EDWARD D. SHERLINE
CONFIRMATION THEORY AND MORAL JUSTIFICATION*
(Received 15 September 1993)
The ultimate aim of a metaethical theory is to establish the objectivity of morality. 1 For one envious of scientific objectivity, assimilating moral justification into empirical justification is a promising way to accomplish this. According to the orthodox view of empirical justification, which I shall call "explanationism," a theory is justified if and only if it is either needed in the best empirical explanation of the world, or is precisely reducible to a theory that is needed in the best empirical explanation of the world, or if it is only vaguely reducible to a theory needed in the best empirical explanation, it is practically useful in the best empirical explanation. 2 If we couple explanationism with a nonreductionist view of morality, then to avoid skepticism we must hold that moral theory is needed in our best empirical explanation of the world. What results is the version of ethical naturalism David Copp calls "confirmation theory" or "CT. ''3 CT, then, has three basic tenets:
(1)
A normative moral theory is justified in the same way that any other empirical theory is justified. I shall call this "the continuity thesis. ''4
(2)
An irreducible empirical theory is justified if and only if it is needed in the best empirical explanation of the world. This is explanationism.
(3)
Some moral theory is justified because it is needed in the best empirical explanation of the world. I shall call this "the need thesis."
Philosophical Studies 73:225-238, 1994. © 1994 Kluwer Academic Publishers. Printed in the Netherlands.
226
EDWARD D. SHERLINE
CT is an archetype of the kind of nonreductive naturalist theory that many contemporary philosophers hold. 5 Even if no one holds CT in precisely this form, however, it is well worth considering because of its purity, simplicity and intuitive plausibility. The best known objection to CT is that our best empirical explanations don't need moral concepts, propositions, etc. 6 As the critical literature has developed, it has become increasingly clear that attacking the need thesis isn't the greatest threat to CT. On problem for such an attack involves settling on an account of the best explanation. Darwall et al. point out that "[s]ince no standard account exists of what makes for "the best explanation," even in paradigm natural sciences, it is unlikely that definitive refutation - or vindication - of nonreductive naturalism [their name for CT] will come from this quarter. ''7 A related problem is settling on an account of explanatory need. An adequate account of explanatory need should be neither too lenient one that recognizes bogus theories and assumptions - nor too strict one that rejects legitimate theories and assumptions. 8 Since there isn't a standard, non-question begging account of bogus and legitimate theories and assumptions, no standard account of explanatory need exists. For these reasons the naturalist can exploit the disputes over the best explanation and explanatory need to her advantage. Recently, however, Sayre-McCord and Copp have identified an objection to CT that avoids the controversies over explanationism. In metaphysical terms, the problem is that though CT can successfully establish that moral properties exist, these moral properties are normatively inert. 9 In epistemic terms, the problem is that although CT can successfully justify moral propositions, it fails to justify moral standards. 10 Because Copp has given the most forceful and systematic statement of this skepticism, I shall concentrate on his discussion. He argues that even if a moral theory, understood as a system of propositions and concepts, were needed in the best empirical explanation, this wouldn't have any tendency to justify any moral standard. 11 Since the aim o f a theory of moral justification is to establish the objectivity of moral standards, CT fails. Copp attacks the combination of the continuity thesis and explanationism, not the need thesis. -
-
CONFIRMATION THEORY AND MORAL JUSTIFICATION
227
Copp's conclusion is, more precisely, that CT at best establishes "incidental moral explanations." "A[n] explanation is incidental to a conceptual structure just when it is an explanation that invokes concepts or principles of the structure, where that structure cannot be justified on the basis of its role in successful explanation. 'q2 Even though our best empirical explanation needs the moral concepts or principles of the structure, this isn't a good reason for concluding that the structure is justified. Other examples of incidental explanations are explanations of law and etiquette. The best explanation of why Alan is in prison might be that the court convicted him of theft, but this doesn't have any tendency to justify the legal standards the court relied on to send Alan to prison. The best explanation of why Brenda was rude at dinner might be that she was talking with her mouth full, but this doesn't have any tendency to justify the standard of etiquette that one does not talk with one's mouth full at the dinner table. 13 Copp supports the thesis that moral explanations are incidental with an argument by counter-example. He identifies a standard that many people hold to be unjustified, and then shows that according to CT, this standard would be justified under certain conditions. Nonetheless, this shouldn't persuade us to change our original judgment that the standard is unjustified. Therefore, CT fails. Here is his statement of the argument: Consider the Nietzschean concept of overman. A Nietzschean might seek to explain Stalin's ruthless behavior on the basis that he was an (approximation to the) overman.
But this would be an ordinary psychologicalexplanation that incidentally invoked a Nietzschean moral concept. Psychologycould even adopt the Nietzscheanconcept and postulate the existence of an "overman" personality, but we would not be tempted in the least to accept Nietzschean morality on this basis. This is because we would not regard the explanatory utility of the concept as justifying any standard that treats being overmanlike as a virtue or ideal.14 In short, "the explanatory utility of the concept would not undermine our conviction that there is no warranted norm which prescribes that we aspire to become overpeople. ''15 This argument is convincing. At least this application of CT fails, and this moral explanation is incidental) 6 The naturalist has two basic ways of responding to this objection. The "so what?" response concedes that all moral explanations are incidental,
228
EDWARD D. SHERLINE
but argues that this doesn't pose a threat to an adequately formulated naturalism. The "oh yeah?" response argues that Copp's counterexampie doesn't establish that all moral explanations are incidental. 17 In this paper I shall consider the worst case assumption for the naturalist, that all moral explanations are incidental, and so develop the "so what?" response. My aim is to explore ways of reconstructing naturalism to include normativity, in spite of the assumption that all moral explanations are incidental.
For the naturalist who concedes that all moral explanations are incidental, there are three alternatives for maintaining normativity: first, keep CT intact; second, keep the continuity thesis, and reject explanationism; third, abandon the continuity thesis, and perhaps keep explanationism.
1. Keep CT Intact The naturalist cannot simply maintain CT in the face of the conclusion that all moral explanations are incidental. For according to Copp, the best explanation of the fact that all moral explanations are incidental is that morality is a system of standards. 18 If morality is a system of standards, then the purpose of a theory of moral justification is to justify moral standards. So if all moral explanations are incidental, CT fails as a theory of moral justification. The naturalist must offer an equally good alternative explanation of why all moral explanations are incidental. The naturalist's explanation of why all moral explanations are incidental is that one cannot use CT to justify moral standards. CT is an epistemic account of justification. Its objects are moral propositions, not moral standards. No reason has been given for thinking that moral explanations are incidental to the justification of moral propositions, although they might be incidental to the justification of moral standards. In summary, Copp's explanation of why all moral explanations are incidental is that morality isn't the kind of thing that one can justify
CONFIRMATION THEORY AND MORAL JUSTIFICATION
229
epistemically. The naturalist's explanation of why all moral explanations are incidental is that moral standards aren't the kind of thing that one can justify epistemically, but moral propositions are. There is good reason to prefer the naturalist's explanation to Copp's, for it presupposes a more thoroughgoing version of CT, one that more consistently applies the continuity thesis. Empirical theories are in large part systems of propositions and not merely systems of standards. One aim of a theory of empirical justification is to provide evidence that empirical propositions are true and not merely to justify empirical standards. This first option simply applies these ideas to moral theory. In contrast, Copp attacks a version of CT that doesn't follow through on the continuity thesis. It conceives of morality as a system of standards although it employs an account of justification that is normally used to establish the credibility of propositions. According to this first "so what?" response, then, the naturalist justifies moral propositions in the same way that she justifies empirical propositions, through CT. How then are moral standards justified? Where does normativity enter the picture? The answer is simple. The naturalist justifies moral standards the same way she justifies empirical standards, however that might be. At this point the defense of CT encounters a slight problem. For if the naturalist cannot justify moral standards through explanationism, then she cannot justify empirical standards through explanationism either. Doesn't CT assume that explanationism exhausts the account of empirical justification? If so, then the naturalist faces the troublesome conclusion that it isn't possible to justify empirical standards through an account of empirical justification. The solution is that explanationism exhausts the epistemic account of empirical justification, rather than all accounts of empirical justification. The general contrast to keep in mind is between epistemic and practical accounts of justification. An epistemic account of justification operates through a theory of knowledge, whereas a practical account of justification operates through a theory or conception of practical reason. 19 In other words, an epistemic account of justification aims to establish truth, whereas a practical account of justification aims to establish rationality. According to this first option the naturalist combines an epistemic justification of moral propositions with a practical justification of moral standards.
230
EDWARD D. SHERLINE
The continuity thesis applies not only to the justification of propositions, but also of standards. Say that empirical standards are hypothetical imperatives. In that case moral standards must also be hypothetical imperatives, and both moral and scientific standards would be justified based on whether following them helps us to better achieve our goals. On the other hand, say that some scientific standards are categorical imperatives, so their justification wouldn't be based solely on their instrumental value. In that case at least some moral standards must also be categorical imperatives, and the naturalist would use a different account of practical justification to justify both. It might help to contrast this new version of CT with the original version. The original version justifies propositions and standards (whether moral or scientific) through explanationism. In contrast, the new version justifies propositions and standards independently. Explanationism justifies scientific and moral propositions. A separate practical theory justifies scientific and moral standards. Sayre-McCord suggests a way to flesh out this first option. He begins with the argument that explanationism (in his terminology "the explanatory criterion") is a legitimate scientific standard. Thus explanationism is on a par with moral standards. Just as explanationism is an account of empirical justification, normative moral theory (a system of moral standards) is an account of moral justification (of our moral judgments and actions). Since there must be some higher-order theory that justifies explanationism, the hope is that the same higher-order theory also justifies moral standards. 2° So the naturalist starts with the higherorder theory of the justification of standards, and uses it to justify both scientific standards (including explanationism) and moral standards. Explanationism, in turn, justifies scientific and moral propositions, and moral standards justify moral judgments and actions. Pushing the continuity thesis this far, however, might strain our intuitions about morality as normative. To evaluate this worry, it is important to distinguish weaker and stronger ways that morality might be normative. The weaker position is that normativity is an essential (but not distinctive) feature of morality. Keeping CT intact provides an obvious and straightforward way for the naturalist to hold this position. She can point out that normativity enters the picture through standards
CONFIRMATION THEORY AND MORAL JUSTIFICATION
231
rather than through concepts, propositions, properties, and facts. This is true not only for morality, but also for science. Both morality and science have normative dimensions in virtue of their respective standards. So pushing the similarities between moral theory and empirical theories to the extreme degree envisioned above doesn't entail that morality lacks normativity. There is a stronger interpretation of the claim that morality is normative. This holds that normativity is a distinctive feature of morality, at least in the sense that the normativity possessed by morality differs from the kind possessed by science. For instance, most likely the distinctive normativity of morality is that moral standards are categorical imperatives, whereas explanationism and other scientific standards are hypothetical imperatives. This position on the normativity of morality poses an insurmountable problem for combining normativity with CT. For according to CT, science is normative in the same way that morality is, moral standards are the same kind as scientific standards, and the same theory justifies both. CT is incompatible with the position that morality possesses a distinctive normativity. Of course the advocate of CT might deny that morality is distinctively normative, but that is an issue for another paper. For the sake of exploration, let's grant that morality is distinctively normative. This shouldn't bother the naturalist too much, for there are still two other options available, if she is willing to give up some element of CT. The continuity thesis frames the naturalist's two choices, since this is the core of CT. These choices are either to keep the continuity thesis, and so to reject explanationism, or to reject the continuity thesis (and perhaps to keep explanationism).
2. Keep the Continuity Thesis In keeping the continuity thesis and rejecting explanationism, the idea is to loosen up the empirical account of justification to fit the conclusion that the naturalist cannot justify moral standards through explanationism. To contrast this response with the one just considered, I shall distinguish between "full CT," the newer version of CT that keeps all
232
EDWARD D. SHERLINE
three claims, and "partial CT," which keeps the continuity thesis but rejects explanationism. The naturalist must reject the element of explanationism which holds that being properly related to our best explanation is individually sufficient for justification. 21 There are two ways that the naturalist can compensate for the deficiency of explanationism. First, she might also give up the element of explanationism which holds that being properly related to our best explanation is necessary for justification, and employ an alternative account of empirical justification. Second, she might hold that being properly related to our best explanation is a necessary or important constraint on justifying standards, but it needs to be supplemented with other constraints. 22 Of course the naturalist doesn't have carte blanche here. Her account of empirical justification must be an accurate depiction, and she must show that the account succeeds in justifying moral standards. Partial CT looks very much like the original version of CT. Both rely on the same theory to justify scientific propositions and standards, on the one hand, and moral propositions and standards, on the other. 23 The only difference is that partial CT expands the epistemic account of empirical justification. It differs more significantly from full CT. Partial CT attempts to justify standards through the same epistemic account used to justify propositions, whereas full CT teases apart the justification of propositions and standards. Of the two "so what?" responses described so far, full CT is preferable to partial CT. This isn't because partial CT's rejection of explanationism is a mistake. Instead, the worry is that if explanationism fails to justify moral standards, then no other account of epistemic empirical justification has any better promise of success. For explanationism is in some sense more lenient than other traditional constraints of epistemic empirical justification, such as testability, observability, verifiability and falsifiability. So even if there are alternatives to explanationism, given its failure they aren't very promising ways to epistemicallyjustify moral standards. It might appear that the naturalist can easily overcome this problem with the use of reflective equilibrium. The constraint of overall coherence - arising from reflective equilibrium - seems more lenient than the
CONFIRMATION THEORY AND MORAL JUSTIFICATION
233
traditional positivist constraints mentioned above (testability, observability, etc.). The problem for this suggestion arises because reflective equilibrium is a coherence theory of moral justification. It is reasonable to hold that coherence isn't merely a matter of logical consistency, but also a matter of mutual explanatory dependence. 24 Very roughly, this constraint holds that A coheres with B only if we need A in B explanations and B in A explanations. Thus if moral explanations are incidental, moral standards don't pass this coherence test even if considered moral judgments support them. For by hypothesis they aren't properly related to the best empirical explanation of the world, and so they don't cohere with the best empirical explanation. So reflective equilibrium fails to justify moral standards if moral explanations are incidental.
3. Reject the Continuity Thesis The naturalist might be uncomfortable with partial CT because it is an unpromising way to justify moral standards, and might also be uncomfortable with full CT because it fails to capture the distinctively normative character of morality. Then her only remaining option is to reject the continuity thesis, that moral justification is a kind of empirical justification. Of course showing that a discontinuity version of naturalism can maintain the normativity of morality is in some sense uninteresting. For once the naturalist rejects the continuity thesis, the constraints on naturalism are so weak that the problem of developing a normative naturalism becomes easier (no one doubts the normative credentials of Hobbesian practical reasoning theories of morality). Nonetheless, a look into discontinuity naturalism isn't without some merit, for it helps the naturalist to identify a theory that differs from traditional discontinuity theories. For one who longs for CT, the most attractive version of discontinuity naturalism looks much like full CT except that it rejects the continuity thesis (so I shall call it "discontinuity CT"). Both use explanationism to justify scientific and moral propositions. Both use some form of practical justification to establish the objectivity of scientific and moral standards. The major difference is that full CT uses the same account of practical justification to justify both scientific and moral standards, whereas discontinuity CT employs different
234
EDWARD D. SHERLINE
accounts to justify scientific and moral standards, since it holds that these two kinds of standards differ. The major difference between discontinuity CT and traditional discontinuity theories is in the former's use of explanationism and its embrace of moral realism and cognitivism. Standard discontinuity theories, such as Hobbesian and Kantian practical reasoning theories, Rawlsian constructivism and noncognitivism, are antirealist and avoid explanationism. The question arises whether discontinuity CT can abandon the continuity thesis without giving up the two major benefits of naturalism: establishing moral objectivity, and insuring compatibility between ethical and scientific presuppositions. Though discontinuity CT gives up the continuity thesis, it justifies moral propositions through explanationism. Therefore discontinuity CT retains the objectivity that comes with the use of explanationism to justify moral propositions. The question lingers of how discontinuity CT establishes the objectivity of moral standards. This, of course, depends on the particular account of the justification of moral standards. Nonetheless this isn't a more pressing issue for discontinuity CT than it is for other versions of discontinuity naturalism, or even for full CT. What about the worry that discontinuity CT permits the development of an ethical theory that contains claims incompatible with the presuppositions of science? Clearly the rejection of the continuity thesis doesn't commit the naturalist to discontinuity across the board. She still might sensibly require that all presuppositions of the ethical theory must be consistent with the presuppositions of our most advanced sciences. The rejection of the continuity thesis is only the rejection of methodological continuity between science and morality, not substantive continuity between the two realms. Even after the naturalist has ruled out partial CT, two "so what?" responses and their reconstructions of CT remain. Which is most plausible depends on several large issues that I cannot adequately address in this paper, starting with whether moraland empirical standards are similar, both being either categorical or hypothetical, or whether they are sufficiently different so the same theory cannot succeed in justifying them. 25
CONFIRMATION THEORY AND MORAL JUSTIFICATION
235
III
I shall finish by addressing an objection Copp might raise for the versions of CT defended in this paper. Copp proposes to explain the normativity of morality in terms of moral standards. Of course both full and discontinuity CT have no problem with this proposal. However, he further stipulates that an adequate account of moral normativity should distinguish normative facts from facts with normative significance, that is, facts necessarily connected with moral standards and facts contingently connected with moral standards. 26 In semantic terms, the account should distinguish normative propositions and propositions with normative significance. Full and discontinuity CT might seem to have a problem satisfying this condition. For they separate propositions, properties, and facts on the one hand, and standards on the other. They justify propositions through explanationism, whereas they justify standards through an entirely unrelated practical conception of justification. Thus a justified proposition isn't necessarily connected to a justified standard, and vice versa. This distinction might seem to commit the naturalist to denying the distinction between normative facts and facts with normative significance. Despite appearances, she can recognize this distinction while at the same time justifying propositions and standards differently. She holds that normative facts and propositions are necessarily connected to standards, but the standards aren't necessarily justified in the same way the facts and propositions are justified. Thus we cannot conclude from a justified normative proposition (or that some moral fact exists) that its connected standard is also justified. So the naturalist can provide accounts of normative propositions and facts while acknowledging that moral explanations are incidental and so she can recognize this distinction. The difficulty with this solution is that it seems to violate a standard inference of logic: If P entails Q, and P is true, then Q must also be true. Notice, however, that the naturalist's position is in terms of justification, not truth, and there is a good reason for this. Standards don't have a truth value. Thus it would be a mistake to use the truth-functional
236
EDWARDD. SHERLINE
notion of entailment to describe the relation between propositions and standards. Instead, the rule that threatens the naturalist's solution must be: I f P is necessarily connected to Q, and P is justified, then Q is also justified. The nonreductive naturalist has a way around this rule. For although the rule obtains when P and Q are identical, it d o e s n ' t obtain when the relation is constitution. For in general it isn't the case that if P is constituted by Q, then all properties o f P are properties o f Q, and vice versa. Thus the naturalist needs to hold that n o r m a t i v e facts are constituted by moral standards, but the two aren't identical. In semantic terms, she needs to hold that n o r m a t i v e propositions are defined using moral standards, but the two d o n ' t have the same meaning. So even if the worse case scenario holds for the naturalist, and all moral explanations are incidental, this d o e s n ' t c o n d e m n either C T or naturalism. There are a n u m b e r of ways that the naturalist can recognize the normativity of morality and maintain the attractive elements of CT.
NOTES * This is a revised version of a paper given at the Pacific Division APA, San Francisco, CA, March 1993. An earlier version of this paper was presented at the Mountain-Plains Philosophy Conference, Colorado State University, Fort Collins, Colorado, October, 1991. I am indebted to Sharron Connor, David Copp, Mary Gore Forrester, Susanna Goodin, Russell Jacobs, Shelly Kagan, David Resnik, and Seana Shifrin. 1 1 use 'objective' in Darwall et al.'s sense "of a kind consistent with a respectable resolution of a range of issues- epistemological, metaphysical, semantic - that in philosophical common sense are characteristically bundled together in the idea of objectivity" (1992, 126 note 29). 2 Harman, 1977,21-23. 3 Copp, 1990,237. 4 Darwall et al., 1992, 130. According to them, the continuity thesis holds that "moral judgments are factual in the paradigm sense afforded by empirical or theoretical judgments in the natural sciences" (1992, 129-30). If moral judgments are factual in this sense, then it must be because moral methodology and epistemology are parts of empirical methodology and epistemology. 5 Copp (1990, 237 note 1) guardedly refers to Peter Railton (1986, 163-207), Richard Boyd (1988, 181-228), and Nicholas Sturgeon (1984, 49-78). It would be safe to add David Brink (1989) to this group, and also to remove Peter Railton from it, since his version of naturalism is reductive, and the version of CT Copp has in mind is non-
CONFIRMATION THEORY AND MORAL JUSTIFICATION
237
reductive. On the other hand, in the following discussion nothing turns on whether the version of CT is nonreductive or reductive. 6 Harman, 1977, 6-9. 7 Darwall et al., 1992, 171. 8 Sayre-McCord, 1988,270-276. 9 Ibid., 276. 10 Sayre-McCord is more sanguine than Copp on the prospects of saving CT. As noted later in the paper, Sayre-McCord thinks there is a way of conjoining normativity to CT. 11 I shall not distinguish normative and evaluative standards, since the skeptic's argument against CT applies to both kinds of standards. Copp's version of the argument aims mainly at normative standards, while Sayre-McCord's worry concerns evaluative standards, standards concerning the evaluation of moral properties as better and worse. lz Copp, 1990, 246. 13 Ibid. 14 Ibid., 248. 15 Ibid. 16 Seana Schifrin raises doubts about the force of Copp's counterexample because it uses the wrong explanandum, Stalin's behavior, instead of considered moral beliefs (Schifrin, 1993, 2-3). Although I am sympathetic to her point, it isn't so much doubting the force of the counterexample, since she concedes that moral explanations of behavior are incidental, as doubting that moral explanations of considered moral judgments are incidental. 17 For the distinction between "so what?" and "oh yeah?" responses, see Sturgeon, 1986, 115. 18 Copp, 1990, 254--55. 19 Copp, 1991a, 216-17. 20 Sayre-McCord, 1988, 277-81. 21 Railton's account of moral justification seems closest to this latter option. For being properly related to our best explanation (in his terms, being "responsive to empirical demands,") is an important condition for justification, but not sufficient. Anotherimportant condition is being responsive to normative intuitions (1986, 205-206). 22 This point is indebted to Sayre-McCord's distinction between two versions of the explanatory criterion, as a necessary and sufficient condition, or merely as a necessary condition, for justification (1988,267). 23 This is a slight oversimplification. It might be the case that the standards (scientific and moral) aren't justified directly through the epistemic account of empirical justification, but instead, are derived from propositions that are themselves justified through the epistemic account. 24 Brink attributes this view to Harman, and subscribes to it himself (1989, 182). 25 For a discussion on whether scientific standards are hypothetical or categorical, see Resnik, 1992, 498-507. 26 Copp, 1991b, 623.
238
EDWARD D. SHERLINE REFERENCES
Brink, David O. (1989) Moral Realism and the Foundations of Ethics, NY: Cambridge University Press. Boyd, Richard (1988) "How to Be a Moral Realist," in Sayre-McCord, Geoffrey, ed., Essays on Moral Realism, Ithaca, NY: Cornell University Press, pp. 181-228. Copp, David (1990) "Explanation and Justification in Ethics," Ethics 100, 237-258. Copp, David (1991 a) "Moral Skepticism," Philosophical Studies 62,203-233. Copp, David (1991b) "Moral Realism: Facts and Norms," Ethics 101,610-624. Harman, Gilbert (1977) The Nature of Morality, NY: Oxford University Press. Railton, Peter (1986) "Moral Realism," The Philosophical Review 95, 163-207. Resnik, David B. (1992) "Are Methodological Rules Hypothetical Imperatives ?," Philosophy of Science 59,498-507. Sayre-McCord, Geoffrey (1988) "Moral Theory and Explanatory Impotence," in SayreMcCord, Geoffrey, ed., Essays on Moral Realism, Ithaca, NY: Cornell University Press, pp. 256-281. Schifrin, Seana (1993) "Response to Confirmation Theory and Moral Justification," American Philosophical Association Pacific Division Meeting, March, 1993, San Francisco, CA, Sturgeon, Nicholas (1984) "Moral Explanations," in Copp, David and Zimmerman, David, eds., Morality, Reason and Truth: New Essays on the Foundations of Ethics, Totowa, NJ: Rowman & Allenheld, pp. 49-78. Sturgeon, Nicholas (1986) "What Difference Does It Make Whether Moral Realism Is True?" The Southern Journal of Philosophy 24, 115-141.
Department o f Philosophy University o f Wyoming Laramie, W Y 82071 USA