DISCUSSION

Ideal Agency and the Possibility of Error* Bradford Cokelet In “Practical Reason and the Possibility of Error,”1 Douglas Lavin claims to have discovered a paradox deep in the heart of Christine Korsgaard’s neo-Kantian project. He starts by investigating various interpretations of the error constraint, which many philosophers have in mind when they assert that there is a connection between practical reason and the possibility of error: “An agent is subject to a principle only if the reasoner can go wrong in respect to it” (425). Lavin’s main conclusion is that the strong imperatival interpretation of this constraint is incompatible with what he calls the constitutivist approach to explicating the binding force of rational norms—schematically put, the constitutivist aims to show that agents are subject to certain norms simply in virtue of the fact that they are capable of rational agency.2 Lavin thinks that Korsgaard, for example, hopes to elucidate the binding force of certain principles by “showing that the capacity to act just is, in part, the capacity to follow the relevant principle[s]” (453). Lavin’s conclusion—that strong imperativalism and constitutivism are incompatible—spells trouble for contemporary Kantians who, like Korsgaard, hope to combine these two doctrines. I aim to offer them some solace by showing that Lavin’s criticism rests on a mistaken conception of ideal rational agency.

* Many thanks to Ryan Doran, Richard Kraut, Douglas Lavin, Cullen Padgett Walsh, the members of the Northwestern Graduate Student Ethics Group, and two anonymous referees for Ethics for helpful comments on earlier versions of this article. 1. Douglas Lavin, “Practical Reason and the Possibility of Error,” Ethics 114 (2004): 424–57. 2. Lavin uses “principle,” “standard,” and “norm” more or less interchangeably. I will use “norm” throughout, but nothing I say hinges on my doing so. Ethics 118 ( January 2008): 315–323 䉷 2008 by The University of Chicago. All rights reserved. 0014-1704/2008/118020001$10.00

315

316

Ethics

January 2008

I. LAVIN’S ARGUMENT Translated into premise-conclusion form, Lavin’s strongest argument for the incompatibility claim looks like this:3 (P1)

Strong imperativalism rules out the concept of perfect rational agency (449–52). (P2) Without a conception of perfect rational agency, one cannot be a constitutivist (454). *(C) Constitutivism and strong imperativalism are incompatible. My criticism will focus on Lavin’s argument for P1; I grant P2 and the argument’s validity. However, before turning to P1, some preliminary remarks about the nature of strong imperativalism, constitutivism, and P2 are in order. In section 3 of his paper, Lavin defines the strong imperatival interpretation of the error constraint (SIEC for short) as follows: “[According to] the strong imperatival interpretation . . . an agent is subject to a principle only if there is some kind of action such that if the agent did it she would violate the principle and it is possible for the agent to do it and her doing it would be a genuinely practical defect” (436). Although one could raise questions about several of the concepts deployed in this definition (especially A being subject to P), all of the concepts except for that of a genuinely practical defect are clear enough for our purposes, and Lavin’s rough characterization of that type of defect will suffice for our discussion: he holds that a defect in reason is genuinely practical just in case it is not simply the result of a mistake or error in theoretical reasoning. Quoting Korsgaard, he mentions cases in which “people’s terror, shyness, or depression is making them irrational and weak-willed” as instances involving genuine practical defects (436). But what about constitutivism? Here Lavin’s remarks are (intentionally) sketchy: Very generally, the constitutivist’s objective is to demonstrate that any particular exercise of rational agency, or action, involves a commitment to comply with certain principles. The objective is to be achieved by showing that the relevant principle is simply a partial 3. Lavin has a second argument that runs as follows: (P3) strong imperativalism entails understanding the concept of freedom as the liberty of indifference, and (P4) if one understands the concept of freedom as the liberty of indifference, one cannot be a constitutivist; therefore, (C) constitutivism and strong imperativalism are incompatible. I have doubts about P3 and P4, but the issues involved are too complicated to deal with here (in part, I think they hinge on Lavin’s contentious formulation of constitutivism). I consider this argument to be weaker because Lavin only gives a couple of sentences in defense of P4 (see 454).

Cokelet

Ideal Agency

317

description of agency itself; that is, the aim is to be achieved by showing that the capacity to act just is, in part, the capacity to follow the relevant principle. If so, then when one does exercise that capacity one is committed to abiding by the relevant principles— it is a standard internal to the capacity (453). Given this gloss, one naturally wants to know what it could mean to say that a capacity and commitment to follow a principle is “part of ” a capacity to act. One possibility would be to replace the mereological terminology with logical terminology; we could say that the constitutivist argues that any particular exercise of rational agency, or action, involves a commitment to comply with certain norms, and that she does so by attempting to show that having the capacity for rational agency entails that one has the capacity and commitment to exercise it in accord with (or response to) those norms.4 Regardless of how he would develop his remarks about constitutivism, I am willing to grant that the constitutivist needs a concept of perfect rational agency (i.e., P2). Nonetheless, to understand that premise, it helps to consider why the constitutivist needs that concept. Lavin does not give an explicit argument, but he does say that it is needed to “fix the standard constitutive of the activity in question” (454). It is not clear what Lavin has in mind, but one possibility is as follows: the constitutivist must, by definition, show that some norm represents a standard that is internal to rational agency, and, the suggestion might be, the constitutivist should accomplish this task by, first, defending a conception of rational agency that shows us how to divide exercises of that agency into perfect and imperfect types (or to into a spectrum of cases ranging from minimally rational to ideally rational) and then arguing that the resulting qualitative measure of exercises of the capacity constitutes a standard that is internal to the capacity. I am not clear why the constitutivist must proceed along these lines, but doing so is not obviously misguided; I will leave questions about this aside because I am granting P2. This brings us to our main topic: Lavin’s claim that strong imperativalism rules out the concept of perfect rational agency (i.e., P1). He 4. This reformulation is appealing because it is more broad—it still allows us to classify the views that stick with the “part of” locution as constitutivist—and because it is in line with the popular “deduction of the moral law” interpretation of what Kant is trying to do in Groundwork III, namely, that he is trying to show that the ability and commitment to act on the moral law is entailed by our actual ability for free, rational agency. See, e.g., Henry Allison, Kant’s Theory of Freedom (Cambridge: Cambridge University Press, 1990), pt. 3; for discussion of debates about whether Kant gave up the aim to provide that deduction in his later work and about whether he even had that aim in writing the Groundwork, see Karl Ameriks, Kant’s Theory of Mind (Oxford: Oxford University Press, 1982), chap. 6.

318

Ethics

January 2008

defends this claim by arguing that SIEC rules out the two most plausible conceptions of a perfectly rational agent—those of an essentially infallible agent and of an ideal finite agent.5 I will grant that these are the only two pertinent conceptions and focus on his argument that SIEC rules out the conception of an ideal finite agent—an ideal agent who, unlike an essentially infallible agent, can possibly fail to live up to the relevant norm (even if she actually does not). According to Lavin, a finite conception of perfect rational agency must explain why an ideal agent conforms to norms of rationality, and it does so by appeal the agent’s ideal, “state of will, a disposition, habitus, or hexis” (451). His basic argument is that commitment to SIEC undercuts this explanatory strategy. To begin, Lavin notes that the explanation must allow for a distinction between “perfect rationality and its imitation” (451). More specifically, it must allow for a distinction between action that accords with a norm by accident and action that accords with a norm by no accident; an ideal agent conforms to norms of rationality by no accident. About this much, I think Lavin is right. But, with only this much said, it is unclear why proponents of SIEC cannot allow for the relevant distinction. The strong imperatival interpretation of the error constraint entails that if A is subject to a norm N, then A could violate N as the result of a genuine practical defect. But that has no obvious bearing on whether A could live up to N by accident or not. It seems, for example, that the proponent of SIEC could say that if A is under N, then, ceteris paribus, three situations are possible: (1) A lives up to N by accident, (2) A lives up to N by no accident, and (3) A violates N as the result of a genuine practical defect. Of course, we may still ask for an account of the difference between living up to N by accident or not, but it is unclear why SIEC makes this task difficult, let alone impossible. So how does Lavin argue that SIEC spells trouble for the attempt to distinguish between living up to a norm by accident and doing so by no accident? He proceeds in two steps. First, after reasonably insisting that we be able to distinguish between actions that accord with N by accident and those that do not, he glosses that demand as follows: “In order to distinguish perfect rationality from its imitation, it seems that one has to say something like this: if X is in an ideal state of will, then it must be no accident that when X acts, X acts correctly.” Next, he argues that, if we endorse SIEC, we can say no such thing: The imperativalist must accept that an agent is liable to error even when in possession of the ideal state of will. That is, within the 5. Lavin expresses skepticism about there being any viable third conception. He writes, “I have not proved that no conception of the ideal agent is available to the imperativalist. But what are the alternatives?” (451 n. 51).

Cokelet

Ideal Agency

319

imperatival framework, given some agent X which is in the “ideal” state, there is some circumstance in which X’s power to err would be exercised. . . . If there is some circumstance in which X, such as he is, would go wrong, and if it is only an accident that X does not find himself there, then for any state of the will that X is in, it must be an accident that X does not go wrong. And so, the imperativalist it not entitled to a conception of an ideal agent as what has a perfectly rational state of will. (451) But Lavin’s conditional claim that “if there is some circumstance in which X, such as he is, would go wrong, and if it is only an accident that X does not find himself there, then for any state of the will that X is in, it must be an accident that X does not go wrong” (my emphasis) is either false or fails to generate the conclusion stated on the next line, depending on what one means by “accident.” Let me explain. His conditional claim is false if the word “accident” in the consequent is meant in the same sense that it is meant when we say that it is no accident that someone lives up to a norm. Consider an example: if Sandy Koufax, while at the height of his powers, pitches excellently and strikes out every batter in a game, we say that it is no accident that he does so, even though there are other circumstances in which Sandy Koufax, while at the height of his powers, would fail to pitch well—for example, circumstances in which Sandy is tired, distracted, depressed, or intentionally throws the game. The claim that it is no accident that he pitches excellently—perhaps backed by a credit-bearing explanation that adverts to his high-grade pitching hexis—is not threatened by the fact that it is an accident (in the counterfactual sense) that he does so. Analogous points could be made about the actions that result from Sandy’s spotless means-end reasoning about how to get home from the ballpark or about my constructing a valid argument in front of my logic class; if I had let my mind wander during class I would not have constructed a valid argument, so it is only an accident (in the counterfactual sense) that I did not do so, but those facts do not show that my actual construction of a valid proof was something I did by accident. More generally, the fact that it is an accident that some defeating circumstance D is not actual does not rule out the claim that it is no accident that X lives up to N. The flaw in Lavin’s argument is further highlighted if we take the word “accident” in the consequent of his key conditional (“for any state of the will that X is in, it must be an accident that X does not go wrong”) to mean that X’s living up to the principle is an inessential or contingent feature of X. In this counterfactual sense, Lavin’s conditional is true; the possibility of my letting myself be distracted by an attractive member of my audience does establish that it is an accident—that is, it a contingent fact—that I constructed a valid proof. But that use of the word

320

Ethics

January 2008

“accident” is not the one we have in mind when we say that someone has lived up to a norm by accident. The distinction we need in order to have a concept of an ideal rational agent is that between living up to a norm by accident and living up to it by no accident, and that distinction could be correctly glossed as being between cases in which the agent can and cannot be given credit for living up to the norm, in the sense that one can or cannot attribute the success to him (or his will, his disposition, his habitus, his hexis, etc.). And although that distinction is hard to clarify—as the literature on responsibility shows—it differs from the distinction between cases in which it is a contingent fact that the agent lives up to the norm and cases in which that is a necessary or essential (no mere accidental) fact. II. IDEAL FINITE AGENCY AND THE “ABILITY” TO FAIL I think that Lavin’s argument is flawed because he conflates what we can call the “possibility” and “credit-bearing” senses of “no accident,” but even if he were to concede that fact, he might contend that I have not reached the heart of the issue. My criticism hinges on the claim that a finite agent can live up to a norm by no accident even if it is possible, in a counterfactual sense, for the agent to fail to live up to it. But I need to do more than that if I am to show that strong imperativalism is compatible with the conception of ideal finite agent. To see why, remember that SIEC entails not just that it is counterfactually possible for an agent to fail to live up to a norm but that it is possible for the agent to fail to live up to the norm and to do so as the result of a genuine practical defect. As Lavin puts it, SIEC entails that if A is subject to some norm N, then (i) A has the ability to violate N as the result of a genuine practical defect, and (ii) given that A is a finite ideal agent, “there is some circumstance in which A’s power to err would be exercised” (451). With this point in mind, Lavin could raise two challenges: first, to do more to establish that SIEC and ideal finite agency are compatible; second, to explain how someone can have the ability to violate a rational norm and still count as an ideal finite agent. I will tackle these challenges in turn. The first challenge is to show that a finite agent can live up to a norm by no accident even if, in some counterfactual circumstance she would fail to live up to the norm and fail to do so as the result of a genuine practical defect. We can show this by fleshing out the Koufax case introduced earlier. Imagine that, at the start of the eighth inning, Sandy experiences elation occasioned by the prospect of perfection and subsequently fails to maintain his concentration—he does not gather his wits by breathing deeply and counting to three, as he normally does before winding up to deliver a pitch. As a consequence, he pitches poorly and gives up a hit. In this counterfactual case, Koufax fails to live up

Cokelet

Ideal Agency

321

to the norm of excellent pitching and he fails to do so because of a genuine practical defect. But imagine that in the actual circumstance Sandy does better. He notices his growing elation, keeps it in check, and, with steadfast concentration, flawlessly executes his pitching routine. What’s more, he performs in like (exemplary) fashion throughout the remainder of the game and does so by no accident. In that case he stands, I think, as a model of ideal finite agency; as the example illustrates, it is an agent’s actual, nonaccidental performance that determines whether she lives up to a norm in ideal fashion, and one can actually, nonaccidentally execute an ideal performance even if one would, as the result of a genuine practical defect, fail to do so in some counterfactual circumstance.6 This response to the first challenge assumes that being an ideal finite agent is a matter of actually living up to norms (by no accident); the second challenge questions this assumption. The challenge can be put as follows: the “ability” to violate some norm N as the result of a genuine practical defect is itself a practical defect or deficiency and— because being ideal is incompatible with having a practical defect—any agent who has such an ability cannot count as an ideal agent; by extension, since SIEC entails that any agent who is subject to a norm has the ability to violate it as the result of a genuine practical defect, it rules out the conception of an ideal agent who is subject to a norm.7 Despite a surface plausibility, this argument is unsound. Contrary to what it assumes, an agent can be an ideal finite agent even if she has an ability to fail to live up to a norm she is under. It is exercising such an ability, not just having it, that counts, in the relevant sense,8 as a practical defect and makes an agent less than ideal. If a finite agent actually always lives up to every rational norm, and does so by no accident (in the credit-bearing sense), then that agent is an ideal finite agent in the sense that matters as far as the debate over constitutivism is concerned; constitutivists can argue that she fixes a standard for the activity in question. And an agent can be ideal in that sense even if she has an 6. What I say here leaves open the possibility that being an ideal agent depends on more than one’s actual performance. If, for example, an agent lives up to a norm by no accident only if she has certain dispositions, and those dispositions are not reducible to categorical bases, then one might argue that being an ideal agent (of the relevant sort) depends on more than the agent’s actual performance. 7. I am grateful to an anonymous referee for raising this objection. 8. Of course one may call the unexercised ability to fail “a practical defect,” if one likes, but that will help Lavin’s case only if there is an argument showing that an ideal finite agent cannot be thought of as having a practical defect in that sense if it is to play the role that constitutivists need it to play, namely, that of fixing a standard for the activity in question. As far as I can tell, nothing that Lavin says indicates that or how such an argument could be constructed.

322

Ethics

January 2008

unexercised ability to fail to live up to the relevant norms (as the result of genuine practical defects). As this brings out, SIEC entails that if an agent is subject to a norm, then it is possible for her to fail to live up to it as the result of a genuine practical defect, but SIEC does entail that her positive ability to live up to that norm is defective. For example, Sandy’s counterfactual poor pitching does not reveal a defect in his ability to pitch well. This point must be kept in mind in order to avoid confusion in cases in which we talk about the ability to X in addition to the ability to X well. In such cases, SIEC entails that any agent who is subject to a norm of Xing has the ability to X defectively, but it does not entail that the agent’s ability to X or her ability to X well is defective. For example, Sandy’s counterfactual failure to pitch well reveals that he has the ability to pitch defectively, but it does not show that his pitching ability or his ability to pitch well is defective.9 By parity of reasoning, the fact that one has the ability or potential to exhibit genuinely defective practical reasoning does not reveal that one’s general ability to engage in practical reasoning or one’s ability to reason well is defective. If one actually deploys one’s ability to reason in ideal fashion, one is a prototype for ideal finite practical agency, despite one’s ability to do worse; for all Lavin has shown, constitutivists like Korsgaard can argue that an agent who, by no accident, performs like that does “fix the standard constitutive of the activity in question” (454). III. CONCLUSION I have argued that Lavin gives us no reason to think that those who endorse the strong imperatival interpretation of the error constraint are unable to make sense of an ideal finite agent—one who could be described as living up to a norm by no accident—but even if I am right about Lavin’s argument, SIEC and constitutivism might turn out to be incompatible in the end. To prove otherwise, a proponent of SIEC would need to offer an account of the distinction between living up to a rational norm by accident and doing so by no accident, to which Lavin helpfully draws attention but that he also unfortunately misconstrues. For my part, I see no reason to doubt that proponents of SIEC can account for that distinction, but I am nonetheless inclined to reject SIEC. Simply put, I do not think that we need to adopt SIEC in order to make sense of excellent practical agency, and I suspect

9. At least not in the sense of “defect” that is relevant to the issues at hand (see n. 8). I am grateful to an anonymous referee for pressing me on this point.

Cokelet

Ideal Agency

323

that debate over the truth of SIEC is misleading because it distracts our attention from more significant differences between Aristotelian and Kantian accounts of acting well, by no accident. But those are thoughts to pursue on another occasion.

Ideal Agency and the Possibility of Error* Bradford ...

My criticism will focus on Lavin's argument for P1; I grant P2 and the argument's .... Of course, we may still ask for an account of the difference between living up to N .... call the “possibility” and “credit-bearing” senses of “no accident,” but even.

43KB Sizes 1 Downloads 110 Views

Recommend Documents

ERROR THEORY AND THE POSSIBILITY OF ...
morally right or good, then normative ethics loses its point. I do not believe .... ment can, in cases, only frustrate desire, and stifle self-interest. Those who .... There is something to be said for his saving the girl, even if Thrasymachus can't

Estimation of the Possibility of Cladoceran Invasion and Survival ...
Aug 5, 2014 - pose was to analyze the survival strategies of the test .... culate arithmetic means from these data, as food con- .... Analysis of Interactions.

On Probability and Systematics: Possibility ... - Oxford Journals
tical phylogenetic techniques do, in fact, conform with. Siddall and ... users in a similar position to make inferential claims con- ..... signing probabilities to the actual states of affairs. ..... ing and building electronic infrastructure in the

1 Knowability, Possibility and Paradox
Jan 8, 2007 - fail because, even if they do block Fitch's formulation of the paradox, not pinning ... Our trouble with knowability may be that we just do not know ... may generalize, since line (5) rests on no assumptions containing the name 'p'. ...