1. Introduction
Recall Saul Kripke’sFootnote 1 strategy in advancing a sceptical argument against rule-following and meaning. A bizarre sceptic (KW) asks whether there is any fact that determines what S (a speaker) means by an expression, let’s say by ‘plus,’ as Kripke does. To mean addition by ‘plus,’ S ought to follow the rule of addition in her uses of it.Footnote 2 But, KW argues, there is no fact that determines that S is not following the rule of quaddition, a mathematical function that is identical to addition for numbers smaller than, e.g., 57, but computes it as 5 for all bigger ones. The sceptic’s reason for this is that all S’s past behaviour for the numbers smaller than 57 is compatible with following the rule of quaddition. Thus, if queried about the sum of ‘68+57,’ S, the sceptic claims, must answer 5, instead of 125, because what S meant by ‘+’ in the past was quaddition and S must stick to the same rule now. KW thus claims that there is no such thing as meaning, since the argument can be applied to all expressions of language.
KW’s next move is to consider all meaning-determining candidates that could be offered to defeat this sceptical scenario. One of these is S’s disposition to use expressions. According to this suggestion, what S means by ‘plus’ is determined by her dispositions in using it. This view is known as the dispositional theory of meaning. KW, however, argues that the dispositional theory fails for a couple of reasons. The first difficulty is that, while S’s dispositions are finite, meaning is a phenomenon with an infinite domain. The second problem is that a dispositionalist fails to distinguish the speaker’s correct uses from her incorrect applications.Footnote 3 In whatever way that S is disposed to use an expression, the dispositionalist must accept this as the meaning of that expression.
These two problems have been developed by Paul A. Boghossian.Footnote 4 Boghossian, I think convincingly, has shown that a dispositionalist fails to solve these problems due to the holistic nature of belief ascription.
I discuss these issues in more detail in Section 2. In particular, first I explain why the dispositionalist must appeal to what is known as ideal conditions; then I outline Boghossian’s challenge to it based on belief holism. Boghossian’s observation shows why an appeal to ideal conditions is of no help to the dispositionalist. Then, in Section 3, I turn to Allen Gibbard’s defence of the dispositional theory against the problems of finitude and error.
In the course of his project of reconciling meaning naturalism (dispositionalism) and normativity of meaning,Footnote 5 Gibbard suggests that what constitutes the meaning of an expression is not solely S’s dispositions, but also the best-fit ascriptions to S’s dispositions. He characterises the best-fit ascriptions on the basis of two desiderata—simplicity and closeness—which, according to him, can be characterised naturalistically. After outlining Gibbard’s suggestions, I argue that his account suffers from the problems raised by Boghossian. In the final section, I argue that Gibbard’s use of simplicity not only is based on a misinterpretation of KW’s argument, but it also has consequences that are far from appealing for a meaning naturalist.
2. Dispositional Theory of Meaning: The Problems of Finitude and Error
As said above, KW makes a bizarre claim that S in the past meant quaddition by ‘plus,’ not addition, and therefore that S is now obliged to use ‘plus’ in accordance with the rule of quaddition. To this sceptical challenge, a dispositionalist responds that S means addition by ‘plus’ because S is disposed, when asked for any sum ‘n+m,’ to give the sum of n and m. In fact, according to the dispositional theory, S was always disposed to mean addition by ‘plus’ and would have responded by using plus if asked about the sum of any two numbers.
Kripke’s first problem with the dispositional account is that, while the totality of a speaker’s dispositions is finite, an expression has conditions of use in an infinite number of places and times. For example, to say that S has a disposition to apply ‘horse’ to the property horse means that S is disposed to apply it to all horses, including horses in the future or those far away. But S has finite brain capacity and cannot have dispositions that apply to all horses. S’s lifespan is also finite, and she may die before being able to use ‘horse’ for all horses. If so, meaning cannot be determined by dispositions. A common response that a dispositionalist gives to these problems, as Kripke himself envisages, is an appeal to ideal conditions: conditions where all the factors that inhibit S using ‘horse’ for horses are absent.
Kripke’s immediate rejoinder to an appeal to ideal conditions is that the invocation of such conditions is circular because it presupposes that S has dispositions to give the correct answer. Since the dispositionalist too takes advantage of an appeal to ideal conditions in responding to the problem of error, I will discuss this in detail after explaining the problem of error.
To outline the problem of error, it is helpful to discuss the causal/informational version of the dispositional theories.Footnote 6 To avoid unnecessary confusion, I also concentrate on causal/informational theories of content, instead of linguistic meaning. It is clear that if a dispositionalist fails to show how the content of a mental symbol is determined, he ipso facto fails to show how the meaning of the linguistic symbol gets constituted. Following Boghossian, I use the following scheme to define the causal/informational theory of content:
(D) S’s mental symbol ‘C’ expresses whatever property to which she is disposed to apply it.
(D) states a causal relation between what S’s mental symbol expresses and the property that causes her to token that symbol. Thus, S’s mental symbol ‘horse’ means horse if and only if it is a law that only horses cause S to token ‘horse.’ But the plausibility of (D) is questionable. The problem, according to Boghossian, is that under circumstances that are sufficiently dark, a horsey-looking cow can cause S to token ‘horse.’ Now, since a horsey-looking cow can cause S to token ‘horse,’ a dispositionalist is forced to say that S’s mental symbol expresses the disjunctive property horse or horsey-looking cow.Footnote 7 A dispositionalist therefore is unable to distinguish between the correct and incorrect applications of symbols.
In response to the problem of error, a dispositionalist once again may appeal to ideal conditions, saying that under ideal conditions only horses would cause the tokening of ‘horse,’ not the horsey-looking cows.Footnote 8 Therefore, under ideal conditions the dispositional theory seems to be unproblematic. Now that we have seen a dispositionalist’s attempt to neutralise the problems of finitude and error by the invocation of ideal conditions, we can discuss this approach in more detail, outlining the problems of circularity and open-endedness that it must face.
These problems arise because, for any concept R, the ideal conditions must be specified without the use of any intentional language, such that the subject’s judgements involving R would be deemed to be correct. The reason the ideal conditions must be specified in non-intentional language is due to the reductive nature of the dispositional theory. It should be noted, as outlined above, that the dispositional theory, characterised as a causal/informational account, defines content in terms of a causal connection between tokening a symbol and the cause of that tokening. Stated thus, the dispositional theory is not allowed to utilise intentional language for the specification of the ideal conditions.
But it is impossible, says Boghossian, to specify the ideal conditions in non-intentional language, because of the holistic character of belief ascriptions. Boghossian expresses this worry as follows:
The point is that under normal circumstances, belief fixation is typically mediated by background theory—what content a thinker is prepared to judge will depend upon what other content he is prepared to judge.Footnote 9
What Boghossian means is that S may come to form a belief that, for example, there is a magpie out there as a result of seeing some non-magpies because of S’s further beliefs about how magpies look, or about the wildlife in the surrounding environment, and so on. This belief holism, however, causes trouble for a dispositionalist.
To illustrate this further, I outline Alexander Miller’s interpretation of Boghossian’s belief-holism worry.Footnote 10 Let us suppose that belief holism is intuitively correct.Footnote 11 Boghossian’s belief-holism objection, according to Miller, consists of three different objections: (1) the circularity objection, (2) the open-endedness objection, and (3) the certification objection. Since our primary aim is to assess the dispositional theory and not necessarily the plausibility of all three of Boghossian’s objections, focusing on the first two should be sufficient for our purposes here.
With the belief-holism thesis to hand, we can say that the dispositionalist must provide a non-intentional analysis of a situation in which any background belief that can potentially cause S to think that there is a magpie out there, even when there is no magpie, is absent. To do so, a dispositionalist must have a set of naturalistic necessary and sufficient conditions for being a belief with certain content. But this is what a dispositionalist was supposed to be providing.Footnote 12 So the dispositionalist’s appeal to ideal conditions ends up in circularity.Footnote 13
What Miller calls the open-endedness objection could be explained as follows. Since the totality of background beliefs that may cause S to have a belief about magpies is potentially infinite, a dispositionalist must identify an open-ended list of those beliefs. This means that a dispositionalist can never succeed in non-intentionally specifying a condition in which only a magpie causes S to have a belief that there is a magpie out there.Footnote 14
In sum, the dispositionalist’s appeal to ideal conditions fails for at least two reasons. First, a causal dispositional account of a content ascription to the speaker is implausible since the belief that S would form in such a situation depends on the background beliefs that S might hold. Second, to have a non-intentional specification of a situation in which the background belief is absent means that we must already have a reductive theory of meaning, and hence have no need of a dispositional theory.
Having provided a brief outline of the problems of finitude and error as they play out against the dispositional theory of content and meaning, in the next sections, I discuss whether Gibbard’s defence of the dispositional theory can adequately address these problems.
3. Gibbard’s Dispositional Theory: The Best-Fit Dispositions
Gibbard begins his defence of the dispositional theory by suggesting that “meaning is a matter of the ascription that gives the best fit to an utterer’s dispositions.”Footnote 15 This suggestion, according to him, is a departure from the version of the dispositional theory that KW rejects because, instead of holding that the meaning of an expression is solely determined by S’s dispositions, it proposes that meaning is determined by the ascriptions giving the best fit to S’s dispositions. This means that only the ascriptions that are deemed to be the best fit to S’s dispositions should be taken as the right candidates for determining meaning, not the totality of S’s dispositions including the disposition for error. How is this suggestion going to solve the problem of finitude and error? The problem of finitude, according to Gibbard, is solved as “the best fit to the subject’s disposition can extend to numbers too high to contemplate,” and the problem of error disappears since “the best fit to the subject’s dispositions need not exactly match them.”Footnote 16
But it seems to me that there are some serious problems with Gibbard’s suggestion. The first is that Gibbard’s account of the best-fit ascription falls prey to the same problems that beset the appeal to ideal conditions. The best-fit dispositions, as we have just seen, must go beyond finite uses and stay free from error. What guarantees that they will satisfy these conditions? Perhaps one way to ensure it is to consider S’s uses of an expression under ideal conditions where the limiting factors and all the background beliefs that can potentially cause her to hold an erroneous thought are absent. But, if this is the way to characterise the best-fit ascriptions, then the problems of circularity and the open-endedness again loom large, precisely due to the reasons discussed above.
Gibbard, however, chooses not to appeal to ideal conditions. He instead suggests two desiderata for defining the best-fit ascriptions: closeness of fit to the subject’s dispositions, and simplicity. Closeness of fit, according to him, can be measured by a scoring function that gives scores in “pairs consisting of a set of dispositions and a system of meaning ascription.”Footnote 17 Simplicity, he suggests, can be defined in terms of naturalistic and mathematical terms. The best fit, he contends, is a matter of a particular balance between these two desiderata. The ascription of addition and quaddition might fit S’s dispositions equally, but addition is the simplest function of the two. So, the ascription of addition to S’s dispositions turns out to be the best fit.
But the problem of circularity still hangs over the head of the dispositionalist unless the desiderata of simplicity, closeness and the best fit can all be characterised in non-intentional language. Gibbard promises—though without showing how—that these problematic terms will be eliminated in favour of clear naturalistic formulations.Footnote 18 But doubts about this suggestion still remain, based, as discussed above, on the holistic nature of belief formation. The problem is that, in order to ascribe the best fit (the closest and the simplest) dispositions to S, a dispositionalist needs a naturalistically specified necessary and sufficient condition in which S forms a belief with a certain content. But, as seen above, this is impossible to provide because of the infinity of background beliefs and the circularity in specifying them. Therefore, I conclude this section by suggesting that Gibbard has left the most formidable objections against the dispositional theory unsolved.Footnote 19
4. Gibbard and KW on Simplicity
In the previous section, I reviewed some objections against Gibbard’s desiderata of closeness and simplicity, on the basis of Boghossian’s belief-holism observation. In this section, I look further at Gibbard’s use of simplicity. Since Kripke himself broached the appeal to simplicity against the sceptic in his discussion of the sceptical argument, I begin with a brief outline of Kripke’s discussion. Reviewing Gibbard’s response to Kripke’s objection will not only disclose some misunderstandings in his interpretation of Kripke’s sceptical argument, but will also show that Gibbard’s deflationary account of facts leaves no room for talking about the meaning in a naturalistic manner.
Kripke’s first objection is that an appeal to simplicity would be useful only if the competing hypotheses about what S means by ‘plus’ stated “genuine matters of fact.”Footnote 20 In response to this objection, Gibbard claims that competing hypotheses about meaning are in fact genuine because they underlie a normative debate at the level of a meta-theory of meaning. Gibbard asks us to imagine two meaning theorists (Jerry and Tyler) who agree with each other that meaning is constituted by a natural property but disagree over the nature of that property. Whereas one theorist (Jerry) holds that it is an individual’s dispositions that determine meaning, the other (Tyler) believes that meaning is constituted by the dispositions of a linguistic community. Such a disagreement, according to Gibbard, has normative consequences because Tyler and Jerry disagree about which utterances to accept if one is the subject.Footnote 21 In short, Gibbard argues that competing hypotheses about meaning, contra Kripke, are genuine because they are the ground of a normative debate between Jerry and Tyler.
The first problem with this argument, however, lies in Gibbard’s (mis)interpretation of Kripke’s sceptical argument. Whereas, in Kripke’s exposition, the sceptic is concerned about the metaphysics of meaning, asking whether meaning-determining facts exist, in Gibbard’s account the constitutive nature of Kripke’s sceptical argument gets lost. It should be noted that, although the sceptical challenge about meaning is standardly presented as an issue about the determination of which competing hypotheses is correct, the sceptic could achieve his goal without invoking the existence of competing hypotheses. Boghossian has expressed this point clearly:
The constitutive problem about meaning—how could there be so much as a correctness condition—can be stated quite forcefully without actual provision of a convincing global reinterpretation of a person’s words.Footnote 22
Thus, I think the reason that Kripke contends that competing hypotheses are not genuine is that they are not essential in developing the sceptical argument against rules and meaning. If this observation is right, then the invocation of competing hypotheses (interpretations) has no essential role in developing the sceptical argument. This causes a problem for Gibbard because his reconstruction of the argument is essentially reliant on the invocation of competing hypotheses as a ground of normative debate. In sum, the claim that competing hypotheses are genuine implies the denial of the constitutive feature of the sceptical debate. Thus, Gibbard owes us an explanation as to how it would be possible to combine the genuine competing hypotheses with the constitutive nature of the sceptical argument.
Closely related to this problem, the second difficulty for Gibbard is as follows. Even if we grant him that the competing hypotheses are genuine, an appeal to simplicity would still be irrelevant. The reason for this is that simplicity is called for when there is no direct access to evidence in favour of one theory. Kripke argues that scientists (who are realists, for example, about electrons) may appeal to simplicity because of insufficient direct evidence in favour of the existence of electrons. But, in the case of meaning, even for someone with access to all facts (an omniscient God), it would be impossible to tell what the speaker means by uttering an expression.
In response to this problem, Gibbard first charges Kripke with viewing meaning as a straightforward property, and then rejects the idea that meaning is a straightforward property. Gibbard writes:
Kripke is addressing someone who thinks that meaning is straightforwardly a property, a property that God, at least, can observe, just as God could presumably observe electrons. … We think that meaning isn’t fact, ordinary or super. Or we think that it’s a fact only in a minimal sense, a sense in which “it’s a fact that ‘dog’ means DOG” means no more that “ ‘DOG’ means DOG.”Footnote 23
Kripke’s point is that meaning facts are not like facts about electrons; whereas facts about electrons could be directly observed by God,Footnote 24 facts about meaning are not observable at all (or more precisely not distinguishable). The point once again is that simplicity consideration is only relevant against an epistemological sceptic who denies that we have direct evidence for hypotheses about meaning. But the sceptic’s concern is not epistemological. The point of Kripke’s discussion of God or an omniscient being is to highlight the epistemological nature of an appeal to simplicity, not to espouse the property view.
More importantly, if Gibbard believes that the meaning property is not an ordinary fact and is a fact only in a minimal sense, how then can one make sense of his goal of locating the property of meaning in the fabric of the natural world? It is widely believed that the natural world is made of ordinary facts, and meaning naturalism can be viewed as successful only if it can locate meaning in this sense in the world. If so, how should we construe Gibbard’s claim that the property of meaning is natural but not an ordinary fact?
Let me explain. Gibbard explicates his meaning minimalism as follows: the statement it is a fact that S means plus by ‘plus,’ according to him, means no more than S means plus by ‘plus.’ However, to defend his claim that the meaning property exists as a natural property, he suggests that there must be a set of naturalistic truth conditions for the statement about S’s meaning plus, otherwise S’s meaning plus must consist in further facts that fail to supervene on how things are naturalistically.Footnote 25 What he means by the failure of supervenience is that there might be two possible situations which differ as to whether S means plus without in any way differing naturalistically. In sum, there must be some facts, according to Gibbard, that constitute the truth of S means plus by ‘plus,’ otherwise the supervenience of meaning properties on natural properties fails.
But there are some issues here. First, it is not clear why meaning properties should supervene on natural properties. While the supervenience thesis seems to be acceptable for moral concepts and evaluatives based on certain pre-theoretic observations, it is not so for meaning concepts. For example, if someone judges that the picture on the wall is good, but refrains from holding the same judgement for an exactly similar picture in an exactly similar circumstance, then he will probably be deemed incompetent with the term ‘good.’Footnote 26 Although Gibbard may wish to justify the supervenience relation between meaning properties and non-meaning properties by drawing some analogy between moral concepts and meaning, reflection on the concept of meaning does not reveal the same sort of supervenience as conceded to evaluatives.Footnote 27
Moreover, it is doubtful whether the acceptance of the supervenience thesis along with a deflationary view of meaning will imply the existence of a meaning property that can be couched in properly naturalistic terms. Suppose that P1 and P2 are two different situations as to whether S means plus: in P1 S means plus by ‘plus,’ but in P2 S means quus by ‘plus.’ The question is whether, in order to accept the supervenience thesis, a minimalist must commit to the existence of natural meaning property. The answer seems negative. In a minimalistic sense of fact, if S means plus by ‘plus,’ then it is a fact that S means plus by ‘plus’ and if S means quus by ‘quus,’ then (again in a minimalistic sense) it is a fact that S means quus. Thus, things differ naturalistically (in a minimalistic sense) if situations differ as to whether S means plus or quus; what is considered a fact (in a minimalistic sense, once again) in P1 is not the same as what is considered a fact in P2. In P1 it is a fact that S means plus whereas in P2 it is a fact that S means quus. Thus, to accept the supervenience thesis, a minimalist needs no more than to accept that there are two different facts corresponding to P1 and P2, and accepting this would entail no commitment to accepting facts in a way that differs from the deflationary view. In other words, if a minimalist is prepared to say that “it’s a fact that ‘dog means DOG” means no more than “‘DOG’ means DOG,” why should she not also be prepared to say that in a situation in which S means plus it is a fact that S means plus and in a situation in which S means quus it is a (different) fact that S means quus?
Gibbard also discusses a different sense of facts, however, and perhaps a more substantial one. The other sense of facts that must be discussed, according to Gibbard, is the state of affairs. On this sense, saying ‘S means quus’ means that there is a state of affairs such that “the attribution of quus maximises a certain balance of simplicity and closeness of fit to dispositions in her community.”Footnote 28 Nevertheless, what this state of affairs comprises, according to Gibbard, cannot be settled. Thus, Gibbard once again fails to specify which natural facts constitute meaning.
5. Conclusion
My aim has been to assess Gibbard’s defence of the dispositional theory of meaning and content. I have argued that Gibbard’s attempt to provide a naturalistic account of best-fit ascription of dispositions in terms of closeness and simplicity fails because of the holistic nature of belief ascription. I have also argued that Gibbard’s interpretation of the sceptical argument not only confuses an epistemic problem with a constitutive one, but also denies him the possibility of giving an account about meaning property in any sense that holds interest for a meaning naturalist.
Nevertheless, it should be noted that Gibbard’s work is an ambitious project that interweaves the key concepts and theories of philosophy of mind and language in order to bring together meaning naturalism and normativity. Singling out one part of the project and expressing doubt about it, it might be objected, is an unfairly simplistic way to challenge such a complex body of work. In other words, someone may object to the arguments set out in this paper that Gibbard’s aim is not really to defend the dispositional theory against some established objections. Rather, his aim is to illustrate how we can legitimately talk about meaning facts in a naturalistic sense (dispositionalism) without making any commitment to the existence of facts.Footnote 29 The plausibility of this reading, however, depends on how our objector would wish to answer some further questions: to what extent could Gibbard’s dispositional account be viewed as a “straight solution” to the sceptical argument?Footnote 30 Shouldn’t doubts arise as to whether Gibbard’s solution is entitled to be called a straight solution to the meaning sceptical argument, if his defence of the dispositional view ultimately turns out merely to be a way of couching meaning in a naturalistic manner without any serious commitment to existence of a meaning property? These, however, are questions for another work.
Acknowledgements:
An earlier version of this paper was presented at the Institute for Research in Fundamental Sciences (IPM), Tehran, in 2017. I am grateful to the audience of that session for insightful comments and discussion. I would also like to thank two referees of this journal for very helpful suggestions to make some arguments clearer. I should express my gratitude to Alex Miller for stimulating conversations on Gibbard’s Meaning and Normativity.
