The questions of ontology, as they are usually understood, are questions about what there is, or about what sorts of entities exist: are there numbers, propositions, tropes, temporal parts, arbitrary mereological sums, etc.? Contemporary analytic philosophers tend to take these questions very seriously and to treat them as questions worthy of sustained inquiry. On the other hand, meta-ontology, the study of what ontology is and whether it is possible, is also flourishing,Footnote 1 and there are a number of meta-ontologists who are sceptical either of the ontological enterprise as a whole, or of certain ontological questions. Some of these see themselves as trying to revive the anti-metaphysical tradition that goes back to the writings of Rudolf Carnap, who held a thoroughly deflationary view of ontology (as defended e.g. in ‘Empiricism, semantics and ontology’ (henceforth ‘ESO’)Footnote 2).
In his article ‘Internal and external questions about God’Footnote 3 (as well as in parts of his book Arguing for Atheism),Footnote 4 Robin Le Poidevin offers a sympathetic, though ultimately critical, discussion of Carnap's position. Interestingly, he applies Carnap's argument to the ontological question that is perhaps least often suspected of being illegitimate or meaningless, namely the question of whether God exists. Although he argues against Carnap's deflationism, he takes Carnap's argument to have two important implications for the philosophy of religion: (1) theists need to take God to be a putative entity within space-time in order for their claim that God exists to be meaningful, and (2) instrumentalists about theology, who take talk about God to be non-descriptiveFootnote 5 have a defensible position.
My aim in this article is to clarify and evaluate Le Poidevin's argument for these interesting and far-reaching conclusions. I argue that although Le Poidevin's response to Carnap's argument is no less problematic than that argument itself, Le Poidevin's position is in fact thoroughly un-Carnapian. The upshot will be that his discussion provides some support to atheism (either with respect to a timeless God only or tout court), but none to either of his two official conclusions.
The article is divided into three sections. In the first section, I summarize Carnap's position as put forth in ESO and briefly outline the particular version of Carnap discussed by Le Poidevin, which Stephen Yablo (Reference Yablo1998) develops in detail. In the second section, I examine Le Poidevin's response to one of the more explicit arguments from ESO and Le Poidevin's own meta-ontological commitments. In the third and final section, I consider the implications of these commitments for Le Poidevin's conclusions with respect to the philosophy of religion.
Carnap's view
Let us consider Carnap's position, as laid out in ESO. Famously, that position makes use of the notion of a ‘linguistic framework’, and of the accompanying distinction between ‘internal’ and ‘external’ existence questions.
If someone wishes to speak in his language about a new kind of entities, he has to introduce a system of new ways of speaking, subject to new rules; we shall call this procedure the construction of a linguistic framework for the new entities in question. (Carnap (Reference Carnap, Benacerraf and Putnam1950), 242)
A linguistic framework is made up of terms and expressions together with rules governing those terms and expressions. For example, the numbers framework consists of the following: number terms such as ‘three’; the general term ‘number’ and sentence forms like ‘three is a number’; expressions for properties (e.g. ‘prime’), relations (e.g. ‘is greater than’), and functions (e.g. ‘plus’), and sentence forms like ‘two plus three is five’; numerical variables (e.g. ‘n’) as well as existential and universal quantifiers governed by the usual deductive rules.
Note that not all frameworks are consciously constructed; nor are all frameworks of a technical nature. For example, another framework discussed by Carnap, which will be important in what follows, is the framework of things and events in space-time (henceforth the ‘things framework’). That framework is adopted by each of us ‘early in life as a matter of course’.
And now we must distinguish two kinds of questions of existence: first, questions of the existence of certain entities of the new kind within the framework; we shall call them internal questions; and second, questions concerning the existence or reality of the system of entities as a whole, called external questions. (Carnap (Reference Carnap, Benacerraf and Putnam1950), 242)
Take the numbers framework. Examples of internal questions would be ‘How much is three plus three?’ or ‘Is there an even prime number?’ These are answered with the help of the rules of the framework; since the framework is a logical one, the method of answering is logical also. In the case of frameworks that are not logical but factual, such as the things framework, internal questions are answered empirically. Thus, examples of questions internal to the things framework would be ‘Did King Arthur actually live?’ or ‘Is there a white piece of paper on my desk?’
Now consider questions such as ‘Do numbers exist?’ or ‘Are there physical objects?’ According to Carnap, such questions must be interpreted as either internal or external questions. If they are interpreted as internal questions, namely as questions internal to their respective frameworks, they have trivially positive answers. Their answers follow analytically from positive answers to more specific internal questions. Thus, from ‘There exists a prime number between three and seven’, the proposition ‘There exist numbers’ can be derived, given the rules of the numbers framework. Similarly, from ‘There is a white piece of paper on my desk’, the proposition ‘There are physical objects’ can be derived, given the rules of the things framework. Such trivial answers indicate that this is not the way in which ontologists asking about the reality of numbers or the external world intend their questions to be interpreted. However, the only other way they could be interpreted is as external questions, and in particular, as external practical questions concerning whether or not we should adopt certain frameworks. Clearly, this is not what ontologists have in mind either: they mean to be asking a non-trivial theoretical question. But it is not clear what meaning external questions might be given that would make them theoretical rather than practical. This is why, in Carnap's view, ontological disputes are simply misguided.
Le Poidevin refers to something like the above as ‘Carnap's argument’, but I think it should be stressed that the above does not yet constitute much of an argument for Carnap's position, only a rough statement of it (I will reserve the term for the argument discussed in the next section). In particular, we have not yet been given any argument for the crucial claim that theoretical external questions are meaningless.
There is also the prior question of what exactly we should take frameworks to be. For my purposes, what matters is the way Le Poidevin construes the notion, as implied by his construal of the internal/external contrast. He suggests that we understand these notions by analogy with fiction (Yablo develops such a proposal in detail; see below). Pieces of fiction correspond to frameworks, and internal and external questions are questions posed from within and without those pieces of fiction, respectively. Thus, a question like ‘Did Dorothea publish Casaubon's Key to All Mysteries?’ would be an example of a question that is internal to the fictional framework of Middlemarch. A question like ‘Did Casaubon exist?’, unless it is taken to ask whether in the novel Casaubon was merely a dream of Dorothea's, is a question that is external to the fictional framework. It means ‘Did George Eliot base the character of Casaubon on someone she knew?’ Similarly, the question ‘Is Middlemarch true?’ is an external question asking whether Middlemarch corresponds to reality, where ‘reality’ means our world, namely the world in which George Eliot is real.
This construal of the notion of a framework is, as Le Poidevin suggests, intelligible as far as it goes; it is also prima facie more promising than what Carnap himself had in mind. Carnap, as we have seen, takes the answer to general internal questions to follow analytically from specific internal statements. Thus, the answer to the question ‘Do numbers exist?’, as posed from within the numbers framework, follows analytically from e.g. ‘There is an even prime number’. That is, the rules of Carnap's frameworks are ‘analytically valid’ in the sense that they give the sentences of the framework their meaning, so that as long as those meanings are held fixed, the rules are incorrigible. As is well known, this reliance on the analytic/synthetic distinction made Carnap's position vulnerable to the Quinean critique, which Yablo neatly summarizes as follows:
The factors governing assertion are an inextricable mix of the semantic and the cognitive; any serious question about the assertive use of ‘X’ has to do both with the word's meaning and the X-ish facts. Accordingly Carnap's external stance, in which we confront a purely practical decision about which linguistic rules to employ, and his internal stance, in which we robotically apply these rules to determine existence, are both of them philosophical fantasies. (Yablo (Reference Yablo1998), 240)
Yablo suggests that if we give up on the idea that the rules of the X-framework are what give the term ‘X’ its meaning, these problems are circumvented. Instead of insisting that the rules of a framework are analytically valid meaning-conferring rules we can allow that ‘X’ has a meaning prior to the adoption of the X-framework. And then, in order to ensure that internal statements about X's are still not answerable to external standards, we construe frameworks as fictions (or ‘make-believe games’).
However, while prima facie more promising, this construal of the notion of a ‘framework’ is perhaps even harder to pin down than Carnap's own: what counts as part of a given fictional framework? The analogy only goes so far. Yablo himself says that compared to the literal/metaphorical distinction (on which his non-make-believe/make-believe construal is loosely based), the analytic/synthetic distinction is ‘a marvel of philosophical clarity and precision’. Moreover, when understood as a global doctrine, this version of Carnap may well imply a thoroughgoing relativism (as may, for all I know, Carnap's own position).Footnote 6 In any case, as we shall now see, these problems are largely irrelevant to Le Poidevin's argument. The reason is that Le Poidevin's own position is in fact thoroughly un-Carnapian.
Le Poidevin's response to Carnap
Consider now one of Carnap's more explicit arguments for his central claim that (supposedly) theoretical external questions, like ‘Are there physical objects?’ (or, equivalently for present purposes: ‘Are physical objects real?’, ‘Is there (really) an external world?’), are meaningless. Carnap mentions questions internal to the things framework, such as ‘Did King Arthur actually live?’, and then says:
The concept of reality occurring in these internal questions is an empirical, scientific, non-metaphysical concept. To recognize something as a real thing or event means to succeed in incorporating it into the system of things at a particular space-time position so that it fits together with the other things recognized as real, according to the rules of the framework.
From these questions we must distinguish the external question of the reality of the thing world itself. In contrast to the former questions, this question is raised neither by the man in the street nor by scientists, but only by philosophers. Realists give an affirmative answer, subjective idealists a negative one, and the controversy goes on for centuries without ever being solved. And it cannot be solved because it is framed in the wrong way. To be real in the scientific sense means to be an element of the system; hence this concept cannot be meaningfully applied to the system itself. (Carnap (Reference Carnap, Benacerraf and Putnam1950), 243)
This argument seems problematic to me, because it simply assumes that the only sense of ‘real’ that might be relevant to the external question at issue is the one associated with the things framework itself (the ‘scientific sense’).Footnote 7 It is only on this assumption that considerations about reflexivity and self-membership are relevant at all.
Yet, it is these considerations that Le Poidevin focuses on in his response. He first points out that self-membership need not lead to paradox (e.g. the set of all abstract objects contains itself), and that fictions can make reference to themselves (e.g. Gulliver's Travels makes reference to itself as a journal). He then says that what is really needed in response is an example of a framework that allows the question of whether it reflects reality to be answered from within itself. He construes this as a framework that ‘compels acceptance of itself’. The framework he settles on for this purpose is ‘the framework about us’. Which framework this is, Le Poidevin says, depends on how we view ourselves, but the framework he goes on to discuss is the one that is based on a view of ourselves as things existing in space-time.
Given the Cartesian conviction that I exist, anything spatially and temporally related to me must also exist. So to define ‘real’ in terms of being located in the spatio-temporal framework is not some arbitrary definition which we could have replaced with any other, it is the most fundamental concept of reality. (Le Poidevin (Reference Le Poidevin1995), 497)
In this way, Le Poidevin means to counter the above argument and, pace Carnap, to provide a meaning for genuinely theoretical external questions. These questions are external to certain fictional frameworks, though internal to a certain privileged fictional framework, namely the spatio-temporal framework that ‘contains us’. (Note that this phrase is slightly peculiar given the definition of a framework as a system of expressions; in an earlier version of ESO, Carnap specifies that by ‘framework’ he means only the expressions, not the entities. This confusion may be partly responsible for some of the problems identified below.) And it is precisely the ‘ontological authority’ of that framework that gives these theoretical external questions their meaning, according to Le Poidevin. Because we cannot doubt that we ourselves exist, we cannot help taking the framework ‘containing us’ to reflect reality.
The first thing to ask about this response is this: isn't, for all we are told, Le Poidevin's ‘us-framework’, at least as it is explicitly developed by him, identical with Carnap's things framework? After all, they are both the framework of things and events in space-time. Le Poidevin explicitly calls the framework in question ‘the spatio-temporal framework’; Carnap, too, speaks of ‘the framework of things and events in space-time’. Perhaps it may be objected that Le Poidevin emphasizes that the ‘us-framework’ is defined indexically (‘to be located in this framework is to stand in some relation to me’), but if this is meant to signal that the framework in question is not simply the things framework, much more needs to be said about its nature. (For one thing, the mere presence of indexicals does not disqualify a question from being internal to, and straightforwardly answerable within, Carnap's things framework: the question ‘Is there a white pen on my desk?’ is, for Carnap, a straightforward example of such a question.)
The second thing to ask is how exactly Le Poidevin has responded to Carnap's argument. That argument was supposed to establish that a question such as ‘Are there physical objects?’, when it is intended as external to the things framework and as theoretical rather than practical, is without sense. Le Poidevin is attempting to supply such a sense to theoretical external questions. The sense he offers is the following: given any fictional framework, we can ask about the framework as a whole whether it reflects, or corresponds to anything, in reality, where ‘reality’ is to be understood in the sense associated with the spatio-temporal framework. That these questions are still internal to the spatio-temporal framework does not threaten their significance (i.e. make them objectionably limited in scope), because of the ‘ontological authority’ of the spatio-temporal framework. Once we adopt this framework, we cannot help taking it to reflect reality. But which sense of ‘reality’ is this? That, after all, was the question.
Given that Le Poidevin thinks there is no more fundamental sense of ‘reality’ than the one associated with the spatio-temporal framework, we may presume that this is the intended sense. That would explain the attention he pays to Carnap's reflexivity objection. However, that objection then still looms large: what sense does it make to take space-time to be real in the sense in which things and events in it are real? It can hardly be ‘incorporat[ed] … into the system of things at a particular space-time position so that it fits together with the other things recognized as real’.Footnote 8 The realization that we ‘can't help’ taking this framework to reflect reality does not answer this objection, any more than it would to point out that metaphysicians can't help asking about the reality of the external world. That we are ‘compelled’ to accept the spatio-temporal framework does not make that acceptance any more intelligible as a theoretical rather than a practical step: in what sense can we thereby be said to have accepted the framework as real? The claim that ‘we cannot coherently doubt that the framework of which we are a part exists’ may even work in Carnap's favour. If we cannot coherently doubt that there is an external world, then perhaps this is all the more reason to think that the question of the reality of the external world is meaningless.
Given this difficulty, one might wonder whether perhaps Le Poidevin, like Carnap, does not actually take the question ‘Are there physical objects?’ to be meaningful. It is true that Le Poidevin favours a deflationist stance on many ontological questions, including all those that ask about the existence of abstract entities like numbers or propositions (see below). But in fact, he does not mean to extend this stance to the question about the reality of the external world. Unfortunately, if that question lacked a sense to begin with, he has not supplied it with one.
Which questions, then, have been supplied with meaning? The answer is: those questions that are external to fictional frameworks other than the special spatio-temporal framework, asked from within the latter. Le Poidevin is suggesting that asking about the reality of a given fictional framework means asking whether the fiction corresponds to (spatio-temporal) reality. The problem with this is that those questions were never suspected of lacking meaning in the first place, nor was any other suggestion as to their meaning ever salient. Given any fiction, we can of course ask whether it corresponds to anything in (spatio-temporal) reality. As we have seen, Carnap did not object to such questions. True, he stressed that they were internal questions, namely internal to the things framework: ‘Did King Arthur actually live?’; ‘Are unicorns and centaurs real or merely imaginary?’ But for Le Poidevin, too, these questions are internal, namely internal to the spatio-temporal framework.
Nonetheless, it is quite clear that Le Poidevin views these questions very differently from Carnap. For Le Poidevin, the spatio-temporal framework is not one among many, nor is its privilege merely one of practical utility or even indispensability. Le Poidevin does stress its indispensability, but only because he takes it to indicate further that the framework also has a special ontological authority. The spatio-temporal sense of ‘real’ is, for Le Poidevin, the most fundamental sense of ‘real’, and the one in terms of which all ontological questions should be interpreted. So Le Poidevin firmly rejects Carnap's characteristic pluralist stance towards frameworks, assigning a unique privilege to one, namely the spatio-temporal framework that ‘contains us’.Footnote 9
The claim that all ontological questions should be interpreted in terms of spatio-temporal reality is controversial;Footnote 10 moreover, it is in no way suited to the deflationist view that ontological questions lack meaning, because it supplies a clear meaning to most of those questions. (As mentioned, the question of the existence of the external world may be an exception.) Questions concerning the existence of abstract objects, far from lacking a theoretical meaning, can then be interpreted as asking whether there are any abstract objects within space-time. Of course the answer will likely be ‘No’, but that is another matter (I return to this point below), as is whether this is the question metaphysicians actually mean to ask.
To sum up: although Le Poidevin's response to Carnap is no less problematic than that argument itself, Le Poidevin's meta-ontological position differs radically from Carnap's. Far from being a pluralist, Le Poidevin takes the spatio-temporal framework to have ontological authority. So one might even wonder what work the talk of ‘frameworks’ is doing for Le Poidevin, who after all, construes them as something like fictions. Perhaps the best thing to say is that the spatio-temporal ‘framework’ is not one fiction among many, because it is in fact nothing like a fiction.
The response applied to theism
Finally, let's consider Le Poidevin's claims regarding theology and the philosophy of religion.
Le Poidevin defines the ‘theistic framework’ as follows.
Firstly we introduce the term ‘God’ and associated expressions: ‘God is good’, ‘God is unique’, etc. Secondly we introduce expressions linking the terms ‘God’ with the physical object framework: ‘God created the universe’, ‘God loves his creation’, etc. Thirdly, we introduce methods for deciding the truth of certain propositions. In part this will consist of an authoritative text, such as the Bible, or Koran. The framework may also distinguish, amongst the statements in the text, between those statements which are intended to be read literally, and those that are intended as metaphorical or allegorical. (Le Poidevin (Reference Le Poidevin1995), 488–489)
Before discussing the relation between a Carnapian stance on ontology and theology, Le Poidevin ably rebuts a number of prima facie objections to the idea that the theological question ‘Does God exist?’ is a candidate ontological question. He then points out what, in his view, would be wrong with a response that simply granted Carnap's argument and settled for construing the question of God's existence as an internal question about the existence of a particular kind of person. I will refer to this strategy as the ‘internal strategy’ below.Footnote 11 (Note that commentators have pointed out that not all internal questions concern a subclass of a more inclusive class (nor vice versa). But nothing will hang on this here.)Footnote 12
Instead of treating God as if he constituted a class of his own, we could treat him as a member of a more inclusive class. The obvious class to pick is that of persons, since traditional theism represents God as having many characteristics of a person … So ‘Does God exist?’ should, according to this strategy, be read ‘Is there a person who is omniscient, omnipotent, and perfectly good, who created the universe, etc.?’ This can be answered non-trivially, provided that the framework of persons includes a logical or empirical procedure by which we can determine the answer. That is to say, provided that there is a sound argument for the existence of God.
However, this strategy will not give us back full-blooded theism. The reason is that, for Carnap, no internal question, even if it has a non-trivial answer, concerns the question of what really exists. So even though we can reconstrue the question of God's existence as a specific question about a certain kind of person, the answer will not (if Carnap is right) give us what the theist wants. (Le Poidevin (Reference Le Poidevin1995), 492)
As we have seen, Le Poidevin does not think that Carnap is right. However, he does draw two far-reaching conclusions from Carnap's argument: (1) theists need to take God to be a putative entity within space-time in order for their claim that God exists to be meaningful, and (2) instrumentalists about theology, who take talk about God to be non-descriptive, have a defensible position.
Le Poidevin's argument for these two claims can now be reconstructed as follows. His discussion of Carnap is supposed to have established that a deflationist attitude towards ontological questions is appropriate in the case of existence questions about abstract entities, or entities outside of space-time, but not appropriate in the case of existence questions about concrete entities, or entities within space-time. Hence, when the question of God's existence is taken to be about a timeless God (one outside of space-time), a deflationist attitude towards the question is appropriate, whereas if it is taken to be about a God within space-time, a non-deflationist attitude towards the question is appropriate.
Unfortunately, the first and key premise of this argument relies on a mistaken appraisal of the results of the above discussion. Le Poidevin's insistence on the ontological authority of the spatio-temporal framework implies not deflationism about abstract existence questions, but rather nominalism concerning abstract entities. Insisting that all ontological questions be construed as questions about the spatio-temporal reality of entities leads not to the view that asking about the reality of abstract entities is meaningless, but to the view that abstract entities are not real:
Suppose we think of ourselves as objects existing in time. Then, if we believe that something stands in a certain temporal relation to us, and so, like us, it is a temporal object, then we cannot but think of that object as real. The problem with abstract object frameworks is that they are not defined in terms of the relations they stand in to us, but rather the lack of such relations: according to one kind of realism, numbers are objects outside time and space and independent of any mind. That is why we are suspicious of conferring reality upon such frameworks: there is nothing to give them ontological authority … What I am suggesting here is that the natural position to adopt is an ontological parochialism, to believe in the existence only of things of our kind, things which exist in time. (Le Poidevin (Reference Le Poidevin1996), 132)
When Le Poidevin claims that adopting the numbers framework ‘does not compel us to think of numbers as real, because they are not “one of us”, so to speak’ (ibid.), he goes beyond what either his or Carnap's argument has established; and as we have seen, he thereby makes a claim that sits uneasily with his suggestion that all ontological questions should be understood in terms of spatio-temporal existence.
If we now follow Le Poidevin in applying his argument to the question of God's existence, the implication is atheism, at least with respect to a timeless God. Le Poidevin himself seems to acknowledge this:
If the fundamental concept of existence is that of spatio-temporal location, then when we ask ‘Does God exist?’ we mean ‘Can we locate God in our spatio-temporal framework?’ But if the theistic framework is one which emphasizes God's ‘otherness’, his being utterly different from any familiar object, and in particular his being outside space and time, then the answer to this question must be ‘no’. (Le Poidevin (Reference Le Poidevin1995), 498)
This is clearly in tension with the conclusion that the question of the existence of a timeless God ‘would best be treated as a question about the advisability of adopting the theistic framework, and not about the reality of God’ (Le Poidevin (Reference Le Poidevin1996), 133).
Le Poidevin's discussion does not, then, support either his claim that in order to claim meaningfully that God exists one must take God to be a putative entity within space-time, or the more general lesson that instrumentalism about theology is viable. I will close with some additional remarks concerning Le Poidevin's recommendation to take God to be a putative entity within space-time.
First, and perhaps less seriously, it is not clear that Le Poidevin's response to Carnap differs significantly from the ‘internal strategy’ he dismisses, namely the response that grants Carnap's argument and construes the question of God's existence as internal to the persons framework. Le Poidevin's position makes the question ‘external’, namely external to the theistic framework, but as mentioned, this still leaves the question internal to the spatio-temporal ‘us-framework’. Perhaps this is as intended, since Le Poidevin (Reference Le Poidevin1995, 492) means to dismiss the internal strategy only conditionally on the success of Carnap's argument (‘the answer will not (if Carnap is right) give us what the theist wants’). But the methodology for determining the answer to the question which Le Poidevin associates with the internal strategy is just what one would expect in the case of his own strategy. It involves using logical or empirical procedures to construct and evaluate arguments for the existence of God, in order to determine whether there is a person such as God. Determining whether God is part of the spatio-temporal ‘us-framework’ presumably involves just such methods.
However, and this is the second worry, it is not clear how likely the methodology appropriate to the spatio-temporal framework is to turn up a positive answer to the question (and thus not clear how good a piece of advice for theists Le Poidevin's recommendation is). I am not here concerned with traditional theological arguments for taking God to be a putative entity within space-time (or time) as opposed to outside of it. There may be independent reasons for taking God to be within space-time. But as I have argued, for all we are told, Le Poidevin's spatio-temporal framework is identical with Carnap's things framework. And in order to be real in the sense of that framework, God would have to be such that s/he can be ‘incorporat[ed] … into the system of things at a particular space-time position so that [s/he] fits together with the other things recognized as real, according to the rules of the framework’.Footnote 13 If we do adopt Le Poidevin's ‘ontological parochialism’, believing only in the reality of objects that are straightforwardly spatio-temporally related to and of the same kind as us, the result may just be atheism (not only with respect to a timeless God, but tout court). Of course this too may be as intended in some sense (the book is entitled Arguing for Atheism after all), but it clearly differs from Le Poidevin's official conclusions.Footnote 14
