Robert Stalnaker’s metaphysical thinking is distinctively modest in two respects. First, while modal logic and its standard model theory are taken seriously, Stalnaker is careful to distinguish between artefacts of the model (unactualised possibilia) and what one is committed to in light of how one reckons the actual world to be. Second, and correlatively, standard metaphysical entities are not theorised as if they were independent objects whose properties are to be discovered but as consequences of the demands made by our descriptive and explanatory frameworks. In this light, Stalnaker offers less a theory of propositions and more a way of understanding the conditions under which we can talk about propositions without any further specification of their nature. It has been noted previously, not least by Stalnaker himself, that, in a sense, Quine resuscitated metaphysics from empiricist depredations by allowing for a realist characterisation of ontological commitment beyond epistemological or otherwise a priori conceptions of what there is. Stalnaker seeks a return to such a stripped-down attitude that is at odds with much contemporary neo-medievalism. The present volume nicely encapsulates this general approach.
The last 60 years or so of semantics and its relation to metaphysics have been shaped by a rejection of Quine’s austere naturalism, which denies a role for propositions, properties, modalities (possibility/necessity), and, perforce, any logic that trades in them. Stalnaker comes to praise Quine, not bury him further, although with a distinctive twist. He offers a theory of propositions with an attendant logic that promises to keep to Quine’s naturalistic strictures but makes use of the kind of conceptual resources Quine rejected. In a nutshell, Quine’s methodology and the suspicions it enjoined were correct, but the package of commitments Quine rejected on such bases can be reconstructed via the very methodology that led Quine to the contrary commitments. This is all very welcome.
The volume is brimming with insight, delivered with Stalnaker’s tidy style: not a word is wasted without any decent into oracularity. It essentially divides into two. The first half deals with broad theoretical motivations and the conceptual nuts and bolts, while the second half offers various technical details and proofs concerning the logic and model theory of the system Stalnaker sets forth. Here, I shall restrict myself to the first half, which should be of greatest interest to linguists and philosophers of language. I shall offer some comments and queries as I go.
Chapter 2 sets out the Quinean legacy. Like Carnap, Quine took questions of ontology to be settled by the framework within which one theorises. Unlike Carnap, Quine considers the choice of framework not to be a simple decision but an attempt, in medias res, to regiment the way we ordinarily think, talk, and theorise about the stuff around us to arrive at a limning of the most general traits of reality. Thus, substantive ontological questions as to what there is are not separable from the choice of framework, which is one way of denying the analytic/synthetic distinction. For Stalnaker, Quine is ambivalent as to whether the goal of such regimentation should lead to a comprehensive and integrated general theory of reality or a more modest pluralism of frameworks, which does not rule out a comprehensive integration, but which does not presuppose still less entail it (14). What is not in question is that Quine and Stalnaker in his turn are realists in the sense that they register a distinction between the truth of a framework (or the claims made via it) and the reasons for endorsing it. At any rate, Stalnaker commends realism and modest pluralism.
Stalnaker is especially good on regimentation. Regimenting ordinary language into fist-order logic is not the discovery of its underlying logical form, or even necessarily the capture of what is said by speakers of the language; rather, it paraphrases the quotidian idiom into an extension of it that provides an explicit rendering of how we take the world to be in line with the developing regimented theory. In this sense, the analysis can be shallow, providing just enough deductive structure for the explanatory purpose at hand. This sort of bootstrapping approach has no a priori justification; it only has the worth of its products to commend it, such as the very development of first-order logic.
Also crucial for Quine on Stalnaker’s view is the division of regimentation between a logic proper, which is topic neutral and only raises the question of ontological commitment via the quantifier-variable relation, and the set of admissible predicates, which engenders a domain of entities that the predicates might describe. ‘Ontological commitment’, therefore, ‘cannot be separated from the decisions about what predicates we judge to be acceptable’ (23). So, bound variables give us the criterion for what exists, but the question of what there actually is remains unsettled.
Stalnaker goes into some classically thorny issues concerning identity, sets, and empty names to highlight the nuances of this approach and also to register some revisions, especially regarding the utility of free logic. One sin of omission in Stalnaker’s discussion is Quine’s elimination of variables at the hands of his own predicate-functor logic or earlier Schörnfinkel techniques or later cylindrical algebras. For Quine, variables are merely a convenient notation (‘abstractive pronouns’) to track the dependence of predicates on the objects of the domain. Thus, Stalnaker is absolutely correct to give priority to predicates, not variables, in the discernment of ontological commitment; later, Stalnaker is clear, reframing Quine’s famous slogan as ‘to be is to be subject of predication’ (129), and all variable binding can be reduced to a predicate-forming operator. Thus, the very appeal to first-order logic or any variable-involving language as a topic-neutral system is inessential; what is essential is the relation between predicates and a domain, but we have no steer on the nature of the denizens of a domain from logic alone.
Chapter 2 spells out a theory of propositions that cleaves to the Quinean methodological precepts. For Stalnaker, the fundamental objection Quine has to propositions is not that they are abstract or conflict with parsimony but that, if propositions are the meanings of sentences, then they presuppose synonymy as the relation that unites the equivalence class of sentences upon which propositional identity rests (49–50). But Quine thinks no good sense can be made of such a relation, because of indeterminacy and the instability of the analytic/synthetic distinction. Stalnaker effectively follows Quine in this reasoning, only departing from him in thinking that while propositions are expressed by sentences, they are independently explained and justified. So, ‘language presuppose[es] the notion of a proposition’, a proposition just being ‘objective information’ that language is used to express (50). In this light, a theory of propositions is not obliged to resolve the complexities that have traditionally burdened the concept concerning the individuation of sentence meaning and attitude content. To the contrary, a regimented language might be constructed along Quinean lines and be taken to express propositions ‘where propositions themselves might or might not be among the things in the world the language can talk about’ (54).
Stalnaker’s fundamental point is that the sentences of a regimentation should compositionally express propositions (although propositions are independently understood as not mere sentence meanings), and if so, propositions should be among the things we can talk about or have a commitment to. On Stalnaker’s view, as explained, this issue turns not on quantification but on the admissible predicates of the language. Chapter 3 lays out the core idea, which is that a predicate should express what condition must hold for an arbitrary individual to satisfy the predicate (73). Further, for two predicates to have the same compositional ‘theoretical role’ is for them to be necessarily coextensive; merely extensional equivalence will fail to distinguish properties contingently unsatisfied, for example. These conditions establish propositions as simply the conditions for predicates to be satisfied.
The rest of the volume is concerned primarily with metaphysical issues arising from this view, with Stalnaker defending a version of contingentism, the thesis that some extant things might have failed to exist and other things of such kinds might have existed instead. For Stalnaker, this is of a piece with Quinean naturalism, where the framework of regimentation on offer makes only conditional claims as to what there must be, given what predicates are judged admissible, but our choice of the admissible predicates is based on our explanatory aims, given how we reckon the world to be; it is not simply a function of the semantic framework itself (152).
In general, Stalnaker does not defend the overall program from doubts or alternatives but is concerned with spelling out its consequences in various metaphysical disputes. One might worry, for example, that no clear sense can be made of ontological commitment, given that neither quantifiers nor predicates nor the bare idea of a domain suffices. As Stalnaker notes, ontological commitment is not separable from ‘decisions about what predicates we judge to be intelligible’ (23). Quite. But we have no neutral sense of what is intelligible in the shape of a regimentation; at any rate, regimenting natural language appears to tell us little about the language itself or how to decide what counts as intelligible, for thinkers as easily take, say, fictional entities as intelligible as not. Such worries do not threaten to scupper Stalnaker’s project but more indicate the pluralism he endorses, a pluralism not inconsistent with realism.
Overall, Stalnaker’s overarching thesis is not entirely new: naturalism need not be wedded to extensionalism. Stalnaker’s fundamental claim, though, is that where we depart from a blanket extensionalism matters, and it should be at the level of the predicate, which records ontological commitment. As pointed out above, this is a move Quine himself fully recognised in his ‘elimination’ of variables, but what he failed to see is that being a predicate is a modal notion. Figuring all this out for current theories of propositions and various metaphysical theses will undoubtedly generate much profitable work.
