Swinburne’s modal argument for the existence of the soul Agnieszka Rostalska Jagiellonian University, Poland arostalska@gmail.com Rafal Urbaniak Gdansk University, Poland Ghent University, Belgium rfl.urbaniak@gmail.com entiaetnomina.blogspot.com Abstract & Introduction Richard Swinburne (Swinburne and Shoemaker 1984; Swinburne 1986) argues that human beings currently alive have non–bodily immaterial parts called souls. In his main argument in support of this conclusion (modal argument), roughly speaking, from the assumption that it is logically possible that a human being survives the destruction of their body and a few additional premises, he infers the actual existence of souls. After a brief presentation of the argument we describe the main known objection to it, called the substitution objection (SO for short), which is raised by Alston and Smythe (1994), Zimmerman (1991) and Stump and Kretzmann (1996). We then explain Swinburne’s response to it (1996). This constitutes a background for the discussion that follows. First, we formalize Swinburne’s argument in a quantified propositional modal language so that it is logically valid and contains no tacit assumptions, clearing up some notational issues as we go. Having done that, we explain why we find Swinburne’s response unsatisfactory. Next, we indicate that even though SO is quite compelling (albeit for a slightly different reason than the one given previously in the literature), a weakening of one of the premises yields a valid argument for the same conclusion and yet immune to SO. Even this version of the argument, we argue, is epistemically circular. Acknowledgements We would like to express our gratitude to all the people who discussed these issues with us and commented on earlier versions of this paper. We are grateful to participants of the events where the paper has been presented: Workshop & Young Researcher’s Day in Logic, Philosophy and History of Science in Brussels, 2008, Jeffrey Ketland’s ω-seminar in Edinburgh, 2008, and Formal Methods in the Epistemology of Religion in Leuven, 2009. The main ideas of this paper originated after a number of discussions about philosophy of religion and mind with Professor Jack MacIntosh (Calgary). Comments provided by Professor Richard Swinburne (Oxford), who was in the audience when this paper was presented in Leuven in June 2009, were also very helpful, and it was interesting to learn that Professor Swinburne agrees with all our main points, apart from our final assessment of the modified argument. It was Lara Buchak (Berkeley) who observed that our version of the argument developed in response to SO results from a weakening of one of the premises. We also owe gratitude to Paul Draper for his invaluable editorial comments. 1 The modal argument First, let’s take a brief look at the original formulation of the argument.1 For mnemonic purposes we changed the letters used to represent sentences from Swinburne‘s original notation to the following:2 C ⇔ Swinburne is a Conscious person and exists in 1984. 1 D ⇔ Swinburne’s body is completely Destroyed in the last instant of 1984. S ⇔ Swinburne has a Soul in 1984. E ⇔ Swinburne Exists in 1985. We will use ‘♦’ as the standard possibility operator. Swinburne introduces a variable p that is supposed to range over propositions of a specific sort. The way he phrases his explanation of the range of this quantification is this:3 p ranges over all consistent propositions compatible with C ∧D and describing 1984 states of affairs. ‘∀p ’ is to be read in the normal way as ‘for all propositions p’. (Swinburne 1996: 69) The first premise of the argument is contingent. It just states that Swinburne is a conscious person and exists in 1984: (1) C The second premise states that for any sentence about 1984 compatible with C and D, it is possible that Swinburne survives the destruction of his body, and yet that this compatible sentence is true:4 (2) ∀p ♦(C ∧ D ∧ p ∧ E) The third premise says that it is not possible (at least for Swinburne) to survive the destruction of his body if he doesn’t have a soul (an immaterial part): (3) ¬♦(C ∧ D ∧ ¬S ∧ E) Premise (2) says that any sentence compatible with C ∧ D and describing 1984 states of affairs is compatible with C ∧D ∧E but premise (3) says that ¬S is not compatible with C ∧ D ∧ E. Therefore, ¬S is not a sentence that is compatible with C ∧ D and describes 1984 states of affairs. Or, in other words, premises (2) and (3) together entail that ¬S is not within the range of p. But if ¬S is not compatible with C ∧ D, C ∧ D entails S, because if ¬S, C, D can’t be simultaneously true, ¬S cannot be true if C and D are true, and hence, if C and D are true, S has to be true. But D doesn’t have any impact on the truth of S. Therefore, C itself entails S. Let‘s restate the last paragraph in plain English. (3) says that it is not possible that Swinburne is conscious and alive in 1984, his body is destroyed in the last instant of 1984, he survives into 1985 and yet he doesn’t have a soul in 1984. (2), on the other hand, says that any sentence that describes 1984 states of affairs and is compatible with the claim Swinburne is conscious and alive in 1984 and his body is destroyed in the last instant of 1984 is compatible with the claim that Swinburne is conscious and alive in 1984, his body is destroyed in the 2 last instant of 1984 and yet he exists in 1985. These two premises, Swinburne insists, entail that the claim that Swinburne does not have a soul in 1984 is not compatible with the claim that he’s conscious and alive in 1984 and his body is destroyed in the last instant of 1984. So the fact that he‘s conscious and alive in 1984 and yet his body is destroyed in the last instant of 1984 entails that he does have a soul in 1984. But whether his body is destroyed in the last instant of 1984 has no bearing on whether he has a soul in 1984 and therefore the assumption that he is alive and conscious in 1984 entails that he has a soul. 2 A few objections to the modal argument Before discussing the SO objection, which will be in the main focus of our paper, let’s take a rather brief look at other objections which have been raised against Swinburne’s argument. Shoemaker (1984) says that de re possibility claims cannot be justified by bare thought experiments, and therefore the claim that it is possible that Swinburne survives the destruction of his body does not follow from the fact that Swinburne can imagine that he survives the destruction of his body. Shoemaker also complains that even if we had good reasons to accept this de re claim, still no conclusion about the actual world can follow from de re possibility claims. (Swinburne 1996) agrees with the last statement, but explains that a crucial role in his argument is played by premise 1 which is contingent and believes that his argument is based not only on thought experiments but also on actual experiments (although he is rather vague when it comes to explaining what those actual experiments are) and hence his conclusion is not solely based on modal claims. As for the first part of the criticism, Swinburne argues that there is no logical incoherence in the story in which he himself survives the destruction of his body and thus the de re possibility claim is true. Hasker (1998) suggests that Swinburne’s argument is epistemically circular in the sense that no reasonably well-informed person would accept the premise who does not already accept the conclusion, although he doesn’t get into much detail. By the time Hasker’s paper was published, Swinburne (1996) attempted to defend against this sort of criticism by indicating that someone may accept premise 2 without even understanding the conclusion, or without accepting premise 3. It is rather unclear whether this is sufficient for an argument to avoid circularity. One might still insist that an argument is not circular if it is possible that a reasonably informed person having understood the conclusion (but not having seen how it follows from the premises) doesn’t accept it, and yet can still accept the premises. Reames (1999) gives an interesting argument, parallel to the modal argument, for the opposite conclusion: he argues that from the assumption that (i) he exists in 1997, (ii) his body is destroyed in the last instant of 1997, (iii) it is possible that he does not exist in 1998 and (iv) that it is not possible that he has a soul and yet does not survive the destruction of his body, it follows that he actually does not have a soul. He also argues that the plausibility of 3 premise 2 relies on an ambiguity between two understandings of logical possibility. He distinguishes between the standard and the restricted understanding of logical possibility (the latter usually being used for talking about the essence of things). Swinburne’s argument at some point, he argues, employs the restricted notion. But in this sense of logical possibility, contrary to Swinburne, conceivability does not imply logical possibility. As for the argument, Swinburne might reject (iv) because he believes that the fact that soul survives the destruction of the body is a purely contingent fact granted by the grace of God. As for the distinction between two kinds of possibility, the objection is really interesting, but discussing it at length lies beyond the scope of this paper. Nagasawa (2005) reads the argument as if Swinburne used two thought experiments in order to support (2): (a) The disembodiment experiment where it is imagined that a person becomes disembodied and continues to exist, and (b) transfer experiment where a person gradually “moves” his ability to control the body from his own body to another body, thus switching bodies. Then he argues that Swinburne cannot use (a) because it already assumes the existence of the soul and that (b) does not support the claim that soul exists, because it hasn’t been proven that the existence of the soul is the only possible explanation of the personal identity in the transfer case. First of all, Swinburne might respond, (a) does not, strictly speaking, assume the existence of the soul. It only postulates the possibility that it exists. Secondly, (b) was not a thought experiment put forward to support (2). Rather, it was a part of quite a different argument (for the view that physical continuity is not a necessary condition for personal identity). The gist of Nagasawa’s criticism seems to be that he disagrees with the so-called quasi-Aristotelian premise which says that there is no identity through time between two objects if they have no part in common. In this sense, Nagasawa’s criticism is the criticism of (3): he points out that maybe one can survive the destruction of their body even if one does not have a body, or at least, that Swinburne hasn’t excluded such a possibility. 3 The substitution objection (SO) An interesting objection has been put forward against the truth of the second premise (Zimmerman 1991; Alston and Smythe 1994; Stump and Kretzmann 1996). Basically, these authors point out that if we substitute for p a sentence which states ‘Swinburne is purely material in 1984’ or ‘Swinburne is identical with his body or some part of it’, premise (2) comes out false, for (arguably) it is not possible that Swinburne is purely material and yet he survives the destruction of his body. The way Swinburne responds to this objection is quite surprising, so we’ll quote his rejoinder at length (modifying the notation slightly): Like all worthwhile arguments, mine purported to start from premises which many an opponent might grant — viz. (1), (2), and (3) as they stand — to establish a conclusion which he did not previously recognize. I suggested that most people not already hav4 ing a firm philosophical position on the mind/body issue will grant my premises. But someone already having a firm philosophical position contrary to mine can challenge my premise (2) by inserting a p which he claims to be compatible with C and D and which he claims will show the premise to be false, where p states a philosophical thesis about the very issue in dispute, contrary to the one which I am seeking to prove. Examples include “I am purely material in 1984” of Alston and Smythe, or “I am identical with my body or some part of it” of Zimmerman. Now of course I claim that no such p is compatible with C ∧ D. Since I put forward premises (2) and (3) as purported necessary truths, my argument was designed to show that (given C) S is a necessary truth. The claim therefore that any p of the above type is compatible with C ∧ D amounts to the denial of my conclusion. Now it is true that my argument will not convince anyone who claims to be more certain that the conclusion is false than that the premises are true. But then that does not discredit my argument — for no argument about anything will convince someone in that position. My argument was designed for those prepared to set aside philosophical dogma concerned explicitly with the mind/body issue, and rely only on philosophical theses and intuitions about logical possibility relating to other or wider issues. (Swinburne 1996: 71) Swinburne seems to simply deny that the substitution instances that would falsify (2) lay in the range of variable p. That is, he defends his argument by saying that sentences like ‘Swinburne is purely material in 1984’ are not sentences about 1984 and compatible with C ∧ D. Before we fully assess the plausibility of this response, let us take a look at the modal argument more carefully, focusing on two things. First, the talk about range restrictions on propositional variables muddles the argument. This can be avoided by moving the range restriction into the formulas (what this means will become clear as we formalize the argument). Secondly, the argument has not been fully formalized: not all premises have been explicitly stated and the “logical moves” that the argument depends on haven’t been explicitly listed. In section 4 we will fix these issues by giving a full formalization of the argument. 4 Modal argument revisited Having surveyed the main lines of criticism, let’s take a look at the logical structure of the modal argument. We’ll start with a few useful abbreviations and then move on to the formalization. 4.1 Abbreviations First, there are two additional abbreviations that we’ll need. 84(p) ⇔ (sentence) p is about 1984 5 By 84(p) we only mean that p states something about an event or a state of affairs in 1984 and it does not state anything about an event or state of affairs “outside of” 1984. What is also important, a sentence does not have to be true in order to be about 1984. The notion of being about 1984 is fairly vague, but in fact we do have certain intuitions about whether a sentence is (purely) about 1984. It seems that such a sentence has to satisfy at least two restrictions: (i) what it states, if it occurs, has to take place in 1984 (i.e. it has to be about 1984), and (ii) the sentence, by itself, should not analytically imply a contingent sentence that states something about a period of time outside of 1984. By analytic consequence, in this context, we, roughly speaking, mean the relation that holds in virtue of the meaning of expressions occurring in the statements involved.5 Another intuitively plausible restriction is that sentences with embedded occurrences of temporal operators, whose truth-conditions are ultimately grounded in events taking place in 1984 should not be counted as not being about 1984. So, for instance, we exclude sentences like: It is the case in 1984 that in 1985 Swinburne will not exist. Although, in a way, this sentence is about 1984, it is not purely about 1984, because it clearly implies a contingent sentence about 1985. However, both sentences: Swinburne is purely material in 1984. Swinburne is not purely material in 1984. seem, on the face of it, to be purely about 1984. For instance, none of them, without any additional premises implies anything about whether Swinburne would survive the destruction of his body in the last instant of 1984. Of course, they might imply sentences of the sort: In 1985 it is true that Swinburne was purely material in 1984. but these aren’t taken to be really saying something about 1985, because their truth-conditions depend ultimately on events taking place in 1984 only. Another piece of notation that will be useful is this: (4) ♦(p ∧ C ∧ D) ⇔ sentence p is compatible with C ∧ D.6 It may seem slightly unclear what sort of compatibility Swinburne has in mind. Swinburne emphasizes that it is the same notion as that of logical coherence. He quite explicitly denies that there is a separate, “metaphysical” kind of necessity: . . . the contrast [between the metaphysical and the logical necessity] is misleading. For not merely is the necessity of both kinds equally hard, but it has the same nature – the necessary is that which holds in all possible worlds, where ‘possible’ means ‘coherently describable’. (Swinburne 1986: 314) 6 In one of the appendices in the second edition of the Evolution of the Soul (p. 324) he also explicitly states that the notion of possibility that he employs in this argument is logical.7 The further development of this paper should make it clear, however, that the notion of compatibility used here is quite strong: it seems to go beyond what we intuitively take to be analytic compatibility and probably the most convenient way to think about it is to interpret it as “metaphysical compatibility,” putting qualms about whether it is the same as logical compatibility aside. 4.2 The argument Now, we can restate the argument more carefully. The first premise is fairly trivial. It only states that Swinburne is alive and conscious in 1984: (5) C The second premise is a modification of Swinburne’s original second premise. It also captures the assumption about the range of variables that was mentioned but not included in the formula. (6) ∀p [84(p) ∧ ♦(p ∧ C ∧ D) → ♦(C ∧ D ∧ p ∧ E)] It is worth noticing that in (6) the universal quantifier is now really universal and implies all its substitution instances.8 Now, (6) says about any sentence p that if it is purely about 1984 and is compatible with C ∧ D, it is compatible not only with the claim that Swinburne’s conscious and alive in 1984 and his body is destroyed in the last moment of 1984, but also compatible with the claim that Swinburne’s conscious and alive in 1984, his body is destroyed in the last moment of 1984 and yet he survives and exists in 1985. Premise three is exactly the same as in the original argument: (7) ¬♦(C ∧ D ∧ ¬S ∧ E) It is obviously equivalent to: (8) (C ∧ D ∧ E → S) Sentence (8) is equivalent to (7) and says that if Swinburne is conscious and alive in 1984, his body is destroyed in the last moment of 1984 and yet he exists in 1895, he must have had a soul in 1984. In other words, (7) claims that having an immaterial part is a necessary condition of Swinburne’s survival of the destruction of his body. Now we need two premises which were not explicitly stated among the assumptions of the modal argument by Swinburne. One says that ‘Swinburne does not have a soul in 1984’ is purely about 1984: (9) 84(¬S) and another claims that if C and D necessarily entail S, so does C:9 7 (10) (C ∧ D → S) → (C → S) That is, if the facts that Swinburne exists and is conscious in 1984 and that his body is destroyed in the last moment on 1984 entail that Swinburne has a soul 1984, the fact that Swinburne’s body is destroyed in the last moment of 1984 has no relevance for this conclusion and the very fact that Swinburne is alive and conscious in 1984 already necessitates the fact that Swinburne has a soul in 1984. These assumptions logically entail S. Let’s see how the argument proceeds. First, we substitute ¬S for p when we eliminate the universal quantifier in (6):10 (11) 84(¬S) ∧ ♦(¬S ∧ C ∧ D) → ♦(C ∧ D ∧ ¬S ∧ E) By modus tollens from (7) and (11) we get: (12) ¬(84(¬S) ∧ ♦(¬S ∧ C ∧ D)) De Morgan’s law applied to (12) gives us: (13) ¬84(¬S) ∨ ¬♦(¬S ∧ C ∧ D) But this, together with (9) entails: (14) ¬♦(¬S ∧ C ∧ D) Quite easily we now get: (15) ¬(¬S ∧ C ∧ D) Since: (16) ¬(¬S ∧ C ∧ D) ⇔ [(C ∧ D) → S] and because we can substitute logically equivalent expressions within the scope of modal operators, we can infer: (17) ((C ∧ D) → S) Now we apply modus ponens to (10) and (17) and obtain: (18) (C → S) With (5), by  elimination and modus ponens , this entails: (19) S which completes the argument. Now we have a detailed formulation of a logically valid argument — we’ve just proven that the conclusion follows logically from the premises. The question is: is it also sound? 8 5 Substitution Objection revisited Let us abbreviate: M ⇔ Swinburne is purely material (at all times) in 1984. As remarked before, the universal quantifier in (6) is now really universal — there is no restriction on the range of sentential variables. Hence, we can substitute M for p when we eliminate the quantifier: (20) 84(M ) ∧ ♦(M ∧ C ∧ D) → ♦(C ∧ D ∧ M ∧ E) Is (20) true? Well, let’s first check if the antecedent of (20) is true. If it comes out false, we know that the conditional is true. So, is it the case that 84(M )? On the face of it, ‘Swinburne is purely material in 1984’ is a sentence that is purely about 1984. It certainly states something that is supposed to happen in 1984, and, prima facie, it doesn’t seem to imply any contingent sentence about events that are not in 1984, unless supported with additional premises.11 Unless one actually shows how M by itself implies a sentence that is clearly not about 1984, we may safely assume 84(M ). How about ♦(M ∧C ∧D)? Is it possible that Swinburne is alive and conscious in 1984, his body is destroyed in the last moment of 1984 and he does not have a soul in 1984? This question is hard to answer decisively because the question whether a description of a situation is logically coherent (in Swinburne’s sense) is not in general decidable. However, there are philosophers tend to believe that it is possible (practically all the materialists that believe that there are conscious events), and by doing so they do not seem to entangle themselves in obvious logical impossibilities. Quite to the contrary, the belief that it is impossible for Swinburne to be alive and conscious and yet not to have a soul could be accepted only by a person who already believes that there is a very strong (well, logical) connection between being conscious and possessing a soul. So to reject ♦(M ∧ C ∧ D) one has to believe not only that Swinburne has a soul in 1984, but also that it is logically impossible for him not to have one, given that he is alive and conscious. Interestingly, this is the strategy that Swinburne takes: [. . . ] someone already having a firm philosophical position contrary to mine can challenge my premise (2) by inserting a p which he claims to be compatible with C ∧ D and which he claims will show the premise to be false, where p states a philosophical thesis about the very issue in dispute, contrary to the one which I am seeking to prove. Examples include “I am purely material in 1984” of Alston and Smythe, or “I am identical with my body or some part of it” of Zimmerman. Now of course I claim that no such p is compatible with C ∧ D. (Swinburne 1996: 71) There are various reasons why this response is not deeply satisfactory. To believe ♦(M ∧ C ∧ D) one does not really have to have “a firm philosophical position” that would be contrary to Swinburne’s view. One can consistently 9 agree that Swinburne actually has a soul in 1984 and yet allow the logical possibility that one can be conscious and alive without possessing a soul. Thus, to use instances like that employed in the objection, one does not have to have a firm position regarding Swinburne’s conclusion. By admitting ♦(M ∧ C ∧ D) one does not admit anything “contrary” to the one that Swinburne is “seeking to prove.” Swinburne’s conclusion was simply S, and ♦(M ∧ C ∧ D) does not imply ¬S. Also, Swinburne may claim that no such p is logically compatible with C ∧ D, but we are in no way obliged to reject a sentence just because Swinburne disagrees with it. Unless he provides and independent argument for such logical incompatibility, we are free to accept ♦(M ∧ C ∧ D) if the claim sounds plausible. Swinburne could, of course, make ¬♦(M ∧C ∧D) (or anything to that effect) an explicit assumption of the argument. But then the argument becomes quite useless, because to reject ♦(M ∧ C ∧ D) one already has to have a very firm position on the philosophical position at issue. One not only has to believe that Swinburne has a soul, but one also has to believe that it is not even logically possible to be alive and conscious without having a soul.12 Now that we granted 84(M ) ∧ ♦(M ∧ C ∧ D), let us take a look at the consequent of (20). Is it possible that Swinburne is alive and conscious in 1984, he is purely material, and yet he survives the complete destruction of his body in the last moment of 1984? Here, the problem is that Swinburne’s notion of logical possibility is quite vague when applied to natural language. ‘Being true in a certain possible world’ or ‘being true according to a certain coherent story’ is not a clear property because (a) it is unclear that there is a single logical framework against which we can evaluate the coherence of stories, and (b) even if there were such a framework, the coherence problem would be quite likely undecidable. Even more importantly, (c) we don’t even know how far we are supposed to go with our sense of “logicality”. Are, for instance, analytic truths logical? From the fact that Swinburne does consider certain situations impossible even though they don’t involve explicit propositional (or first-order) contradiction and their impossibility relies on certain truths that are, at best, analytic, it seems that we are supposed to take “logicality” in this wide sense. In such a case, one can argue that there are conceptual reasons for which even if there existed an object in 1985 which qualitatively would be identical to Swinburne, there would be no reason to claim that it is numerically identical to Swinburne, given that Swinburne was purely material and his body was completely destroyed. Quite the opposite seems to be true: if he really is purely material, by destroying his body we completely destroy Swinburne himself, and we have a pretty clear intuition that according to our concept of persons, persons cannot cease to exist and pop into existence again. What might pop into existence is an exact copy of Swinburne, but not Swinburne himself.13 But we don’t have to discuss the questions raised in the paragraph above in depth. Consider a story according to which Swinburne is alive, conscious and purely material in 1984. Assume also that (according to this story) his body is completely destroyed in the last instant of 1984 and yet he survives into 1985. Is the story coherent? If no, the consequent of (20) comes out false and (20) is 10 falsified. If it is coherent, (20) comes out true, but another problem arises. For admitting this coherence commits one to saying that it is equally well coherent that Swinburne, premortem alive and conscious, doesn’t have a soul and yet he survives the destruction of his body. But this would falsify Swinburne’s assumption (7). Either way, the modal argument comes out unsound. Thus, the SO defeats the modal argument, at least in this formulation, although for a slightly different reason than the one suggested by the authors who simply suggested that M falsifies one of the premises. In section 6 we’ll take a look at a strategy that Swinburne might have employed to give a modal argument that evades this criticism. This strategy would allow him, in a sense, to exclude counterexamples of this sort by weakening one of the premises. 6 Evading the Substitution Objection Swinburne complained that to believe that M falsifies (6) (i.e. the claim that any sentence about 1984 that is compatible with C ∧ D is also compatible with C ∧ D ∧ E) one already has to have a firm philosophical position. We’ve seen that this complaint is illegitimate, because what in fact is sufficient is the belief that being conscious and alive does not logically entail possessing a soul. One can, however, give a version of the modal argument which is not susceptible to counterexamples of this sort. Basically, what one needs to do is to suggest that in (6) the condition in the antecedent instead of ‘is about 1984’ should contain ‘is true about 1984’. In that interpretation, one can’t take M to falsify the sentence because it really is the case that to accept that M is true about 1984 one already has to disagree with Swinburne on the question at issue. Let’s make sure we can actually run the argument in this interpretation. We define ‘being true about 1984’ (tr84): (21) tr84(p) ⇔ 84(p) ∧ p We still assume (5), (7), (9) and (10), but instead of (6) we take: (22) ∀p [tr84(p) ∧ ♦(p ∧ C ∧ D) → ♦(C ∧ D ∧ p ∧ E)] We then mimic the original argument, eliminating the universal quantifier from (22): (23) tr84(¬S) ∧ ♦(¬S ∧ C ∧ D) → ♦(C ∧ D ∧ ¬S ∧ E) from (7) and (11) substituting ¬S for p we obtain: (24) ¬(tr84(¬S) ∧ ♦(¬S ∧ C ∧ D)) De Morgan’s law applied to (24) gives us: (25) ¬tr84(¬S) ∨ ¬♦(¬S ∧ C ∧ D) At this point we split the disjunction into a proof by cases. Suppose ¬tr84(¬S). By (21), this means that either ¬84(¬S) or ¬¬S. But (9) says that 84(¬S). So 11 ¬¬S and hence S. Suppose on the other hand that ¬♦(¬S ∧ C ∧ D). In this case we land back at (14) and proceed with the argument exactly the same way as before, obtaining S. Hence the whole disjunction in (25) entails S. Let’s rephrase the last step in a more accessible manner. (25) tells us that either ¬S (‘Swinburne doesn’t have a soul in 1984’) is not a true sentence about 1984, or is not a sentence that is compatible with C ∧ D (‘Swinburne is alive and conscious in 1984 and his body is destroyed in the last instant of 1984’). If ¬S is not a true sentence about 1984, either it is not about 1984 or it isn’t a true sentence. But we know it is about 1984. So ¬S is not a true sentence and therefore S is a true sentence. Thus, if the first disjunct of (25) is true, S is true. If, on the other hand, it is the case that ¬S is not compatible with C ∧ D, this means that C ∧ D entails S. But, pretty much like in the original argument, D in this case has no bearing on the truth value of S and therefore C by itself entails S. We now have to see why the SO doesn’t work. To believe that M falsifies (22), one has to believe that the antecedent of (22) is true and the consequent false. But the antecedent of (22) is tr84(M ), which entails S. Thus, to believe that M falsifies (22), one must have a firm position about the issue at stake. One of the reactions to this argument that we have encountered is the complaint that this modified version is not a version which Swinburne would accept. First of all, (22) is actually a weakening of (6), because any formula of the form φ → χ entails φ ∧ ψ → χ, but entailment in the opposite direction doesn’t hold. That being the case, anyone accepting (6) is committed to (22) anyway. Secondly, Professor Swinburne is actually familiar with our modified version and agrees that it constitutes an improvement of his original argument. Alas, we need to observe that even though the present strategy makes Swinburne’s initial response to SO more convincing, the argument might be seen as epistemically circular in the following sense. Suppose we indeed replace (6) with (22). Basically, the assumption employed now says that no true sentence compatible with Swinburne‘s being alive and conscious in 1984 and his body being destroyed excludes the possibility of him surviving the destruction of his body. This is a fairly strong claim, because it is equivalent to the claim that any sentence compatible with C ∧D that excludes the possibility of Swinburne‘s survival is already false. Thus, for instance, if one believes that M excludes such a possibility and is compatible with C ∧ D, it is already false and its negation is true.14 But if this is the case, by accepting (22) we commit the same fallacy of which Swinburne was accusing his critics: to accept it we already have to have a firm philosophical position on the issue: one has to believe that no true sentence makes Swinburne’s survival of the destruction of his body impossible. A possible way to react to this criticism is to say that no sentence purely about 1984 has any bearing on Swinburne’s survival in 1985. This hinges on the unclarities surrounding the notion of being purely about 1984. Indeed, in the sense of ‘analytic consequence’ used in our specification of ‘being about 1984,’ no sentence purely about 1984 analytically entails anything about 1985. This, however, doesn’t mean that no sentence purely about 1984 is incompatible with the negation of any sentence about 1985 in the relevant sense of compatibility. 12 For indeed, if we assume that our notion of compatibility overlaps completely with analytical coherence, we have a reason to reject premise (7) which says that Swinburne cannot survive the complete destruction of his body and continue to exist in 1985 if he doesn’t have a soul in 1984, and the argument comes out unsound anyway.15 Professor Swinburne argued (in personal communication) that the truth of (22) is available even to children when he tries to explain the argument to them. They tend to agree when faced with statements like ‘Look, it is at least logically possible, whatever else is true and doesn’t exclude us being conscious and our bodies being completely destroyed, that we survive the complete destruction of our bodies’. There are two points to make to explain these intuitions away. First, there is a notion of possibility on which (22) comes out true, but which falsifies (7). It is quite likely that some people when faced with the sentence in question use this notion to assess its truth. Then, when they’re faced with (7) they use quite a different metaphysical notion of possibility not noticing the difference. Second, there is an important scope distinction to be kept in mind. One one reading, the sentence in question says exactly what (22) says, and yields a rather strong statement. On the other reading, it rather says that no matter what is true about 1984 and doesn’t exclude C and D, it is still possible to survive the complete destruction of one’s body: ∀p [tr84(p) ∧ ♦(p ∧ C ∧ D) → ♦(C ∧ D ∧ E)] On the second reading, anyone who admits ♦(C ∧ D ∧ E), a rather weak claim of the mere possibility of conscious beings surviving the complete destruction of their bodies, is committed to this claim. On this reading, however, the claim is too weak to constitute a premise of a valid modal argument.16 7 Summary We presented Swinburne’s modal argument for the existence of human souls17 in its original formulation. SO, put forward against the argument, was our point of departure. To be able to assess its strength in detail, we formalized the argument fully. Our formulation doesn’t contain tacit premises, vague remarks about range of propositional variables are avoided, and the conclusion follows logically from the premises. In this setting, SO turns out to falsify at least one of the premises and Swinburne’s response to it is quite unsuccessful. Then, we put forward another version of the modal argument that seems to fit better with Swinburne’s rejoinder. It is a valid argument and SO does not falsify any of the new premises. The problem arises, however, that even though we cannot easily falsify a premise, we also cannot assume it to be true unless we’ve already made up our mind regarding the issue at question. In this sense, unless a successful strategy of evading SO is given, the modal argument is either unsound or epistemically circular. 13 Notes 1 The argument can be found in (Swinburne 1986: ch. 8). It also occurs in (Swinburne and Shoemaker 1984: ch. 2). (Swinburne 1996) is meant as a defence of the modal argument against certain objections raised in the literature of the subject. 2 I use ‘⇔’ as standing for definitional and logical equivalence. 3 We just changed the notation in order to accommodate our conventions. 4 “Premise 2 says that it is possible that I survive into 1985, given that I am conscious in 1984, even if my body is totally destroyed and whatever else might be the case in 1984, compatible with these last two suppositions.”(Swinburne 1996: 69) 5 We decided to ignore Quinean qualms about analyticity. For a recent defense of the respectability of this notion, see (Russell 2008). 6 Observe that being compatible with C ∧ D is a stronger condition than being compatible with C and with D. For instance, the sentence ‘Socrates likes beer’ is compatible with ‘Socrates likes wine’ and with ‘Socrates doesn’t like wine’, even though ‘Socrates likes beer’ is not compatible with ‘Socrates likes wine and Socrates doesn’t like wine’. This doesn’t have to have anything to do with the two conjoined sentences being mutually exclusive. For instance ‘Socrates likes beer’ is compatible with ‘Socrates likes wine’ and with ‘If Socrates likes beer, he doesn’t like wine’ separately but is not compatible with the conjunction of these sentences, even though they do not exclude each other logically. 7 For this reason it is surprising that he wanted to deny that Zimmerman’s counterexample was compatible with C ∧ D, because the sentence seems logically compatible with C ∧ D. 8 Of course, variable p is still restricted, in the sense that it ranges over propositions. This fact, however, is fairly clearly captured by the choice of variable symbols. Explicating this restriction even further would not increase the clarity of the argument and would bring to light any philosophically relevant issues. 9 Technically speaking, a weaker premise, (C ∧ D → S) → (C → S) would suffice to make the argument work, but it seems that Swinburne prefers the stronger, modal reading. 10 Note that since now the quantifier ranges over all sentences, the question whether a sentence is a legitimate substitution instance will not concern us at all, neither now, nor later when we will be discussing the plausibility of the argument. 11 Most importantly, together with certain philosophical background assumptions, it entails that Swinburne would not survive the destruction of his body in the last moment of 1984. This, however, does not mean that the sentence is not purely about 1984. 12 In the original formulation of the argument (Swinburne and Shoemaker 1984; Swinburne 1986), (12) is pretty much treated as an assumption of the modal argument. In (Swinburne 1996), additional argument is brought up to support the claim that one has to have a soul if one survives the destruction of one’s body. Swinburne (1996) refers to what he calls a ‘quasi-Aristotelian’ assumption, which says that an object x can be identical through time with an object y only if they have a part in common or its parts have been obtained by gradual replacement of the original parts. This extended version of the modal argument can be briefly formulated as follows: Ar1 For all x, if x is a material object and x is totally destroyed in 1984, then no part of x exists in 1985 (i.e. physical objects and their parts can’t be annihilated and then come back into existence at a later time). [Assumption] Ar2 For all x and for all y, if x exists in 1984 and y exists in 1985, then x and y are one and the same thing only if either x and y share at least one part or y’s parts are the result of gradually replacing all of x’s parts. [the quasi-Aristotelian assumption] Ar3 So ¬S is not compatible with C ∧ D ∧ E. [from Ar1 and Ar2, so long as D implies that Swinburne’s body is totally destroyed – that is, that all of its parts are annihilated] Ar4 For all x, if x is a sentence purely about 1984 and x is compatible with C ∧ D, then x is compatible with C ∧ D ∧ E. [Assumption] Ar5 Either ¬S is not a sentence purely about 1984 or ¬S is not compatible with C ∧ D. [from Ar3 and Ar4] Ar6 ¬S is a sentence purely about 1984. [Assumption] 14 Ar7 ¬S is not compatible with C ∧ D – that is, C ∧ D entails S. [from Ar5 and Ar6] Ar8 If C ∧ D entails S, then C entails S. [Assumption] Ar9 So C entails S. [from Ar7 and Ar8] Ar10 C [Assumption] Ar11 Hence, S. [from Ar9 and Ar10] 13 We can, on the other hand, take a very strict understanding of logical necessity that doesn’t treat analytic truths to be logically necessary. For instance, we can say that a description of a situation is coherent iff its translation into a first-order language of classical logic (if such a translation is possible) is classically consistent. On this reading, the consequent of (20) comes out true. Alas, if this is the reading of logical possibility that we take, (7) comes out false and the modal argument is unsound. 14 See the discussion in section 5. 15 This is one reason to think that the modalities used in the argument go beyond what we normally would take to be logical or analytical factors. 16 When we eliminate the universal quantifier and substitute ¬S for p, we won’t be able to use modus tollens, because ♦(C ∧ D ∧ E) won’t contradict (7). 17 Technically speaking, for the existence of Swinburne’s soul, but the argument, if successful, can be easily generalized. References Alston, W. P. and Smythe, T. W. (1994). Swinburne’s argument for dualism. Faith and Philosophy, 11(1):127–133. Hasker, W. (1998). Swinburne’s modal argument for dualism: Epistemically circular. Faith and Philosophy, 15:366–370. Nagasawa, Y. (2005). Critical notice of Richard Swinburne’s “the evolution of the soul (revised version)”. Posted on The Secular Web website. Reames, K. (1999). A response to Swinburne’s latest defense of the argument for dualism. Faith and Philosophy, 16:90–97. Russell, G. (2008). Truth in Virtue of Meaning: a Defence of the Analytic/Synthetic Distinction. Oxford University Press. Stump, E. and Kretzmann, N. (1996). An objection to Swinburne’s argument for dualism. Faith and Philosophy, 13:405–412. Swinburne, R. (1986). The Evolution of the Soul. Clarendon Press. Page references to the second edition published in 1997 by Clarendon Press. Swinburne, R. (1996). Dualism intact. Faith and Philosophy, 13(1):68–77. Swinburne, R. and Shoemaker, S. (1984). Personal Identity. Basil Blackwell. Zimmerman, D. W. (1991). Two cartesian arguments for the simplicity of the soul. American Philosophical Quarterly, 3:217–226. 15