Quine’s Metametaphysics

W. V. Quine stands out as one of the foremost figures of 20th century analytic philosophy. This chapter aims to show that a significant part of his work’s enduring value lies in its contribution to metametaphysics, which will include showing how some more contentious aspects of Quine’s thought can be seen as indispensable to it; we will problematise the widespread belief that one can isolate basic elements of Quine’s metametaphysics without eroding their warrant.


1: The view from a distance
There is widespread agreement that Quine's ideas, for better or worse, had a substantial impact on metaphysics, in particular on that part of the discipline called metametaphysics.i The term seems to postdate the period of Quine's greatest influence, though not Quine's life as a whole; ii metametaphysics has, however, come into greater focus in the 21 st century, prompted especially by the 2009 collection Metametaphysics.That collection in turn was partially inspired by Peter van Inwagen's 'Meta-ontology' (1998), which purported to articulate Quine's methodology for ontology and thereby to expose the foundations of a popular -by some lights, the dominant -tradition in analytic philosophy.
Quine's contribution, however, is often reduced to one paper -'On what there is' [OWTI] (1948) -treating it as the locus of critical consideration of method in analytic metaphysics.This view is not restricted to those merely acquainted with Quine, having been encouraged even by those who are intimately familiar with his work.For instance Hilary Putnam, who studied under and engaged extensively with Quine, writes that when Quine published a famous paper titled 'On what there is'... [he] single-handedly made Ontology a respectable subject.(2004: 78-9) iii This focus is understandable -without doubt the paper is significant -but can lead to the assumption that the wider body of Quine's work is irrelevant.It can also slip into a yet narrower focus which further reduces Quine's contribution to metametaphysics to a few easily misinterpreted sound-bites.Analytic metaphysicians are familiar with the adage '[t]o be is to be the value of a variable' (Quine 1948: 34), typically called the Quinean criterion of ontological commitment, but this brief claim may conceal much, and may even lead to trouble if we rely on the radically incomplete picture it provides.So I will claim.
Here, therefore, is our strategy: unpack that sound-bite along with other elements of Quine's metametaphysics from OWTI ( §2), before discussing ( § §3-5) the metametaphysical theses that feature elsewhere in the 52(!) years of work that Quine published after this seminal paper.This helps us to evaluate the viability of adopting elements of Quine's metametaphysics piecemeal.It is hoped by many analytic metaphysicians that this can be done easily; I contend that it's far from easy.
Two brief warnings before proceeding: first, for those familiar with the debate, we will be exploring Quine's metametaphysics, not Quinean metametaphysics (whose locus is van Inwagen (1998van Inwagen ( , 2009)), but see also Eklund 2006, Jenkins 2010and Berto and Plebani 2015).The task of comparing these must be left for elsewhere (on this subject see Price 2009; see also my 2015).I will therefore take a more historical focus than has been the norm.By approaching the question of the methodology for metaphysics via Quine's developing body of work, we will be able to focus more heavily on Quine himself than on self-proclaimed Quineans.
Second, as we explore later, metaphysics and metametaphysics aren't strictly separate enterprises for Quine as they might be for other thinkers, primarily because of his holism, but insofar as is possible without going astray, we will here put aside Quine's specific metaphysical views.More in-depth exploration would involve considering Quine's attempts to dispense with properties, possibilia, etc., but we'll have to leave that for elsewhere.

2: 'On what there is'
A curious thing about the ontological problem is its simplicity.It can be put in three Anglo-Saxon monosyllables: 'What is there?'It can be answered, moreover, in a word -'Everything'.(Quine 1948: 21) Thus begins OWTI -with a deceptively simple statement about how to understand the ontological/ metaphysical project.We want to know what there is, or what exists.And we can say what there is without engaging in any difficult work, as long as we don't mind trite (facetious?) answers -everything exists.The problem, as Quine acknowledges, is that we don't know what that 'everything' comprises, so we must put in more hours.
Of course it's not that simple anyway -objecting to the one-word answer, someone might insist that not everything exists -not, say, Pegasus, or round squares.At first Quine seems stuck should he deny that some kind K of entity exists, because even saying 'Ks do not exist,' seems to admit that Ks are something.Were this the case, Quine's only way to resolve ontological disputes in the negative would be to refuse to say anything whatsoever, but this looks unsatisfying, indeed implausible.Surely it's coherent to deny that Ks exist!To see why non-existent objects don't disturb Quine's easy answer, why he can insist on this simple statement, we must delve further.
The solution is to invoke quantification.Quine follows Russell (1905): there's nothing strange in saying 'The current King of France does not exist,' provided we treat this as disguised quantification.We don't intend to say 'You know the current King of France?It turns out he doesn't exist,' but rather 'There is no such entity as the current King of France,'.To speak more formally, what we mean to say can be represented as '¬ x(KoF(x)) ∃ ' or equivalently ' x(¬KoF(x)) ∀ '.We can conclude our ontological task negatively by using a description and stating either that there is no such thing, or that each thing there is fails to satisfy the description.This is the significance of the aforementioned phrase 'to be is to be the value of a variable' -to say that something of kind K exists just is for it to be a replacement-instance of a bound variable in a statement that says, or entails, ' x(Kx) ∃ .' This is a basic, but crucial, component of Quine's metametaphysics.It forms the beginnings of a methodology, accepting which cuts off a range of putative ways of investigating ontology and renders the questions 'Are there Ks?', 'Do Ks exist?' and 'Is " x(Kx) ∃ " true?' equivalent.Other notions in the vicinity, like 'subsistence', are either subsumed within this equivalence or disallowed as having no clear sense.This is the core of how Quine 'made Ontology a respectable subject' -by formulating a proposal about how to interpret the ontological question and thereby limiting the threat of disputants talking past one another.In order to disagree with Quine on ontology, one has to either lay out their differences in terms of entities quantified over (which requires that they regiment their language enough to enable an answer to the quantificational question) or reject Quine's conception of ontology outright.iv But this treatment of ontological debate is far from complete.For all the above version of the criterion says, mere use of the term 'K' might be enough to make the statement ' x(Kx) ∃ ' true, and thereby to make the statement 'Ks do not exist' a contradiction.Perhaps mere ability to meaningfully use 'K' demonstrates that Ks are something.This is not an option Quine wants available, so we need a second, stronger statement of the Quinean criterion for ontological commitment: a theory is committed to those and only those entities to which the bound variables of the theory must be capable of referring in order that the affirmations made in the theory be true.(1948: 33, my emphasis) This adds a further condition -if you don't need to speak of Ks, you are not committed to Ks' existence.We should accept only the minimum number of entities (or kinds) to allow our theory to be true.To maintain my overall theory I don't need to be capable of referring to anything fitting the description 'current King of France' because I need the description only in the context of disputes with those who are confused about history and/or politics, and ' x(KoF(x)) ∃ ' is by no means a consequence of my claims in such disputes because I simply provide the condition King of France (or entity identical with Pegasus) and charge my opponent to find the entity meeting the condition in the list of things 'to which the bound variables of the theory must be capable of referring'.
We find ourselves, then, in a position to represent ontological debate.Two interlocutors start with a term 'K', and one claims to hold a theory which must accept (explicitly or implicitly) statements like ' x(Kx) ∃ ' in order to be true, while the other does not.This also permits some understanding of how to resolve ontological debates: for when we investigate what sentences require the truth of ' x(Kx) ∃ ', we may find that those sentences have different consequences from those first suspected, or that we don't wish to accept them after all.In the simplest cases we can avoid foolish debate about the existence of, e.g., sakes, or heebie-jeebies.Nevertheless much remains unanswered about how we resolve ontological debates.
OWTI closes with some relevant remarks that are seldom considered with care -Quine asks how we adjudicate between rival ontologies, and after initially saying that adopting an ontology is 'similar in principle to our acceptance of a scientific theory' (ibid.: 35) he goes on to say that we may pursue many options in tandem, and in the absence of strong reasons to prefer one ontology, 'the obvious counsel is tolerance and an experimental spirit' (ibid.: 38).Unless we encounter trouble with one or the other, we can allow them to sit side-by-side as alternative theories.This result might look unwelcome: for in the debate over the existence of Pegasus, I'm not satisfied with the result that I can now maintain its non-existence if this just means that my opponent and I retreat to our respective theories and mind our own business.After all, we are opponents; each thinks that the other ought to come around to their way of thinking.Furthermore, the mention of tolerance naturally evokes Carnap's 'principle of tolerance' (see, e.g., Carnap 1937, §17), a pluralist position which has been generally understood as seriously in tension with Quine.
In OWTI, Quine says little by way of clarification.Alongside his gestures toward scientific theory, he also makes what looks a merely dogmatic claim: he says that an 'overpopulated universe … offends the aesthetic sense of us who have a taste for desert landscapes' (ibid.: 23).Relative to its actual significance, this quote has played a greater role in establishing Quine's reputation than any other.While Quine's vague remarks about tolerance are forgotten, his vague remarks about sparseness are taken with utmost seriousness -hence the widespread view that '[t]he Quinean method is eliminativist by design' (Schaffer 2009, 372), inherently biased in favour of casting out entities.If we work only on the basis of what is given here, though, it's equally consistent with Quine's metametaphysics to exercise tolerance!The degree of openness found in OWTI may be virtuous (to form the foundation of a broad school, one might think, it would have to be consistent with various developments).Nevertheless Quine himself certainly had a more developed programme that manifested elsewhere in his work, so to treat his remarks about a 'desert landscape' as crucial is as short-sighted as it would be to take on wholesale his remarks about tolerance.Instead we should broaden our horizons and consider Quine's more mature work.

3: 'Two dogmas of empiricism'
In order to develop Quine's metametaphysics beyond OWTI, we will first consider perhaps his only work that rivals it for fame: 'Two dogmas of empiricism' [TD] (1951a).Here Quine famously critiques empiricism through its 'dogmas' of analyticity and reductionism, and introduces the epistemological holism that is now well-known under the label of the 'web of belief'.Why, though, insist that this work is relevant to metametaphysics?First, it develops OWTI's merely gestured-at 'scientific' attitude to our metaphysical (among other) theories, so if we want to clarify this, TD is a good place to look.This might seem outweighed by the fact that the dogmas introduce new philosophical territory: we're now engaging in philosophy of language (analyticity) and epistemology (reductionism), so why bring in metaphysics?But aside from the fact that the holism here introduced shows that precise subject divisions are not Quine's style, it should also be clear that it's part of investigating the nature of metaphysical questions to ask how one comes to know their answers, and what it means to ask/answer one.Interestingly, Quine later describes himself as doing 'the epistemology of ontology ' (1983, 500).This is natural, for we can't properly understand a research programme without some notion of what would constitute a significant result within it.Let us, then, examine what TD introduces to Quine's metametaphysics.
Here, as elsewhere, Quine demonstrates a complex relationship with the empiricist tradition.On one hand, Quine is widely held to undermine at least one strand of the empiricist project -logical positivism -and TD substantially contributes to that.On the other, Quine clearly has extensive sympathies with their project: in the closing section, titled 'Empiricism without the dogmas', he stresses '[a]s an empiricist I continue to think of the conceptual scheme of science as a tool, ultimately, for predicting future experience' (41).Later he is even more committal: I haven't thought of myself as destroying [logical empiricism, but] as contributing to what it seemed to me needed further development … What I was taking issue with was pretty much, I think, in the domain of logical rigour, and also of being more completely empiricistic.And certainly I felt that I was insisting on the ideals of the Vienna Circle more than they, and saying what I thought they ought to be saying.(Quine and Fara, 1994) To see how the tension can be resolved, we must examine TD's move from empiricism to naturalism.
In OWTI Quine already demonstrated commitment to empiricist ideas: when speaking of deciding between ontologies, he sees the 'phenomenalist' option as important because it takes 'epistemological priority ' (1948: 38).Quine's conflicted comparison of phenomenalism (privileging sense data) and physicalism (privileging physical objects) showcases his empiricist scruples: the former has better epistemological credentials because its most basic ingredients -sense data v -are immediate to experience, but the physicalist conceptual scheme inherits the virtue of association with the successes of physics despite relying heavily on unobservable theoretical posits whose very positing requires extensive assumptions.This conflict continued to bother Quine and in TD the reason emerges.In critiquing the 'dogmas' of analyticity and reductionism, Quine crystallises a growing loss of faith in the idea that our words, concepts, or sentences can be understood as being traceable back to their ultimate implications for our experience.We'll briefly consider why.
Quine starts by examining the notion of analyticity, or truth in virtue of meaning.The idea that some sentences are known to be true simply in virtue of the meanings of their constituent words, like 'Vixens are female foxes,' or 'Bachelors are unmarried,' has a long history, and played an important role for the logical positivists.For them it validated a distinction between empirical knowledge, whose source is clear (the senses), and apparently a priori knowledge.The latter had been an empiricist stumbling block because it was mysterious how such knowledge was acquired if not through the senses -but if based purely on meaning, analyticity would involve no special content, being founded merely on linguistic competence.However Quine finds truth in virtue of meaning impossible to unpack properly, since (i) we lack reason to believe in special discoverable entities, meanings, appeal to which would be a marker of truth, and (ii) the notion of sameness of meaning, which is needed to isolate meanings, is hard to clarify.We cannot make the notion more manageable by applying epistemic standards, e.g.universal willingness to assent to the sentence under the same conditions, since this doesn't exclude empirical claims that inspire universal assent (compare 'Vixens are female foxes' and 'There are dogs').After discussing several other candidates for making sense of analyticity, e.g., semantic rules, Quine concludes that the notion cannot do the work the logical positivists require of it.
The way this feeds into Quine's metametaphysics becomes clearer with the second dogma: reductionism.Quine insists that we must abandon the idea that there corresponds to each statement a selection of possible empirical data that count as the evidence for or against it.If this were true one would be able to definitively give each statement's implications, but it doesn't survive scrutiny.Even for apparently purely observational statements, I might reject parts of my theory to hold those observational statements constant, or vice versa.If I make an observation of a neutrino that implies that it travelled beyond light-speed, I must choose between abandoning an important, deep-seated principle of my theory (that faster-than-light travel is impossible) and claiming that my observation was faulty somehow.I will choose based on the circumstances and wider implications, and for Quine, the rational choice is the path of least disturbance.The manoeuvres needed to settle the disturbance of rejecting what one has seen will often be less drastic than those needed to revise a central principle, though the balance may shift if enough observations accumulate.This is a statement of epistemological holism: we cannot rationally accept or reject any claim without reference to our wider theory, so the siloed enquiry reductionists require is a distortion.Quine regards reductionism and analyticity as really the same dogma viewed from different angles: just as observing that matters of meaning and of fact both contribute to a statement's truth can trick us into thinking we can isolate that sentence's meaning-giving and fact-stating aspects, awareness that observations bear on some statements more than others can trick us into thinking we can pinpoint what's necessary to finally confirm a statement and conclusively settle its truth value.Rather everything faces the ongoing test of coherence with our best theory, which is overall science -with this step empiricism develops into naturalism.This importantly applies to all statements, including apparently metaphysical ones: we cannot perform any study into the nature of things that floats free of the enquiries comprising our broad science, but just as there can be no isolated metaphysical enquiry, there can be no wholesale ruling-out of metaphysics.This aspect of Quine's work was taken to scuttle the logical positivists' project, since their verification principle had ruled out metaphysical statements as meaningless because neither verifiable nor meaning-giving.Now, on Quine's picture, one cannot rule out a category as meaningless in advance since a lack of relevant implications for the remainder of our theory cannot be guaranteed.Yet importantly, a version of the verification principle survives, applied to whole theories: later, Quine asks of a hypothetical, completely empirically confirmed, theory, 'In what sense could the world then be said to deviate from what the theory claims?' and his answer is simple: 'Clearly in none ' (1981, 22).
What, then, should we now add to our stock of ideas from Quine on the nature, or methodology, of metaphysics?I think it is clear: In TD, epistemological holism becomes inevitable because naturalism becomes a key part of Quine's metametaphysics.In saying this I go against van Inwagen especially, who says that Quine's attitude to science was 'a consequence of certain of his epistemological commitments and not of his metaontology ' (2009: 506, n. 53).As I've already mentioned, that this is epistemological is no reason to suppose that it's not metametaphysical.
And we need it in order to understand two points briefly introduced above.Without naturalism we get neither the result, important for Quine, that metaphysical statements are (within certain restrictions) candidates for inclusion in our theory, nor the result that metaphysical statements are meaningful only insofar as they bear genuine connections to the rest of our theory.For the holism required by Quine's naturalism tells us that all respectable knowledge is continuous with science, which is governed by what is often called the Quine-Duhem thesis.
We've already uncovered a significant additional principle in Quine's metametaphysics by looking beyond the core of OWTI.Now we'll move on to an idea that's developed more extensively in Quine's great constructive project, Word & Object.

4: Word & Object
In Word & Object [WO] (1960), Quine develops the insights of TD with a more constructive focus.The book centres on a project of rational reconstruction -through the thought experiment of radical translation, Quine considers what a stranded linguist could learn about a wholly alien language, with this developing into an account of the theory that could be regarded as underpinning our own language.The lack of fixity in this theory is the notorious finding of the indeterminacy of translation.
We'll postpone exploring this, though, as on the way we find another idea of metametaphysical importance introduced.So far we've just seen that the ontological question requires us to establish what we must quantify over if our statements are to be true, and that our statements are to be taken as a corporate body rather than divided into independent clusters.But who's to say that we shouldn't take large quantities of the statements we habitually utter to be false?Our corporate body of truths might be very small, very distant from what's typically taken to be true, or both.However, in a further development of Quine's naturalism, we see in WO an attempt to preclude that.Our construction of an overall theory needs a guiding principle, otherwise we can make sense of nothing at all -and that is that by and large people are getting things right.This guiding principle is the principle of charity.Some would be surprised to see this treated as central to Quine's ideas, since it has been more often associated with the work of Donald Davidson (see his 2001).However it is indeed important for Quine.When exploring the creative revisions required in the course of radical translation Quine says that '[t]he maxim of translation underlying all this is that assertions startlingly false on the face of them are likely to turn on hidden differences of language ' (1960: 59), and in an accompanying footnote identifies this as the principle of charity.He goes on to claim that 'the more absurd or exotic the beliefs imputed to a people, the more suspicious we are entitled to be of the translations ' (ibid., 69).The idea is that since interpreting someone requires me to attribute beliefs to them, I must impute some degree of coherency and rationality to them in order for the attributions themselves to be coherent or rational.Otherwise I will have no reason to hypothesise even that two occurrences of a symbol are more likely to signify the same than are two occurrences of different symbols, and without such hypotheses I can't even start interpreting.
As Davidson helpfully puts it, there are two directions of pull to charity.We need to assign sense to utterances, and that means assuming some sharedness in what is believed true -'we must maximize agreement, or risk not making sense of what the alien is talking about ' (2001, 27), but we also need to understand why certain sentences and not others are assented to, and this means imputing rationality -'we must maximize the self-consistency we attribute to him, on pain of not understanding him' (ibid.).So if I take there to be a distinction between hawks and handsaws, but on my interpretation my interlocutor recognises no such distinction, then ceteris paribus my interpretation is probably wrong.On the other hand if my interlocutor seems to routinely display some attitude but my interpretation attributes to them something radically at odds with that attitude, that suggests that ceteris paribus my interpretation is likely to be wrong even if I think the first, more internally coherent, attitude is obviously false.
Again, one might ask, why is this relevant to Quine's metametaphysics?The reason is that it advises us how widely our responsibilities range during metaphysical enquiry.We can't draw the limits of our theory within a restricted area, for instance saying that fundamental physics is the most respectable picture of reality so metaphysicians can safely ignore the question of how to reconcile it with talk of the macroscopic world.Within Quine's approach we may find talk of macroscopic objects to be misguided, but only with sufficient reasons to overturn the weight of our useful discourse about macroscopic objects.This also identifies more clearly what we're interested in when doing metaphysics: truth.We're not especially interested in entities, or essences, or fundamentality, but in what is the case.
One of the most significant aspects of the work Quine does in WO we have so far neglected -indeterminacy -but this is because it is more comprehensively explored elsewhere.While indeterminacy plays a significant role in WO, the response to indeterminacy is clearer in Quine's infamous paper 'Ontological relativity'.

5: 'Ontological relativity'
The thesis of the indeterminacy of translation is well-known in analytic philosophy, as is the thesis of the inscrutability of reference, and these ideas, introduced in WO, become the central concern in 'Ontological relativity' [OR] (1968).In order to see their role in Quine's metametaphysics, however, we must first clear up a terminological issue.It's easy to get the mistaken impression that the indeterminacy and inscrutability theses are fundamentally different, but while they should be distinguished, they are at root similar.Indeed Quine himself later indicates that his choice of words was unfortunate and that 'indeterminacy of reference' would have better stated what he wished to convey (1992: 50).We'll now see why.
The thesis of the indeterminacy of translation is that two translation manuals for a language might agree on how we would expect language-users to behave on the basis of their translations despite those two manuals being incompatible with one another.We're invited to imagine two radical translators, operating independently, who each generate hypotheses about how to translate the utterances of an entirely alien community and amend them based on observation, refining hypotheses until they build up a vocabulary they match with their own to render them capable of communicating with that community.Given the many ways that one could consistently interpret and systematise a community's behaviours, Quine sees it as implausible that there wouldn't be multiple translations that successfully tracked all behaviour yet differed somewhere in what they took statements to signify.
The thesis of the inscrutability/indeterminacy of reference is that two translations of a language might agree even on the above and yet diverge regarding reference.That is, what entities the translations take the language to be picking out can differ without one assigning any statement the value 'true' where the other assigns that statement 'false'.A simple example of this idea is complement-based interpretation.Take the sentence 'My cat is an animal': one can ask why this should be analysed as speaking about my cat rather than everything but my cat.The obvious answer, that the statement would turn out false if understood as speaking of the my-cat-complement (which is not an animal but a vast aggregate of physical space), goes nowhere because by stipulation the complement-based interpretation also takes predicates like 'is an animal' to range over complements.So when I say 'My cat is an animal,' I can consistently be interpreted as saying 'The my-cat-complement complement-is an animal.'Both sentences are true, but on the latter interpretation what I am talking about is not my cat, and what I am attributing is not an intrinsic property.Rather I am talking about everything but my cat, and attributing a clearly extrinsic property.vi   One view of the indeterminacy theses is that they have sceptical results: I cannot know what someone else means, nor even what they refer to!This would suggest that whenever I try to understand anyone I'm taking a leap, performing a new radical translation.But this negative interpretation, on which I might be succeeding at every task that seems constitutive of communicative success yet still be getting my translation wrong, sits poorly with Quine.Since Quine holds that language is inherently public -'[i]n psychology one may or may not be a behaviorist, but in linguistics one has no choice' (1992, 37-8) -the standards for communicative success we set are all that's required to be right.Rather than saying that either translator may be failing despite apparent success, Quine wants to say that both can succeed despite their different interpretations.Hence in both cases, we are not dealing with mere inscrutability (inability to tell), we are dealing with indeterminacy.
Furthermore, the indeterminacy of translation may be based on a controversial claim that linguists have felt they can challenge, but the indeterminacy of reference has a much sparser base.The former is a hypothesis about what remains unfixed by observations yet is implied by an interpretation.It may therefore be too restrictive, too permissive, or both.It may be too restrictive since the resources Quine allows his radical translators aren't all that rich, and good translators might do more to pin down which statements should come out true; it may be too permissive since it allows good translations that differ on remote parts of theory to pronounce on things toward which the community might bear no attitudes whatsoever.However the indeterminacy of reference is secured on simpler, technical grounds.We need only appeal to the basic representative machinery of model theory.vii If we can offer any model on which a theory comes out true, we can offer multiple models, some of which will assign to the expressions of that theory different referents.One such model would be mathematical; we can preserve truth while taking our ordinary-object terms to range over not medium-sized dry goods but the natural numbers, as long as we reinterpret systematically.This might ring alarm-bells -does Quine's metametaphysics now collapse, bringing the whole notion of his methodology to nothing?Quine is aware that the situation looks dire: We seem to be maneuvering ourselves into the absurd position that there is no difference on any terms, interlinguistic or intralinguistic, objective or subjective, between referring to rabbits and referring to rabbit parts or stages; or between referring to formulas and referring to their Gödel numbers.(1968: 200) This would be absurd if nothing else on Moorean grounds -we couldn't be justified believing this with anything like as strong a conviction as that rabbits are not rabbit-complements.Accordingly Quine does not think this the end of the story, just a twist.This is where the last of Quine's key metametaphysical principles comes in: ontological relativity: '[s]pecifying the universe of a theory makes sense only relative to some background theory, and only relative to some choice of a manual of translation of the one theory into the other ' (1968: 205).We make sense of reference relative to a background theory, in effect by accepting that theory as the metalanguage to the object language under scrutiny.
For Quine an absolute reference relation is untenable because any attempt to give it would simply set up more truth-preserving reinterpretations, but instead of conceding that reference is nonsense, Quine treats it as theory-relative.When speaking of interpretations of one theory in another, we treat the interpreting theory as fixed and understood; when speaking of interpretations of our own theory, we treat the theory through which we clarify the original as understood.There is then a difference between referring to rabbits and referring to rabbit-complements, because there is a difference from the perspective of our theory: as Quine playfully puts it, '"rabbit" denotes rabbits, whatever they are' (1992, 52: emphasis in original).Whether the theory is indeed understood will then depend on our success in satisfying the demands of (charitable) naturalism.
The scale of the metametaphysical implications should immediately be clear.We must hold our metaphysics relative to the theory in which we situate it.In one sense this isn't worrying -after all, it was our theory that we wanted to better understand.In another, though, it's highly disruptive.Much traditional metaphysics trades on the notion of investigating the deepest nature of things, with the assumption that this will lead to a result that is not parochial but all-encompassing.Certain notions are effectively incompatible with ontological relativity -fundamentality as featured in much metaphysics, for instance, since this is supposed to identify what is most basic in a special metaphysical sense intended to float free of standard theorising.Strictly speaking we might allow fundamentality a smaller role, as a theoretical term of the theory under discussion, but that would substantially change the notion in play.
But for Quine we have just recognised something that's obvious elsewhere: to ask what reality is really like … apart from human categories, is self-stultifying.It is like asking how long the Nile really is, apart from parochial matters of miles or meters.(1992,9) Furthermore, to return to our starting point, while we have moved far afield of the sparse beginnings of OWTI, OR really works from that base -by accepting that our interest is in quantification, and taking on the representative machinery, we furnish the tools for recognising that countless theories could systematise our language, provided those theories retained the right structure.The concerns motivating the principle of ontological relativity are not local to a later, special project of Quine's, but extend right back to the base that has been seen by some as uncontroversial.
Quine was aware throughout that his view re-imagined metaphysics -as indicated earlier, he usually avoided the term.Tellingly, when Carnap criticises Quine for using the term 'ontology' to describe his project, protesting that the term is meaningless, Quine responds that 'meaningless words … are precisely the words which I feel freest to specify meanings for ' (1951b: 66).As is gradually being recognised through reevaluation of the history of logical positivism and the Quine-Carnap debate by historians of analytic philosophy, Quine develops the logical positivists' project, abandoning the untenable goal of ruling out metaphysics wholesale and instead finding a niche for something legitimately describable as metaphysics.That niche is helping to clarify -picking up a term that held great traction for both Carnap (1947: 8) and Quine (1960: §56), to explicate -elements of our theory.Quine was aware of this trajectory from early on: he states in TD his intention to oversee both 'a blurring of the supposed boundary between speculative metaphysics and natural science' and 'a shift toward pragmatism ' (1951a: 20).Quine's introduction of his criterion for ontological commitment, supported by charitable naturalism and underpinned by ontological relativity, transforms the subject-matter of metaphysics: it moves from the pursuit of deep truths underlying all theories to the project of clarifying our theories without stepping outside them.

Conclusion
Far more could be said about we've explored above: it has been possible only to trace the shape of Quine's metametaphysics across several stages of articulation, without delving into either the opportunities it presents or the serious challenges it of course faces.However this first step has at least three dividends.First, by drawing out the implications of Quine's metametaphysics we illuminate it, allowing comparative work that is blocked if the theory remains a shadowy background presence.Second, we broaden our understanding of what metaphysics can be by acknowledging a metametaphysics that adopts a serious methodology despite circumscribing the subject's ambitions, opening up a space between 'fundamental' metaphysics and wholeheartedly deflationary projects.Finally, we introduce a challenge for those availing themselves of parts of Quine's machinery.If they embrace the more radical underpinnings, this is in itself interesting (and, I hazard, for the better!), but if they refuse, it raises the question whether they can construct a coherent alternative that retains the appeal of Quine's version of naturalism.