1 Introduction

In this paper, I am going to argue that there is a genuine problem for Lewis’s theory, and it is not obvious how this problem could be solved.

The Lewisean counterpart theorist– despite not defending a genuinely essentialist view of what is possible, de re, of individuals – has a way to make, if at all possible, essentialist claims come out as true, in those contexts in which they are endorsed by a committed essentialist. For instance, one can imagine two philosophers disputing about essentiality of origins. Lewis tells us that they might be both right: both the Kripkeans and the non-Kripkeans might speak truly in the contexts of their essentialist assertions. The same holds for other debates in essentialism – for instance, sortal essentialism. According to sortal essentialism, if Gregor Samsa is a human being, he could not have been a cockroach, let alone become one. What is said to happen to Gregor according to Kafka would be thus deemed to be impossible. Well, Lewis would say that, if the committed sortal essentialist thinks that Gregor could not have been a cockroach, then she can speak truly in the context of her assertion. On the other hand, if Kafka thought it really possible for Gregor to be a cockroach, then, for Lewis, in the context of his assertion, Kafka can be truly saying that Gregor could have been a cockroach. In other words, when it comes to the matter of which properties are determined to be essential to individuals, almost anything goes for Lewis.

However, I will show that this does not hold for Kit Fine’s essentialist claims (Fine, 1994a). Now, the neo-Aristotelianism that Fine advocates was not predominant when Lewis was writing about essentialism, so that what Fine means by ‘essence’ is something different from what Lewis means by it. Nonetheless, if for Lewis almost anything goes when it comes to essentialism, it still might come as a surprise that Lewis’s theory cannot make Fine a truth-teller when he wants to say apparently reasonable things, like that Socrates is not essentially such that there are sets. I will argue, then, that this is a genuine problem for Lewis’s theory.

After having proposed a diagnosis of such problem, I will propose a strategy for solving it. My aim is to show that the Lewisean has a way to account for Fine’s essentialist beliefs, in at least some context, consistently with her own counterpart-theorist reading of them.

2 Lewis’s Account of Essentialist Claims

According to the modal account of essentialism (hereafter MAE), the properties that are determined to be essential to an individual a are the properties that a has in all the possible worlds in which she exists. That is to say, the properties that a has as a matter of essentiality coincide with the properties that a has as a matter of necessity de re.

Lewis equates questions of essentialism to questions of necessity de re too. However, Lewis explains de re modality through counterparts (Lewis, 1968, 1986). Let us consider sentence type 1:Footnote 1

  1. 1)

    Socrates is essentially human.

According to Lewis, 1 is true if and only if (hereafter iff) every counterpart of Socrates is human, where a counterpart of Socrates is a thing that resembles Socrates, according to a relevant similarity relation. It is a matter of context which similarity relation is deemed to be relevant and, thus, it is a matter of context which counterparts Socrates has.

The general form of the truth-conditions for an essentialist sentence type is thus incomplete for Lewis: it needs to be completed with the input of the counterparts. Therefore, Lewis believes in the semantic inconstancy of essentialist claims: different tokens of the same essentialist sentence type about Socrates might be produced and evaluated in different contexts and evoke different relevant similarity relations and, thus, different counterparts of Socrates; accordingly, they might have different truth-values.

Now, Lewis believes in a principle of accommodation holding that “what you say makes itself true, if at all possible, by creating a context that selects the relevant features so as to make it true” (Lewis, 1986: 251). This principle is also at work when it comes to essentialism (see, for instance, Lewis, 1983: 43). Accordingly, almost any essentialist claim a person makes creates a context within which the claim is true. This happens because, in the context of an utterance of an essentialist statement, we can project backwards, as it were, the kind of similarity relation that must be selected in order to make that sentence true. Therefore, the system Lewis adopts, if he wants to make an essentialist speak truly in the context of her own speaking, would be to say that, in the context of her utterance, we need to pick out the kind of similarity relation that makes that sentence true.

Nonetheless, such method does not work with Fine, as I will show in the next Sections. And the source of the problem is that Fine wishes to take essentialist predications to be more fine-grained than de re modal predication.

3 Kit Fine’s Essentialist Beliefs

Fine (1994a) has developed a strong objection to MAE. Fine claims that the concept of necessity de re is too wide to capture the narrower concept of essence. Fine agrees with MAE that if a property P is essential to a, then P is also necessary to a. What Fine claims, against MAE, is that being necessary to a is not sufficient for a property P to be also essential to a. In addition, for P to be a’s essence, P has to define a’s nature, namely, it has to tell us what a is. And there are many properties, Fine claims, that a has as a matter of necessity de re which are, nonetheless, irrelevant to her nature. For instance, properties like being a member of singleton Socrates, being distinct from the Eiffel Tower, and being such that there are sets are determined to be necessary to Socrates. However, contrary to what MAE claims, although had as a matter of necessity de re by Socrates, these properties should not count as essential to him, for they are irrelevant to Socrates’s nature.Footnote 2 Therefore, since MAE equates questions of essentialism to questions of necessity de re, then, Fine concludes, MAE should be rejected and essences should be understood in terms of real definitions.

4 Lewis’s Standard System does not Work with Fine’s Counterexamples

As anticipated, the standard method that Lewis uses, if he wants to make essentialists truth-tellers, does not work with Fine. On the one hand, this is not surprising, since the Finean counterexamples involve hyperintensional matters that a modal approach to essentialism cannot account for.Footnote 3 On the other hand, since for Lewis almost anything goes as far as essentialism is concerned, it still might come as a surprise that Lewis cannot make Fine a truth-teller.

I will consider three examples to illustrate why Lewis cannot account for Fine’s essentialist beliefs, in any contexts whatsoever.

Let us start with one of the best-known Finean counterexamples to MAE. According to Fine (1994a: 4, 5), sentence 2 should come out false:

  1. 2)

    It is essential to Socrates to belong to {Socrates}.

First of all, the standard counterpart theoretic truth-conditions previously provided must be regimented in order to guarantee that no source of counterpart theoretic content is suppressed. In sentence 2, there are two sources of such content. So, the truth-conditions for 2 to be considered are the following:

  • 2 is true iff every counterpart of Socrates belongs to some counterpart of {Socrates}.

We know that it is a flexible matter which things count as counterparts, and it is not easy to understand which the counterparts of {Socrates} might be in a context. At any rate, according to Lewis (1991: 37) (see also Divers, 2002: 290), the most natural way to select the counterparts of {Socrates} would be to pick out those singleton sets that have a member that is a counterpart of Socrates. It seems indeed plausible that what makes one singleton a counterpart of another singleton is their having members which are similar to one another. Therefore, the truth-conditions for 2 become:

  • 2 is true iff every counterpart of Socrates belongs to some singleton set that has as a member a counterpart of Socrates.

Now, Lewis’s way to make sentence 2 false, as Fine wants, would be to say that, in the context of its utterance, we should pick out the kind of similarity relation that makes that sentence false. Sentence 2 would be made true in the context of its utterance whenever the same similarity relation was invoked for both the occurrences of ‘Socrates’. That is to say, if 2 is true iff every counterpart of Socrates belongs to its own singleton set, then 2 would come out true. Indeed, for any selection of counterparthood whatsoever, if, say, a is a counterpart of Socrates, then a does belong to the set whose sole member is a. So, the only way to make sentence 2 false would be if different similarity relations were invoked by the two occurrences of ‘Socrates’. However, I do not think that this would be the natural way to interpret 2 in counterpart terms. Indeed, let us compare sentence 2 with a claim of necessity of identity such as ‘a is necessarily identical to b’. In this case, Lewis tells us that “an identity pair has the de re possibility of being a nonidentity pair” (Lewis, 1986: 263). The reason being that, given an object a/b, it might evoke different relevant similarity relations and, so, different counterparts in some world so that, according to that world, a would not be identical to b. Thus, the identity between a and b might not be necessary. Therefore, when the very same object is presented in two different ways within the same context of utterance, it can evoke two different similarity relations. However, it seems that when an object is presented twice with the same name, as in ‘a is necessarily identical to a’, we should interpret the sentence as to make the two occurrences of ‘a’ in it evoke the same similarity relation and, so, the same counterpart relation. In this way, Lewis is also able to make sense of the instances of the property of necessary self-identity. Indeed, while Lewis is willing to deny the necessity of identity, he does not intend to deny the necessity of self-identity (Lewis, 1986: 263).

Therefore, I take it that, according to the most natural way to interpret 2 – namely, the one according to which 2 is true iff every counterpart of Socrates belongs to its own singleton set – the Lewisean cannot account for Fine’s belief about the falsity of 2. The falsity of 2 can only be obtained by making a seemingly arbitrary selection of similarity relations: one that is not justified by any aspect of the case other than that it gets the right result.

I am aware, however, that this example is quite complicated.Footnote 4 So, let us move to some simpler and clearer examples. Let us thus consider sentence 3, that Fine believes to be false (Fine, 1994a: 5):

  1. 3)

    Socrates is essentially such that there are sets.

Well, can Lewis account for Fine’s belief in this case? The standard system would consist in selecting a kind of similarity relation so as to make the sentence false. However, there is no context that can select such kind of similarity relation. Indeed, since 3 is true iff all the counterparts of Socrates are such that there are sets, then in order to make it false, Lewis should find a context that selects a similarity relation, according to which at least one counterpart of Socrates is not such that there are sets. Now, since it is a necessary truth that there are sets, any counterpart of Socrates, no matter according to which similarity relation has been picked out, will inhabit a world in which there are sets. Thus, no matter which similarity relation is deemed to be relevant, it will be true, in all the contexts, that all the counterparts of Socrates are such that there are sets.

Therefore, Lewis also fails to account for Fine’ belief about the falsity of 3. And this holds in all the cases in which we have to do with necessary truths:Footnote 5 for instance, Lewis cannot make Fine a truth-teller if he wants to say that Socrates is not essentially such that there are infinitely many prime numbers, and so on.

Finally, let us consider sentence 4, that Fine believes to be false (Fine, 1994a: 5):

  1. 4)

    Socrates is essentially distinct from the Eiffel Tower.

Well, the only selection of counterparthood that would make 4 false is the selection according to which at least one counterpart of Socrates is identical to the (counterpart of the) Eiffel Tower. This means that the only way to make a token of 4 false is to regard, in that very same context, Socrates as not essentially human. However, I take this way of accommodating the falsity of sentence 4 to be unsatisfying. Indeed, Fine does not want to deny 4 because he believes that, in some world, Socrates is identical to the Tower, since he believes Socrates to be necessarily distinct from the Tower. In fact, in Fine’s view, not only Socrates is necessarily human, but he is also essentially so.Footnote 6 Therefore, reading 4 the way Lewis needs to in order to make it false does not do justice to Fine’s intuitions. And I believe that a satisfactory answer to Fine aims to account for his intuitions, rather than misconstruing them. Therefore, I take it that Lewis, in order to account for Fine’s beliefs, needs to make 4 false (that is, he needs to make true that Socrates is not essentially distinct from the Eiffel Tower) in the intended context in which Socrates is essentially human. So, the counterexample Lewis needs to account for, and which makes Fine’s intuitions explicit, is given by sentence 5, that Fine wants to be true:

  1. 5)

    Socrates is essentially human and not essentially distinct from the Eiffel Tower.Footnote 7

The challenge for the Lewisean is to find one similarity relation that makes both conjuncts of 5 true. But, as was said, there is no such similarity relation that can help the Lewisean to achieve this result.Footnote 8

Therefore, the Lewisean seems not to be able to account for Fine’s essentialist beliefs: there are essentialist sentences Fine believes to be false (like 2, 3) and others that he believes to be true (sentence 5) that Lewis’s theory, as it stands, fails to account for, in any context whatsoever.

5 Is there Really a Genuine Problem for Lewis’s Theory?

In this Section, I would like to consider two lines of thought to the effect that the inability to make Fine a truth-teller is not a problem that the Lewisean should be really worried about.

Firstly, one might rightly observe that what Fine means by ‘essence’ is something different from what Lewis means by it (that is, de re necessary properties). Therefore, it is not obvious that it is a problem for his theory if his system to make the essentialist a truth-teller does not work with Fine. Surely, it would not be a problem if his system would not work with the essentialist beliefs of someone, s, who decides to use ‘essence’ as a synonym of a non-modal notion, for indicating all the properties that make their bearers happy. The point is that, if Lewis takes ‘essence’ to be a de re modal notion, anyone who attempts to argue against this conception of ‘essence’ should just be ignored: she is simply talking about something different. So, it is not a problem for Lewis’s theory if her statements are not able to create any context in which they are true.

What I would like to claim in response to this is that, while it is true that Lewis does not need to account for any use whatsoever of ‘essence’, it is clear that Fine’s proposal is not as deviant as the one proposed by the unwise s. First of all, Fine has strong and compelling arguments for his understanding of ‘essence’. Moreover, Fine’s use of ‘essence’, contrary to s’s use of it, fits very well with the use of this notion in the history of philosophy.Footnote 9 Finally, recall that Lewis is able to make someone who wants to say that Gregor Samsa could have been a cockroach a truth-teller; hence, if he is able to account for such an essentialist judgment, he should also be able to account for apparently more reasonable essentialist judgments, like that Socrates is not essentially such that there are sets. Therefore, I think that, if Lewis is not able to account for such a suitable use of the notion of ‘essence’, then this is a problem for his theory.

The second line of thought takes into account what Lewis says with regard to his principle of accommodation: “what you say makes itself true, if at all possible, by creating a context that selects the relevant features so as to make it true” (Lewis, 1986: 251, my emphasis). One might, thus, appeal to the provision Lewis makes. As far as I know, Lewis imagines two cases, related to essentialism, in which it is not possible to create a context that selects the relevant counterparts so as to make a sentence true: (i) “if you gave an especially silly answer, such as that Humphrey could have been a poached egg, yet he could not have been a human born to different parents” (Lewis, 1986: 251); and (ii) if an essentialist sentences attribute to an object a contradiction as a de re possibility (Lewis, 1986: 253). One might then object that Fine’s essentialist claims cannot be accounted for, simply because they are silly or contradictory. For instance, thinking about essence as a broadly modal notion might make 5 a ‘silly’ claim, as well as it might make the negations of 2 and 3 subtly contradictory: they would amount to the claims that at least one of Socrates’s counterpart exists in a world that lacks some necessary existents.

However, reading such claims from a purely modal point of view would mean to misunderstand Fine’s beliefs. So, I think that it would be more satisfactory for the Lewisean if she could accommodate them. And I will show that she can.

6 My Proposal for the Lewisean

We know that the core problem for both Lewis and MAE is that Fine’s counterexamples require certain hyperintensional resources that purely modal accounts do not have.Footnote 10 Nonetheless, I find it disappointing that the semantic inconstancy of essentialist claims, that only Lewis endorses and that allows him to make almost any essentialist claim true in the context of its utterance, does not allow him to accommodate Fine’s counterexamples. And, indeed, it is precisely at the source of such a semantic inconstancy that the Lewisean needs to look in order to make Fine a truth-teller.

We saw that essentialist sentences are semantically inconstant because, according to different contexts, different similarity relations are deemed to be relevant. In the standard cases in which essences are understood in purely modal terms, in order to obtain the truth-conditions of an essentialist sentence, all that the Lewisean need rely on is the context selecting the relevant similarity relation. Once we have such relation, we will have the counterparts of, say, Socrates, and all the facts about those counterparts and the worlds that they inhabit will be relevant. But, as we know, if we stop here, we cannot account for Fine’s essentialist beliefs.

I claim that the Lewisean should make the context play more than one role. And the second role the context is asked to play is already included in the usual role that it plays. Indeed, besides selecting the relevant similarity relation, the context can also discriminate among the properties that happen to be shared by all the counterparts picked out by that relation, by selecting which of these properties helped to determine the relevant similarity relation.

Let us take Quine’s famous example. There is Andrew, who is a mathematician and a cyclist. Let us suppose that a context selects as relevant the cyclist-wise similarity relation. Accordingly, that similarity relation will select the counterparts of Andrew, namely all and only cyclists-counterparts. However, why does that context select one similarity relation (the cyclist-wise similarity relation) over the other (the mathematician-wise similarity relation)? Well, because it makes salient Andrew’s cyclist aspects over his mathematician aspects, namely, because it stresses some of Andrew’s properties (the cyclist-related properties, like his being two-legged) over others (the mathematician-related properties, such as his being rational).Footnote 11 This means that the properties that are made salient by the context (the cyclist-related properties) go toward determining the relevant similarity relation (the cyclist-wise similarity relation) that will pick out the counterparts (the cyclist-counterparts). After all, it is not by chance that the cyclist-wise similarity relation selects cyclist-counterparts rather than, say, planet counterparts. Suppose now that all these cyclist-counterparts are both two-legged and rational. Well, it is the property of being two-legged that helped to determine the relevance of the cyclist-wise similarity relation, which picked out the cyclist-counterparts. Therefore, it is the property of being two-legged that is relevant in this context.

Let us call ‘relevant’ the properties that are made salient by the context and that help to determine which similarity relation is relevant in that context. Then, my proposal for the formulation of truth-conditions for essentialist sentences goes as follows. A sentence like ‘a is essentially P’ is true iff:

  1. a)

    all the counterparts of a are P, and.

  2. b)

    P is relevant.

Then, it can be said that, if P meets only the first condition, then P is determined to be necessary to a. If P instead meets both conditions, then P is determined to be essential to a.

We know that the general form of the truth-conditions for an essentialist sentence type is incomplete: it needs to be completed with the input of the counterparts. Now, what is said is that the Lewisean might add that it needs to be completed also by the input of the relevant properties.

Note that, by adding a condition of relevance to the modal criterion - namely, by adding condition (b) to Lewis’s original account - I am not using more context-sensitive notions than before. Indeed, according to my proposal, the context needs to do two jobs, instead of one: to pick out both the relevant similarity relation and the relevant properties. However, as I said, the second job that the context is asked to do is already included in the job that it usually does: it is by stressing some properties that the context selects the relevant similarity relation. Therefore, according to different contexts, the relevant similarity relation can change, but, if it changes, also the relevant properties change. Better: as the matter as to which properties are determined to be relevant changes depending on the context, the question as to which similarity relations are regarded as relevant changes accordingly. Therefore, at the end of the day, we only need the context to select the relevant similarity relation, as in Lewis’s original account, because in doing so, the context has already given us the relevant properties.

Needless to say, the Finean would not accept such a solution. From her point of view, it would not be enough for Socrates to be, say, not essentially such that there are sets, in some context: she might want Socrates to be not essentially such that there are sets in a context-independent way. However, I aim to find a solution that the Lewisean might employ to make Fine a truth-teller, which is acceptable for the Lewisean,Footnote 12 and the Lewisean does accept the semantic inconstancy of essentialist claims.Footnote 13

6.1 My Strategy at work

Let us see now in detail how my strategy works, by starting with sentence 3:

  1. 3)

    Socrates is essentially such that there are sets.

My aim is to account for Fine’s belief that 3 is false. Let us suppose that the context selects a similarity relation that works in a very generous way, demanding nothing more of counterparts than they be things of the same species of Socrates.Footnote 14 I will call such a similarity relation ‘the species-wise similarity relation’. This similarity relation guides us to pick out, as Socrates’s counterparts in this context, all and only human beings. However, we know that all the human counterparts will exemplify the property of being such that there are sets (let us call such a property ‘S’). But, now, the context is asked to tell us whether or not S is relevant, namely whether or not S is one of the properties that helped to determine the species-wise similarity relation. Well, clearly, the context selected the species-wise similarity relation, by making salient Socrates’s species-related properties, such as his property of being human, over other Socrates’s properties (such as S). And precisely these species-related-properties are the relevant ones. Accordingly, the fact that the counterparts of Socrates happen to share also S is irrelevant. Therefore, S is not determined to be essential to Socrates: 3 is false in this context, as Fine wants.

One could make the same argument in other terms, as follows. There are many objects in all the possible worlds. For instance, there are a and b that share with Socrates the property S introduced before, the properties of being human (H), of belonging to their own singletons (O), of being distinct from the Eiffel Tower (D), of being a human or a mountain (H v M) and so on. There is the Eiffel Tower, t, which shares with Socrates S, O, and so on. In order to obtain the truth-value of sentence 3, we need to know which respect of similarity is relevant and, thus, which of these objects are counterparts of Socrates.

If we want to deny 3, as Fine demands, we need to look for a context that selects a kind of similarity relation suitable to make 3 false, and, as I said, the context that picks out the species-wise similarity relation introduced before could do the job. Indeed, in a context that selects the species-wise similarity relation as relevant, then only a and b are chosen as counterparts, and only the property H is deemed to be relevant. There are also other properties that Socrates, a and b share (for instance, S), but these properties did not help to determine the species-wise similarity relation. Now that we have both the input of the counterparts (Socrates, a and b) and the input of the properties that are relevant (H), we can attain the ‘correct’ truth-value of 3, in this context.

6.2 Other Sentences

Let us see now, briefly, how this strategy would work with the other essentialist sentences that Fine discusses.

Recall sentence 2:

  1. 2)

    It is essential to Socrates to belong to {Socrates}.

In the context described above, in which the species-wise similarity relation is invoked, even though all the human counterparts of Socrates exemplify O, O is not relevant. Therefore, sentence 2 comes out false, as Fine wants.

Let us consider our sentence 5, that Fine wants to be true:

  1. 5)

    Socrates is essentially human and not essentially distinct from the Eiffel Tower.

We already know that, in the context described above in which the species-wise similarity relation is relevant, Socrates comes out as essentially human. Let us see now if, in that very same context, Socrates can also be determined as not essentially distinct from the Eiffel Tower (D). Well, in that context, all the human counterparts of Socrates (Socrates, a and b) are D. However, D is not relevant, because it is not among the properties that helped to determine, as relevant, the species-wise similarity relation.

However, the reader might think that, among Socrates’s species-related properties, there is also D. After all, she might say, it is because Socrates belongs to a certain species that he exemplifies D. Well, if the reader thinks so, then she can just look for a more accurate context that evokes a more precise similarity relation. For instance, a context might select the human-wise similarity relation. In this context, no appeal to species is made. Therefore, even though, as a consequence, all the counterparts of Socrates are D, this is not significant: the only property that this context made salient and that helped to determine the human-wise similarity relation is H.

Therefore, according to my strategy, Lewis is able to account for Fine’s beliefs.

7 Fine’s Objections Against the Relevance Condition

Fine thinks that appealing to a condition of relevance, as my proposal does, would not do the job. He says:

One might try to add a condition of relevance to the modal criterion. One would demand, if a property is to be essential to an object, that it somehow be relevant to the object. However the case of Socrates and his singleton makes it hard to see how the required notion of relevance could be understood without already presupposing the concept of essence in question. For we want to say that it is essential to the singleton to have Socrates as a member, but that it is not essential to Socrates to be a member of the singleton. But there is nothing in the ‘logic’ of the situation to justify an asymmetric judgement of relevance; the difference lies entirely in the nature of the objects in question. (Fine, 1994a: 6, 7).

I will divide my answer to Fine’s concern in three parts.

Firstly, I show that my proposal can guarantee the truth of the following conjunctive Finean assertion:

Asymmetry

It is essential to {Socrates} that it have Socrates as a member and not essential to Socrates that he is a member of {Socrates}.

When the Finean asserts Asymmetry, she is doing so in a single context – that is, the sentence is said ‘in one breath’. Accordingly, the Lewisean needs to make both conjuncts of Asymmetry true in the same context. Let us start from the second conjunct that we have already discussed. Recall indeed what was said about sentence 2:

  1. 2)

    It is essential to Socrates to belong to {Socrates}.

We know, from the previous discussion, that, in 2, there are two sources of counterpart theoretic content (‘Socrates’ and ‘{Socrates}’), and that the most natural way to select the counterparts of {Socrates} would be to pick out those singleton sets that have a member that is a counterpart of Socrates (let us call such a similarity relation ‘the membership-wise similarity relation’). We also saw that, in this context, both the occurrences of ‘Socrates’ (also the one within ‘{Socrates}’) need to evoke the same similarity relation. Then, as I previously argued, in a context that selects the species-wise similarity relation for Socrates, we can make 2 false and, accordingly, the second conjunct of Asymmetry true. Indeed, while it is true that all the human counterparts of Socrates belong to their own singletons, namely exemplify O (condition (a) of my proposal is met), O is not relevant, since O did not help to determine the species-wise similarity relation (so, condition (b) is not satisfied).

With this in mind, let us consider now the first conjunct of Asymmetry, and let us see if, in this very same context, we can make it true.

  1. 6)

    It is essential to {Socrates} that it have Socrates as a member.

We cannot change the context from one conjunct to another, and this means that we cannot change the parameters of similarity. Accordingly, all the occurrences of ‘Socrates’ will evoke the species-wise similarity relation (also the occurrence within ‘{Socrates}’), and ‘{Socrates}’ will evoke the membership-wise similarity relation. Well, this very same context also determines the ‘correct’ truth-value of 6: (i) all the counterparts of {Socrates} will have, as a member, (a counterpart of) Socrates (selected by the species-wise similarity relation); and (ii) having (a counterpart of) Socrates (selected by the species-wise similarity relation) as a member is relevant, since it is this property that helped to determine, as relevant, the membership-wise similarity relation. Hence, in this context, it will be essential to {Socrates} that it have Socrates as a member. Therefore, Asymmetry can be made true, in the Lewisean perspective.

The second part of my answer argues that there is something in the ‘logic’ of the situation to justify an asymmetric judgement of relevance. Asymmetry is the conjunction of two sentences, with two different subjects: the first sentence is about {Socrates}, as it is represented in a context C, while the second is about how Socrates is represented in the very same context C. In other words, Asymmetry is about two different subjects in the same context C. My condition of relevance (b) applies to a property, and the relevance of this property depends on what is evoked by the subject of a sentence. Therefore, if the subject changes in the passage from one sentence to the other, also the condition of relevance to apply changes. But, as long as the similarity relations invoked do not change from one sentence to the other, we are not changing the context: indeed, as I said, the job of the context is to select the relevant similarity relation (once we have it, we have, a fortiori, the relevant properties). So, without changing the context from one sentence to another, while the first sentence focuses on the membership-wise similarity relation evoked by ‘{Socrates}’ in C and on which properties helped to select this relation as relevant, the second sentence is about the species-wise similarity relation evoked by ‘Socrates’ in C and about which properties helped to pick out such relation. But in C, say, ‘{Socrates}’ always evokes the membership-wise similarity relation and, hence, the property of having (a counterpart of) Socrates as a member is always relevant: the sole change from the first to the second sentence is that only the former sentence focuses on this similarity relation and on which properties were relevant to determine it. Consequently, the difference is in the ‘logic’ of the situation, and it does not lie in the nature of the objects in question: it lies indeed in the similarity relations contextually evoked, and the similarity relations were already invoked in this analysis in order to account for essences.

Finally, I believe that, in giving the analysis of the concept of essence, there would be nothing wrong in being guided by prior opinions about what essences are expected to be. I think, indeed, that we are unavoidably guided, in the construction of our analyses of the concept of essence, by prior opinions about what an essence should be (see, for instance, Divers, 2002: 110). And this kind of guidance is not enough to justify the charge of circularity, as it emerges from Fine’s concern. What is more, it is not clear to me how Fine himself might discern what is in the real definition of an object and what is not, without already presupposing the concept of essence.

8 Other Counterexamples to MAE in Fine’s Style

The Lewisean, with my account, can take into account all the Finean essentialist beliefs. This also holds with those problematic sentences that not even Fine, with his account, can always account for (see Fine, 1994b). I will show only some examples.

As partially anticipated, the fact that a property is a logical consequence of another does not affect my proposal. For instance, if, in a context C, it is made true that Socrates is essentially human, it does not need to follow that it is also true, in C, that Socrates is essentially human or a mountain (H v M). Indeed, as we saw, if the species-wise similarity relation is deemed to be relevant in C, while all Socrates’s human counterparts will exemplify (H v M), this property is not relevant. The same goes for properties such as ‘to be human or not human’ or ‘to be human, if human’ (for any property whatsoever), and so on.

Note that the reason why Socrates can be determined to be essentially H and not essentially (H v M) in C is that H and (H v M) are two different properties: they are different sets of possible individuals (for instance, the Mount Blanc is included only in the second set). Accordingly, it makes sense to draw a demarcation between them and say that only H was relevant to determine the similarity relation. Let us consider now another counterexample in Fine’s style, where the property which is deemed to be relevant in a context is conjoined with a necessary truth. For instance, let us take sentence 7, that Fine would accept:

  1. 7)

    Socrates is essentially human and not such that there are sets.

In analogy with what was said before, I might claim that, in order to make 7 false, the Lewisean might select a context that makes relevant the species-wise similarity relation. In this context, indeed, while all the human counterparts of Socrates have both H and (H ∧ S), (H ∧ S) is not relevant: only H helped to determine, as relevant, the species-wise similarity relation. However, in this case, H is cointensional with (H ∧ S) and, for Lewis, cointensional properties are identical. So, either both properties are relevant, being both salient in determining the species-wise similarity relation, or neither is. Hence, in order to favor one property (H) over the other (H ∧ S), as the relevant one, the Lewisean needs to accept hyperintensional differences between properties. But this is fine. Indeed, recall that I aim to find a way to make Fine a truth-teller, which is acceptable from the Lewisean perspective. And Lewis recognizes that we might want to make use of hyperintensional distinctions, as long as they can be ultimately explained in terms of possible worlds and set-theoretic constructions out of them (Lewis, 1986: 55–59). For instance, he considers a structured version of properties that would allow us to distinguish between two cointensional properties that have different ‘meanings’ associated with their standard names. Hence, if the Lewisean follows Lewis’s hint on structured properties, she is able to distinguish between H and (H ∧ S), which plausibly have different structures. Accordingly, if Socrates is determined to be essentially H in C, the Lewisean does not need to derive that he is also essentially H ∧ S in C.

Also, Socrates can be determined not to be essentially identical to himself in some context, if Fine wants it: all his human counterparts will be identical to themselves (so, of course, Socrates will be necessarily identical to himself in any context), but being identical to himself is not relevant to determine, say, the species-wise similarity relation.

Finally, I will consider one last Finean counterexample to MAE, that the Lewisean was able to account for even without my modification. However, I will show that my proposal makes the solution safer. Fine claims that, according to MAE, it will erroneously be part of the essence of any object that every other object has the essential properties that it has (Fine, 1994a: 5, 6). So, contrary to MAE, Fine would want sentence 8 to be false:

  1. 8)

    Socrates is essentially such that an object a is essentially P.

Well, the Lewisean might account for Fine’s belief with Lewis’s original account: there might be a context C in which not all the counterparts of Socrates are such that a is essentially P: something which is a counterpart of Socrates in C may inhabit a world in which some counterpart of a is not-P (this reply relies on the semantic inconstancy of essentialist claims). However, since whether or not some counterpart of Socrates inhabits a world in which a counterpart of a is not-P depends on how the worlds are made, with my suggested modification, the Lewisean can make sure that 8 is false, in the context in which she wants it to be false, without relying on the failure of condition (a). Indeed, let us suppose that condition (a) is met: all the counterparts of Socrates are such that all the counterparts of a are P (and P is relevant to determine the similarity relation between a and its counterparts). Well, being such that a is essentially P did not help to determine, as relevant, say, the species-wise similarity relation between Socrates and his counterparts. Not even if the counterparts of a are selected according to the same species-wise similarity relation. For instance, let us talk about Plato, and let us suppose that this context selects the species-wise similarity relation for both Socrates and Plato. Therefore, it would be true, in this context, that all the counterparts of Socrates are such that Plato is essentially human. However, it is not relevant to determine the species-wise similarity relation between Socrates and his counterparts, that they are all such that Plato is essentially human. In other words, it is clear that the property of being such that Plato is essentially human did not help to determine the species-wise similarity relation. Therefore, 8 is false in this context.

9 Conclusion

In this paper, I have argued that there is a genuine problem for Lewis’s theory. This problem is posed by Fine’s belief that questions about essentialism are not at one with questions about necessity de re. The standard Lewisean account for essentialist claims cannot account for such a belief.

I offered a diagnosis of the problem, by explaining the reason why the standard way that Lewis adopts when he wants to make essentialist sentences true in the context of their utterances does not work with the Finean cases. I also suggested a way to amend Lewis’s theory in order to solve such a problem. My proposal allows the Lewisean to account for Fine’s essentialist beliefs.

As a result, I have shown that the semantic inconstancy of essentialist claims, which only the Lewisean endorses and which allows her to make almost any essentialist claim true in the context of its utterance, also allows her to accommodate Fine’s counterexamples.