A new book from Cambridge University Press, Interpreting Carnap, edited by Alan Richardson and Adám Tuboly, contains some interesting papers that I hope I will get around to discussing here. I will start with one whose title is calculated to arouse, well, interest, shall we say: “Carnap is not against metaphysics” by Vera Flocke. (Just as one’s attention would naturally be drawn to a headline “Pope advocates contraception” or “Mike Johnson to propose mandatory teaching of evolution in public schools.”)

Flocke makes clear she has only the “late” Carnap in mind, i.e. the Carnap of the principle of tolerance, though she never makes that an explicit criterion for the lateness of the late Carnap; in her mind, the dividing line between early and late Carnap is marked by Gödel’s incompleteness theorems. And it’s certainly true that they did play a role (as discussed in great detail in the now-ancient papers by Steve Awodey and myself about this transition), but the only role she assigns them is that “Gödel’s discovery showed verificationism to be false” (p. 37). Hm, really? It’s not clear, though, that Carnap ever went in for verificationism, and if he did, it’s not clear what he might have meant by it. When he discusses verificationism in retrospect (in Testability and Meaning), he doesn’t actually say anywhere that he once endorsed it — as he does admit there e.g. to having adhered to finitism in the Aufbau.

Anyway, Flocke says that an opposition to metaphysics can be understood two ways; one can object to metaphysics as subject matter, or one can object to the methods employed; in the latter case, metaphysics can be made acceptable if the methods are changed. The early Carnap, she says, was a subject-matter anti-metaphysician. But the later, post-Gödel Carnap was a merely methodological anti-metaphysician, she argues, and since metaphysicians have changed their methods since his day, Carnap would approve of a lot of present-day metaphysics, such as the examples she gives in her final section.

She does not address the question how and why the pre-Gödel verificationist turned into the open-minded advocate of “a distinctive metaphysical methodology” (p. 45). Still, she discerns three differences between the “early” and the “late” Carnap; the late Carnap, she says, unlike the early one: (1) “distinguishes between framework principles and other sentences”; (2) regards analyticity as a “framework-dependent concept”; and (3) “distinguishes between internal and external statements” rather than regarding any sentence as meaningless that can’t be verified (“or falsified”). The principle of tolerance, it seems, doesn’t enter into it in her view.

Actually, only the second of these distinguishes the earlier from the later Carnap, and that only marginally. Regarding (1), the Carnap of the Aufbau certainly distinguished the components of the constitution system itself (the basis plus logic, including quasi-analysis; these are what individuate constitution systems) from other sentences. And regarding (3), Carnap distinguishes in the Aufbau between “scientific” and “metaphysical” realism. Something is real in the “scientific” sense, Carnap defines, if it can be constituted in the Aufbau‘s constitution system; otherwise it’s metaphysical. This distinction obviously maps very closely onto its successor, the ESO distinction between internal and external statements or questions. Regarding (2), analyticity was certainly framework-dependent in the Aufbau (constitution-system-dependent), but in the text of the Aufbau itself, this is not made explicit, nor is there even implicitly any room for alternative logical and mathematical frameworks. Shortly after the Aufbau was published, though, Carnap made “analytic equivalence” (definability from the rules of the system) the basis of his “New Foundation of Logic” composed in Davos in early 1929 (see my book, pp. 196-203, with the background on the Allgemeine Axiomatik on pp. 191-6), and also used this new approach and terminology when Eino Kaila visited Vienna a few months later. Whereupon Kaila, in his 1930 book about the Vienna Circle — actually just about the Aufbau — organized his discussion of the book around “analytic equivalence,” to Carnap’s delight (Ch. 8 of my book). So Flocke’s (2) doesn’t really distinguish the early from the late Carnap, either.

The almost universal consensus about what distinguishes early from late Carnap is that it was something to do with the principle of tolerance. To which I would add the larger context of what I’ve called the “Carnapian linguistic turn” (to distinguish it from the Fregean one highlighted by Dummett), which began with the well-known “sleepless night” in January 1931 that resulted in the first sketch of the syntax idea, the “Versuch einer Metalogik,” in which Carnap gave up his series of efforts to fuse Hilbert and Wittgenstein and went entirely over to the Hilbert side. And since “meaning” was therefore no longer an issue, of course the critique of metaphysics had to change, since the diagnosis of “meaninglessness” no longer had any traction in the new program (intended initially as purely syntactic). Carnap pointed this out when he first presented his new idea to the Vienna Circle, in June of 1931, and suggested a new approach to the critique of metaphysics. Without yet having invented the terminology, he replaced the classification of metaphysics as “meaningless” with its classification as framed in the “material mode of speech” (inhaltliche Redeweise). Carnap was realistic enough to understand that ordinary language was committed to the material mode, though, and that the Vienna Circle was not going to change people’s ingrained speech habits, so the material mode wasn’t condemned outright — it was portrayed only as dangerously misleading, since it suggested that the meta-scientific items it referred to (such as numbers, or functions, or causes) amounted to more than just language. The criterion of acceptability became translatability into the formal mode of speech, where only linguistic artifacts are referred to (including those of the scientific object language), not (extra-linguistic) things. So the material mode was okay for everyday use provided everything was translatable into the formal mode.

With this new criterion the diagnosis of metaphysics as non-cognitive, in the Syntax, became a matter of classification rather than a claim that it lacked “meaning” and was therefore “nonsense.” But when a straightened form of “meaning” was reintroduced a few years later with Carnap’s embrace of Tarski’s semantics, there was no concerted reversion to the previous exclusion of metaphysics for reasons of meaninglessness. Carnap recognized that by the principle of tolerance he could not, on cognitive grounds, exclude ad-hoc frameworks for just about any metaphysical garbage anyone could dream up. He could still classify a lot of metaphysics as external to the generally accepted frameworks employed in mathematics and empirical science, as practiced in universities and research institutions — and therefore of no cognitive significance by those standards. But as the principle of tolerance had made explicit, anyone can construct any framework they like. So the emphasis shifted, as Cohen and Marschall point out, to the question of the value of an inquiry, and the value judgement that metaphysics has little or no value.

Flocke’s taxonomy of possible Carnapian reasons for rejecting metaphysics (whereby subject-matter and methodological reasons exhaust the possibility space) is therefore seriously incomplete. As we’ve just seen, it’s incomplete both with respect to the early Carnap and with respect to the later Carnap. The early Carnap began with a meaning-theoretical exclusion of metaphysics (as failing to meet a criterion of meaningfulness), then in 1931 moved to a classificatory demotion of metaphysics (as failing the test of translatability into the formal mode of speech), then after gradually assimilating the consequences of adopting the principle of tolerance in late 1932, the later Carnap eventually dropped explicit criteria of meaning and classification altogether (as far as metaphysics was concerned) and focused rather on the low (or negative) value of metaphysics. NB: None of these three successive stances vis-à-vis metaphysics concerns either the subject matter or the method of metaphysics.

The arguments Carnap gives for his value judgement about metaphysics in ESO are articulated as essentially utilitarian. In a stand-off between a realist and a nominalist about mathematical entities, he says:

I cannot think of any possible evidence that would be regarded as relevant by both philosophers, and therefore, if actually found, would decide the controversy or at least make one of the opposite theses more probable than the other. (To construe the numbers as classes or properties of the second level, according to the Frege-Russell method, does, of course, not solve the controversy, because the first philosopher would affirm and the second deny the existence of the system of classes or properties of the second level.) Therefore I feel compelled to regard the external question as a pseudo-question, until both parties to the controversy offer a common interpretation of the question as a cognitive question; this would involve an indication of possible evidence regarded as relevant by both sides. (ESO, p. 219)

The two sides can’t agree on a common standard for the appraisal of arguments for or against their respective positions, therefore both positions, neither internal to any framework acceptable to both, must be excluded as addressing an external pseudo-question. (Which just follows from the definitions.) The value judgement is based, in other words, on the pointlessness of comparing the relative merits of two positions if the holders of those positions can’t agree on what “merit” consists in. This can be interpreted as utilitarian in a narrow (Gradgrind-style) sense, i.e. people shouldn’t waste their time and mental resources on metaphysics rather than something socially useful and productive. That interpretation isn’t exactly wrong, but it is such a small part of the answer as to constitute a fundamental misunderstanding.

In fact, it’s not so distant from the — hostile — misunderstanding entertained by Horkheimer and Adorno about logical empiricism back in the 1930s. They saw Carnap & Co as trying to restrict human dreams and aspirations, so as to force them to conform with dominant social and scientific norms and keep them tame. Since the Frankfurt School rightly thought that unruly human dreams and aspirations need to go beyond what is, and try to imagine what could be, they saw metaphysics as an essential vehicle for the articulation of such aspirational visions. But this was exactly backwards. Carnap not only thought, like Wittgenstein (Investigations 118), that metaphysics failed in this role because it consisted of Luftgebäude (buildings in the air); his main objection to metaphysics was its authoritarian subordination of human aspirations, their imprisonment in a particular version of what is. It did the opposite of what Horkheimer and Adorno imagined it could achieve — by trying to put us in a cage of what really and ultimately is, in realms where humans are actually free to imagine and decide for themselves. (It’s sort of tragic that Adorno of all people set so much store by metaphysics in this aspirational role, since he was after all himself an artist — a composer and student of Alban Berg, as well as close musical collaborator with Thomas Mann on Doktor Faustus — and would have agreed entirely with Carnap that art was the superior vehicle for the articulation of transcendent aspirations.) It’s this positive aspect of Carnap’s conception that really mattered to him, and this aspect was the whole point of, and motivation for, the critique of metaphysics: the positive idea of liberation from authoritative versions of how things really are and forever have to be.

But, Flocke might object, current metaphysics isn’t the sort that Horkheimer and Adorno had in mind; metaphysics has become much more science-friendly and forward-looking. But the tendency of analytic metaphysics has precisely not been aspirational and imaginative of undreamt new possibilities; on the contrary, as Ladyman and Ross complain, its primary role has been to “domesticate” ideas in recent science that are utterly non-intuitive, i.e. to try to assimilate them somehow to familiar human intuitions, to tame them and render them less disturbing. Instead of utopian imagination we have reactionary conformity to the familiar.

Of the three examples Flocke gives for present-day metaphysics that she thinks would meet with Carnap’s approval, the first (Haslanger’s “ameliorative analysis”) doesn’t fall within what Carnap would have considered metaphysical — it’s just a case of explication. With a different goal from Carnap’s own projects of explication, to be sure (a goal of social progress rather than increased precision), but still basically just explication. Flocke’s other two examples (metaphysical grounding and Williamson’s “necessitism”) lie squarely within the philosophical kingdom of darkness that Carnap wanted to liberate us from. They try to use logical and conceptual tools to argue that the world is fundamentally a certain way, a certain way that in both those cases lies outside the boundaries of what any current science can say anything about. Not only, therefore, does it lie outside of what we can coherently say anything about in explicit language, Carnap would have said, but it tries to imprison us in somebody’s conception of how things have to be. Why would we voluntarily go along with that?