In a recent thought-provoking paper on skepticism concerning meaning (1997), Scott Soames claims that Kripke's and Quine's arguments that there are no facts about meanings are flawed for similar reasons. According to Soames, both of them are based on a confusion about how a certain kind of fact determines another (for instance, what it takes for a dispositional fact to determine a particular linguistic meaning). Soames' strategy to refute the skeptical arguments advanced by Kripke and Quine involves distinguishing two notions of determination both of which, if applied unambiguously and consistently throughout the formulation of the above skeptical reasonings, would fall short of licensing the far-reaching and devastating skeptical conclusions that their proponents intended them to have.

This paper is an attempt to vindicate the problem raised by the meaning skeptic, and to show that Soames' suggested dispositional account cannot even partially solve it. I leave the problem of the indeterminacy of translation aside for lack of space as well as because of my greater familiarity with the literature related with Kripkean skepticism. In section 2, the skeptical problem is introduced from a slightly different perspective from which it is usually presented. I interpret Kripke's problem as possessing both constitutive and epistemological dimensions; it requires of the prospective meaning-constitutor to satisfy two conditions: a) to be able to account for the kind of normativity that is attached to meaning; b) to allow for an explanation of our knowledge of meaning. Section 3 contains Kripke's most damaging objection to dispositionalism as a solution to Kripke's problem: the objection of the normative non-adequacy. In section 4, Soames' version of the skeptical problem is presented; he separates it into two distinct questions: a) that of finding non-intentional facts which epistemologically determine meaning, and b) that of finding non-intentional facts which metaphysically determine meaning. According to him, although there may be no solution to the first question, there are dispositional facts that correctly answer the second question. Section 5 contains my criticism of Soames' rendering of the skeptical problem, and finally section 6 expounds my argument against his claim that there are dispositional facts which metaphysically determine meaning.

Kripke introduces the problem of meaning skepticism with the following simple example. He invites us to imagine a subject (S) who is a competent English speaker and has access to all there is to know about his present and past mental life, behavior and brain. One could almost certainly say that S uses the expression `+'Foot note 5_1 to refer to the arithmetical operation of adding two natural numbers; whenever S uses `plus' he is applying the mathematical rule: add the numbers m and n. The idea that naturally comes to mind when we think of a rule like this is that of an algorithm determining a unique value for any two arbitrary natural numbers. But the rule of addition is not the one that interests Kripke; what concerns him is rather the linguistic rule: S uses the word `plus' to refer to addition. As S will have applied this latter rule to finitely many cases, we can always imagine him in a situation where he has to apply the rule to a new case. Kripke suggests that this situation is that of answering the question «68+57=?»; S is supposed to have never used `plus' in situations involving numbers equal to or greater than 57 before.

Suppose, Kripke says, that S is presented with the sign «68+57=?» and his reaction consists in uttering the word `125'. Suppose also that S is quite confident of having given the right answer. Now imagine that S encounters a skeptic -- let us call him K -- who questions the certainty of S's answer in what Kripke calls the metalinguistic sense; K raises doubts about S's knowledge of the meaning of `+' and not about his knowledge of arithmetic. This last knowledge is taken for granted in Kripke's dialectic. Skepticism about S's understanding of `+' creeps in as follows. One of the hypotheses of the example is that S uses `+' now in the same way as he did before; if he intended it to refer to addition in the past then S means the same in the present. But how does S know, K insists, that, in applying the term `+' in the past, he meant the adding function and not, say, the quadding function (that we will henceforth symbolize by `*')? The definition of `*' goes like this:

x*y = x+y, if x, y < 57

x*y = 5, otherwise.

After all, K goes on, all his previous uses of `+' fit perfectly well his meaning quaddition by `+' rather than addition. But if there is room for skepticism concerning what `+' meant for S in the past, then it obviously extends to its present meaning for him as well.

According to Kripke, two aspects must be distinguished in the skeptical challenge. First of all, there is the question of whether any fact about S determines which function -- addition or quaddition -- he means when he uses `+'. That is, what constitutes S using `+' to mean addition rather than quaddition? This is the metaphysical or constitutive aspect of the skeptical problem. But there is also the problem of what justifies S in believing that the word `+' in his idiolect means plus rather than quus; how does he know that it means plus rather than anything else? This is the epistemological aspect of the skeptical problem. Kripke also describes the two sides of the problem in the following way:

An answer to the skeptic must satisfy two conditions. First, it must give an account of what fact it is (about my mental state) that constitutes my meaning plus, not quus. But further, there is a condition that any putative candidate for such fact must satisfy. It must, in some sense, show how I am justified in giving the answer `125' to `68+57'. (Kripke 1982, p. 11)Foot note 5_2

It seems to me that the separation between a constitutive and an epistemological questions within Kripke's problem is extremely relevant for a thorough understanding of the reasons why some of the suggested solutions to it are not acceptable. However, most commentators have claimed that the appearance of an epistemological dimension to meaning skepticism is misleading.Foot note 5_3 Paul Boghossian, for example, uses the following argument against the purported epistemological character of Kripkean skepticism. One of the assumptions of Kripke's problem is that S is an idealized subject; he is by hypothesis not subject to the limitations of our cognitive capacities. If his sense organs never deceive him, his memory works perfectly, his mind does not create illusory representations such as dreams or hallucinations and so on, then there can be, according to Boghossian, no room for epistemological skepticism concerning S's access to what he means by his words.

Yet, Boghossian's argument loses sight of a more radical variety of skepticism which, in my opinion, underlies Kripke's problem. After all, is it not possible that S, although not subject to our cognitive limitations, found himself at a loss in his attempt to justify his belief about what `+' means to him? Imagine, for example, that in response to the skeptic S mentioned the mental state he is in while using `+' with understanding. Given the faultlessness of his memory, perhaps S could appeal to it in order to justify his beliefs about his past understanding of `+'. The suggestion is that the constitutive fact would be an occurrent mental state that accompanies every competent use of `+', and that the subject's grasp of this state is mediated by memory for all his past applications of `+' and by introspection for his present uses of this word. But how could memory or even introspection enable someone to recognize a certain state as that of meaning addition by `+'? Someone might be tempted to say that such a state possessed some qualitative feature which could distinguish it from similar states of meaning like that of using `+' to refer to quaddition. It is plausible to think, however, that, unlike occurrent mental states (e. g. a sensation), dispositional mental events such as that of understanding a linguistic expression in a specific way do not exhibit any phenomenology that could allow the subject to distinguish it via introspection from, say, a slightly different understanding of the same expression. This objection to the suggestion that linguistic understanding is constituted by a qualitative mental state can be found in Kripke.Foot note 5_4 It poses an epistemological difficulty to those who believe that states of understanding are qualitative. Now, if meaning skepticism did not have an epistemological dimension -- that is, if the question of the subject's access to that which constitutes his linguistic understanding was no part of the skeptical problem -- then the above objection could not be raised. The fact that it is part of Kripke's strategy against the advocate of the qualitative character of states of understanding confirms, I think, the claim that there is also an epistemological aspect to meaning skepticism.

That this suggested separation between a constitutive and an epistemological question correctly represents Kripke's problem is also corroborated by his insistence on the requirement that the putative meaning-constituting fact justifies the subject's beliefs about what he means. Thus, while considering the question of whether appeal to a linguistic disposition can justify the subject in answering as he did to the addition problem, Kripke says:

I know that `125' is the response you are disposed to give (...), and maybe it is helpful to be told -- as a matter of brute fact -- that I would have given the same response in the past. How does any of this indicate that -- now or in the past -- `125' was an answer justified in terms of instructions I gave myself, rather than a mere jack-in-the-box unjustified and arbitrary response? Am I supposed to justify my present belief that I meant addition, not quaddition, and hence should answer `125', in terms of a hypothesis about my past dispositions? (Kripke 1982, p. 23)

Another important clarification to make at this stage is that, in assuming that S's cognitive capacities are always reliable, we are not automatically committing ourselves to the assumption that the objective criteria for the truth of his beliefs about what he means or about how he intends to use a word are available to him. On the contrary, that from the perspective of the speaker there can be no criteria for objective attribution of meaning and intention is precisely one of the lessons of Kripke's variety of skepticism. This means that the distinction cannot be drawn between a speaker meaning something by a word and him merely thinking that he means it, if the perspective of the speaker is isolated from that of any external observer. Since, according to the meaning skeptic, the speaker himself cannot ultimately justify his own semantic beliefs, a solution to the skeptical problem, if there is one, would also require an explanation of first-person knowledge of linguistic meaning and intention which can account for the lack of objectivity of such a knowledge.Foot note 5_5 If these considerations are correct, then it is hard to see how one could deny an epistemological dimension to meaning skepticism.

One of Kripke's preferred targets with respect to the skeptical problem is the so-called dispositional account. Against it, he aims a number of objections the most harmful of which is what I will be calling here the objection of the normative non-adequacy. It goes like this. Linguistic meaning requires norms for the correct use of words and sentences. Yet, linguistic dispositions fail to capture these norms. Therefore, linguistic dispositions cannot constitute meaning.Foot note 5_6

Kripke certainly does not exhaust nor claims to have exhausted all the possible versions of dispositionalism. He considers, however, two quite representative varieties of the dispositional approach. The first, which I will call straightforward dispositionalism, consists in spelling out S's add-disposition towards `+' in terms of the following counterfactual: if S were presented with any expression of the form «m+n=?» (where `m' and `n' stand for any two numbers) he would have responded with `p' (where `p' stands for a number which is the sum of m and n).Foot note 5_7 According to Kripke, straightforward dispositionalism fails to capture the normative relation between meaning and use because from the supposition that S is add-disposed towards `+' no norms concerning how he should use the word in any given case follows (that is, how it would be correct for him to apply it in these cases). What we can infer from the disposition's corresponding counterfactual is only how S would have used it, had he been exposed to the relevant stimulus.Foot note 5_8

The second kind of dispositional account discussed by Kripke is what I shall call here cæteris paribus dispositionalism. This account is tentatively proposed in his book as a way of responding to the well-known objection to straightforward dispositionalism that human dispositions are finite. The new approach (cæteris paribus dispositionalism) eliminates the gap between human capacities and the infinite dispositions postulated within the straightforward approach by idealizing human dispositions. The trouble is that by doing this, says Kripke, the theorist of dispositions has rendered his account of that which constitutes meaning circular. In order to see why, let us look at how the new approach constructs the counterfactual associated with the add-disposition towards the word `+'. Kripke represents it in the following way: if S were given the means to carry out his intentions towards numbers that are presently too large for him to add (or to grasp), and if he were to carry out these intentions, then if queried about the result of m+n for some large m and n, he would respond with their sum.Foot note 5_9 This modified variety of dispositionalism is not subject to the normative non-adequacy objection but only because it builds into the relevant counterfactual an intention to use a word in a certain way, an entity of the same problematic kind as that of meaning something by the word. This new intentional item would be another easy target to Kripke's skeptic.Foot note 5_10 Besides, the explanation of linguistic meaning in terms of a linguistic intention would constitute no advance in our understanding of what constitutes meaning because, in order to account for what fulfills a linguistic intention, one would have to appeal to a linguistic meaning, and that would render such an explanation evidently circular.

In apparent agreement with most commentators, Soames seems to admit only a constitutive dimension to Kripke's problem. According to him, the problem would be solved if we could exhibit some fact that determined what S means by his uses of the word `+'. This is one of his formulations of the problem:

So, if it is a fact that we mean so and so by a given word w, then some fact about us must determine in advance how w properly applies in new cases. This much seems undeniable. The surprise comes when we examine potential candidates for such a determining fact and find that none fills the bill. Because of this, the skeptic concludes, we have no choice but to admit that it is not a fact that we mean anything by w after all. (Soames 1997, p. 212)

The formulation of the constitutive problem in terms of the determination of facts about meaning or intention in terms of facts of another nature plays a crucial role in Soames' discussion of meaning skepticism. He sees Kripke's problem as that of specifying non-intentional facts which determine (in a sense that we are going to consider below) the mental facts that correspond to the meaning the subject assigns to his words and to his linguistic intentions towards these words. In the specification of these non-intentional facts, Soames claims, appeal can be made to intentional facts (like representations, mental images, sensations and so on) provided that their content is not assumed to be already established.Foot note 5_11 A way of understanding this proviso is the following: intentional facts are allowed to occupy a meaning-determining position if they are described non-intentionally. Here I will suppose that this is what Soames has in mind; the reasons for this assumption will be apparent in the ensuing discussion.

The skeptical problem about meaning seems to be forcing upon us a pessimistic solution, namely: the conclusion that there are no such things as meanings or linguistic intentions. According to Soames, the argument produced by Kripke in order to justify this conclusion is the following:

P₁If there is a fact that S meant addition by `+' in the past, then either: i) this fact is determined by non-intentional facts of such and such kinds (for example, the set of all his past linguistic dispositions towards `+')Foot note 5_12 or ii) the fact that S meant addition by `+' in the past is a primitive fact (i. e. not determined by any non-intentional fact).

P₂Non-intentional facts of the kind mentioned in (i) do not determine that S meant addition by `+'.

P₃What S meant by `+' is not a primitive fact -- that is, it is determined by non-intentional facts.

Therefore, there is no past fact about what S meant by `+' and no fact either about what he means by it now; and generally there is no fact about what he or anybody else means by his or her words.Foot note 5_13

Soames complains that the above argument falls into a fallacy of equivocation: it trades illicitly on two concepts associated with the term `determination', that occurs in premises 1, 2 and 3. This is why the skeptical argument seems so forceful. Once we distinguish the two concepts of determination as below and employ either of them consistently throughout the argument, its apparent high persuasiveness vanishes. The first is the notion of epistemic determination which is characterized by the following constraint: a fact that P determines a priori (or epistemically) the fact that Q only if knowledge of P allows one to demonstrate Q without recourse to any other empirical facts; Q is said in this case to be an a priori consequence of P.Foot note 5_14 Applied to the case under consideration, the constraint of epistemic determination generates the following condition that any candidate non-intentional fact for determining the meaning fact which corresponds to using `+' to refer addition must satisfy:

(N_E)The non-intentional fact that P determines epistemically that one means addition by `+' only if knowledge of P provides one in principle with a sufficient basis for concluding that one ought to give the answer `125' to the question «What is 68+57?»Foot note 5_15

N_E states that a necessary condition for the epistemic determination of a meaning fact by a given non-intentional fact is that knowledge of the latter be sufficient for the speaker to the derive a priori (i. e. without the help of any bit of empirical knowledge) the norms of meaning.

Now, if `determination' is understood epistemically (i. e. if the relation of determination is identified with that of a priori consequence), then, according to Soames, while P<sub>2</sub> may be accepted as true, it is not plausible to say the same of P<sub>3</sub>. As far as the truth of P<sub>2</sub> is concerned, it is reasonable to accept it, he goes on, because the non-intentional facts mentioned in P<sub>1</sub> most probably will not satisfy condition N<sub>E</sub>. The reason why such facts will not satisfy N<sub>E</sub> is that S might be aware of all his linguistic dispositions towards `+' without having any clue about whether his use of `+' in the new case is correct or not. As to P₃, Soames justifies its implausibility by saying that the norms of meaning are probably not deducible a priori from the conjunction of propositions describing any set of non-intentional facts.Foot note 5_16

Soames claims that the second notion of determination involved in the skeptical argument is that of metaphysical determination. According to him, the fact that P determines metaphysically the fact that Q only if Q is a necessary consequence of P, which means that all possible worlds containing P must also contain Q. From this general characterization of metaphysical determination, Soames extracts the following condition that any putative non-intentional fact must satisfy if it is to determine metaphysically the fact a speaker uses `+' with the intention of adding:

(N_M)The fact that P metaphysically determines that one means addition by `+' only if in any possible world in which it is the case that P, `125' is the answer one ought to give to the question «What is 68+57?».Foot note 5_17

N_M says that a necessary condition for the metaphysical determination of the fact that the speaker means addition when he uses `+' by a non-intentional fact is that the norms corresponding to this specific meaning must be a necessary consequence of the proposition that expresses P; in all possible worlds where P obtains, the norms for the use of `+' are those which accord with the attribution of the content addition to this word.

Now, if the relation of determination is to be equated with necessary consequence (that is, if the skeptic is talking about metaphysical rather than epistemic determination), then the reverse is the case, namely: it makes sense to attribute truth to P₃ but not to P₂. The former is plausibly true because, although, as Soames admits, some dispositional facts probably will not comply with condition N_M, he is convinced that there must be non-intentional facts (for instance, the complex fact mentioned in P₁) which would finally satisfy N_M.

Soames grounds his firm belief that the above-mentioned dispositional facts will satisfy N_M, and therefore render P₃ true, on an alleged supervenience of meaning facts on these non-intentional facts. According to him, if a meaning fact (M) supervenes on certain non-intentional facts (P), then in all possible worlds where P are the case M is also the case, and therefore the proposition which describes M follows necessarily from the ones that describe P.Foot note 5_18 Moreover, P₂ must be false if `determine' in it means the same as in the true P₃.

From these considerations, Soames draws at least two conclusions. The first is that the alleged skeptical argument is unsound. The second is that the skeptic has not demonstrated that there are no facts about what we mean by our words. What he has probably established is the epistemological non-determination of facts about meaning by any kind of non-intentional facts. But, as this sort of non-determination is compatible with the metaphysical determination of meaning facts by strongly construed dispositional factsFoot note 5_19 and as, according to Soames, the skeptic has not shown that no non-intentional fact can determine meaning metaphysically, the claim that there are no meaning facts is so far unjustified. Soames goes even further: this claim is not merely unjustified but altogether false, since there is a version of dispositionalism that will meet the condition of metaphysical determination (N_M). If Soames is right, then Kripke must have misjudged the merits of dispositionalism with respect to its adequacy for generating the norms of meaning.

According to Soames, skepticism about meaning is just based on a confusion; the thesis actually established by the skeptic is much weaker than he intended it to be. The bold and far-reaching character of Soames' anti-skeptic conclusions calls for a more careful examination of his argument. Let me start with the question of the correction of his interpretation of Kripke's problem.

Perhaps what Soames wants to contemplate with his distinction between an epistemic and a constitutive condition of determination (N_E and N_M, respectively) is the existence of both an epistemological and a metaphysical (or constitutive) challenge of explaining the kind of normativity that attaches to meaning. Since linguistic meaning and intention are essentially normative notions, and, moreover, notions called upon by others and by ourselves in order to justify our intentional behavior, any prospective explanation of meaning and intention must account not only for what constitutes the norms that are associated with them but also for our access to these norms. If this is so, then Soames and myself would be on the same side against interpretations of the skeptical problem which recognize only a constitutive dimension to it. Recall, however, that the constitutive and the epistemological facets of Kripke's problem are intimately related; they should be seen as two constraints on prospective solutions to it rather than constituting two different problems.

Yet, Soames seems to think that there are two separate problems -- the problem of showing that facts of some kind metaphysically determine meaning and the problem of demonstrating that the former facts determine meaning epistemologically -- such that a solution to the first problem does not depend upon the solution one gives to the second problem. Thus, he claims against Kripke that a version of dispositionalism can solve the problem of the metaphysical determination of meaning even though the problem of the epistemological determination of meaning is most probably insoluble. And he also believes that, in order to convince us that dispositions in general cannot constitute meaning, the skeptic would need to show, although he does not, that these dispositions fail to satisfy the constitutive condition of determination.

However, to conceive the skeptical problem as two unrelated questions misrepresents Kripke's purposes; it leads to the false view that in order to give a negative answer to the problem you need to answer both questions negatively. If we see meaning skepticism as Kripke does -- that is, as a problem with two inseparable conditions -- then the rejection of a prospective meaning-determining fact requires only a demonstration that it does not comply with one of the conditions of the problem. Hence, if, as Soames maintains, the skeptic had concluded that a dispositional fact falls short of constituting meaning because it cannot accord with the epistemological requirement of determination, his argument would be quite correct.

Nevertheless, the skeptic does not argue this way! For Kripke dispositions are no use for explaining meaning precisely because they cannot adequately generate the norms which we associate with the notion of linguistic meaning.Foot note 5_20 But even if it was granted that the skeptic has not demonstrated that any disposition will fail to metaphysically determine meaning, Soames' concession that such a disposition would probably not meet the condition of epistemological determination indicates that his interpretation of meaning skepticism diverges a lot from Kripke's. Otherwise, Soames should have concluded that, because dispositionalism cannot meet one of the conditions of the problem, it therefore is not apt to solve the skeptical problem.

Soames also believes that the meaning skeptic demands that the prospective meaning-determining fact be non-intentional. The motivation for such a demand is already familiar: to suppose that an intentional fact could determine meaning is to invite a new skeptical problem concerning the determination of the intentional content of such a fact.Foot note 5_21 If we are not after an empty explanation of the normativity of meaning, then the candidate for meaning-determining fact should be required not to possess the kind of normativity we are seeking to explain (that is, the normativity that is associated with the intentional items such as beliefs, meanings and desires). On the other hand, to require that the meaning-constituting fact be non-intentional is to open oneself to the objection that facts with which no normativity of the relevant sort is associated can neither generate nor explain the norms of meaning. Call this latter the objection of the normative non-adequacy for its similarity with Kripke's most important charge against dispositionalism.Foot note 5_22 Philosophers who are more sensitive to the problem of the reiteration of meaning skepticism have proposed a reductionist account of the normativity of meaning and the other intentional items (i. e. an account of the intentional in terms of non-intentional items). This is the position of Michael Dummett in his first paper on what a theory of meaning is (1974). Others have felt the power of the normative non-adequacy of the non-intentional items much more acutely; they have thereby suggested a non-reductionist account of the normativity of the mental. For example, Colin McGinn in his book on Wittgenstein (1984). The two requirements -- that the meaning-determining fact should not be subject to a new skeptical challenge and that it be such as to generate and explain the norms of meaning -- seem to be in blatant conflict. Neither intentional nor non-intentional facts appear to be capable of satisfying these requirements simultaneously.

Some theorists would nonetheless claim that their suggested meaning-determining facts attend the above apparent conflicting requirements and hence escape the dilemma. I assume Soames would want to make this claim about his robustly construed linguistic dispositions. Hence, his position deserves to be evaluated in the light of the question of whether it actually satisfies these two requirements.Foot note 5_23 But, as a prolegomenon to the evaluation of Soames' position, something more substantial needs to be said about the sort of normativity that is specific to meaning and the other mental items.

We mentioned in section 3 that there is a normative relation between the meaning a speaker assigns to a word of his idiolectic and this person's linguistic performances involving the word; some of these performances will be evaluated as correct, while some others will be assessed as incorrect uses of the word. Moreover, in cases of meaning attribution as well as the attribution of other mental items the subject must be credited with the awareness of such norms (or criteria of correctness), as he can often be observed to respond to correctness. This reflects the idea that the intentional behavior of human beings is normally within the field of their consciousness so that they are capable of voluntarily and purposefully guiding this sort of behavior. This feature of the normativity characteristic of human speech and action does not belong to the norms with which other, non-intentional activities are evaluated. For instance, the behavior of a heart or that of a carburettor can be judged as in accord or in conflict with the function these objects were designed and built to discharge. Thus, a criterion for the correct operation of a carburettor is that it mixes petrol and air in a certain proportion; otherwise we will normally say that it fails to perform its function. It would not make sense, however, to demand of the latter type of norms that the object whose performances are supposed to be under its jurisdiction should be aware of, and intentionally guide its behavior by, them. Awareness of the norms for assessing the correctness of their performances can only be assumed in the case of the intentional behavior of human beings.

Let me consider first the following question: do Soames' two normativity conditions take into account the desideratum discussed in the last paragraph? The constitutive condition (N_M) does not even mention this epistemological constraint of the normativity that applies to human intentional states. Perhaps his epistemic condition (N_E) is better suited to represent the latter type of normativity. N_E states that knowledge of the meaning-constituting fact corresponding to any word of his language must be sufficient for the speaker to infer a priori the norms for the correct use of the word.

But is it plausible to suppose that first-person knowledge of meaning and motive is inferential? Ordinary intuition seems to point in the opposite direction, namely: that normally the speaker's access to the norms associated with what he means or to how he intends to use an expression is non-mediated rather than inferential. Wittgenstein was perhaps the first to capture the intuition of the ultimate immediacy of first-person knowledge of meaning. Thus, in discussing the question of whether reasons are needed in order to justify first-person access to linguistic rules, he says the following:

How can he [the rule-follower] know how he is to continue a pattern by himself -- whatever instructions you give him? -- Well, how do I know? -- If that means «Have I reasons?» the answer is: my reasons will soon give out. And then I shall act, without reasons. (Wittgenstein 1953, § 211)Foot note 5_24

Wittgenstein's denial that first-person knowledge of the norms of linguistic meaning is inferred from some other piece of knowledge the speaker possesses -- a priori or a posteriori -- relates to another thesis of his that self-knowledge about meaning is practical rather than theoretical. I cannot discuss this latter thesis here.Foot note 5_25 Suffice it to say, however, that the speaker's impossibility of ultimately justifying his linguistic behavior does not, according to Wittgenstein, license the conclusion that such a behavior cannot be justified altogether. As he put it: «to use a word without a justification does not mean to use it without right» (Wittgenstein 1953, par. 289). I take Wittgenstein to be thereby suggesting that, although the speaker cannot definitively justify his use of words because he lacks objective criteria to judge about the correctness of such a use, someone else -- a suitably positioned and informed observer -- might be able to produce the reasons that will finally justify the speaker's intentional behavior towards the words of his language.Foot note 5_26

Davidson has also emphasized the normally non-inferential and criterionless character of self-knowledge about the mental.Foot note 5_27 One of his main concerns was to show -- much more explicitly than Wittgenstein -- that the ungroundedness of such knowledge from the perspective of the subject undermines neither the authority of the first person with respect to the majority of the subject's mental states (including those of understanding a word) nor the entitlement of his sincere avowals about these states to constitute knowledge.

So, if Wittgenstein and Davidson are right about the normally non-inferential character of first-person knowledge of meaning then N_E cannot be taken as providing a satisfactory requirement with which to evaluate the adequacy of a prospective meaning-constituting fact for explaining self-knowledge about meaning. N_E must be deemed unsuitable for functioning as such a requirement because it misrepresents first-person access to the norms of meaning; our access to these norms is usually not inferred a priori from another piece of knowledge.Foot note 5_28

Let us move now to the question of whether the enhanced non-intentional facts proposed by Soames can actually evade Kripke's normative adequacy objection. Recall that according to Soames the skeptic fails to demonstrate that these dispositional facts do not metaphysically determine meaning; he does not prove that such facts do not satisfy the metaphysical condition of normativity (N_M). If, as Soames claims, meaning facts really supervene on this sort of dispositional facts, then the latter must comply with N_M.

The normative non-adequacy objection against dispositions, let us not forget, was that these items cannot constitute the sort of normativity that is attached to meaning. So, quite independently of whether they could comply with the epistemological constraint of the skeptical problem, dispositions can be safely dismissed as meaning-constituting facts, this is what Kripke would say, because they fail to accord with its constitutive constraint. Yet, Soames insists that at least the dispositions he is proposing can indeed satisfy this constraint -- understood as the condition of metaphysical determination. Who is right here? Let us take a more detailed look at the issue.

According to Kripke, the reason why dispositions in general -- excluding the cæteris paribus ones, which are subject to another problem (see section 3) -- cannot constitute meaning is that the notions of correct or incorrect use of a word, which are crucial for linguistic meaning, remain uncaptured by dispositional facts. Suppose, for example, that it makes sense to attribute to Kripke's subject -- call him Smith -- a certain understanding of the word `+'; he means addition by `+'. Suppose, moreover, that in the case in which Smith has to answer the question «68+57=?» -- i. e. in the new use of `+' -- he responds with the word `125'. A rational justification of Smith's intentional behavior might be the following: Smith answered that way because he understood the expression «68+57=?» as the question of what the sum of the numbers 68 and 57 amounts to. If Smith understands the sign `+' as referring to addition then there is a normative relation between this meaning of the word and a specific group of uses of numerical expressions (which the use of `125' in the above context belongs to): these latter uses are correct answers to questions involving the sign `+' if it means addition. The normative relation between meaning and use is what licenses the appeal to a certain meaning in the rational justification of Smith's linguistic behavior. Now, suppose that the justification of Smith's response resorts to a disposition to add. Could such a disposition rationalize Smith's linguistic deed? No, Kripke would say, because the relation between a linguistic disposition and any of the uses of words that are in accord with that disposition is merely causal; a disposition simply describes how the subject would respond if exposed to stimuli of a certain type.

Soames might protest that the dispositions discussed by Kripke are too simple; human linguistic dispositions, Soames might insist, are much more complicated than those. Let us construe the disposition to use in order to add then in the way he envisages, that is: as a disposition to produce numerals in response to questions `What is n+m?' plus «dispositions covering cases in which I `check and revise' my work, dispositions to insist on one and only one `answer' for any given question, dispositions to strive for agreement between my own answers and those of others, and so on».Foot note 5_29 Recall that, according to Soames, this complex disposition must be conceived as non-intentional. The question is whether the resulting dispositional factFoot note 5_30 could finally capture the notion of a criterion (or norm) for the correct use of a word.

Soames is positive about the prospects of ED-facts to generate the norms of meaning. And we know already his reason for maintaining that such dispositional facts cannot fail to capture the norms of their respective meaning facts: the supervenience of the latter facts on ED-facts guarantees that ED-facts determine metaphysically their corresponding meaning facts.Foot note 5_31 The notion of metaphysical determination is explained in terms of possible worlds: a fact that P determines another fact that Q metaphysically if and only if all possible worlds containing P must also contain Q. Another way of describing the metaphysical determination of Q by P would be to say that the sentences which express Q follow necessarily from the linguistic expressions of P and only from these. Notice that the relation of necessary consequence being used here is parasitic on that of a possible world: a sentence q follows necessarily from p if and if in all possible worlds where p is true q is also true. What is being asserted by the above thesis of metaphysical determination is that the norms of meaning corresponding to the word `+' must follow necessarily and solely from the sentences which express the above suggested complex disposition towards `+'. But is it true that the supervenience of M-facts on ED-facts is as strong as to imply that the norms associated with the meaning of a certain word follow necessarily from the expressions of the ED-fact which corresponds to this word? Why, if mental facts supervene on physical facts, should it be concluded that the latter metaphysically determine the former?

A more or less consensual account of the supervenience of the mental upon the physical would be this: a class of mental properties is said to supervene on a class of physical properties with respect to a certain domain of objects -- events, states or individuals -- if and only if it is impossible for any two elements of the domain to be indistinguishable relative to all their physical properties and yet differ in at least one mental property. In other words: any change in the mental attributes of an object must correspond to a change in some of its physical attributes.Foot note 5_32

One source of divergence among philosophers who defend the supervenience of the mental upon the physical concerns the way the modal operator should be understood. For some, the above impossibility must be understood metaphysically. This is probably the way Soames wishes supervenience to be taken. For other philosophers, the impossibility that helps to define supervenience is weaker than metaphysical impossibility; in this sense, two individuals of the mentioned domain cannot differ in all their mental attributes without being distinguishable in at least one of their physical attributes provided that the method of assignation of mental properties to human subjects remains the same. It is not inconceivable though that such a relation of supervenience might fail to hold if the procedure for identifying psychological properties was different from the one we currently use. This is the view associated with the philosopher who introduced the term `supervenience' into the philosophical debate: Donald Davidson.Foot note 5_33 Of course, there will be as many notions of supervenience as different notions of necessity can be distinguished. But the purposes of the present discussion do not require an exhaustive examination of every one of them; a review of the consequences of adopting each one of these two alternative notions of supervenience will suffice.

Let me start with the Davidsonian supervenience. According to the author of Inquiries into Truth and Interpretation,Foot note 5_34 the mental attributes of an individual (or event) supervene upon all of his (its) physical attributes, including the relational attributes with the environment around him (it). In spite of admitting this sort of determination between mental descriptions or facts upon the physical facts conceived broadly, Davidson has always insisted that the former descriptions are not reducible to the latter: there can neither be empirical laws connecting psychological properties to physical ones nor definitions affirming the synonymy or the co-extensionality between them. The reasons Davidson offers for maintaining that these co-relations between the mental and the physical are not forthcoming come down to the following: there is a normative, externalist and holistic dimension to the attribution of mental properties (in the interpretation of speech and action) with no parallel in the physical domain.Foot note 5_35

The stronger supervenience -- that which entails the metaphysical determination of the physical upon the mental -- requires something like a definitional relation between mental properties and the physical properties that constitute the former's supervenience basis. Neither the co-extensionality nor the extensional inclusion of properties of the second kind in those of the first kind could guarantee that all possible worlds where a set of physical properties are simultaneously instantiated by an object will contain the instantiation by the same object of the mental properties which supervenes upon them. Co-extensionality or extensional inclusion between properties in this world is obviously compatible with their not being co-extensional, or the extensional inclusion not obtaining, in other possible worlds.

The following example will make things clearer. Suppose someone said that the fact that John's C-fibers are firing metaphysically determines the fact that John is in pain. This could only be the case if the proposition that pain is C-fibers firing is metaphysically necessary. That is, if the link between these two properties were anything less than that of metaphysical necessity, no relation of metaphysical determination could obtain between facts like the firing of John's C-fibers and John's being in pain. A way of spelling out this strong relation between the property of being in pain and that of the firing of a brain's C-fibers would be to say that the second defines the first. Another would be to say that the predicates `X is in pain' and `X has his C-fibers firing' are synonymous. How specifically one chooses to spell out the metaphysically necessary connection between pain and C-fiber firing is not so important for our discussion. What matters is that mere co-extensionality between them would not suffice for the metaphysical determination of a mental property like pain by the property of a brain's C-fibers firing.

The trouble is that the strongest relation one can hope to establish between the properties of the supervenient class and those of its supervenience basis is that of co-extensionality, and this only under the special circumstance where the domain of objects to which both sets of properties apply is finite (or equivalently, that there are finitely many properties in the supervenience basis).Foot note 5_36 But if this is so, then the prospects of the supervenience of the mental on the physical securing the metaphysical determination of the mental by the physical look really dim.

The ball is therefore on Soames' side; he must show that supervenience entails not only that for each mental property there is a physical property with the same extension, but also that their co-extensionality obtains across possible worlds. Meanwhile I think we are warranted in being skeptical about the ability of the facts proposed by him (ED-facts) to satisfy the constitutive condition of the skeptical problem and therefore, if my reading of the skeptical problem is correct, to solve the problem.

If I am right about how to conceive meaning skepticism, then the way Soames construes it is mistaken. Fistly, there are no two separate problems -- one of finding non-intentional facts that epistemically determine meaning and the other of finding non-intentional facts that metaphysically determine meaning -- but just one problem with a metaphysical and an epistemological constraint: to find a fact that can, on the one hand, constitute meaning and, on the other hand, justify our beliefs about what we mean. Secondly, his proposed meaning-determining facts are not up to the task of generating meaning facts (i. e. of conforming to the metaphysical condition of the problem). Thirdly, even if they were, that would not have solved the skeptical problem because such non-intentional facts cannot satisfy the problem's epistemological condition.

Boghossian, P. 1989; «The Rule-Following Considerations» in Mind (98): 507-549

Davidson, D. 1970: «Mental Events» in Davidson 1980b

Davidson, D. 1973: «The Material Mind» in Davidson 1980b

Davidson, D. 1980a: «Towards a Unified Theory of Meaning and Action» in Grazer Philosophische Studien (II): 1-12

Davidson, D. 1980b: Essays on Actions and Events (Oxford: Oxford University Press)

Davidson, D. 1984a: «First Person Authority» in Dialectica (38): 101-11

Davidson, D. 1984b: Inquiries into Truth and Interpretation (Oxford: Oxford University Press)

Davidson, D. 1987: «Knowing One's Own Mind» in Proceedings and Addresses of the Aristotelian Society (60): 441-58

Davidson,, D. 1995: «Could There Be a Science of Rationality?» in International Journal of Philosophical Studies (3): 1-16

Diamond, C. 1975: Wittgenstein's Lectures on the Foundations of Mathematics, Cambridge 1939 (Chicago: The University of Chicago Press)

Dummett, M. 1974: «What is a Theory of Meaning?» in Guttenplan, S 1975: Mind and Language Wolfson College Lectures, 1974 (Oxford: Clarendon Press)

Fodor, J. 1987: Psychosemantics (Cambridge: The MIT press)

Kim, J. 1978: «Supervenience and Nomological Incommensurables» in American Philosophical Quarterly (15): 149-156

Kripke, S. 1980: Naming and Necessity (Cambridge: Harvard University Press)

Kripke, S. 1982: Wittgenstein on Rules and Private Language (Cambridge: Harvard University Press)

Lycan, W. 1990: Mind and Cognition (Oxford: Basil Blackwell)

McGinn, C. 1984: Wittgenstein on Meaning (Oxford: Basil Blackwell)

Millikan, R. 1990: «Truth Rules, Hoverflies and the Kripke-Wittgenstein Paradox» in Philosophical Review (99) 3: 323-353. Reprinted in Millikan 1993

Millikan, R. 1993: White Queen Psychology and other Essays to Alice (Cambridge: the MIT Press)

Pinto, S. 1998: Wittgenstein, Meaning and Mathematics PhD dissertation (London: London University)

Smith, B. 1998: «Meaning and Rule-Following» in The Encyclopedia of Philosophy ed. E. Craig (London: Routledge)

Soames, S. 1997: «Skepticism about Meaning: Indeterminacy, Normativity, and the Rule-Following Paradox» in Canadian Journal of Philosophy Supplem. Vol. 23: 211-249

Soames, S. 1998: «Facts, Truth-Conditions, and the Skeptical Solution to the Rule-Following Paradox» in Philosophical Perspectives (12): 313-348

Stich, S. 1978: «Autonomous Psychology and Belief-Desire Thesis» in The Monist (61): 573-91. Reprinted i Lycan, W. 1990

Stich, S. 1983: From Folk Psychology to Cognitive Science (Cambridge: Bradford Books/the MIT Press)

Wittgenstein, L. 1953: Philosophical Investigations (Oxford: Basil Blackwell)

Wittgenstein, L. 1978: Remarks on the Foundations of Mathematics 3^rd. Edition (Oxford: Basil Blackwell) 1^st. Edition 1956

Wright, C. 1980: Wittgenstein on the Foundations of Mathematics (Cambridge: Harvard University Press)