A few weeks ago I argued first that Chalmers’s conception of internal and external questions bore little relation to the Carnapian one it’s supposed to explicate, then that the Chalmers version is actually incompatible with the Carnap one. Chalmers says Carnap’s internal-external distinction needs to be replaced (p. 80 of his paper in the Metametaphysics volume) because the idea of a framework is too philosophically tendentious to be allowed to burden that distinction, which must therefore also be replaced by a supposedly more neutral one. In my earlier posts, I focused on Chalmers’s replacement of the internal-external distinction, and bracketed the (in a sense more fundamental) question of his replacement for Carnap’s notion of a framework. That it needs some replacement I take to be obvious; you can’t very well have any notion of “internal” if there isn’t something for concepts or questions to be internal to, a representational medium of some sort in which questions or concepts can be stated. Carnap called that medium a (linguistic) “framework.” Why does Chalmers consider this tendentious? Not clear.
But Chalmers does actually, later on in his paper, propose a replacement for Carnap’s conception of a framework, and here we begin to see what he might have had in mind. Instead of frameworks, he proposes a broader notion of “domains” (or “contexts of utterance”) that list the objects existing in possible worlds. Each of these domains consists, in his proposal, of a corresponding “furnishing function” that specifies the list of nouns, or, in Chalmers’s preferred idiom, the “ontology” of its world:
Intuitively, a furnishing function is a technical counterpart of a Carnapian ontological [sic] framework. Different ontological views will correspond to different furnishing functions. There will be a nihilist furnishing function, mapping worlds to domains representing a class of simple entities that are taken to exist on a nihilist view. There will be a universalist furnishing function, mapping worlds to domains representing a class of simple and complex entities that would be taken to exist on universalist views. There will be furnishing functions that admit abstract objects, and furnishing functions that admit only concrete objects. . . (p. 108)
The idea then is that the truth-conditions of an ordinary existence assertion are determined by the (furnishing function more or less automatically specified by the) context of utterance (pp. 109-11), while the truth-conditions of an ontological utterance will depend on various conditions of “admissibility” for furnishing functions, such as consistency constraints. This just shifts the problem; Chalmers himself admits that in those terms, the issues among various ontological views “will all recur as issues concerning the proper constraints on admissibility” (p. 113). Looser constraints on admissibility will result in wider ranges of indeterminacy of ontological claims (and thus in various gradations of anti-realism, i.e. skepticism about ontological claims), while tighter constraints lead to more determinate ontological claims.
For Carnap, a framework was a candidate language of unified science (i.e. for all knowledge), while Chalmers’s “domains” determined by furnishing functions result merely from various gradations of ontological assertion, unconnected to any larger bodies of knowledge (pp. 114-16). The point of classifying a question as “internal” to a framework, for Carnap, was to regard it as, in principle, answerable, with the resources specified by that framework — i.e. to distinguish what we can in principle know from what we in principle can’t. In Chalmers’s terms, the Carnapian constraints on admissibility are explicitly supplied by the framework itself, and have no need of any supplementation by ontological fiat. So from a Carnapian point of view, the best sense one can make of Chalmers’s supposed “replacement” for the internal-external distinction is that he is attempting to create a space for a “third realm” of statements that are neither answerable in the cut-and-dried, scientific or mathematical sort of way (i.e. internal), nor are fully indeterminate (i.e. external), but, let us say, possibly-answerable by looser constraints in a not-quite-scientific, ontological dialect of ordinary language (and the intuitions it supports) that is somewhat regimented but whose boundaries are unclear (Chalmers refers to Cian Dorr and Ted Sider in this context, p. 100). Between properly behaved frameworks and the outer space of indeterminateness, that is, Chalmers wants to introduce a space for quasi-frameworks.
Now this sounds very loosey-goosey and anti-Carnapian — but Carnap was actually willing to contemplate an alternative to his own preferred crisply-defined, forbiddingly Fregean frameworks. He was sympathetic to Neurath’s proposal of a scientific vernacular or “universal vernacular” as a lingua franca among scientists in all fields and scientifically-inclined practitioners, in which the meanings of all terms to which scientific attention has been given are to be understood not in their ordinary-language senses but in their explicated ones (in Carnap’s sense of “explicated”). Now this language was of course much messier than Carnap’s own, explicitly specified “frameworks.” There would still be unexplicated “clots [Ballungen]” in these Neurathian kludges for which no explication was available in any part of science, nor were even the explicated parts deductively unified in the way Carnap preferred his “language of science.” Still, Carnap was willing (as Thomas Uebel has eloquently stressed on various occasions) to compromise with Neurath to the point of countenancing such a rough-and-ready lingua franca as a provisional medium for ordinary human, especially for scientific and political, communication.
However, Neurath’s loose vernacular, even with its Ballungen, has much clearer and more sharply defined criteria of membership (i.e. of Carnapian internality and externality) than Chalmers’s “domains.” At least the users of the scientific vernacular are clear that the Ballungen are prime candidates for future explication, while Chalmers’s ontological interlocutors advance no further than to stake their claims relatively to each other. There is no program for rational adjudication of these claims, no room for convergence on resolutions to their disagreement — nor even a sense that either the disagreement or an eventual resolution would matter for anything in the wider world outside what Chalmers calls the “ontology room.” And remember that Chalmers allows vague intuitions, in some cases, to determine whether a statement is internal or external (whether it is, in his supposedly neutral replacements, “ordinary” or “ontological”).
In short, nothing been preserved of Carnap’s internal-external distinction. The explicatory replacement Chalmers proposes for it bears no relation to the explicandum; whatever purpose it may serve, it does not even attempt to distinguish answerable from unanswerable, tractable from intractable, fruitful from sterile questions.
One thought on “What is a “framework,” and why does Chalmers have an opinion about this?”
“Chalmers’s “domains” determined by furnishing functions result merely from various gradations of ontological assertion, unconnected to any larger bodies of knowledge (pp. 114-16).”
Now THIS seems like an assertion. Chalmers clarifies, “A domain is then admissible at a world if its domain sentence is consistent with that world” (p. 114) and also says that this consistency can be checked through standard logical rules: “We can instead suggest that S1 and S2 are consistent when they are logically compatible with a certain subset of necessary truths. To say that sentences are logically compatible is to say that no contradiction can be derived from them in a standard logical framework, in which logical rules for the quantifiers are exhausted by the standard introduction and elimination rules and interdefinitions” (p. 115).
That doesn’t sound like it’s “unconnected to any larger bodies of knowledge” to me. Nor does it sound like “a not-quite-scientific, ontological dialect of ordinary language (and the intuitions it supports) that is somewhat regimented but whose boundaries are unclear”. Is logical consistency really less clear than colloquial “ballungen” for you?
Comments are closed.