Reflective Equilibrium and Moral Consistency Reasoning

Abstract

It is more than a half-century since Nelson Goodman [1955] applied what we call the Reflective Equilibrium model of justification to the problem of justifying induction, and more than three decades since Rawls [1971] and Daniels [1979] applied celebrated extensions of this model to the problem of justifying principles of social justice. The resulting Wide Reflective Equilibrium model (WRE) is generally thought to capture an acceptable way to reconcile inconsistency between an intuitively plausible general principle and an intuitively plausible judgment about a particular case. Recently a different model for reconciling moral inconsistency has emerged: Moral Consistency Reasoning [Campbell and Kumar 2012, 2013a; Kumar and Campbell 2012; Campbell 2009: 86–7; Campbell and Woodrow 2003; Wong 2002]. MCR applies when two moral judgments give opposing assessments of (what appear to be) relevantly similar particular cases. Though WRE and MCR are strikingly different, each arguably captures a rationally acceptable method for reconciling moral inconsistency. Moreover, as will be shown, they function in complementary ways. Are they parts of a more comprehensive model of moral reasoning in the face of inconsistency that would explain the attractions of each?

This essay first spells out the relevant differences between the models and then formulates a more general model of moral reasoning in the face of inconsistency. §1 reviews the emergence of Goodman's model that he offers in the spirit of epistemology naturalized, almost a decade before Quine coined the term [1969a]. §2 analyses six salient features of WRE to be compared with six contrasting features of MCR in §3. §4 presents the general model.

Notes

¹ That is, the study of knowledge empirically as a natural phenomenon. See Quine [1969a: 82–3]. For an up-to-date defence of the view that ‘knowledge’ is a natural kind term, see Kumar [2013a].

² Since the universal validity of any inference can be in doubt on this method if it leads to contradiction, judging the validity of universal instantiation to be an exception in such cases is surely normative.

³ For example, classical utilitarianism implies that slavery can be just only given certain assumptions about marginal utility; see Rawls [1971: 158f].

⁴ At least Rawls does not give a realist interpretation of moral truth. See Rawls [1980].

⁵ Moral internalists, such as McNaughton [1988], maintain that moral beliefs necessitate desires and emotions. Even so, the reflective equilibrium is to be among beliefs, however beliefs are interpreted.

⁶ For example, one can maintain that no moral beliefs exist, strictly speaking, but also hold that moral claims, understood expressively, can be formally inconsistent. See Kalderon [2005].

⁷ ‘I shall not even ask whether the principles that characterize one person's considered judgments are the same as those that characterize another's’ [Rawls 1971: 50]. In later work, when Rawls [2005] is interested in ‘overlapping consensus’ among persons with different values, a measure of abstractness in WRE is critical. I am grateful to an anonymous referee for this important observation.

⁸ The story, based on first-hand accounts, is taken from Howarth [1955: 98–114].

⁹ Some but significantly not all of these points of contrast are addressed by Campbell and Kumar [2012].

¹⁰ Note that to deny that the explicit-principle feature is present is not to say that there are no principles operating implicitly in their thinking. For example, in finding themselves repelled by the prospect of the Oslo family's surrendering Marius, they may be implicitly responding to what is called the Principle of Double Effect, once they have the comparison of cases before their minds, since the sacrifice of Marius and Jan too would be seen as their being used simply as a means to avoiding a bad result, rather than as a foreseen but unintended bad side-effect. (The principle forbids using someone directly as a means to avoiding a bad effect but permits causing harm as an unintended but foreseen effect.) There is empirical evidence to suggest that people tend to be unconsciously guided by this principle in their judgments [Cushman et al. 2006; but also see Mikhail 2008]. Whether this is so or not, neither this principle nor any other enters explicitly and consciously into their locating the inconsistency, unlike what would be true in a strict application of WRE. Precisely for this reason our example does not exemplify the explicit-principle feature of WRE.

¹¹ I owe this point to Chike Jeffers.

¹² Richard Brandt [1979: 169] makes this point in defining a social moral system. For a recent important discussion of the relation of moral belief to justified moral motivation, see Kernohan [2010]. As I indicate in §4, a moral response can be a moral belief or a combination of motivation and desire that normally is part of a moral judgment or both [Campbell 2007b].

¹³ This definition of practical inconsistency applies also when R and R* are moral beliefs. Note that in the example R and R*, even as beliefs, are not formally inconsistent, since they are about distinct cases.

¹⁴ In §4 I offer a schematic representation of the logical structure of this reasoning.

¹⁵ Precisely, D is a difference between x and y if and only if D is a feature of x and not y or vice versa.

¹⁶ The points to follow on this example are adapted from Campbell and Kumar [2012].

¹⁷ The logical structure of this ramifying is represented in the recursive schema given in §4.

¹⁸ Where moral responses are to the same case (see §4), they can be also formally inconsistent.

¹⁹ But then, for the same reason, persons’ being numerically distinct could be morally relevant in any comparison. Though it is theoretically possible for someone to be relativist in this respect, arguably it would be difficult practically for members of any group to treat mere numerical difference as morally relevant without undermining the ability of consistency reasoning to enhance social coordination of those members. For an elaboration of the coordination function of consistency reasoning, see Campbell and Kumar [2012: 299–304] and Campbell [2009: 85–90]. I owe this objection to Darren Abramson.

²⁰ Sue Campbell [1997] has argued that our experiences of emotion, even before we reason about them, are inherently social, implying that consistency reasoning among persons is doubly social.

²¹ See, for example, Haidt [2001, 2007], Greene and Haidt [2002], Greene [2008], who give evidence of the emotional basis of moral judgment. They argue, however, that emotion creates bias in morals. For the contrary view that emotion in moral judgment can enhance their rationality, see Woodward and Allman [2007], Allman and Woodward [2008], Campbell and Kumar [2013a; 2012], and Kumar and Campbell [2012].

²² In the narrow sense of holding that moral judgments are neither true nor false.

²³ For a review of problems and solutions regarding the concept of moral knowledge, see Campbell [2011].

²⁴ Expressivism and naturalized epistemology are often linked; see Campbell and Hunter [2000: 1–28]. For a suitably broad understanding of naturalized moral epistemology see Walker [2000].

²⁵ Kumar [2013b, 2013c] argues that these elements form a unified natural kind with an empirical identity.

²⁶ For a detailed interpretation of this realist qualification, see Campbell [1998: 155–60].

²⁷ Campbell and Kumar [2012: 311–12] sketch a unified normative model without working out the details.

²⁸ For elaboration see Campbell and Woodrow [2003], Campbell [2009], and Campbell and Kumar [2012].

²⁹ Campbell and Kumar [2013b] explore these examples in relation to consistency reasoning.

³⁰ I offer a less partial picture of the conditions for full justification of moral responses in [2009].

³¹ For an insightful discussion of the role of trust in moral justification, see Tobin [2011].

³² I am grateful to Victor Kumar for his encouragement and critical assessment of multiple drafts and to three anonymous referees for their detailed constructive comments. The faults remaining are mine.