Expressivism and irrationality

Philosophical Review

Volumn 105, Issue 3, 1996, Pages 311-335

  Van Roojen, Mark ^a  

^a UNIVERSITY OF NEBRASKA LINCOLN   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 26644445133     PISSN: 00318108     EISSN: 15581470     Source Type: Journal    
DOI: 10.2307/2185703     Document Type: Article

References (57)

1. The relevant texts are, "Moral Realism," in Morality and Moral Reasoning, ed. Casey (London: Methuen, 1971), 101-24;
2. SpreadingThe Word (Oxford: Oxford University Press, 1984);
3. and "Attitudes and Contents," Ethics 98 (1988): 501-17.
4. The first and third of these texts were republished as essays 6 and 10 of his Essays in Quasi-Realism (Oxford: Oxford University Press, 1993).
5. Blackburn continues to endorse the general approach, most recently in "Gibbard on Normative Logic," Philosophy and Phenomenological Research 72 (1992): 947-952
6. Gibbard's treatments are found in his Wise Choices, Apt Feelings (Cambridge: Harvard University Press, 1990), especially chapter 5
7. and in his "Reply to Blackburn, Carson, Hill and Railton," Philosophy and Phenome-nological Research 72 (1992): 969-980
8. By 'norms' I mean rational standards, principles, or regulative ideals that prescribe how a person should operate within the domain in question.
9. In addition to Schueler, whose paper (cited in note 11) occasioned Blackburn's response, prominent critics include Bob Hale, "The Compleat Projectivist," Philosophical Quarterly 36 (1986): 65-84;
10. M. H. Brighouse, "Blackburn's Projectivism-An Objection," Philosophical Studies (1990): 225-33
11. and Nick Zangwill, "Moral Modus Ponens," Ratio (1992): 177-193
12. Geach first raises the point in a footnote in "Imperative and Deontic Logic," Analysis 18 (1958): 49-56, at 54.
13. It is raised again by John Searle in "Meaning and Speech Acts," Philosophical Review 71 (1962): 423-32.
14. Geach develops the point in greater detail in "Assertion," Philosophical Review 74 (1965): 449-465
15. The example is from Geach, "Assertion," 463
16. While embedding is the issue here, Geach himself thought this was an example of a general kind of problem for theories of many sorts: A thought may have just the same content whether you assent to its truth or not; a proposition may occur in discourse now asserted, now unasserted, and yet be recognizably the same proposition" (ibid., 449).
17. "Moral Realism," in Morality and Moral Reasoning, ed. Casey (London: Methuen, 1971), 101-24, at 119;
18. this paper was reprinted in Blackburn, Essays in Quasi-Realism.
19. Essays in Quasi-Realism, lbid., 120.
20. Essays in Quasi-Realism, Ibid., 121.
21. This point is made by G. F. Schueler, in "Modus Ponens and Moral Realism," Ethics 98 (1988): 492-500, at 496.
22. Spreading the Word, 193.
23. See Spreading the Word, 194-195
24. This idea is already suggested by Blackburn in the earlier article, where he says that to make the claim that one attitude involves another is to make a moral claim ("Moral Realism," 121- 22).
25. Spreading The Word, 195.
26. Blackburn, "Attitudes and Contents," Ethics 98 (1988): 501-517, responding to Schueler's "Modus Ponens and Moral Realism."
27. "Attitudes and Contents," 512. Actually, it isn't clear that it was part of the content of the conditional in the second proposal cither, given that the judgment expressed that attitude rather than stated that the speaker had it. After all, that is the expressivist proposal regarding moral utterances generally. As a result, it isn't clear how the new proposal distinguishes between expressing the moral attitude and merely expressing the commitment one has because one has the moral attitude or outlook.
28. For example, Blackburn's fourth rule for generating approximations to an ideal relative to a set of moral judgments applies only when a set of sentences is itself the "next approximation" to some ideal. But we need to know the content of the fourth rule before we can setde the issue of which sets of sentences are such approximations (see "Attitudes and Contents," 513-14).
29. Bob Hale does a heroic job, both of explaining the problems with the proposal as presented and of suggesting ways of fixing them, in "Can There Be A Logic of Attitudes?" in Reality, Representation and Projection, ed. Haldane and Wright (Oxford: Oxford University Press, 1993), 337-363. He also presses a line of objection to the proposal as amended.
30. At least that seems to be the intended effect of the rules he defines. The actual proposal involves applying the rules to generate sets of sentences that are supposed to be closer and closer approximations to worlds in which obligations are satisfied and tolerable actions brought about. If in any of the descriptions of those worlds a well-formed formula and its negation is propounded, then that set is unsatisfiable. Judgments are inconsistent if every route to a final ideal via application of the logical rules contains a well-formed formula and its negation. Since H!A' is a well-formed formula if 'A' is, unsatisfiability is generated even where an attitude and its negation are expressed. But since the negation of 'H!A' is interpreted as T!not(A) (expressing the idea that not(A) is tolerable) the "inconsistency" is generated precisely because A and not(A) are not co-satis-fiable. Actually, given the rules as Blackburn actually defines them, tilings do not always work out like this. 'H!A,' and 'T!A' (the translation of 'not(H!A)') can be made "satisfiable" if considered together with a disjunctive premise. Clearly, Blackburn intended otherwise.
31. Since the operators introduced in Spreading The Word merely formalize what is already part of the noncognitivist proposal, we can make use of that formalization to capture the idea behind even the earliest proposed analysis.
32. The argument was suggested by a letter to advice columnist Abigail Van Buren, author of "Dear Abby."
33. At least that seems a perfecdy good way to formalize it, given that Blackburn indicates that 'B!' and 'H!' are to "attach to descriptions of things to result in expressions of attitude" (Spreading The Word, 193). Immediately following, however, he introduces a special notation for talking about attitudes or beliefs by "putting its expression inside bars: /H! (A)/ refers to the approval of (X)" (ibid., 194). So perhaps the sentence should be formalized as 'B! /Believing that (my father is unfaithful to my mother)/'. Since attitudes are a kind of thing, it isn't clear which way is the preferred translation. In any case, it matters little for the point being made, since either way the attitude in question, as interpreted by Blackburn, involves a higher-order disapproval of a lower-order attitude which one oneself holds. And that, he claims, is logically inconsistent.
34. BIackburn might wish to resist this conclusion, but I don't see how he can, given that the argument relies on just the aspect of these attitudes that he uses to count the noncontroversially inconsistent attitudes as inconsistent, despite their nonrepresentational status. It is worth noting that nothing turns on the first-order attitude in question being a belief rather than a pro or con attitude. For we might have constructed a parallel argument for the claim that my father had done something deeply wrong, and for the judgment that it is wrong to believe that one's friends have done something deeply wrong.
35. In any case, the expressivist owes us an account of the functioning both of expressions regarding lightness and of expressions regarding goodness, a requirement that is generally ignored. Since on many views it can be good that someone did the wrong thing, the task would involve some complication.
36. In fact one well-known argument against "realism" and in favor of some kind of noncognitivism starts with the idea that such conflicts are possible. See Bernard Williams, "Consistency and Realism," in Problems of the Self (Cambridge: Cambridge University Press, 1973), 187-206.
37. Gibbard's analysis is found in his Wise Choices.
38. Wise Choices, 172-73. Gibbard goes on to discuss the conditions under which one can make conversational demands of others to accept the norms towards which one is oneself expressing acceptance. It isn't entirely clear whether the analysis further requires one to believe oneself in such conditions before one can sincerely express the requisite attitude. If so, the analysis is more complicated yet. As I read the argument, the ensuing section is merely an interesting discussion of when it might be sensible to make, or to heed, such demands; the actual analysis is complete at page 173.
39. Ibid., 102.
40. As Gibbard formulates the proposal, the norms apply to naturalisti-cally described kinds of objects, so that they will never treat two naturalis-tically identical objects differently. Failure to respect the supervenience of the normative on the non-normative will, therefore, be treated as a logical error. Since Gibbard could allow norms applying to particulars, this is but an incidental feature of the approach, and does not show that any similar view would illegitimately treat such errors as logical errors when they are not.
41. As is apparent when he states: We can characterize a system N of norms by a family of basic predicates 'N-Forbidden', 'N-optional', and 'N-required'. Here 'N-forbidden' simply means "forbidden by system of norms N", and likewise its siblings. Other predicates can be constructed from these basic ones.. [W]hen a system of norms N applies in a definite way to an alternative, that results in an alternative's having exactly one of the three basic properties, being N-forbidden, N-optional, or N-required. We can call a system complete if these predicates trichotomize the possibilities.. (So long as Nis consistent, nothing will be more than one of these things.) (Wise Choices, 87)
42. Ibid., 88. In a footnote he cites van Fraassen and Marcus (no logical slouches) as opponents of the requirement under consideration. Gibbard appears not to have noticed that his own proposal is not really any different, in that it rules out a person's being both required to do and forbidden from doing the very same thing. That seems no different from requiring that there always be an outcome that is permitted.
43. Why am I not equally charitable towards the analogous problem in Blackburn's third proposal? Because given that Blackburn wants to derive the inconsistency of the attitudes in question from the purported irrationality of having desires that would be self-defeating if all realized, the commitment is essential to his approach. Since Gibbard is not quite so clear on why the attitudes in question bring with them the relevant logical commitments, it is harder to judge whether the commitment in question is essential to his approach.
44. It really does seem odd that a statement could have two kinds of content, both of which contribute to its logical functioning. But that idea has been part of noncognitivist proposals for some time. Stevenson, for example, distinguishes the emotive meaning of expressions from their descriptive meaning, and claims that moral terms have both. Thus, the proto-analysis he offers of judgments of goodness-'I approve of this; Do so as well-contains both a descriptive and a prescriptive component (Ethics and Language (New Haven: Yale University Press, 1944), chap. 2; see also chap. 3, sec. 7).
45. R. M. Hare similarly distinguishes prescriptive meaning from descriptive meaning. Ethical terms have both; for the most general the prescriptive meaning is primary, whereas for more particular terms (such as 'industrious') the descriptive meaning is primary. See The Language of Morals (Oxford: Oxford University Press, 1952), 118-125 It is always a bit of a puzzle how such contents are supposed to be connected to one another, but given the commitments of the noncognitivist, it is not too surprising that the view yields such dual contents. Once one has postulated a special nondescriptive meaning for terms that makes them apt for the expression of attitudes, one has only to note that there are terms with clear descriptive meaning that are apt to express attitudes to wonder whether an expression might have both kinds of meaning.
46. Obviously, the logical behavior of utterances with overlapping content need not be precisely alike. For my argument to work, all I need is that a judgment will be inconsistent with anything not consistent with judgments having only a subset of its content. In other words, adding content to a judgment will not make it consistent with any judgment that it is inconsistent with. And that seems obviously correct.
47. One kind of reply that won't work is to plead that one is offering reforming analyses, rather than analyses of the actual meaning of the terms as currendy used. The idea here might be that if the proposed analysis is not meant to capture what we currently mean, but is meant instead to offer a reform of our current practice that captures most of what is important, then what appear to be counterintuitive consequences of the theory won't be reasons to reject it. They will be counterintuitive only given that we use the words in an unreformed way. Once we adopt the reforms suggested by the expressivist theory, the currendy counterintuitive consequences should lose their odd flavor as we adjust to new linguistic practices. (See Gibbard, Wise Choices, 32) The reason this reply won't work is that the expressivist account of logical contradiction does not leave our nonevaluative judgments undisturbed. The account involved two components: (1) an analysis of evaluative judgments as expressions of certain complex attitudes, and (2) an account of logic that counts certain expressions logically inconsistent. A reforming definition is in effect a stipulation that we will use our evaluative terms in the manner suggested by the analysis that constitutes the first component. But the trouble complained of here was really trouble for the second component, the account of what is logically consistent and inconsistent with what. It was because the account must find certain expressions of attitudes logically inconsistent that it could not count the arguments in question consistent. And this was not dependent on the words with which those attitudes were expressed. (To see this, notice that the account will have to count expressions of certain attitudes logically inconsistent, whether they use a form of words including evaluative vocabulary or not.) A redefinition of moral terms should not affect the logical behavior of judgments not employing those terms. I suppose Gibbard might be proposing a reforming definition of logical terms ('consistent', 'valid', and so on) as well. This would be more ambitious, and hence would come at the cost of likely additional departures.from common sense over and above those required by reforming definitions of only evaluative vocabulary. My discussion below of a possible "bullet-biting" response by Blackburn (or Gibbard) details some of the departures from common sense required if we extend the notion of logical inconsistency in the evaluative realm so as to class arguments like those of the examples logically inconsistent. Finally, it is worth mentioning that Gibbard's use of "open-question" arguments to refute candidate cognitivist analyses of evaluative terms (see 16-22) does not sit well with a reforming analysis approach. A reforming analysis would itself be subject to open-question arguments.
48. Does Gibbard's most recent account avoid the problem? There he seems to abandon the analysis of the judgments in question as the acceptance of sets of higher- and lower-order attitudes. Instead, he claims (somewhat cryptically) that they stand to the acceptance of norms as fully factual judgments stand to the factual apprehension (Gibbard, "Reply to Blackburn, Carson, Hill, and Railton," Philosophy and Phenomenological Research 51 (1992): 970-71).
49. Until that relation is filled out, it will be hard to tell whether the objection, as formulated here, applies. For without more details it is not clear what the state of mind in question is supposed to be. In any case, the more recent account seems to make my first criticism of Gibbard harder to elude, in that it ties the inconsistency of conflicting moral judgments to the inconsistency of any fully detailed contingency plan-for what to do, think, or feel for every imaginable circumstance- that the norms might commend (ibid., 973-74.) Thus, ruling out conflicting obligations becomes essential to the approach.
50. The explanation very likely would involve explaining why expressions of the attitudes with contents of the sort provided by the analysis must have the "logical form" exhibited by the logic defined in terms of norm-world pairs. Gibbard does not say very much about this. One approach, the one taken by Blackburn, is to argue that one is committed by the contents of the attitudes to making the appropriate inferences, on pain of irrationality. I argue below that this will not work.
51. This response on Blackburn's behalf was forcefully urged against me by both Frances Howard-Snyder and Joe Mendola on independent occasions. My response to it has benefited enormously from suggestions by a reader for the Philosophical Review.
52. See, G. E. Moore, Ethics (New York: H. Holt, 1912), 125 (or reset ed. (London: Oxford University Press, 1947), 78)
53. G. E. Moore, "A Reply to my Critics," in The Philosophy of G. E. Moore, ed. Schilpp (Evanston: Open Court, 1942), 543.
54. This particular point is reason to reject the kind of bullet biting that would save Blackburn's first account from the example I employ against it. The general point I'm urging here counts against the sort that would save the other two.
55. These costs will be costs whether the analysis is offered as an analysis of the concepts we already possess or as a "reforming analysis" of the sort I consider on Gibbard's behalf in note 34. The latter sort purport to offer us reformed definitions for terms like 'logical inconsistency' that capture most of what we could reasonably hope to say using our old unreconstructed vocabulary. The considerations above just underline some of the things we had hoped to say, but which would be barred to us using the reformed terms.
56. Gilbert Harman has long urged this and similar conclusions. See for example, Change in View (Cambridge: MIT Press, 1986), especially chap. 2.
57. For example, one kind of pragmatic inconsistency occurs when the very assertion of a claim refutes the truth of the claim, as in "I have written nothing."