Artículos

 

An Evaluation of Kripke's Account of the Illusion of Contingency

 

Una evaluación de la explicación kripkeana de la ilusión de contingencia

 

Manuel Pérez Otero

 

Departamento de Lógica, Historia y Filosofía de la Ciencia, Facultad de Filosofía, Universidad de Barcelona, LOGOS: Research Group in Logic, Language and Cognition. E–mail: perez.otero@ub.edu

 

Received: July 4, 2006
Revised: February 14, 2007
Accepted: December 5, 2007

 

ABSTRACT

Kripke argued for the existence of necessary a posteriori truths and offered different accounts of why certain necessary truths seem to be contingent. One of these accounts was used by Kripke in an argument against the psychophysical identity thesis. I defend the claim that the explanatory force of Kripke's standard account of the appearance of contingency (the account used to argue for psychophysical dualism) relies on the explanatory force of one of the more general accounts he also offers. But the more general account cannot be used to undermine the psychophysical identity thesis. Specifically, a crucial feature in Kripke's standard account, which is needed to argue for dualism, is explanatorily superfluous. Alternative accounts that are similar to Kripke's original one but lack that trait would also explain the phenomenon. Consequently, the Kripkean dualist argument is blocked.

KEY WORDS: explanatory relevance, rigidity, metaphysical necessity, aprioricity, psychophysical identity.

 

RESUMEN

Kripke ha argumentado que existen verdades necesarias a posteriori, y ha ofrecido diferentes explicaciones de por qué ciertas verdades necesarias parecen contingentes. Una de esas explicaciones se usaba en un argumento con el que Kripke criticaba la tesis de la identidad psicofísica. En este trabajo sostengo que la fuerza explicativa de la explicación kripkeana estándar de la apariencia de contingencia (la explicación utilizada para apoyar el dualismo psicofísico) depende de la fuerza explicativa de una de las otras explicaciones que Kripke también propone. Pero esa otra explicación, más general, no sirve para rechazar la identidad psicofísica. Concretamente, un rasgo crucial de la explicación kripkeana estándar (esencial para su defensa del dualismo) resulta explicativamente superfluo. Explicaciones alternativas, similares a la de Kripke pero que carezcan de ese rasgo, explicarían también el fenómeno. Como consecuencia, el argumento dualista kripkeano queda bloqueado.

PALABRAS CLAVE: relevancia explicativa, rigidez, necesidad metafísica, aprioridad, identidad psicofísica.

 

Kripke convincingly argued for the existence of truths that are necessary but can be known only a posteriori. He also tried to explain why certain necessary truths seem to be contingent, offering in fact different accounts of this illusion of contingency, which were related to the distinction between necessity and aprioricity. One of these accounts was used by Kripke to construct an argument against the psychophysical identity thesis (cf. Kripke 1980). In this article I will defend that the explanatory force of Kripke's standard account of the appearance of contingency (the account used to argue for psychophysical dualism) relies, as far as we know, on one of the more general accounts he also offers. But the more general account cannot serve to reject the psychophysical identity thesis. The elements in the standard account that favour it over the more general account are explanatorily superfluous. They would only provide a more detailed description of how we evaluate the epistemic status of some necessary truths, and this is insufficient to grant the dualist conclusion Kripke extracts from them. Specifically, a crucial trait in Kripke's standard account, which is needed to argue for dualism, adds nothing to the explanation. Therefore, alternative accounts that are similar to Kripke's original one but lack that trait would also explain the phenomenon. Consequently, the Kripkean dualist argument is blocked. (As it will be clear, my aim here is not to vindicate the psychophysical identity theories. I am concerned with the logical structure of the Kripkean argument.)

A preliminary comment. The philosophical background assumed in this article is notably Kripkean. I accept many of Kripke's metaphysical and semantic views on necessity, rigid designation, essence and natural kinds. So, my criticism to the argument for dualism is not connected to other possible objections based on a rebuttal of some of these ideas. There are also replies to the dualist argument that contest some Kripkean premises on the analysis of qualitative mental states. I am reluctant to grant these premises. But my case will not depend on a rejection of them either.

 

1. The Illusion of Contingency and Kripke's Dualist Argument

The class of statements that, according to Kripke, are necessary includes all of the following truths:

(1)  Cicero = Tully

(2)  Hesperus = Phosphorus

(3)  heat = molecular motion

(4)  water = H₂O

Kripke's semantic and metaphysical theory implies that all the aforementioned sentences are necessarily true if they are true at all. The referential terms occurring in them are rigid designators; therefore they designate the same entity (a particular object, a natural phenomenon or a substance) in every possible world. This being the case, the sentences have the same truth value in every possible world. Now, it is a fact that sentences (1)—(4) seem to be contingent (or at least, they seemed to be so, before the works of Kripke and other defenders of direct reference theory argued against that). An obvious desideratum of Kripke–style theories, which attribute necessity to statements like (1)—(4), would be that they could explain why intuitively the statements seem to be otherwise. In fact, Kripke was concerned with this issue, and wanted to explain that illusion or appearance of contingency. He offered two main accounts of the phenomenon, on which I want to focus in this article: what I will call the general account, and what I will call the specific account. The specific account stands out as the official or standard Kripkean explanation of the appearance of contingency. Still, I am going to defend that —contrary to what Kripke himself seems to think— the real explanatory force of the phenomenon is contained in the first, general account.

This general account or general explanation, to which I will also refer as "GE", states that we confuse two properties: necessity and aprioricity; or, at least, we take them to be extensionally equivalent. To be more precise, this and later allusions to the necessity/aprioricity confusion should be understood as just abbreviations for a general confusion between, on the one hand, the structure of metaphysical modal concepts and distinctions made thereof and, on the other, a similar structure in which the epistemic concept of aprioricity takes on the role played by the concept of necessity in the first structure. According to GE, when we encounter a truth that cannot be known a priori, independently of experience, the truth is deemed a contingent one. The explanation applies straightforwardly to sentences such as (1)—(4), which can be known only a posteriori. And this is the end of the story. The mistake consisting in taking necessity for aprioricity accounts for our inclination to find elements of contingency in a posteriori necessary truths.

Let's turn now to the specific account or specific explanation, that I will also call "SE". It can be seen as a more detailed specification of GE, the general account. When applied to sentences like (1)— (4), containing rigid designators, SE proceeds in two stages. The first stage establishes that a necessary statement R seems contingent because (at least in some cases) we associate with R another really contingent statement D that is equivalent to R in respect of the empirical evidence considered qualitatively. Therefore, there is a possible world where we have the same qualitative evidence and D, an statement qualitatively analogous to R, is false (cf. Kripke 1980, pp. 104, 142 and 150). The other stage is a further specification of the previous one. Basically it concerns identity statements (but —as I will point out later— it could be extended to some other necessary truths containing rigid designators):¹ in some of the relevant cases, R is an identity statement, 'R₁ = R₂', composed with two co–referential rigid designators, 'R₁' and 'R₂', and the corresponding statement D takes the form of 'D₁ = D₂', where 'D₁' and 'D₂' are non–rigid designators that serve to fix the reference of 'R₁' and 'R₂', respectively. In one of the main examples, R is 'Hesperus = Phosphorus' —statement (2) above— and D would be 'the heavenly body in such–and–such position in the sky in the evening = the heavenly body in such–and–such position in the sky in the morning' (cf. Kripke 1980, pp. 143— 144). On the face of it, it would appear that SE is nothing more than an instance of GE. But we are going to see that this is not how Kripke himself takes it to be (cf. section 2 below) and, in any case, SE is not a straightforward instance of GE for, as I will defend, its explanatory force requires GE.

Kripke uses SE to criticize a materialist view about the relation between mental properties and physical or neurological properties of the brain: the identity theory, a physicalist approach to the mind–body problem that was advanced in the second half of the fifties. The identity theory holds that mental states are identical to states of the brain, so that identity statements correlating, for instance, pain with some neurological property are true. Let's suppose, following Kripke's example, that (5) were such a statement:

(5) pain = stimulation of C–fibers

Now, it is plausible to think that the two referential terms occurring in (5), 'pain' and 'stimulation of C–fibers', are rigid designators. In the framework of Naming and Necessity, the identity theory implies, therefore, that (5) is a necessary truth. However —like many other psychophysical identifications— (5) seems to be non–necessary, no less so than any other of the paradigmatic examples of a posteriori necessary truths enumerated above: (1)—(4).

But, Kripke points out, the specific account of the illusion of contingency, SE, cannot be applied to (5).² In the case of pain, it is impossible to imagine an epistemic situation qualitatively identical to that of having a pain but in which some property or phenomenon different to pain is selected by a non–rigid designator. Because in every qualitatively alike situation the phenomenon would be felt as painful; and that property of pain, its being painful, is not just essential to pain, but possessed only by pain (what is felt like pain is pain) (cf. Kripke 1980, pp. 150–151). The apparent contingency of the hypothetical connection between pain and C–fibers stimulation is not explainable away invoking such an analogy with the other cases, and we have no other explanation for it (cf. Kripke 1980, p. 100, and Kripke 1977, p. 101). We should conclude, therefore, that the apparent non–necessity is real: (5) is not a necessary truth. So, it is not true at all.

A number of dualist arguments similar to Kripke's have been developed by other philosophers. The one in Chalmers (1996, chapters 2 and 4) is especially close to Kripke's original argument. My objections below may also be relevant to oppose Chalmers' theses; but here I will restrict myself to a discussion of Kripke's approach.³

 

2. The Straight Story: Confusion of Necessity with Aprioricity

It is tempting to think that GE, what I have termed the general account of the illusion of contingency, is not an explanation at all, and —in any case— Kripke does not endorse it as an alternative way to explain that illusion. It would be just an introduction, like a preliminary stage (without independent explanatory relevance) of SE, the specific account. According to this belief, Kripke cannot think of GE as a plausible explanation by itself; otherwise, in his reasoning against the identity theory, he would not say that the account provided by SE is "the only model I can think of for what the illusion might be" (Kripke 1977, p. 101), and, therefore, he would not have a principled objection to the identification of mental states with brain states.

Nevertheless, there is evidence against that view. Regarding first the exegetical issue, Kripke (1977) seems to endorse GE as a relatively independent acceptable account of the illusion of contingency. There, he asks the question why people believe that (1) and (2) —'Cicero = Tully' and 'Hesperus = Phosphorus', respectively— are contingent. Kripke considers three answers:

Aside from the identification of necessity with a priority [...]. There are two things which have made people feel the other way [feel that (1) and (2) are contingent]. Some people tend to regard identity statements as metalinguistic statement, to identify the statement "Hesperus is Phosphorus" with the metalinguistic statement "'Hesperus' and 'Phosphorus' are names of the same heavenly body". And that, of course, might have been false. (Kripke 1977, pp. 89–90)

After rejecting this metalinguistic reading of identity statements, Kripke asks what people who still say Hesperus might not have been Phosphorus can mean:

Now there are two things that such people can mean. First, they can mean that we do not know a priori whether Hesperus is Phosphorus. This I have already conceded. Second, they may mean that they can actually imagine circumstances that they would call circumstances in which Hesperus would not have been Phosphorus. (Kripke 1977, pp. 90–91)

These last words are the prelude to a more extensive comment, roughly equivalent to the specific account, SE. I take it then that Kripke is providing three different answers to the question for the illusion of contingency, which work as explanatorily relevant accounts of the phenomenon. The third answer, SE, is the most important. But GE, the confusion of necessity with aprioricity, is also mentioned, on two occasions: before and after discussing the second answer, the confusion based on the metalinguistic reading of identity statements. Those three answers to the question are schematic explanations of the illusion of contingency. Therefore, it is sensible to assume that Kripke has considered the general account, GE, as one of the relevant accounts, relatively separate from his specific account, SE, even if they are related.⁴

Someone may think that what I am counting as the second Kripkean answer to the question for the illusion of contingency, ME (the metalinguistic account), is just a particular case of the third one, SE. For the metalinguistic statement (m), "the heavenly body named 'Hesperus' is the heavenly body named 'Phosphorus' ", could be thought a relevant instance of the scheme 'D₁ = D₂' postulated by SE as the epistemic surrogate for 'Hesperus is Phosphorus'. So, in classifying ME and SE as two separates accounts, I would be misunderstanding Kripke, or Kripke himself would be making a mistake.

My reply will make clear why the metalinguistic account Kripke is referring to (ME) is not a case of SE, and why Kripke doesn't dedicate too much time to discuss it (and he doesn't even mention it in Kripke 1980). In order to take the metalinguistic statement (m) as an instance of the scheme 'D₁ = D₂' (where 'D₁ = D₂' is equivalent to 'Hesperus is Phosphorus' in respect of the empirical evidence considered qualitatively, and 'D₁' and 'D₂' are non–rigid designators that serve to fix the reference of 'Hesperus' and 'Phosphorus', respectively) we need to assume the following restriction: 'Hesperus' and 'Phosphorus' must be used (in each one of the circumstances which are relevant to evaluate (m)) in the same way we use them (maintaining fixed their meaning, let's say). Relatively to circumstances or possible worlds where, for example, 'Hesperus' is a name of the Moon and 'Phosphorus' is a name of Socrates, (m) cannot be considered a plausible epistemic equivalent to our 'Hesperus is Phosphorus'. Now, it is very clear that when Kripke criticizes the confusion postulated by ME, he is not assuming this restriction (for his criticism is that the semantic equivalence between the original identity statement and the metalinguistic statement would require, at least, such a restriction; cf. Kripke 1977, p. 90). Therefore, the metalinguistic account mentioned by Kripke is not a case of SE. But then, this ME is too rude, and it would be more sensible to pay attention instead to a wiser metalinguistic account that incorporated the restriction. That's right. But —as the objection reveals— this wiser metalinguistic account may be a case of SE; it could not be counted as a separate explanation. That's why —I think— Kripke, when counting the distinct possible answers to the question for the illusion of contingency, mentions ME —under the rude reading— and quickly discards the confusion it describes.⁵

Whatever the accurate interpretation of Kripke's thought may be, I want to compare GE with SE and defend the thesis that GE is, by itself, a good account of the illusion of contingency. SE is neither more explanatory, nor is it a better explanation than GE, because the explanatory power of SE lies in GE. (My reasons, if correct, will also constitute evidence for accepting that Kripke has thought of GE as an independent plausible account of the illusion of contingency; although —as I hinted above— that introduces some tension in his dualist argument.)

The important consequence of this discussion is that, contrary to Kripke's suggestion, SE is not the only imaginable account of the appearance of contingency in the connection between pain (or any other mental state) and a physical state. GE is a good alternative. Like any of the sentences (1)–(4), the statement (5), 'pain = stimulation of C–fibers', seems to be contingent because it is a posteriori, and we have confused its epistemic status (a priori/a posteriori) with its metaphysical status (necessary/contingent). In other terms, GE applies perfectly well to (5), as to any other paradigm of necessary a posteriori truth. But from GE —I will argue— it is not possible to construct the dualist argument Kripke constructs from SE.

 

3 . Explaining the Explanans

My main thesis can be put in the following terms: SE does not explain the illusion of contingency better than GE, because the explanatory force of SE is contained (as far as we know) in GE. I start now the argument for that claim, which will continue in the three following sections.

Let's compare the two accounts. GE postulates a confusion on our part: we are liable to confuse two properties: necessity and aprioricity; and we confuse them, in this case, or we think, at least, that the two properties are extensionally equivalent. Therefore, in order to evaluate a statement's modal status (its necessary or contingent character) we assume, explicitly or implicitly, that it is right to evaluate its epistemic status (its a priori or a posteriori character). That's why typical a posteriori truths seem to be contingent. We come to know their a posteriori character, and conclude that they are contingent.

Prima facie, SE does not postulate such a confusion of two properties that statements can have. But SE also postulates a confusion on our part. In evaluating the modal status of a certain necessary sentence R, we associate —SE claims— with R some other sentence D, which is contingent; and we conclude, on the basis of the real contingency of D, that R is contingent. Therefore, we mistake, explicitly or implicitly, R for D; or, at least, we mistake the modal status of R for the modal status of D.

Now, the criticism to the alleged explanatory power of SE starts with this question: why do we confuse the modal status of R with the modal status of D? It seems that the proponent of SE should answer that question. In my view, there are just two interpretations of what Kripke has done to respond to it: (i) He offers no answer at all. (ii) He appeals implicitly to GE: R and D are epistemically equivalent; therefore, they share the epistemic status; since we mistake —GE claims— the epistemic status for the modal status, we transfer the modal status of the contingent truth D to the epistemically equivalent truth R. It may be that more options are open to a potential defender of SE. But I want to check the explanatory value of SE as it is stated in the works of Kripke. He endorses SE without any further explicit attempt to clarify why we confuse the modal status of R with the modal status of D. His comments have to be taken, at most, in the sense of (ii). In this context, it would be inappropriate to attribute to him any answer significantly different to (i) and (ii).⁶

I will consider now what results from our comparison of GE and SE when we assume alternative (i). Alternative (ii) is discussed in sections 5—6. If a defender of SE does not answer the crucial question mentioned above about the confusion of the corresponding modal status of R and D, then she has just substituted a mystery for another. The original explanandum was the appearance of contingency of necessary statements such as (1)—(4). In accounting for that explanandum, SE proposes that we mistakenly believe the original statement to be modally equivalent to some other, contingent, statement. But it leaves us without any account for this other mistake.

What about the other account of the original explanandum? Is GE in a better position? It is. GE also clarifies the illusion of contingency by giving us in return another phenomenon that would need clarification: why do we confuse necessity with aprioricity? or why, at least, do we think that all necessary truths are a priori? But now the situation is not the same as it was regarding SE. Kripke himself addresses the last question, and tentatively proposes two hypothesis to explain the traditional assumption that necessity is aprioricity:

I think people have thought these two things [the terms 'necessary' and 'a priori'] must mean the same for these reasons:

First, if something not only happens to be true in the actual world but is also true in all possible worlds, then, of course, just by running through all the possible worlds in our heads, we ought to be able with enough effort to see, if a statement is necessary, that it is necessary, and thus know it a priori. [. . . ]

Second, I guess it's thought that, conversely, if something is known a priori it must be necessary, because it was known without looking at the world. If it depended on some contingent feature of the actual world, how could you know it without looking? (Kripke 1980, p. 38)

Further relevant data could also contribute to explaining our inclination to confuse necessity and aprioricity. Most of the diverse linguistic resources utilized to express modal concepts (in the metaphysical sense) also have standard epistemic interpretations, as we can recognize in Kripke's considerations on some uses of "it could have turned out that P" or "might".⁷ Even if this is a merely pragmatic —not a semantic— trait of the relevant expressions, it also makes more comprehensible the confusion of necessity with aprioricity. Now, as we have indicated, Kripke has provided two explanatory hypothesis that may also (partially) explain the mistake described by GE. This fact makes an important difference with respect to SE, for which no analogous reasons have been given. SE replaces an explanandum by another, and says nothing about the latter explanandum. GE also replaces an explanandum by another, but is backed by a (partial) explanation of the second one. So far, GE should be considered the best of both.

 

4 . The Explanatory Scope of the General Account

The data favouring GE over SE do not finish here. GE has another advantage: the range of cases potentially explainable by GE is broader than the range of cases covered by SE. In principle, GE can be applied in relation to any necessary a posteriori truth that seems to be contingent. SE, in contrast, would be restricted just to identity statements (this claim will be qualified below). Indeed, SE does not even account for all necessary apparently contingent identity statements, at least not according to Kripke. The examples of our initial list, (1)—(4), illustrate the characteristic illusion of contingency we want to explain. Kripke himself presents and discusses these examples. Nevertheless, when he comes to the specific paradigm provided by SE, he uses as illustrations (2) and (3) —'Hesperus = Phosphorus' and 'heat = molecular motion', respectively— but not (1) —'Cicero = Tully'. This is hardly surprising. Kripke accepts that, in some cases, the reference of a proper name is fixed by a description. Let's call descriptive to names belonging to this category (paradigmatically exemplified by 'Julius', Gareth Evans' famous case). According to Kripke, it is plausible to suppose that the two names occurring in (2), 'Hesperus' and 'Phosphorus' are —in that sense— descriptive.⁸ This fact is essential for applying SE to (2), because SE requires, in its second stage, associating with (2) some other sentence, epistemically equivalent to (2), like (2),

(2) the heavenly body in such–and–such position in the sky in the evening = the heavenly body in such–and–such position in the sky in the morning,

constructed with two non–rigid designators. Let's assume, for the sake of argument, that when an expression fixes the reference of another, their co–referential condition is known a priori; that would make the two sentences (2) and (2) epistemically equivalent, in the sense suggested by Kripke.

Now, it is clear that Kripke thinks ordinary proper names are not descriptive; their reference is not fixed by a description. Therefore, the application of SE to identity statements made with co–referential ordinary (non–descriptive) proper names is hampered. Kripke cannot specify, not even sketch, what kind of non–rigid designator D₁ is associated with a rigid designator R1, in the way required by the paradigm (the second stage in SE), when R₁ is an ordinary proper name. Non–descriptive proper names occur presumably in (1), and in many other sentences expressing necessary apparently contingent identities. That being so, Kripke's use of SE would allow him to explain the illusion of contingency only of sentences following the model of (2)—(4). Sentences like (1) are left out.⁹

Besides being able to apply to more identity statements than SE, GE can also explain the illusion of contingency in necessary a posteriori truths of any other form. In fact, the paradigm provided by SE can also be naturally extended to apply to other necessary a posteriori truths containing descriptive proper names, not just identity statements. For instance, truths derived from the Kripkean theses about the necessity of origin, which instantiate the form: x is father of y, where x and y are descriptive proper names. But this strategy will not work when non–descriptive proper names are involved. Moreover, SE does not account for necessary apparently contingent true statements that do not contain rigid designators. It is plausible that truths of this kind exist. Some of them, for instance, will be entailed by Kripke's conception of possible worlds as alternative histories of the actual world, identical to it up to a certain time and diverging forward from the actual course. This branching conception of possibility (akin to the principle of necessity of origin) may imply purely descriptive or structural necessary truths only cognoscible a posteriori.¹⁰ As an illustration, I tentatively propose the following statement:

(6) there exist at least n simple individuals

The simplicity of an individual should be taken as entailing that the individual is (metaphysically) indivisible and eternal. Let us now suppose that there are indeed simple individuals, and that n is a certain positive number, not greater than the number of simple individuals. Then, I think (6) is a necessary a posteriori truth. But it is also apparently contingent in the relevant sense that concerns us here. The idea of simple individuals is close to Armstrong's position (cf. Armstrong 1989, p. 38). Armstrong, however, allows contracted possible worlds, lacking some simple individuals (1989, p. 47). It is here that the branching conception of possibility matters. That view would entail that the n (or more than n actual simple individuals exist (and exist as simple individuals) in every possible world branching from the actual world (i.e., every world which has some initial segment of its course of history in common with the actual world), and that these are all the possible worlds.¹¹

To sum up, in last section we have identified two interpretations of what Kripke has done to respond to the question "why do we confuse the modal status of the necessary sentence R with the modal status of the epistemically equivalent D?": (i) He offers no answer at all. (ii) He appeals implicitly to GE. If we assume alternative (i), then GE counts as a better explanation than SE for two reasons. First, SE has not been complemented with any answer to the mystery it substitutes for the original explanandum (the illusion of contingency of some necessary statements). Kripke himself provides a complement of this kind in the case of GE. Second, GE has a broader range of application: while SE could only explain the appearance of contingency of necessary truths containing descriptive referential terms, GE can also account for the cases involving ordinary proper names and even any other necessary apparently contingent truth. Our next task is to examine alternative (ii).

 

5 . The Specific Account Relying on the General Account

Let's now consider alternative (ii). Under that option, the defender of SE resorts to GE to explain why we confuse the modal status of R with the modal status of D. Since the two statements share the epistemic status, and —according to GE— we mistake the epistemic status for the modal status, we also think that the modal status of R has to be the same as that of the contingent statement D.

Now, if this appeal to GE is assumed, then we should recognize that the explanatory power of SE is entirely contained in GE. In particular, and this is my main claim, the assumption that the statement D we associate with a certain necessary apparently contingent identity statement R (epistemically equivalent to it) is contingent is a non–explanatory trait of SE. In other terms, it is explanatorily superfluous to postulate a difference in the modal status of R and D. (It's worth remembering that this difference in modal status is crucial in Kripke's dualist argument.) I rely on that point to ground this other claim: assuming the alternative (ii), the phenomena purportedly explained by SE are at least equally well explained by GE. And, as noted in the preceding section, GE has the further advantage of being able to explain some other phenomena beyond SE's scope.

The irrelevance of the difference in modal status of R and D is enhanced by the following reflection. Let's remember: R is the original necessary truth that is apparently contingent; and D is the epistemically related statement, which is —SE says— really contingent. If the modal status of R is going to be evaluated by focusing on an epistemically equivalent statement, D (since our confusion of modal and epistemic status makes us believe that epistemic equivalence entails modal equivalence), then (just by the same token) in our evaluation of the modal status of D we will be —mistakenly— confident that ascertaining the epistemic status of D is sufficient. Therefore, we can judge (correctly) that D is a posteriori, derive from that judgement (combined with the wrong assumption described by GE) the thesis that D is contingent, and finally derive from that thesis (again with the mistaken identification of epistemic and modal equivalence implied by GE) the belief that R is contingent. In this chain of inferences there is no need to suppose that D is more necessary than R. The necessity/aprioricity confusion can operate several times. It operates when we take R and D to be modally equivalent. Why should it not also operate when we check the modal status of D? Once GE is crucially invoked to account for the transition from R to D, we can afford another invocation of GE to account for the apparent contingency of D itself. On this account, the apparent contingency of R is not explained by the real contingency of the associated D. In that sense, the assumption that D is contingent is explanatorily superfluous; another element essential to SE —interpreted according to the alternative (ii)—, namely GE, makes this assumption pointless.

I am not claiming that in most of the examples considered the corresponding statement D is necessary. Kripke may be right on the descriptive condition of some proper names, like 'Hesperus', or natural kind terms, like 'heat'. Their references may well have been fixed by non–rigid descriptions, in such a way that the truths obtained replacing some of the original, rigid, referential terms with these descriptive non–rigid designators (statements like (2), 'the heavenly body in such–and–such position in the sky in the evening = the heavenly body in such–and–such position in the sky in the morning', or 'that which is sensed by sensation S = molecular motion') are really contingent.¹² But the issue I am addressing is how the illusion of contingency is explained; the point is where exactly the explanatory force of the account proposed lies. And what I deny is the explanatory power or relevance of the fact that the statement D in each case involved is contingent, independently of whether or not some of them are indeed contingent.

 

6. Which Dualist Argument Is Supported by the Specific Account?

If the foregoing remarks are correct, should we conclude that every element in SE besides its implicit appeal to GE (according to option (ii), which is the one we still will assume in this section) is completely useless? I think not, and it does not follow from any thesis I have defended so far. There is a sense in which SE —or parts of SE— can rightly be thought of as constituting a more complete description of what happens in some of the cases in which we take the modal status of a statement as its epistemic status. For some of the relevant necessary apparently contingent statements, SE provides details of how such a confusion takes place; details on which GE remains silent. In this sense, we find an advantage of SE over GE: it is more informative.

Let's put it in other terms. Even having at our disposal GE as a good general account, it is still legitimate to ask for further specific information about how the necessity/aprioricity confusion operates in relation to concrete statements leading us to classify them wrongly as contingent. Once we accept GE, this question is basically a question about the cognitive procedure or "mechanism" underlying our assessments of the epistemic status (in particular, the a posteriori character) of necessary apparently contingent statements. It is here that SE enters into scene, postulating additional data about this procedure. If these data are right, SE provides valuable information to complement GE. On reflection, we find —SE hypothesizes— that for a subclass of the necessary apparently contingent truths (the subclass illustrated by the examples (2)—(4) in our initial list), the following claims about the relevant necessary statement R are right:

(a)  We associate with R another statement, D, that is equivalent to R in respect of the empirical evidence qualitatively considered.

(b)  If R is an identity statement constructed with rigid designators, then in the corresponding statement D the original rigid designator, R₁ ('Hesperus', 'heat', 'water'), does not occur. A descriptive expression, D₁ ('the heavenly body in such–and–such position in the sky in the evening', 'that which is sensed by sensation S', 'waterish stuff'), occurs instead. This expression D₁ has served to fix the reference of R₁, and is thereby a priori, conceptually linked with R₁.¹³

(c) The only descriptive expressions —satisfying the constraints (b) imposes— we can think of express properties contingently linked with the entity (object, natural kind, phenomenon) rigidly designated by R₁, with the result that D is a contingent truth.¹⁴

Now, with respect to the class of statements for which the clauses (a)—(c) hold, SE indeed offers a more complete description of what is involved in the illusion of contingency than GE does. But it does not follow that the further information contained in SE is explanatorily relevant to account for the illusion of contingency. SE does not remove any perplexity or mystery not already removed by GE. It is the GE–component in SE that does the explaining. SE is just descriptively more detailed, in relation to some necessary apparently contingent truths. Therefore, SE is not an unavoidable model that must be fit by all necessary apparently contingent identities, contrary to Kripke's suggestions (cf. Kripke 1980, p. 100, and Kripke 1977, p. 101).

To clarify these remarks I will now examine what kind of argument for the psychophysical dualist thesis could indeed be constructed from the data compiled in clauses (a)—(c). It is an argument similar to Kripke's original one, but SE plays a different role in it. SE can be identified with the thesis that clauses (a)—(c) hold for all necessary apparently contingent identity statements. Then, instead of thinking of SE as the result of an inference to the best explanation where the phenomenon (best) explained by SE is the illusion of contingency (as I believe it is usually understood), we can use the sub–thesis that clauses (a)—(c) hold for statements (2)—(4) as granted data, as part of the phenomenon to be explained. And SE would follow from that sub–thesis by a mere enumerative inductive inference. In general terms, the dualist argument would go like this. Premise: clauses (a)—(c) hold for necessary apparently contingent identity statements such as (2)—(4). Conclusion (inductively supported by the premise): (a)—(c) hold for every necessary apparently contingent identity statement. Further conclusion: any psychophysical identification that fits the pattern of (5), 'pain = stimulation of C–fibers', is false. That follows from the previous conclusion in conjunction with two other assumptions I do not question here: if (5) were true, it would be necessary (and it certainly is apparently contingent); clauses (a)— (c) do not hold for (5), since any descriptive way of fixing the reference of 'pain' according to (b) has to select the same phenomenon, pain, in every possible world, and this goes against (c).

The conclusions of that inductive inference (as those of any other inductive inference) can be rejected while accepting its premise. There are several alternatives compatible with the truth of the premise and the falsehood of the conclusions. Some necessary truths will closely follow the pattern of statements (2)–(4), and clauses (a)–(c) would presumably apply to them. But, on the one hand, it is probable that there are necessary a posteriori identities that do not involve any of the clauses (a)–(c), or, at least, do not involve the clauses (b)—(c), as may happen with necessary truths expressed without rigid designators. We cannot rule out the possibility that the procedure or mechanism (underlying our misjudgement of the modal status) operating in relation to necessary a posteriori truths not containing rigid designators operates also in relation to some necessary a posteriori identities, which, in consequence, would not follow the paradigm (a)—(c). (Remember that Kripke himself can hardly assume that (b) holds for identity statements like (1), 'Cicero = Tully', constructed with ordinary non–descriptive proper names.)

On the other hand, even if the procedure involved in cases of necessary identities always entails (a) and (b), it might turn out that (c) had exceptions. In other words, for some necessary apparently contingent identity truth R containing a rigid designator R₁, maybe we associate a corresponding epistemically alike statement D containing, instead of R₁, a descriptive expression D₁ that designates also rigidly the same entity designated by R₁. And (5) could be a case of this sort. In short, we have no overwhelming reason to presuppose that (5) must, if true, follow the model of (2)—(4).

Consequently, there are at our disposal alternative accounts —other than SE— of why (5) seems to be non–necessary. GE is one of them. A more informative account could be GE complemented with something like the part of SE constituted just by clauses (a) and (b). This new account we are considering now is more specific than GE but more general than SE. Let us call it "IE", the intermediate explanation. It is SE minus clause (c): IE is the result of dropping the crucial assumption in SE that D is contingent. I have pointed out that this assumption is explanatorily superfluous; but it is essential to the dualist argument. According to IE, when we evaluate the modal status of (5), we evaluate the modal status of an epistemically equivalent statement, (5), where 'pain' is replaced with a descriptive expression that fixes its reference. This other statement, (5), is also necessary (against the dictates of (c)), but seems to be contingent. This appearance is explained by the general confusion described by GE: we judge it contingent because we see that it is a posteriori. Any of these accounts, GE or IE, may be appropriate, especially given that some of them (or another similar account) have to work for the existing cases of necessary truths that do not entirely fit the (a)— (c) paradigm.¹⁵

In short, Kripke suggests that the only explanation of why (5), even if it were true, seems to be non–necessary must follow the SE model, which involves clauses (a), (b) and (c). Clause (c) cannot apply to (5); so, the alternative would be the trivial explanation: (5) seems to be non–necessary because it is non–necessary. I point out the possibility that the proper explanation —of the appearance of contingency regarding (5)— does not involve clause (c), or does not even involve (b). It may be difficult for me to provide concrete examples of how applying IE —clauses (a) and (b), but not (c)— to (5). But we can justifiably accept that some explanations of illusions of contingency will not involve (c), nor even (b): the explanations corresponding to necessary a posteriori truths not containing rigid designators, or the explanations corresponding to statements like (1), where the rigid designators are not descriptive. So, we cannot discard that there is an alternative explanation for (5) too. The dialectic situation we have arrived to is the following. Kripke presents a controversial argument for dualism, and he must justify each of its premises. A crucial premise is that any explanation of why (5) seems non–necessary (compatible with the assumption that it is true) should involve clause (c) of SE. Since we understand that in relation to some necessary a posteriori truths the proper account of the illusion of contingency may not involve clause (c), it is Kripke who has the main onus of proof in demonstrating his crucial premise.

 

7. The EPP Principle

In this section two possible objections will be examined and rejected. They object to the interpretation of SE I have just proposed, and eventually to the main thesis I have been defending: the explanatory force of SE does not depend on D's contingency (so that alternative accounts without this assumption would also explain the appearances of contingency).

The objector claims that SE cannot reasonably be conceived without clause (c), and that the proper explanation of the illusion of contingency requires in any case that the crucial thesis underlying (c) must hold: the apparent contingency of R has to be explained by a real contingency of some other item, in some way related to R. The two objections are closely related, since as we shall see the former probably relies on an application of the principle explicitly invoked by the latter.

First objection. Taken as an explanation of the illusion of contingency, SE cannot be thought of in the way suggested by the threefold division in separate clauses (a)—(c). The division suggests —and I have remarked— that for some necessary identity statements it is conceivable that clause (a) holds without (b) and (c) holding, or that (a) and (b) hold without (c). But —the objection goes on— this would be contrary to the spirit of SE. In particular, SE makes little sense without the last step: clause (c). The difference in modal status between R and D is crucial to SE. The whole point of moving from the necessary but apparently contingent identity statement R to the qualitatively equivalent truth D —the move described by clause (a)— is to have a more basic case where any appearance of contingency should correspond to real contingency, where a mistake about the modal status would hardly be explainable except by the trivial account: D seems to be contingent because it is contingent. If that is not essential to the move from R to D, what is the transition for? If the appearance of contingency of D could be satisfactorily explained invoking again the general confusion GE describes —as it has been invoked to explain the confusion between the modal status of R and D— it seems that the transition from R to D was vain.

I would reply to the objection by stating, first, that my preferred explanation of the illusion of contingency is the general one: GE. I can agree that clause (c) may be —in a sense— crucial to SE. But to account for the illusion of contingency we do not need clause (c). And that is why I have defended that SE is not a better explanation than GE; to repeat my claim: the explanatory power of SE is contained in GE, which does not commit us to the difference in modal status postulated by (c).

Now, I have also claimed that those further elements in SE, despite being explanatorily irrelevant (in relation to the phenomenon —the illusion of contingency— to be accounted for), may provide a more complete description (more complete than the very general assertion provided by GE) of how the necessity/aprioricity confusion operates in relation to some identity statements: the identity statements illustrated by (2)—(4), which follow the paradigm (a)—(c). Finally, in the previous section, I have suggested that for other necessary truths —including possibly (5)— clause (a) may hold, or even clauses (a) and (b). Therefore, we can imagine other accounts containing clauses (a) and (b) —IE, for instance— which are more informative than GE, but which like GE (and unlike SE) cannot serve against psychophysical identity.

But —the objection says— what sense does the presence of clauses (a) and (b) make in the absence of (c)? What sense does it make to move from R to D without assuming a difference in their modal status?

The transition from R to D could be worthwhile whenever, and insofar as, the epistemic status of D (its a priori or a posteriori condition) turns out to be more easily or more reliably ascertainable than that of R. The point of associating to R an epistemically alike statement D containing some descriptive expression instead of the original referential term is not to reach a surrogate for R that is nevertheless really contingent. Since the evaluation of the epistemic status of R is thought to be sufficient to know its modal status (according to GE, that is invoked —assuming the option (ii), mentioned in section 4— to account for the confusion —postulated by SE— between the modal status of R and D), what we want to know is whether R is a priori or a posteriori. And the point of moving from R to D can be to deal with a statement epistemically equivalent to R, but whose epistemic status is easier to evaluate. This epistemic status is easier to evaluate probably because ascertaining the epistemic status of a statement requires us, at least in normal cases, to have an idea of what its justification would be; and we understand more clearly what would justify D than what would justify R.

The defender of SE, and of its clause (c), can insist on the requirement that D has to be contingent making explicit a certain methodological principle that would grant her stance. The second objection is based on this principle, which I will call "EPP Principle" (for epistemic priority of properties).

We can express the content of the EPP Principle in the following terms. When a particular entity a seems to have a certain property P, if a is not really P, then, ceteris paribus, the most reasonable explanation of that appearance has to postulate some other particular entity, b, related with a in some way, which is really P. Closely linked to the principle is the thesis that a appears to be P because we confuse a and b, or —at least— because we assume that a is P if and only if b is P.

The idea underlying the EPP Principle is this: if we are wrong in judging that a specific item has a certain property, it is a methodologically good strategy —when we discover our mistake— to assume that we are not wrong about which property is being instantiated. It is wiser to attribute the mistake to a misidentification of the item that really instantiates the property. If I believe that a is P, then even if that belief is false it is quite likely that something is P; it's just that maybe it is not a, but some other entity, b, which is P.¹⁶

It is probably not the bare EPP Principle as I have formulated it here that would gain the support of some philosophers, but just specific applications of it regarding particular cases. At any rate, I will reject both the EPP Principle in general and its application to the phenomenon we are concerned with: the illusion of contingency of some necessary truths.

Let's see how EPP applies to this phenomenon. The relevant particular entity a is now a certain necessary truth R; and the property, P, that R appears to have, is being contingent. The EPP Principle recommends endorsing the thesis held by SE: since R is not contingent, another item, b (D, which is related to R by a certain relation of epistemic–qualitative equivalence), is really contingent.

Kripke's defence of SE may be partly based on an implicit adoption of the EPP Principle. Specifically, the following passage from Kripke (1980) seems to presuppose the principle, given the use of the definite article in the second definite description used (which I have highlighted):

In sum, the correspondence between a brain state and a mental state seems to have a certain obvious element of contingency. [. . . ] the element of contingency [. . . ] cannot lie, as in the case of heat and molecular motion, in the relation between the phenomenon (= heat = molecular motion) and the way it is felt or appears. (Kripke 1980, p. 154)

In using that definite description, Kripke is claiming that the element of contingency he refers to exists. In other words, he postulates the existence of something that really is contingent. Indeed, it is as if he inferred this existence from the mere fact that something seems to be contingent.

As I said, my rejection of the strategy of invoking the EPP Principle to uphold SE is twofold. On the one hand, I do not see why such a principle should be adopted in the first place. On the other hand, even if the principle were a good general guide, it would not be appropriate to apply it to the case we are concerned. Let's discuss the two issues in turn.

Consider a false belief that a particular entity a has a certain property P. The EPP Principle recommends supposing that some other entity b, different to a, is P. Now, there is another option that is, prima facie, no less simple or reasonable than that: a has some other property F, different to P. The confusion–about–which–particular–instantiates–the–property strategy, dictated by the EPP Principle, presupposes a correct identification of the property allegedly instantiated, P, and postulates that we mistake the particulara for a different particular, b, which is really P (or it postulates that we assume that a is P if and only if b is P). The confusion–about–what–property–is–instantiated–by–the–particular strategy, dictated by the alternative option, would presuppose a correct identification of the particular, a, allegedly instantiating a property, and would postulate that we mistake the property P for a different property, F, which is really instantiated by a (or it postulates that we assume that a is P if and only if a is F). This other strategy is dictated by what we could call "EPI Principle" (for epistemic priority of individuals).

What strategy is, ceteris paribus, methodologically preferable? I think that an epistemological argument would be needed to choose between them. In its absence, we should not opt beforehand for either one. If neither strategy guides our search for an explanation of the mistake (in believing that a is P), we should address specific cases looking for other elements that could make a significant difference between alternatives accounts. In other words, we should pay attention to whether the ceteris paribus condition is not met after all.

That is precisely what happens in relation to the phenomenon we are primarily concerned with, the illusion of contingency. The alternative explanations, GE and SE, are not on a par. SE follows the con–fusion–about–which–particular–instantiates–the–property strategy, ruled by the EPP Principle. GE follows the confusion–about–what–property–is–instantiated–by–the–particular strategy, ruled by the EPI Principle. Now, even if the EPP Principle had some general plausibility to recommend it, it cannot serve to support SE. Remember the two alternative interpretations of how SE approaches the problem it substitutes for the original illusion of contingency: (i) and (ii). If SE is silent about that problem —option (i)— the EPP Principle would support, ceteris paribus, SE. But the conditions are not equal (the ceteris paribus condition is not met): GE would explain a wider range of data; and some indications have been given (by Kripke himself) to partly account for the problem GE substitutes for the original illusion of contingency (the problem of explaining why we mistake necessity for aprioricity). Suppose —option (ii)— SE confronts the problem by resorting implicitly to GE. Again, the EPP Principle would support, ceteris paribus, SE. But, again, the conditions are not equal: GE postulates just a confusion of properties; SE postulates a confusion of particulars, and also (because of its appeal to GE) a confusion of properties. So, after all, SE appears to follow the two strategies (the confusion–about–which–particular–instantiates–the–property strategy, and the confusion–about–what–property–is–instantiated–by–the–particular strategy). GE is methodologically simpler than SE.

 

8. Conclusions

I have defended that the Kripkean dualist argument for rejecting psychophysical identities can be blocked. The argument assumes that the specific account, SE, is the best explanation of the illusion of contingency; but this thesis is controversial at least. As a first alternative account, another explanation mentioned by Kripke himself is available: the general account, GE. The comparison suggests two advantages of GE over SE: 1) it covers a greater number of successfully explained cases (necessary apparently contingent truths); 2) the explanatory force of SE relies, as far as we know, on GE.

SE is more informative than GE. But, given its explanatory dependence on GE, the elements of SE not contained in GE are explanatorily irrelevant (to account for the illusion of contingency), precisely because of GE. In particular, a crucial assumption in SE, required to argue for psychophysical dualism, is explanatorily superfluous or irrelevant: the thesis that D (the epistemically equivalent statement associated with the original necessary apparently contingent truth R) is contingent. Therefore, we can conceive another alternative account, richer than GE, and closer to SE in its informational content: IE, the result of omitting from SE the crucial thesis about D's contingency. In short, we have alternative accounts of the phenomena, which cannot be used against psychophysical identities.¹⁷

 

REFERENCES

Armstrong, David M., 1989, A Combinatorial Theory of Possibility, Cambridge University Press, Cambridge.         [ Links ]

Block, Ned and Robert Stalnaker, 1999, "Conceptual Analysis, Dualism, and the Explanatory Gap", Philosophical Review, vol. 108, pp. 1–46.         [ Links ]

Chalmers, David, 1996, The Conscious Mind, Oxford University Press, New York.         [ Links ]

Chalmers, David and Frank Jackson, 2001, "Conceptual Analysis and Reductive Explanation", Philosophical Review, vol. 110, pp. 315–360.         [ Links ]

Evans, Gareth, 1979, "Reference and Contingency", Collected Papers, Clarendon Press, Oxford, pp. 178–213.         [ Links ]

Forbes, Graeme, 1985, The Metaphysics of Modality, Clarendon Press, Oxford.         [ Links ]

Fumerton, Richard, 1992, "Argument from Illusion", in J. Dancy and E. Sosa (eds.), A Companion to Epistemology, Blackwell, Cambridge, Mass., pp. 23–27.         [ Links ]

Hill, Christopher S., 1997, "Imaginability, Conceivability, Possibility and the Mind–Body Problem", Philosophical Studies, vol. 87, pp. 61–85.         [ Links ]

Kripke, Saul, 1980, Naming and Necessity, Harvard University Press, Harvard.         [ Links ]

––––––––––, 1979, "A Puzzle About Belief", in A. Margalit (ed.), Meaning and Use, D. Reidel, Dordrecht, pp. 106–146.         [ Links ]

––––––––––, 1977, "Identity and Necessity", in S.P. Schwartz (ed.), Naming, Necessity and Natural Kinds, Cornell University Press, London, pp. 66–101.         [ Links ]

Levine, Joseph, 1993, "On Leaving Out What It Is Like", in M. Davies and G. Humphreys (eds.), Consciousness: Psychological and Philosophical Essays, Blackwell, Oxford.         [ Links ]

Pérez Otero, Manuel, 2002, "Aplicaciones filosóficas del bi–dimensionalismo: modalidad y contenido epistémico", Theoria, vol. 17, no. 45, pp. 457–477.         [ Links ]

––––––––––, 1999, Conceptos modales e identidad, Edicions Universitat de Barcelona, Barcelona.         [ Links ]

Shoemaker, Sidney, 1984, "Causality and Properties", Identity, Cause and Mind, Cambridge University Press, Cambridge, pp. 206–233.         [ Links ]

Soames, Scott, 2006, "Saul Kripke, the Necessary Aposteriori and the Two–Dimensionalist Heresy", in M. García–Carpintero and J. Macià (eds.), Two–Dimensional Semantics, Oxford University Press, Oxford, pp. 272–292.         [ Links ]

––––––––––, 2005, Reference and Description. The Case Against Two–Dimensionalism, Princeton University Press, Princeton.         [ Links ]

––––––––––, 2002, Beyond Rigidity: The Unfinished Semantic Agenda of Naming and Necessity, Oxford University Press, Oxford.         [ Links ]

 

NOTES

¹ It is not clear whether Kripke wants to include under this stage of his account all necessary apparently contingent identity statements.

² Strictly, the appearance of contingency we talk about is indeed mere appearance of non–necessity, which, combined with the supposition —hypothetically accepted, for the sake of argument— that (5) is true, would produce the alleged apparent contingency of (5). With respect to cases (1)—(4), it is not controversial (in the Kripkean framework we are assuming) that they are true, and it is thereby justified to say that (1)—(4) are apparently contingent. As for (5), its truth value is under dispute; therefore, when we allude to its apparent contingency (as Kripke does himself; cf. Kripke 1980, p. 150), we suppose hypothetically that it is true. What would not be controversial regarding (5) is the fact that it is apparently non–necessary, like any of statements (1)—(4).

³ In the debate about dualism, some authors (Chalmers 1996, Chalmers and Jackson 2001, Levine 1993) rely strongly on the alleged fact that the original rigid designator and the non–rigid term that fixes its reference are conceptually related. For a criticism of that appeal to the idea of a priori conceptual analysis, see Block and Stalnaker 1999. This is another issue I will not address here, at least not directly.

⁴ Kripke says, in a footnote, that the confusion based in the metalinguistic reading (the confusion described in ME) and the confusion described in SE, "especially the second, are both related to the confusion of the metaphysical question of the necessity of 'Hesperus is Phosphorus' with the epistemological question of its aprioricity [the confusion described in GE]" (Kripke 1977, pp. 89—90). In the next footnote, he clarifies his third answer a little more, putting it in terms equivalent to what I have called the first stage of SE (cf. Kripke 1977, fn 15). On the issue of how GE is related to SE, I have already claimed that we can see SE as a more detailed specification of GE; I return to it now and in sections 4–6.

⁵ I am thankful to two anonymous referees of Crítica, whose comments helped me to clarify the points discussed in the last two paragraphs.

⁶ An objector could say that I am missing some other options. For instance: speakers confuse the modal status of R and D because they take them to be semantically equivalent or as stating the same truth–conditions, or because they think that the possible situation that makes false the statement D makes false the statement R too. (I am echoing replies made to me by different referees.) My view is that insofar as these options could work as proper enlightening answers to the question they seem to amount to answer (ii), or something very close to it. Let's take the first one. (Something analogous to what I will say could be applied to the other suggestions.) Speakers confuse the modal status of R and D because they think R means the same as D. Now, should we assume that speakers think that semantic equivalence entails modal equivalence? If not, then taking R and D to be semantically equivalent is irrelevant. If they accept the entailment, we can ask: why? A possible answer is that they take semantic equivalence to entail epistemic equivalence and this, in turn, to entail modal equivalence. (Soames 2002, p. 21, attributes to Kripke the assumption that semantic equivalence entails epistemic equivalence when considering what he call Kripke's epistemic arguments against descriptive theories of names.) This answer drives us to option (ii). Another answer: speakers link meaning more directly to modal status, without the intermediation of epistemic notions. But the objector faces a kind of dilemma here: if we construe the notion of meaning very close to the notion of modal status (in terms of, for instance, truth–conditions), then the proposal (speakers confuse the modal status of R and D because they think that R means the same as D) is hardly enlightening as explanation; if, on the other hand, the notion of meaning has sufficient independent ground, then we still have a legitimate worry: why do speakers think that semantic equivalence entails modal equivalence?

⁷ Cf., for instance, Kripke 1980, pp. 141–143, especially footnote 72: "If I say, 'Gold might turn out not to be an element', I speak correctly; 'might' here is epistemic and expresses the fact that the evidence does not justify a priori (Cartesian) certainty that gold is an element."

⁸ Cf. Kripke 1980, pp. 57–58 and 78–79. The 'Julius' example appeared in Evans 1979.

⁹ This criticism suggests that there is a tension between the Millian conception of meaning apparently favoured by the main semantic theses in Naming and Necessity and the passages in this work where Kripke advances his SE account of the illusion of contingency. (It would seem that Kripke draws back from Millianism, since —for instance— he is worried about the substitutivity of proper names in propositional attitude contexts. On this issue, it is interesting to look to the summary of the relevant views in Naming and Necessity that we find in his later "A Puzzle about Belief" (Kripke 1979, section I). In this writing Millianism is more clearly defended —still indirectly—, and one of Kripke's main theses is that alleged failures of substitutivity cannot be used against Millianism.) Those passages constitute one important source of (some versions of) two–dimensionalism, and are criticized by a radical Millian, Soames, in his attack on what he calls ambitious two–dimensionalism, where he sees an attempt to revive descriptivism (cf. Soames 2005, pp. 71–83; see also Soames 2006). In fact, the issues I discuss in this article can be approached using the terminology of two–dimensional modal semantics; since nothing substantial depends on it, I have decided otherwise.

¹⁰ The branching conception of possible worlds sustained by Kripke (1980, pp. 113 and 115) is also embraced by Shoemaker (1984, p. 218), and —combined with the necessity of origin— has been worked out and explicitly justified in Forbes 1985. In Pérez Otero 1999, I have also defended that view and proposed the correction of what I think are some mistakes in Forbes' theses.

¹¹ The situation is in fact a little more complicated. The thesis that all possible worlds branch from the actual world is a strong reading of the branching view. There is a weaker reading: if v and w are possible worlds which have some existent object in common, then v branches from w (cf. Forbes 1985, p. 151). If we grant only this weak branching thesis, our illustration will require not (6), but another statement, something along the lines of (7): For any x, if x is an actual individual, then if x exists there exist at least n simple individuals. (The ideas mentioned in this note, and in the corresponding text, are developed in Pérez Otero 1999, chapter IV, where I also present some methodological considerations favouring the strong branching thesis.)

¹² Cf. Kripke 1980, p. 136. See our non–problematic qualification in note 12.

¹³ Indeed, having rigid designators is enough —according to SE— for this condition (b) to hold, if we bear in mind the possibility —mentioned in section 4— of extending SE to some other non–identity statements that contain rigid designators.

¹⁴ Strictly speaking, if the two rigid terms in the identity statement R are substituted for, the non–rigid character of the substituting terms, 'D₁' and 'D₂', does not entail that D ('D₁ = D₂') is contingent. The properties expressed by 'D₁' and 'D₂' must also be contingently related to each other. This seems to be the case regarding 'the heavenly body in such–and–such position in the sky in the evening' and 'the heavenly body in such–and–such position in the sky in the morning'. It is an implicit premise we can grant.

¹⁵ Further possible accounts, compatible with (5)'s necessity, are proposed in Hill 1997. Hill also claims that "Kripke nowhere provides any defence of the assumption that [. . . ] the explanatory paradigm illustrated by the heat example is the only paradigm for explaining appearances of possibility away" (Hill 1997, p. 65).

¹⁶ The argument from illusion, usually intended to establish representative realism, has a strong reading, favouring an analysis of sensory experience sometimes called the act/object theory. According to that theory, in having an experience we are directly aware of internal subjective entities that actually bear the corresponding perceptual apparent properties (e.g. red, square, tough) that ordinarily we ascribe to external objective entities. The EPP Principle would support such a reading of the argument from illusion. (Cf. Fumerton 1992, p. 26.)

¹⁷ Partial versions of this paper were presented in guest lectures at the Societat Catalana de Filosofia (Barcelona, May 2002) and at the Congress I Jornadas Hispano–Portuguesas de Filosofía Analítica (Santiago de Compostela, November 2002). Thanks are due to the participants for their comments and suggestions, especially Agustín Arrieta, José A.D. Calzada, Manuel García–Carpintero, Dan López de Sa, Genoveva Martí, Joan Pagès, David Pineda, José Luis Prades, Daniel Quesada and Agustín Vicente. I am also grateful to three anonymous referees of Crítica. Pérez Otero 2002 contained a previous, shortened exposition of the main ideas, located in a broader context. This work, as part of the Research Project "The Constitution of Representational Content. Semantic and Epistemic Aspects" (HUM2005—07539— C02—01), was supported by the Dirección General de Investigación of the Spanish Ministerio de Educación y Ciencia.