Florian Steinberger’s paper “How Tolerant Can You Be?” was originally part of Georg Schiemer’s “Carnap on Logic” conference a couple of years ago, and largely shaped the direction my own paper took in its final form, since it contains one of the clearest statements I’ve come across of the basic dilemma facing any attempt to make the principle of tolerance the basis for an overall conception of rationality.
The problem is essentially the same as the one articulated by e.g. Alexander George who argues that the price Carnap pays for his external perspective on scientific knowledge (which Quine eschews) is that rationality — which requires a rationality-constituting framework — can’t be brought to bear on the choice among such frameworks:
. . .rational relevance only gets a grip once a particular linguistic framework has been adopted. Of course, we might be inclined one way or another in our choice of frameworks. But for Carnap, the forces that so incline us are, as he puts it, ‘practical’; they make one choice of framework more useful than another. . . The bearing such forces have on our choice of framework is of a different kind from the rational bearing empirical considerations have on hypotheses formulated within a framework. In fact, . . . Carnap is even inclined to withhold the term ‘judgement’ to describe our embrace of a particular framework. Judgements are cognitive acts that presuppose the adoption of a framework. . . (p. 9)
Florian’s way of putting it turns the picture the other way around: “what would such a framework-selection framework have to look like?” But he arrives at the same conclusion, since as he points out, you’d need a selection framework to choose the framework-selection framework, and so on to an infinite regress. I agree that this is the central dilemma regarding the principle of tolerance, and I think Florian has spelled it out with luminous clarity.
But of course the hallmark of philosophical achievement is to raise problems, not to solve them. And sure enough, I find Florian’s solution (which is really just a kind of coda to his paper) unconvincing. My problem with it (which won’t surprise him — he’s heard this from me before) is that the “soft foundationalism” which he thinks the most promising regress-blocker doesn’t actually do any better than his (rightly) rejected alternative of “hard foundationalism” — i.e. just pounding the table and insisting that some meta-framework or another is (despite the principle of tolerance) ultimately “correct.” Hard foundationalism is clearly inconsistent with the principle of tolerance and with everything we know about Carnap, so I certainly agree with Florian’s rejection of that option. What he calls “soft foundationalism” is suggested to him by what Gilbert Harman (2002) calls “general foundations” theories, such as those of Goodman and Rawls:
The idea is that one starts where one is, trying to eliminate conflicts among one’s beliefs and experiences, trying to make one’s overall view more coherent, seeing implications of one’s opinions for questions one is interested in answering, seeing what one thinks of these implications, and in general aiming at the no doubt impossible ideal of achieving a perfect “reflective equilibrium” in one’s opinions. (Harman, p. 421)
This approach, Florian says, shifts the burden of justification from the “selection framework” to the initial belief endowment of the individual or community, which is held “innocent until proven guilty.”
I agree that there is a kind of kinship between the Rawls-Goodman notion of “reflective equilibrium” and the “dialectical” element in Carnap identified by Howard Stein, but there is also a crucial difference between them: for Carnap, there is explicitly, and indispensably, a self-conscious intermediating step of reflection about, and choice among, competing frameworks or explications, i.e. among concepts and systems of concepts. In this view we can’t just decide what to believe (or know) in isolation from having first decided on criteria for knowledge or belief, and on frameworks for relating our beliefs to each other (or not).
This is analogous to the intermediating step in certain political theories (such as Rawls’s or Hayek’s) of reflection about, and choice among, different possible “constitutions”; in such theories, we don’t just have individuals on the one hand and choices among rulers on the other, we also, ultimately, in the long run, have choices among institutional systems. In both these kinds of “intermediating” or “two-level” approaches, there is a hierarchy of values corresponding roughly to the Kantian hierarchy of Vernunft and Verstand; we make our substantive decisions about what to believe/know (in the one case), and what to do (in the other), within a system or framework chosen in the light of higher-level “framework values.” This is reflected in the classic Enlightenment sentiment (attributed to Voltaire) “I despise what you are saying but am prepared to die for your right to say it.” (Which most people even in liberal societies still find schizophrenic, and are uneasy with.)
There are two related problems with Florian’s soft foundationalism. First, if it is to be understood as having any positive content at all (as opposed to just the avoidance of hard foundations) — i.e. if it is to stop the regress — then it must be construed as prescribing the initial adoption, by each person, of the values they inherit from their socialization of origin. But this seems extremely authoritarian. What about those who rebel against their parents (yes, I know, that isn’t fashionable any more), and can’t wait to escape to the Big City? But if soft foundationalism is only to be construed as prescribing that we start somewhere, then it would appear to lack positive content; in what sense would it be prescribing anything at all?
The second and deeper problem, though, is that soft foundationalism leaves the envisaged convergence toward “reflective equilibrium” unconstrained. The process described by Harman, with its elimination of “conflicts” among beliefs, its striving toward “coherence,” seeing “implications” of opinions, and so on, requires that we define “reason,” i.e. that we know what we mean by “conflicts,” “coherence,” “implications,” and so on. Soft foundationalism, in other words, presupposes a choice of framework.
First off, I want to say that I’m really enjoying what you’re doing on this blog. Though I’m no Carnap expert, I’m always interested in reading about his work, and it’s been nice to see new stuff every couple of days. Thanks for that.
I read Steinberger’s paper as well, and I think I was more convinced by his solution than you were. So I thought I’d just state what I took his view to be and see what you think.
My understanding was that Steinberger’s soft foundationalism could be understood in the following way: There just AREN’T any framework-external rational norms governing framework selection. (That, I take it, is why ‘our framework selections … do not stand in need of any ultimate foundation’ (p. 19).) Any time we’re deliberating about whether to adopt a linguistic framework–i.e., whether to change our language in a particular way–we’re doing so from within a particular language, which has built into it its own rational norms. (It may be that those norms recommend their own revision in some cases, but that seems as it should be.) If that’s right, then it will be the case that we start wherever we are, but this is a descriptive claim rather than a normative one. The point of soft foundationalism, I thought, is not that we SHOULD start wherever we are but that the question of where we SHOULD start is a bad question, since it presupposes, wrongly, that there are rational norms available before we select any framework at all.
I took another look at the relevant part of the paper, and it’s not clear to me that this is precisely what Steinberger has in mind. But in any case, it seems to me that, on this interpretation, neither of your worries really sticks. I feel like I may be missing something, but I’m not sure what.
Well, I don’t think that’s what Florian had in mind, since it would put him in the same boat with Alexander George (in the paper I refer to, building on his Mind paper about Carnap and Quine), and I think Florian wants to suggest that we can get beyond that diagnosis, which leaves Carnap with no rational resources whatever for framework selection. Not that this isn’t a perfectly respectable position; it was essentially Quine’s, after all, and although I disagree with it, George’s papers (which haven’t gotten the attention they deserve among Carnap-types) articulate the trade-offs between the Carnap and Quine positions very clearly and even-handedly.
In exploring the options for getting beyond that position, though, I think Florian thinks he can distinguish between what he calls hard and soft foundationalism. Perhaps it’s possible, but on the basis of what Florian has said so far, at least, it doesn’t work; soft foundationalism is either empty (has no normative bite at all, and so essentially becomes the George position, as you suggest), or degenerates immediately into hard foundationalism, as I argue in my last paragraph.
My own suggestion for getting beyond the George position (based on Howard Stein’s reading of Carnap) superficially resembles Florian’s soft foundationalism, but differs from it (i.e. differs from Rawlsian “reflective equilibrium”) by inserting that “crucial intermediate step,” as I put it in the above post, of reflection on the language or framework in which we reflect substantively on real stuff. The difference, in other words, is the specifically “Carnapian linguistic turn” I’ve discussed in another post here.