Abstract
This paper takes up a neglected problem for metaethical noncognitivism: the characterization of the acceptance states for agent-centered normative theories like Rational Egoism. If Egoism is a coherent view, the non-cognitivist needs a coherent acceptance state for it. This can be provided, as Dreier (Aust J Philos 74: 409–422, 1996) and Gibbard (Thinking how to live, Harvard University Press, 2003) have shown. But those accounts fail when generalized, assigning the same acceptance state to normative theories that are clearly distinct, or assigning no acceptance state to theories that look to be intelligible. The paper makes the case for this and then asks: What should we conclude if the problem cannot be solved? We might conclude that since Egoism is clearly a coherent (if mistaken) view, the argument amounts to a refutation of noncognitivism. But we suggest another possibility. There is, on reflection, something incoherent, or at least odd, in standard formulations of Egoism; noncognitivism predicts this and so provides an intriguing explanation for this fact.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
It is always a bit unclear what it means to say that a statement type S expresses a given attitude. The idea is roughly that S expresses A when a competent sincere assertoric utterance of S is caused in the right way by a token of A on the part of the speaker. But a sincere assertoric utterance of S will normally be caused by many attitudes, not all of which count as attitudes expressed by the statement. To our knowledge there is no standard characterization of the expression relation with which this sort of attitude-theoretic semantics is concerned; and yet there is no way to formulate noncognitivism as a thesis in the philosophy of language without invoking some such notion.
As is familiar, the noncognitivist should not deny that normative statements express normative beliefs. Someone who judges that Fred should keep his promise ipso facto believes that Fred should keep his promise. So it is always true, if unilluminating, to say that a sincere competent utterance of ‘X should φ’ expresses the belief that X should φ. The nontrivial task is to provide a characterization of the normative belief that reveals the nature of the noncognitive state that underlies it (Rosen, 1998).
As is less familiar, the noncognitivist can understand this task in either of two ways. She can say that the judgment/belief that Fred should keep his promise is identical to the noncognitive state she associates with the judgment. (This has the awkward consequence that the state is technically cognitive after all, since beliefs are paradigmatically cognitive: truth-evaluable, aimed at truth, etc.) Alternatively, as we prefer, she can say that when someone judges/believes that Fred should keep his promise, she is in that cognitive state in virtue of being in a more fundamental, noncognitive state. This noncognitive state is distinct from the belief—as it must be if it is to ground it. However the relation is not causal or contingent, but rather constitutive: the belief is grounded in the non-cognitive state that underlies it in roughly the same sense in which the flower’s being red is grounded in its being crimson.
This is a bit quick. This simple case is consistent with taking ‘X should φ’ to express a weak preference for X’s φing: a preference ranking according to which no alternative to < X φs > is ranked ahead of <Xφs>. Views of this sort will generally entail that if I am indifferent between going to the movies and staying home, I can express my normative view by saying ‘I should go to the movies’, and that’s quite wrong. See Silk, 2015 for a sophisticated defense of weak preference expressivism for certain deontic modals.
To be clear: Dreier defends his view only as the best option for the noncognitivist. He does not defend noncognitivism against the cognitivist alternatives.
In what follows we speak freely of possible worlds and possible individuals à la Lewis. But all of that is dispensable for our purposes. We could identify properties with sets of centered worlds, i.e., pairs of the form < c, w > where c is an individual who would exist if w were actual. Or we could adopt an autonomous theory of properties and propositions that makes no foundational reference to worlds. So far as we can tell, these metaphysical issues are orthogonal to our concerns.
For doubts about the framework and Lewis’s reasons for accepting it, see Magidor (2015).
If a more eccentric Smith wants to be an egg, then he wants to have a different property: x. x is an egg. Since this is a property that many things have, the view has the added bonus of giving non-empty content to certain eccentric desires that would have trivial contents if contents had to be sets of possible worlds.
As Gibbard emphasizes, the states he calls plans are not exactly intentions as we ordinarily understand the notion. Most importantly, Gibbard-style plan for what to do in C may forbid certain options while expressly permitting others (Gibbard 2003, 55). A plan that permits both A and B is inconsistent with a plan that forbids A. By contrast, the intention to do either A or B is not inconsistent with the intention not to do A. This point is important to Gibbard’s ingenious solutions to a range of problems. However it will not matter in what follows. For discussion of this aspect of Gibbard’s view, see Schroeder, 2008, Silk, 2015 and Ayars, (forthcoming).
Why not say instead that my judgment that Harry should φ consists in a plan for Harry to φ? Gibbard rejects this view on the ground that it is conceptually impossible for one person to plan for another in the sense in which a person can plan for herself. For extensive discussion of this aspect of Gibbard’s view and a version of plan expressivism that rejects it, see Ayars (forthcoming). This revised version of plan expressivism solves many of the problems we have raised for Gibbard here.
The content of a Gibbard-style plan is naturally given by a set of centered worlds, with the plan to φ in C having as its content the set of worlds at which the center φs if the center is in C. A world in which Harry is the center and Harry wins is a world at which both P1G and P2G are satisfied, so the plans are jointly satisfiable.
We say “schematic” for the following reason. The amnesiac egomaniac’s egomania does not consist in a finite list of plans to benefit Harry if he is Harry, Tom if he is Tom, etc., so it is general in a sense. But it can sound odd to characterize his mental state by saying that for all x, he plans to benefit x if he is x. This seems to imply that he plans to benefit Anastasia if he is Anastasia; but he has no such plan if he has never heard of Anastasia and can’t think about her. Rather his mental state seems to be one that commits him to forming such a plan should Anastasia come to his attention. For want of a better term, we may call this form of generality schematic, by analogy with the sort of generality that attaches to schematic principles in formal logic.
Though the preface paradox may show that even full belief is not inconsistency transmitting without restriction.
More exactly, the attitude expressed by a normative judgment is a mix of de se preference and belief, since the judgment that Harry should win expresses a preference for benefitting oneself together with the belief that Harry can only benefit himself by winning. There are several ways to extend the notion of consistency to these hybrid attitudes, but however it is done, Dreier’s proposal will entail that normative judgment is inconsistency-transmitting.
The complacent cognitivist solves the problem without going de se. For the complacent cognitivist, normative judgments express beliefs, the paradigmatic inconsistency-transmitting attitude. But the contents of the beliefs expressed by (1) and (2) are normative propositions—Tom should win, Harry should win; and these propositions are (on the face of it) straightforwardly consistent.
A slightly more plausible version of the view would identify normative judgment, not with ordinary desire, but with some sort of beefed-up desire, e.g., the desire that X φ together with a non-instrumental desire to retain that desire, as in Lewis’s (1989) account of valuing.
It is not obvious that desire noncognitivism renders Egoism coherent if it identifies Egoism with the desire that everyone benefit himself. After all, in the cases that generate Medlin’s problem it is impossible for everyone to do what would most benefit himself, and it is arguably incoherent to desire a manifestly impossible state of affairs. The most interesting response to this worry is to identify Egoism, not with a desire for the universally quantified state of affairs—Everyone benefits himself—but rather with a schematic desire: a state that commits the subject to a desire that x benefit himself for each individual x who comes to his attention, without committing the subject to a desire for the conjunction of these states (n.10 above). A desire of this sort is general in sensu diviso in Abelard’s sense, as distinct from the universally quantified desire, which is general in sensu composito. Since it is not incoherent to desire two states of affairs that cannot be realized together, this would save the egoist from the charge of incoherence.
In Dreier’s case the judgment that X should φ involves a preference for λx.φx together with the belief that that X can only by φing. The preference for λx.φx may motivate the subject to see to it that she has this property; but it will not motivate her to see to it that others have it. So Dreier’s view respects the asymmetry. In Gibbard’s case, the judgment that X should φ expresses (roughly) the plan to φ if one is X. This planning will motivate one to φ provided one believes that one is X; but it will not motivate one to see to it that others φ. So Gibbard’s view likewise respects the asymmetry.
It is sometimes said that the analogous principle in the moral case is falsified by dilemmas, in which the agent is morally required to perform each of two incompatible actions. But whatever one says about this, it is harder to believe that the practical ‘should’ can give rise to such dilemmas. One standard gloss brings this out: If to say that X should φ is to say that X has most reason to φ, then it clearly can’t be the case that X should φ and that X should do something incompatible with her φing.
The pertinent accounts are as follows:
approvalD S approves of X’s φing iff S prefers to and believes that X can only by φing.
approvalG S approves of X’s φing iff S plans to φ if he is X in X’s situation.
It is easy to verify that these attitudes are inter- but not intra-personally inconsistency-transmitting.
The explanation for this cannot be that ‘should’ implies ‘can’. For we can stipulate that in every relevant sense, Tom can beat Harry and Harry can beat Tom. If this requires that the race be set in an indeterministic world, so be it.
The odd cases are cases in which two people—say conjoined twins—share the circuitry necessary for choosing or intending, so that if the one tries to do what’s best for her, the other cannot even try to do what’s best for him.
References
Ayars, A. Deciding for Others: A non-cognitivist account of third-personal normative judgments. Forthcoming in Philosophy and Phenomenological Research.
Dreier, J. (1996). Accepting agent centered norms: A problem for non-cognitivists and a suggestion for solving it. Australasian Journal of Philosophy, 74(3), 409–422.
Frege, G (1918–1919). Der Gedanke. Beiträge Zur Philosophie des Deutschen Idealismus 2:58–77.
Gibbard, A. (2003). Thinking how to live. Harvard University Press.
Gibbard, A. (1990). Wise choices, apt feelings: A theory of normative judgment. Harvard University Press.
Hare, R. M. (1963). The language of morals. Oxford University Press.
Lewis, D. (1979). Attitudes de dicto and de se. Philosophical Review, 88(4), 513–543.
Lewis, D. (1989). Dispositional theories of value. Aristotelian Society Supplementary, 63(1), 89–174.
Magidor, O. (2015). The myth of the de se. Philosophical Perspectives, 29, 249–283.
Medlin, B. (1957). Ultimate principles and ethical egoism. Australasian Journal of Philosophy, 35(2), 111–118.
Parfit, D. (1984). Reasons and persons. Oxford University Press.
Rosen, G. (1998). Blackburn’s essays in quasi-realism. Noûs, 32(3), 386–405.
Schroeder, M. (2008). Being for: Evaluating the semantic program of expressivism. Oxford University Press.
Silk, A. (2015). How to be an ethical expressivist. Philosophy and Phenomenological Research, 91(1), 47–81.
Acknowledgements
For invaluable comments and discussion, we thank James Brown, Alex Worsnip, Christopher Howard, Michael Smith, and the participants at the 2020 Chapel Hill Normativity Workshop.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Ayars, A., Rosen, G. Noncognitivism and agent-centered norms. Philos Stud 179, 1019–1038 (2022). https://doi.org/10.1007/s11098-021-01704-2
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-021-01704-2