysiological state-type N indicates sarcasm, then the interpreter can use the presence of this state, given the rest of her evidence, to determine what A means. But to what does one commit oneself in offering this reply, especially if it must be invoked for a wide range of prima facie counter-examples? Suppose, for example, that A is asked whether she is coming to some event and utters in reply, ‘Bill is coming’. It may be obvious that A is implicating something, but not obvious on the basis of non-neurophysiological evidence, even to an optimal interpreter, just what is being implicated: that therefore A is not coming, that therefore A is coming, that therefore it’s hard for A to decide, or what? Perhaps A ought to ensure that her addressee can recover her intent; but she may fail to – or be uncooperative, or have no addressee other than perhaps herself (she might mumble her answer under her breath as she removes herself from the prying questioner). Again, (N) does not require of A that she be optimal from the perspective of the interpreter’s task. Here too, the present reply would have it, we must hope that A’s being in a certain neurophysiological state might provide at least all things considered reason, given the rest of the evidence, for an optimal interpreter to reach a (correct) judgment concerning what A implicates. Clearly, this reply commits one to a very substantial empirical claim about what humans can come to know. It is a widely held view that intentional properties supervene on non-intentional properties. It is another matter whether we can, even at our best, always come to know the facts about the former on the basis of knowing facts about the latter. (Non-scalar) implicatures are particularly hard cases, since they arguably involve the sort of general reasoning 525
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
abilities that some doubt we will ever bring within the purview of explanatory empirical theorizing, at least of the sort and to the extent required by our interpreter.14 As goes without saying, we have of course learned a fantastic amount about the mind-brain and will continue to learn more. But whether optimal interpreters can learn enough to turn aside putative counter-examples to (N) is anyone’s guess.
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
b. assessing (S) Let’s turn now to (S). Do we have reason to believe that no optimal interpreter in optimal conditions will arrive at a false belief, on the basis of her interpretive activities, concerning what S means? For instance, might two such interpreters disagree? Note first that the amount of possibly relevant evidence is enormous, so that, even if each piece of evidence taken individually is available to an optimal interpreter, it may be impossible for any such interpreter to survey the whole, let alone think through the evidential bearing of each on the rest, when ‘all the evidence is in’ (Davidson 1979, p. 139).15 This raises the threat of optimal interpreters reasonably disagreeing owing to a reliance on different evidence. But any such disagreement would entail that at least one was incorrect in violation of (S). Similarly, a full sharing of evidence among reasonably disagreeing interpreters might not be possible. Optimal human interpreters still have their limits. What is needed is some assurance that whatever evidence an optimal interpreter takes into account suffices to ensure convergence on a single set of ascriptions, or at least sets that are not incompatible. But why think that we have such assurance? Hopkins (p. 122) acknowledges the possibility of ‘competing hypotheses’ that agree over a finite range of data. So, even if optimal interpreters make use of the exact same evidence, one may wonder whether they are forced to the same conclusions. Hopkins suggests that no special problem arises here: we’re faced merely with an instance of the familiar under-determination of theory by evidence. By this he must not mean that no problem arises from allowing that these competing hypotheses are equally reasonable, so that an optimal interpreter might put forward either. For this would not only allow violations of (S), it would imply that there is a fact of the matter beyond what an optimal interpreter might conclude from the evidence, in violation of (B). If optimal interpretability constitutes A’s meaning that P, then there is no space for an optimal interpreter to get wrong what A means owing to empirical under-determination. The suggestion might rather be that under-determination is no special problem, because, even if finite data are consistent with multiple hypotheses, optimal interpreters, like hypothesizers generally, bring to bear constraints that ensure a single best hypothesis – in the case at hand, perhaps some version or relative of Davidson’s Principle of Charity.16 But it is far from obvious that 526
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
THE METAPHYSICS OF MEANING
such constraints will assure unanimity among optimal interpreters. Davidson of course denies that they do, owing for example to the possibility of trading-off true beliefs and alternative meanings in explaining subjects’ behavior (e.g., Davidson 1973, p. 327). And Hopkins seems to agree (p. 132, fn. 16, quoted below). Davidson concludes that we have here, not just under-determination, but indeterminacy. Whether this is consistent with (S) – and thus with (C) – depends on how one views indeterminacy and construes conditionals. For example, suppose it is indeterminate whether A means that P, and suppose also that an optimal interpreter judges that it’s indeterminate whether A means that P (a best case scenario, since the judgment would be true ex hypothesi). Then (N), (S), and their conjunction are all also indeterminate according to both the weak and the strong Kleene 3-valued semantics for conditionals and conjunctions.17 Some other understanding is necessary if one is to maintain (C). Davidson compares interpretive indeterminacy to the choice among measurement scales (e.g., Davidson 1989). This suggests that we modify (C) to make explicit a relativization to interpretation. Hopkins (1999a, Appendix 3), however, explicitly rejects both this move and Davidson’s measurement analogy, at least in the case of indeterminacy arising from the possibility of belief/meaning trade-offs (cf. also Lepore and Ludwig 2005, pp. 243–7). One might think that the indeterminacy can somehow be quarantined and thus minimized. In fact, since (S) is a general claim about what speakers mean, itself standing in for an even more general claim concerning all thought and language-use, it is unclear what help this would offer: a single counter-example will do. But in any event Hopkins (p. 132, fn. 16) does not believe that the indeterminacy remains local: Our interpretive understanding of language and of others more generally is liable to empirical indeterminacy, as illustrated by Wittgenstein’s use of algebraic formulae, and later particularly stressed by Quine. The embedding of an interpretee’s sentences in the ascription of all propositional attitudes, as emphasized on Wittgenstein’s account, both spreads this indeterminacy across all sententially described states and events, and by this same means minimizes its practical importance, by making the test of each ascription at the same time a test of the interpreter’s and interpretee’s understanding of the sentence as well as the state of mind involved. Note that it’s only the practical importance of indeterminacy that is said to be minimized, not its spread. Indeed, the spread is precisely what minimizes the practical importance, since it is what allows incompatible interpretations to nonetheless accrue equal empirical support. Are there other options for a defender of (S) who would accept under-determination (even with constraints) but deny indeterminacy? It might be thought that an optimal interpreter will recognize the possibility of alternative sets of 527
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
ascriptions consistent with her evidence and applicable constraints, and therefore refrain from entering any judgments: she will maintain an agnostic position, not issue a judgment of indeterminacy. But this would save (S) only at the expense of (B) and (N). If optimal interpreters do not judge of the matter, then their judgments cannot constitute what A means, violating (B). Moreover, we would have cases where A meant something, but was not judged to have meant it by optimal interpreters, violating (N). (We will have saved (S) by reintroducing, writ large, a version of the problem discussed above in relation to (N): apparent cases where the evidence doesn’t tell one way or the other.) It might be replied that, if optimal interpreters remain agnostic about what A means, then it’s not the case that A means anything (and thus there’s nothing to be constituted).18 But this result is problematic in itself, especially if the under-determination is widespread. Since it involves denying what otherwise seem obvious instances of meaning, it arguably amounts to a form of meaning-skepticism. The result is thus also dialectically unattractive for Hopkins. If the choice is between abandoning (C) and there potentially being widespread meaning-skepticism, why not choose the former? (C), after all, was meant to provide a straight answer to the questions that threatened meaning-skepticism. In sum, if we are to construe interpretability in terms of Davidsonian radical interpretation, Hopkins must (contra p. 132, fn. 16) deny Davidson’s contention that, even ‘when all the evidence is in,’ the constraints imposed upon radical interpretation leave room for competing hypotheses. But then we need reason to think this is so and, further, reason to think that competing hypotheses don’t remain even after optimal interpreters have merely assimilated all the evidence they can. It may well be that, for example, in each case it is possible for humans to know neurophysiological facts that, together with other evidence about A’s behavior and environment, knowably entail (or at least reasonably support) the relevant intentional fact – knowably excluding incompatible ascriptions. But it may not be. Again, this is an empirical matter – concerning, not the constitution of meaning, but the extent to which we can come to know it. c. assessing (B) Finally, let’s turn to (B), the claim that, when A means that P, it’s because it’s possible that others so interpret her – so that interpretability is supposed to be in some sense explanatorily more basic than meaning. It is worth noting that the worry I will raise does not rely on any of the particulars discussed above concerning how to construe interpretability. This is important because I certainly have not canvassed all the possibilities. Given the supporting text (and because it puts some meat on the bones), I wound up focusing on Davidsonian radical interpretation, as performed by optimal interpreters in optimal conditions. But when Wittgenstein in §206 mentions an explorer coming to an 528
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
THE METAPHYSICS OF MEANING
unknown land with a language unfamiliar to him – a passage central to Hopkins’ discussion – there is no indication that the explorer must base his interpretation ultimately on non-intentional evidence or indeed that there must be an isolable evidentiary base at all. If we were to drop these Davidsonian elements, interpretability might then refer more generally to intentional ascription (by an optimal interpreter in optimal conditions) in the absence of prior knowledge of the subject’s language and without the aid of a translator. (cf. Davidson 1994, p. 128, fn. 2). Indeed, we might consider dropping any constraint on how the judgments are reached, beyond that they be the best and reached by others.19 It would remain incumbent on the proponent of (C) so construed to defend (N) and (S), both of which would remain far from obvious. But no matter how (C) is understood, even if (N) and (S) could be justified, (B) is open to objection. The obvious question is why the possibility of being interpreted to mean that P should be considered explanatorily more basic than meaning that P. Since both involve the intentional, it would seem prima facie plausible to place them rather at the same explanatory level.20 Indeed, because the interpreter’s judgment is a second-order meta-representation concerning someone else’s first-order representational state, there’s prima facie reason to think that it is, if anything, explanatorily less basic because representationally more complex. It does not seem promising to defend (B) by maintaining that non-linguistic intentionality is explanatorily more basic than linguistic intentionality. First, this would have the later Wittgenstein reverting to an earlier view that Hopkins takes pains to portray him as rejecting. Second, it neglects the fact that meaning-ascriptions in (C) are merely standing in for thought and language-use more generally. Nor does it seem plausible to lodge the asymmetry in some special status optimal interpreters have in comparison with more ordinary folk. Interpretability by another, pending some argument to the contrary, should constitute what optimal interpreters mean too. But if the other is another optimal interpreter, we seem caught in a constitutive circle, rather than having the intentional explained by something more basic. Similarly, that (C) adverts only to the possibility of optimal interpreters and their judgments doesn’t seem to relieve us of our worry so long as we can still ask what makes it the case that such judgments would be made. (And, in any event, it’s not precluded that an optimal interpreter be actual.) This is problem enough, but it is worth noting as well that a further related problem might arise if we are forced to cash out what optimality amounts to. The worry is that optimality might involve certain intentional states being realized or not. For instance, believing without warrant that everyone, including A, is out to make one look like a fool by planting false evidence presumably rules one out as an optimal interpreter. But if we must specify optimality in terms of interpreters’ intentional states, we again face the problem that our constitution base includes that which it is supposed to constitute. To avoid this worry, it 529
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
would have to be shown that optimality can be cashed out without adverting to such states or may be taken as primitive so far as (C) is concerned. However this may go, of course neither strategy is available in dealing with our original worry, since what optimal interpreters judge about A is already – and indelibly – written into (C).
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
V. Regular Connections As Constitutive (C) faces some challenges. Unless those challenges can be met, (C) does not provide a compelling answer to the constitutive question. We can’t quite leave the matter there, however. For, although (C) is the dominant claim in Hopkins’ paper, some remarks suggest an alternative answer. I conclude by briefly discussing it. Recall that, in the quote in which he advances (C), Hopkins (p. 121) ascribes to Wittgenstein the view that a necessary condition of interpretability is the existence of ‘regular connections’ between one’s behavior and one’s environment: … that I admit of understanding by others (via my participation in ‘the common behaviour of mankind’ …) … entails that there are regularities in my verbal and non-verbal behaviour. Some wording perhaps intimates that it is supposed to be a sufficient condition as well, as when Hopkins writes of “‘the common behaviour of mankind” which renders interpretation possible’ (p. 131). Indeed, in earlier work the claim is quite explicit: Just as in §206-7 the radical interpreter is required to find regular connections between utterances of orders and actions which are obeyings of those orders, so in §243 the radical interpreter is required to find regular connections between expressions of intention and actions which are fulfillings of those intentions. In the earlier remarks Wittgenstein claimed that such correlations are necessary for interpretation, and here he adds that they also suffice for it. For given such correlations the explorer-interpreter can both ‘predict these people’s actions correctly’ and ‘succeed in translating their language into ours.’ (Hopkins 2004, p. 132) Now, the claim that shared regularities are necessary and sufficient for being interpretable at all does not yet tell us what is necessary and sufficient for being interpretable as meaning that P. So, it cannot itself begin to address the constitutive question. Still, it might suggest an answer other than (C), viz.: (C*) What makes it the case that A means that P (and that it is possible for others to interpret A as meaning that P) is that there exist certain ‘regular connections’ among A’s verbal behavior, non-verbal behavior, 530
THE METAPHYSICS OF MEANING
and environment, together with certain facts about A’s current state and her environment. (We need the further facts about A’s current state and her environment because, whatever regularities there might be, A may not be going around all the time meaning that P.) That is, given our working understanding of constitution,
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
A means that P, and it is possible for others to interpret A as meaning that P, just in case – and because – there exist certain ‘regular connections’ among A’s verbal behavior, non-verbal behavior, and environment, together with certain facts about A’s current state and her environment. Above, we followed Hopkins in treating interpretability as constituting meaning, with the regularities serving as an evidential base necessary for interpretability. The suggestion here is that these regularities themselves are (in part) what constitutes meaning. And, in a parenthetical remark, Hopkins in fact seems to affirm something like this alternative reading of Wittgenstein: Third-person interpretations may ascribe meaning in describing the use of words, but … it is the real regularity under description, and not the describing of it, which determines the meaning so ascribed. (p. 124, Hopkins’ emphasis)21 What should we think of (C*), and the corresponding (N*), (S*), and (B*)? A first problem is that, if our worries about (C) have merit, then (C*) lacks sufficient generality to answer the constitution question. For it’s not always the case that, when A means that P, it is possible for others to interpret A as meaning that P – at least according to the construal of interpretability on which we settled. (C*) would thus not apply to such cases, leaving their constitution unaccounted for. A natural response is just to drop interpretability altogether from (C*). The price, however, is to increase our distance from Hopkins’ discussion. A second problem concerns (B*). How should we understand this talk of ‘regular connections’? Hopkins (pp. 125 and 127) writes of the ‘order’ a radical interpreter discerns in A’s behavior in simultaneously advancing hypotheses concerning (i) the intentions in A’s non-verbal actions, (ii) the correctness of A’s sentences in expressing his intentions, and (iii) the meaning of those sentences. A radical interpreter thus ‘can make these regularities explicit’ (p. 129). So, the regularities include what interpreters (correctly) ascribe – e.g., that when A utters ‘The day is warm,’, she means that the day is warm (cf. Hopkins 1999a, p. 43), that A believes that P on the basis of seeing that P (Hopkins 1999b, p. 270), that A’s utterance and action are ‘correlated’ in that each is typed using the same that-clause (the utterance as an expression of an intention to V, the action as a fulfillment of the intention to V – 1999b, 531
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
p. 275), etc. The regularities the interpreter discerns are interlocked, intentionally characterized patterns of belief-formation, theoretical and practical reasoning, and action. But then (B*) is obviously false: our constitution base includes just the sort of intentional facts it is supposed to constitute. What if we purge the constitution base of intentionality? Might there be other regularities that can serve (C*)’s needs? If the regularities can’t be characterized intentionally and thus in terms of rational connections, it seems we are left with an appeal to statistical regularities. But then there is room to wonder whether regularities among our verbal and non-verbal behavior and the environment, together with relevant facts specific to the occasion, really suffice to determine what A means. Perhaps there are fewer regularities than would be needed. Are my utterances of ‘I’m going to …’ (and the like) reliably correlated with certain behaviors? Perhaps more often than not I don’t bother to express my intention at all. And perhaps often enough, even when I do, the utterance is not interestingly correlated with a non-intentionally characterized type of behavior: perhaps sometimes my utterance of ‘I am going to grade the papers’ is followed by an episode of grading, sometimes by an episode of procrastination; perhaps my utterances of ‘I’m going to explain to X what I mean’ are always followed by an explaining, but there is no non-intentional way of seeing these actions as of a type. Again, my tokening of ‘car’ does not reliably co-vary with the presence of cars: I discuss them when they aren’t around and often don’t remark upon them when they are. Further, there is the familiar worry that these regularities will require ceteris paribus clauses themselves cashed in intentional terms. This is all very quick, but I mean only to point to a potential problem. There is arguably a poverty of regularities in the stimulus. Indeed, this is just what Davidson intends the Principle of Charity and other aspects of a radical interpreter’s procedure to overcome.22 But why restrict the regularities to the kind of would-be correlations used as examples above? We have dropped interpretability from (C*). This means that it no longer matters whether anyone could know the facts in the constitution base or deduce from them what is thereby constituted. So, the regularities aren’t limited to those an interpreter could discover. In principle, we could just include all non-intentionally characterized regularities – and indeed all nonintentional facts. Then (S*), at least if modally strengthened, simply asserts the supervenience of the intentional on the non-intentional. This is unexciting, but at least plausible. (S*) would be more interesting if we identified a narrower supervenience base, but this is more challenging as well. In any event, (C*) would still be controversial on account of (N*). If our supervenience base is just the non-intentional facts, then we must deny, implausibly, that A could mean that P under different non-intentional conditions. If we specify the base in terms of a disjunction of sets of non-intentional facts, each sufficient for A to mean that P, there is the familiar worry that the base ceases to be explanatory. Attempts to specify a more interesting base would have to be taken up individually. 532
THE METAPHYSICS OF MEANING
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
VI. Concluding Remark Let’s take stock. Hopkins has forcefully developed a ‘straight’ interpretation of Wittgenstein’s rule-following considerations. We argued that it is best construed as a form of interpretivist judgment-dependence about meaning. On this view, meaning and content are constituted by what idealized human interpreters would judge concerning meaning and content. In particular, that idealized human interpreters would render such judgments is both necessary (N) and sufficient (S) for our meaning what we do; moreover, we mean what we do because (B) they would so judge. We argued, however, that each of these components of Hopkins’ constitutive claim is problematic. We examined as well an alternative claim according to which meaning and content are constituted by regular connections among our verbal behavior, non-verbal behavior, and environments; it was found problematic as well. There is obviously much more to be said on all these matters. But I conclude, for now, that it remains unclear what makes ‘meaning or content … real’ (p. 118). Johns Hopkins University, USA Acknowledgements Warm thanks to Jim Hopkins for the stimulation of his work and for helpful conversation. I hope any misreadings or misunderstandings contained herein are nonetheless productive. Thanks also to an anonymous referee for extremely helpful comments. This paper began life as a response to Jim’s talk at the 2007 Wittgenstein and the Philosophy of Mind conference at UC, Santa Cruz. My thanks to the organizers Jonathan Ellis and Daniel Guevara.
Notes 1 Unadorned page references are to Hopkins 2012. Unadorned section references are to Wittgenstein 1953. 2 Hopkins (p. 130, fn. 15) cites Diamond (1991) and McDowell (1984) as proposing illusion-dispelling readings and mentions (p. 118, fn. 7) as well his demurral from Kripke’s (1982) skeptical response. 3 My introduction, for ease of expression, of the schematic that-clause ‘that P’ seems licensed by Hopkins’ discussion. It is asked, for example, whether ‘we really think and mean as we take ourselves to’. One can take oneself, on some particular occasion, to have meant that the door is open. Thus it seems that at least among the questions raised is whether one really meant that the door is open. (Hopkins [p. 120] describes such ‘judgments as to what I intend or mean’ as ‘continuation[s] and amplification[s]’ of the practice that includes such intending and meaning.) 4 We turn presently to whether the parenthetical correctness requirement should be included. 5 Further, one might ask whether the claims themselves are supposed to be a priori, necessary, conceptually necessary, analytic, etc. As I am not sure what Hopkins might say and don’t believe anything in my discussion turns on this, I will prescind from these matters. But see note 14, below.
533
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
6 On Hopkins’ view, understanding another does not require explicit interpretation, but considering explicit interpretation (‘interpretive understanding’ as the long quote above from p. 121 has it) lays bare Wittgenstein’s straight answers to our questions. 7 Two referee’s remarks suggest that alternative understandings of Hopkins might avoid this problem. The first suggests an alternative construal of the correctness requirement, according to which ‘it’s possible for another to correctly interpret A as meaning that P’ might be glossed: it’s possible for another to interpret A as meaning that P and that A means that P is the best explanation of the relevant data – where this in turn might be cashed in Bayesian terms (cf. Hopkins 2013). Our focus is on ‘constitutive’, metaphysical questions (‘what makes it the case …?’), not epistemic questions (‘how can one know …?’). So, the issue is not whether or how an interpretation is warranted but rather whether its being the best explanation ‘makes it the case’ that A means that P. ((B), on the current construal, would then have to obtain in virtue of the distinctive subject matter, as one would not want to endorse such constitutive claims generally: the theory of evolution’s explaining the data is not what makes it true, though it warrants our belief in it.) Various of the issues discussed below would seem to apply just as much to the current suggestion: Must the best explanation be true? (The referee suggests that (S) becomes trivial, but that does not seem so – unless we revert to the natural construal of ‘correctly’ that I discuss in the main text.) What if there are ties for the best? Does the constitutive base presuppose what it’s meant to constitute? But there’s also a question specific to this proposal – viz., whether we really understand what it would be for X’s explaining Y to make it the case that X. (It might be replied that Hopkins doesn’t mean to raise this sort of constitutive question at all, despite the cited text apparently to the contrary. But if such questions should be rejected, not answered, one would, at least to this extent, ally oneself with the illusion-dispelling readings from which Hopkins distances himself.) The referee’s second remark suggests that, given the explanatory context, we should understand (B) to say that ‘it is possible for another to (correctly) interpret S as meaning that P is constitutive of the fact that S means that P, in the sense that the latter is so – that it is a fact that S means that P rather than merely seeming to himself to mean that P – because of the former’. But adding the ‘rather than’ clause does not change the fact that the explanatory base would contain that which it is supposed to explain. 8 Hopkins adverts to radical interpretation – and argues that Wittgenstein does as well (e.g., in §§206-7 and 243) – as a device for revealing the regularities on which all third-person interpretation (radical or not) depends. In Hopkins 1999a, 1999b, and 2004, he is more explicit that radical interpretation thereby plays a crucial role in answering the questions Wittgenstein poses. Hopkins 1999a and 1999b acknowledge that, although on his view Davidson articulates more fully what is merely implicit in Wittgenstein, Wittgenstein’s attitudes towards reference (satisfaction) and formalization seem in tension with Davidson’s. But he suggests that: ‘There is … a case for saying that the theoretically expanded and action-engaging role which Davidson assigns to truth actually serves ends which Wittgenstein himself might (or ought to) have regarded as desirable’ (Hopkins 1999a, p. 37). Though Hopkins’ and Davidson’s views are clearly very close (cf. Davidson 1999), I’ll note below a few places where it’s possible there might be some divergence. One other note: There’s room to question my characterization of radical interpretation as resting upon a nonintentional base, since Davidson (1973, p. 322) suggests that a ‘good place [for radical interpretation] to begin is with the attitude of holding a sentence true’. But elsewhere (e.g., 1975), Davidson indicates that even such ascriptions, though a good starting place, must ultimately be based upon behavioral evidence. (Davidson’s reason for recommending such attitudes – that often there is fairly direct evidence
534
THE METAPHYSICS OF MEANING
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
9
10
11
12
13
14
linking what the speaker utters and her attitude towards it – bears a family resemblance to Wittgenstein’s talk of ‘characteristic signs’ [§54].) In any event, Hopkins’ (1999a, pp. 11–13) own remarks on radical interpretation suggest he accepts the non-intentional characterization of the evidence base. Below, I mention another understanding of radical interpretation. The clause about access to evidence contains a scope-ambiguity. The more plausible reading gives ‘any possibly relevant evidence’ wide scope – see below. Requiring that our optimal interpreters be human may raise the worry that (C) so construed will be vulnerable to counter-examples involving space aliens whom we are poorly equipped to interpret. Rather than face such cases head-on, one might restrict the scope of (C) to human interpretees as well. Alternatively, one might retain the generality of (C) by maintaining that what counts as an optimal interpreter might vary with the kind of subject. Cf. the kind of private language Wittgenstein allows in §243. At least in this passage, Wittgenstein’s concern is arguably with the alleged possibility of a mental state only one person could be in, not (contra Hopkins) with privacy construed as uninterpretability – i.e., the possibility of being in a state no else could discern one is in. (And readings could then diverge again on whether Wittgenstein’s point is to deny the alleged possibility or to question whether we so much as understand it.) I’m assuming that what A means is determined by her intention, not by what is recovered or recoverable. Those who disagree with this assumption could consider instead (N) as applied to what A intends. Recall that (C) is standing in for a more general claim about thought and language. The problem is only worse for unexpressed mental states. Wright (1992, pp. 119–20) responds to ‘conditional fallacy’ worries by ditching biconditionals similar to Hopkins’ in favor of ‘provisional equations’ that restrict their focus to conditions where there’s no such ‘altering’. It’s thus claimed, roughly, that, if optimal, non-altering conditions obtain, then an optimal interpreter will judge that P iff P. Wright notes that this loss of generality has its costs and correspondingly qualifies his claims of judgment-dependence. The states may be characterized abstractly – for example, computationally – so long as they are not characterized intentionally. Hopkins (p. 140, my emphasis) talks here of ‘neural activities which constitute our sensations and … our thoughts’. Unless this is a different sense of ‘constitute,’ there is the threat that adverting to interpretation is in fact explanatorily otiose. Cf. below. Davidson, it should be noted, does not permit so broad an evidential base. For example, he says he assumes that a radical interpreter ‘hasn’t learned what someone thinks or means by opening up his brain’ (Davidson 1994, p. 125). Perhaps Davidson’s point here is rather to rule out the possibility of coming to know what someone thinks or means by inferring it only from brain states. But he also says: ‘I restrict the evidence to what would be plainly available to an observer unaided by instruments’ (1994, p. 127). Cf., e. g., Fodor (1983, 2000) and Chomsky (2003, p. 323). For some evidence that scalar implicatures are another matter, see Chierchia, Fox, and Spector (2013). Sperber and Wilson, who used to maintain that implicature calculation recruits general reasoning abilities, now postulate a dedicated pragmatics module. But they still maintain that general reasoning capacities are recruited when the pragmatics module fails to yield an interpretation. See, respectively, Wilson and Sperber 1986, and Sperber and Wilson 2002. Their discussions focus on implicature comprehension, not production; but arguably successful communication typically requires some form of anticipation of the addressee’s likely comprehension. Finally, note that interpretivist and judgment-dependent views about intentionality are often advanced as a
535
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
15
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
16
17
18
19
20 21
22
priori claims. If (C) were advanced as an a priori claim, it could not be defended by an appeal to empirical speculation on what optimal humans might know. Again, this may be so even if the intentional facts about S supervene on this evidence. The obtaining of a supervenience relation entails neither that we can know the base nor that, if we do, we can deduce the superstructure. Concerning the amount of possibly relevant evidence, note that it includes a lifetime of behavior in various environments, and that the possibly relevant environmental facts may extend well beyond what is in the immediate vicinity of or is readily observable by the subject – e.g., the many facts about A’s neurophysiological states. There has been much debate concerning how best to formulate such constraints. See, e.g., Grandy 1973. How Hopkins might formulate one is perhaps best suggested by his development of Wittgenstein’s remark (§207) about the need to find others’ actions intelligible and ‘logical’. (The reason we can’t say that the people of §207 have a language, even though we find their non-verbal behavior intelligible, is that we can’t render their verbal behavior intelligible.) Cf. p. 126 and Hopkins 1999a and 1999b. According to the weak tables, a logically complex claim with an indeterminate constituent is itself indeterminate. According to the strong tables, such a claim is indeterminate if replacing indeterminate constituents with true or false ones sometimes renders the whole true and sometimes false. Given (C), this seems forced if we are avoiding indeterminacy. In any event, for reasons related to those already given, it would do no good to maintain rather that, if optimal interpreters remain agnostic about what A means, then it’s not true, but possibly indeterminate (and so not false), that A means anything. Excluding A is meant to respect Hopkins’ view that it’s the possibility of being interpreted by another that constitutes A’s meaning that P. But, as Hopkins makes clear, what’s crucial is that the interpretation be from a third-person perspective. Presumably, A could adopt that perspective. Crispin Wright (1989) floats a first-person variant of the current proposal, focusing on intention: roughly, A intends to X iff (and because) in ideal cognitive conditions, A herself would judge that A intends to X. This is proposed, however, not as a reading of Wittgenstein (whom Wright views as a quietist), but as a better reply to Wittgenstein’s concerns. Note that it’s built into the view that a being with propositional attitudes – at least intentions – must be capable of meta-representation. (I say ‘floats’ because, as noted above, Wright ultimately conditionalizes on ‘ideal cognitive conditions’ rather than leave them within the biconditional.) Cf. Boghossian (1989, pp. 546–7) on Wright, and Wright (1992, pp. 138–9, fn. 47) for a brief reply. One might mention as well Davidson’s (1998) claim that knowledge of others’ minds is not more basic than knowledge of one’s own. I include the cautionary qualification ‘seems’ because Hopkins’ point in this passage might be rather just to underscore that it’s the possibility of interpretation that constitutes meaning, not its actuality – as opposed to offering the ‘real regularity’ as a constitution base alternative to the possibility of interpretation. That the possibility of interpretation is in turn determined by the ‘real regularity’ is not incompatible with the possibility’s constituting meaning. Still, the thrust of the remark seems in tension with the ascription of (C) that dominates his paper. (Hopkins 1999a and 1999b are more ambiguous.) Note, however, that, if constitution is transitive, one might ascribe or affirm both (C) and (C*). This would remove the tension between them. Actual human first-language learners likewise bring a host of (arguably innate) assumptions and constraints to bear in learning word-meanings. Bloom (2000) argues that this includes the ability to deploy ‘mind-reading’ abilities to discern the
536
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
THE METAPHYSICS OF MEANING
intentions of speakers. In this way, for example, he explains why children as young as 15 months exhibit a ‘mutual exclusivity’ bias, in effect assuming that an object has only one label: if a new label is used, the child assumes it must label something else, since, had the speaker wanted to direct the child’s attention to the original object, the speaker would have used the other label. Bloom’s hypothesis is relevant to Hopkins’ empirical speculation that there is a cognitive basis for our dualistic proclivities (e.g., p. 142). For Bloom (2004) argues that the nature of our folk psychology and folk physics modules renders us ‘natural born dualists.’ (Though some remarks of Hopkins – e.g., 1999b, p. 267 – might be read as suggesting that he would resist the characterization of 15-month-olds as discerning intentions, his work on psychoanalysis, as a referee points out, endorses contentful mental representations in pre-linguistic infants.) Bloom (2000), incidentally, reports some corpus work relevant to assessing just how ‘regular’ the correlations among utterances, behavior, and environment are. The results do not look good at least for crude associationist models of the acquisition of lexical meaning.
References Bloom, Paul (2000) How Children Learn the Meaning of Words, Cambridge, MA: MIT Press. ——— (2004) Descartes’ Baby, New York: Basic Books. Boghossian, Paul (1989) ‘The Rule-Following Considerations’, Mind 98: 507–49. Byrne, Alex (1998) ‘Interpretivism’, European Review of Philosophy 3: 199–223. Chierchia, Gennaro, Danny Fox, and Benjamin Spector (2013) ‘The Grammatical View of Scalar Implicatures and the Relationship between Semantics and Pragmatics’, in P. Portner, C. Maienborn, and K. von Heusinger (eds) Semantics: An International Handbook of Natural Language Meaning, vol. 3, Berlin: Walter de Gruyter. Chomsky, Noam (2003) ‘Reply to Gopnik’, in L. Antony, and N. Hornstein (eds) Chomsky and His Critics, Oxford: Blackwell. Davidson, Donald (1973) ‘Radical Interpretation’, Dialectica 27: 314–328. ——— (1975) ‘Thought and Talk’, in S. Guttenplan (ed) Mind and Language, Oxford: Oxford University Press. ——— (1989) ‘What is Present to the Mind?’, in J. Brandl and W. Gombocz (eds) The Mind of Donald Davidson, Amsterdam: Rodopi. ——— (1994) ‘Radical Interpretation Interpreted’, Philosophical Perspectives 8: 121–8. ——— (1998) ‘The Irreducibility of the Concept of the Self’, in M. Stamm (ed) Philosophie in Synthetischer Absicht, Stuttgart: Klett-Cotta. ——— (1999) ‘Reply to Jim Hopkins’, in L. E. Hahn (ed) The Philosophy of Donald Davidson, Chicago, IL: Open Court. Diamond, Cora (1991) The Realistic Spirit, Cambridge, MA: MIT Press. Fodor, Jerry (1983) The Modularity of Mind, Cambridge, MA: MIT Press. ——— (2000) The Mind Doesn’t Work That Way, Cambridge, MA: MIT Press. Grandy, Richard (1973) ‘Reference, Meaning, and Belief’, Journal of Philosophy 70: 439–52. Hopkins, Jim (1999a) ‘Wittgenstein, Davidson, and the Methodology of Interpretation’ [a longer, earlier version of Hopkins 1999b], http://www.kcl.ac.uk/kis/schools/hums/ philosophy/handouts/hopkins/wittdavid.pdf ——— (1999b) ‘Wittgenstein, Davidson, and Radical Interpretation’, in L. E. Hahn (ed.) The Philosophy of Donald Davidson, Chicago, IL: Open Court. ——— (2004) ‘Wittgenstein and the Life of Signs’, in Wittgenstein’s Lasting Significance, eds. Max Kölbel and Bernhard Weiss, London: Routledge.
537
Downloaded by [Johns Hopkins University] at 12:11 17 September 2015
INTERNATIONAL JOURNAL OF PHILOSOPHICAL STUDIES
——— (2012) ‘Rules, Privacy, and Physicalism’, in Jonathan Ellis and Daniel Guevara (eds) Wittgenstein and the Philosophy of Mind, Oxford: Oxford University Press. ——— (2013) ‘Psychoanalysis, Philosophical Issues In’, in B. Kaldis (ed.) Encyclopedia of Philosophy and the Social Sciences, Thousand Oaks, CA: Sage. Kripke, Saul (1982) Wittgenstein on Rules and Private Language, Cambridge, MA: Harvard University Press. Lepore, Ernie, and Kirk Ludwig (2005) Donald Davidson: Meaning, Truth, Language, and Reality, Oxford: Oxford University Press. McDowell, John (1984) ‘Wittgenstein on Following a Rule’, Synthese 58: 325–63. Shope, Robert (1978) ‘The Conditional Fallacy in Contemporary Philosophy’, Journal of Philosophy 75: 397–413. Sperber, Dan, and Deirdre Wilson (2002) ‘Pragmatics, Modularity and Mind-Reading’, Mind & Language 17: 3–23. Wilson, Deirdre, and Dan Sperber (1986) ‘Pragmatics and Modularity’, Chicago Linguistic Society 22: 67–84. Wittgenstein, Ludwig (1953) Philosophical Investigations, London: Macmillan. Wright, Crispin (1989) ‘Wittgenstein’s Rule-Following Considerations and the Central Project of Theoretical Linguistics’, in A. George (ed.) Reflections on Chomsky, Oxford: Blackwell. ——— (1992) Truth and Objectivity, Cambridge, MA: Harvard University Press.
538
