Download Handout #2: Moral Motivation and Moral Semantics

Draft of 10-­‐15-­‐14 PHIL 202: Core Ethics Fall 2014; Classics in Metaethics David O. Brink Handout #2: Moral Motivation and Moral Semantics There is a puzzle about moral motivation that we encounter when we try to reconcile intellectual and practical aspects of morality. Cognitivists interpret moral judgments as expressing cognitive attitudes, such as belief. Moral judgments ascribe properties – axiological, deontic, and aretaic – to persons, actions, institutions, and policies. Internalists believe that moral judgments necessarily engage the will and motivate. We expect people to be motivated to act in accord with their moral judgments and would find it odd for people to be systematically indifferent to what they judge morally significant. It is also a common view that motivation involves pro-­‐attitudes, such as desires. We explain intentional action as the product of informational states, such as beliefs, and practical states, such as desires. But beliefs and desires are logically independent states; no belief entails any particular desire. These assumptions are in tension. 1.
2.
3.
4.
Moral judgments express beliefs. Moral judgments entail motivation. Motivation involves a desire or pro-­‐attitude. There is no necessary connection between any belief and any desire or pro-­‐attitude. Different views in metaethics and moral psychology can be understood as responses, explicit or tacit, to the tension created by this quartet of claims, responses that avoid the tension by denying at least one of the constituent assumptions on the strength of others. Noncognitivists (e.g. Ayer, Stevenson, Blackburn, and Gibbard) reject (1), claiming that moral judgments express desires or other pro-­‐
attitudes, rather than beliefs. Externalists (e.g. Foot and Brink) reject (2), claiming that motivation is not internal or necessary to moral judgment, as such. Rationalists about moral motivation reject (3) or (4), claiming either that normative beliefs can motivate without the benefit of desires (e.g. McDowell) or that normative beliefs necessitate the desires that are necessary for moral motivation. (e.g. Nagel). SMITH'S MORAL PROBLEM Michael Smith's book is focused on a related tension, which he calls "the moral problem." His problem consists of three claims (12). 1. Moral judgments of the form 'It is right that I φ' express a subject's beliefs about an objective matter of fact, a fact about what is right for her to do. 2. If someone judges that it is right that she φs then, ceteris paribus, she is motivated to φ. 3. An agent is motivated to act in a certain way just in case she has an appropriate desire and a means-­‐end belief, where belief and desire are, in Hume's terms, distinct existences. It's worth noting some differences between Smith's problem and my puzzle. • Smith's (3) combines the elements in my (3) and (4). o These two elements are independent. Because rejection of these two elements provides two distinct solutions to the puzzle, it is useful to separate them. • Smith's version of internalism – his (2) -­‐-­‐ is weaker than mine – my (2). o Several central figures (e.g. Price, Stevenson, Harman, Mackie, Blackburn) commit themselves to the stronger internalist thesis. o The trio/quartet is not inconsistent if we employ the weaker internalist thesis, which renders arguments that reject one element on the strength of others invalid. 2 NONCOGNITIVISM One way to motivate noncognitivism (e.g. emotivism, prescriptivism, expressivism) is as a solution to the puzzle about moral motivation that preserves (2)-­‐(4). We can argue against cognitivism and for noncognitivism by appeal to internalism and belief-­‐desire psychology. There are various possible noncognitivist views. • Emotivism: moral judgments express (noncognitive) feelings. • Emotivism: moral judgments express desires and aversions. • Prescriptivism: moral judgments express (universal) prescriptions about how to feel or act. • Expressivism: moral judgments express plans for action (actual or hypothetical). All versions of noncognitivism contrast with rational intuitionism. As we noted, the British school were all rational intuitionists who combined a realist and non-­‐naturalist metaphysics of morals with an intuitionist moral epistemology. It was Moore’s OQA that forced their realism to take a non-­‐
naturalist form. 1. If moral judgments express the appraiser’s beliefs about the moral properties of people, actions, and situations, then these properties must be either natural (or metaphysical) properties or non-­‐natural properties. 2. Moral properties are not natural (or metaphysical) properties, because moral predicates are not definable in non-­‐moral terms. [OQA] 3. Hence, if moral judgments ascribe properties, they must be non-­‐natural sui generis properties. Of course, the rational intuitionists affirm the antecedent in (1) and (3) and so defend non-­‐naturalism. But one person’s conditional is another person’s contrapositive. Many noncognitivists, including the emotivists, accepted the OQA and so accepted (3). But they found non-­‐naturalism a metaphysically and/or epistemologically extravagant commitment, and so they rejected the antecedent of these conditionals. 1. If moral judgments express the appraiser’s beliefs about the moral properties of people, actions, and situations, then these properties must be either natural (or metaphysical) properties or non-­‐natural properties. 2. Moral properties are not natural (or metaphysical) properties, because moral predicates are not definable in non-­‐moral terms. [OQA] 3. Hence, if moral judgments ascribe properties, they must be non-­‐natural sui generis properties. 4. Non-­‐naturalism is false. 5. Hence, moral judgments do not express the appraiser’s beliefs about the moral properties of people, actions, and situations. Instead, moral judgments express noncognitive affective or conative attitudes. So began the development of noncognitivism. Its original form was emotivism, as found in A.J. Ayer (1910-­‐89) in Language, Truth, and Logic (1936/46) and C.L Stevenson (1908-­‐79) in his early paper “The Emotive Meaning of Ethical Terms” (1937), in his book Ethics and Language (1944), and in a series of essays collected in Facts and Values (1963). It would also be developed as prescriptivism by R.M. Hare (1919-­‐2002), especially in The Language of Morals (1952) and Moral Thinking (1981), and later as expressivism by Allan Gibbard in Wise Choices, Apt Feelings (1992) and Thinking How to Live (2003) and as quasi-­‐realism by Simon Blackburn in Ruling Passions (1998). TWO KINDS OF EXPRESSIVISM There is an extended sense of expressivism in which emotivism, prescriptivism, and quasi-­‐
realism are all forms of expressivism insofar as they claim that moral judgments fundamentally 3 express noncognitive emotions, desires, commitments, or plans. But it is worth distinguishing two different kinds of expressivism, corresponding roughly to early and late expressivism. Revisionary expressivism self-­‐consciously rejects the realist and cognitivist assumptions of the moral philosophy of the British school. It insists that despite the indicative or assertoric syntax and semantics of moral discourse, moral judgments are really expressive, rather than assertoric. Moral judgments are expressions of feeling, and are not truth-­‐apt. There is no such thing as moral truth or moral knowledge. The most we can hope to do is make sure that our moral feelings are well-­‐informed as to the non-­‐moral beliefs on which they rest. Ayer is the clearest representative of this kind of expressivism, but there are important echoes of this brand of expressivism in Stevenson and Hare. Ayer appeals to his empiricist theory of meaning to claim that cognitively meaningful claims must be either (a) analytic or (b) synthetic and empirically verifiable. He concludes that moral claims are neither; they are synthetic but unverifiable claims and, hence, are without cognitive significance. As a result, he rejects the possibility of a moral science (as distinct from a branch of psychology) and concludes that there is no distinctive contribution to ethics that philosophy can make once it establishes the truth of emotivism. Stevenson and Hare draw related conclusions about the limits of moral philosophy. Interestingly, even less revisionary expressivists, such as Gibbard and Blackburn, retain the revisionist’s idea that moral philosophy is the study of “the logic of the moral concepts” (Hare’s phrase). At least, their own contributions to ethical theory talk about moral talk.1 But when expressivism takes this highly revisionary form, it may seem suspect. For the syntax and semantic of moral discourse is assertoric. Moreover, if we treat moral assertion as disguised expressions of noncognitive attitude, we encounter problems with the meaning of moral phrases in unasserted contexts, such as the antecedents of conditionals. Further, we recognize moral truth when we assert that some moral proposition or judgment is true. We can perhaps offer expressivist reconstructions of such ascriptions of truth to moral claims by adopting a deflationary or redundancy conception of ascriptions of truth, according to which to call a moral judgment true is just to agree with it. But it seems we need truth, and not just ascriptions of truth, to explain valid inferences involving moral contents. Consider this argument. 1. If it is wrong to murder innocent children, then it is wrong to pay someone else to murder innocent children. 2. It is wrong to murder innocent children. 3. Hence, it is wrong to pay someone else to murder innocent children. We can illustrate two concerns with this argument. This first issue concerns the unasserted context in (1). The conditional judgment may express an attitude toward paying someone to murder innocent children, although this attitude would be conditional on the wrongness of murdering innocent children. It expresses no attitude toward the wrongness of murdering innocent children. This is why the moral claim in the antecedent of the conditional in (1) involves an unasserted context. But what then is its meaning? Frege believed that an adequate semantic theory should recognize the univocity of terms across asserted and unasserted contexts of utterance. Peter Geach thought that Frege’s point posed problems for noncognitivist analyses of the meaning of moral judgments.2 The expressivist construes moral assertion as the expression of the appraiser’s attitudes, rather than a description of the way the world is. But then it is not clear how the noncognitivist understands the meaning of moral predicates in unasserted contexts, such as the antecedents of conditional statements. And this is relevant to the validity of the argument. For the argument to be valid, the relevant moral claims must have the same meaning in asserted and unasserted contexts. It’s hard to see how this could be true if we accept the expressivist analysis of 1 It is perhaps significant that Gibbard’s second book is entitled Thinking How to Live, rather than How to Live. 2 Peter Geach, “Assertion” Philosophical Review 74 (1965): 449-­‐65. 4 moral assertion. It’s easy to explain if moral propositions have the same indicative meaning in asserted and unasserted contexts. The second issue is that to explain the validity of the argument we apparently need to invoke truth. The reason the argument is valid is that it is not possible for the premises to be true and the conclusion false. This explanation of validity is truth-­‐theoretic. Here, it seems, we need truth and not just a deflationary semantics for ascriptions of truth. This is a condensed summary of some standard worries about the semantic and logical resources of revisionary expressivism. The conditional/contrapositive gambit can be recycled. If expressivism leads to all these semantic and logical problems, then that might be reason to reject expressivism and accept cognitivism and realism after all. Some might want to resurrect non-­‐
naturalism. Others will think that this makes a strong case for ethical naturalism, especially since we should reject the OQA in any case, in particular, rejecting the semantic test of properties, the assumption that meaning should be transparent, or both. But, of course, the expressivist might think that these semantic and logical problems for revisionary expressivism are not decisive. These problems motivate a different kind of expressivism. Accommodationist expressivism tries to show how there can be semantic and logical relations among our noncognitive attitudes toward moral contents, so that we can reconstruct or mimic validity, truth, and knowledge in moral matters within an expressivist framework. Revisionary and accommodationist expressivism are poles in a spectrum. Though there are accommodationist elements in both Stevenson and Hare, I think their form of expressivism is still predominantly revisionary. Expressivism has assumed a robustly accommodationist character only in the work of Gibbard and Blackburn. This is not the place to explore or assess the adequacy of accommodationist expressivism, which is an extremely complicated matter.3 But I thought it might be useful to situate the more revisionary form of expressivism contained in the emotivism of Ayer and Stevenson within this larger dialectical picture. AYER In Language, Truth, and Logic Ayer applies his empiricist theory of meaning to moral judgments and concludes that they lack cognitive significance, because they are neither analytic nor empirically verifiable. They are not analytic, as shown by a variation of the OQA (104-­‐05). Whereas it is “self-­‐contradictory” to deny that M-­‐things are M, it is never self-­‐contradictory to deny that N-­‐things are M. Ayer assumes that moral claims are not empirically testable, apparently without argument. We might dispute this. We might think that moral claims are empirically testable, but only in conjunction with suitable auxiliary hypotheses. 1. Good people keep their promises even when this requires modest personal sacrifice. 2. Ben made a promise to Sam that can only be kept at modest personal sacrifice. 3. Ben is a good person. 4. Hence, Ben will keep his promise to Sam. Moral claims have no observational consequences by themselves but only in conjunction with auxiliary hypotheses, which include moral claims for which we can have independent evidence. The situation here is no different than it is in non-­‐moral matters. If Ben does not keep his promise to Sam, then something will have to go -­‐-­‐ I will need to revisit my assumptions about the promise-­‐keeping behavior of good people, my assumptions about Ben’s character, or perhaps my assumptions about the sacrifice required of Ben. This is just a familiar feature of holism about confirmation. 3 See my review of Gibbard’s Thinking How to Live in Philosophical Review 116 (2007): 267-­‐72 and Mark Schroeder, Being For: Evaluating the Semantic Program of Expressivism (Oxford: Clarendon Press, 2008). 5 Moreover, some ethical naturalists have wanted to resist Harman’s suggestion that moral claims are explanatorily impotent.4 If moral facts had explanatory value, this would apparently also cast doubt on Ayer’s assumption that moral claims are not empirically testable. If moral judgments are cognitively meaningless, Ayer concludes, they must have some other kind of meaning. They must have a purely emotive meaning, expressing the appraiser’s emotions. As such, moral judgments would be neither true nor false (103, 107, 108). In addition to expressing the appraiser’s noncognitive feelings, Ayer claims that moral judgments can aim and succeed at arousing similar feelings in others (108). Ayer contrasts his emotivist metaethics, according to which moral judgments express the appraiser’s attitudes, with a subjectivist metaethics, according to which moral judgments report the appraiser’s attitudes. According to subjectivism, when I say Torturing babies for fun is wrong. I am reporting that I have a negative feeling toward the conduct in question. In effect, I am saying I don’t like the torturing of babies for fun. Or perhaps I disapprove of torturing babies for fun. By contrast, according to emotivism (and, more generally, expressivism), I am not reporting that I have this attitude, but rather expressing it. For example, I often do things that express my anger, such as swearing or grimacing, without reporting that I am angry, though, in normal circumstances, you might well be able to infer that I am angry from my expression of anger. On this view, I am expressing an attitude of disapproval toward the conduct in question. In effect, I am saying Booh! (torturing babies for fun) If we look for ordinary linguistic expressions of these attitudes, we might look to the imperatival form (103). Don’t torture babies just for fun! Ayer notes that the subjectivist cannot represent moral disagreement. Consider a disagreement between Axel and Bert. Suppose Axel affirms and Bert denies our moral proposition. Torturing babies for fun is wrong. The subjectivist believes that Axel and Bert is each reporting his attitude toward the conduct in question. Axel is saying that he disapproves, and Bert is denying that he disapproves. But there is no disagreement here. They disagree no more than if Axel said he likes sushi and Bert said he detests it. Disagreement seems to require asserting inconsistent things -­‐-­‐ things that can’t both be true. But what Axel and Bert assert, according to the subjectivist, are not inconsistent. Stevenson will claim that the emotivist has a better account of moral disagreement than the subjectivist. But Ayer thinks that the emotivist should agree with subjectivist that genuine moral disagreement is not possible (110-­‐12). Ayer thinks that there can be moral disputes that reduce to 4 See Gilbert Harman, The Nature of Morality (OUP, 1977), chs. 1-­‐2. 6 disagreements about the non-­‐moral facts on which our attitudes depend and that we can argue with others about moral matters insofar as we can provide evidence for non-­‐moral claims on which our moral attitudes depend. Thus, we can make sense of moral reasoning within a shared scheme of attitudes as reasoning about the non-­‐moral facts relevant to applying our shared attitudes. But it is not possible, Ayer thinks, to reason or disagree about fundamental attitudes. These are not genuine disputes. In making these claims, Ayer ignores the realist idea that we can be mistaken not only about derivative moral commitments, as the result of mistaken non-­‐moral assumptions with which we apply fundamental evaluative commitments, but also about fundamental evaluative commitments. He also ignores the idea that we can reason about our fundamental evaluative commitments in a dialectical fashion. STEVENSON Stevenson develops his conception of emotivism in his early paper “The Emotive Meaning of Ethical Terms,” in his book Ethics and Language, and in a series of essay collected in Facts and Values. We will concentrate on general features of Stevenson’s views. Stevenson sets out three desiderata for any conception of the meaning of moral terms and the function of moral judgments (EMET 16-­‐17). 1. It must make sense of moral disagreement. 2. It must make sense of the “magnetic” or “dynamic” aspects of moral judgments, their relation to motivation or the will and their role in interpersonal engagement. 3. It must vindicate the autonomy of ethics in relation to the sciences. Like Ayer, Stevenson contrasts emotivism with subjectivism or what he sometimes calls “the interest theory” (15). Subjectivism fares badly, Stevenson thinks, on all three grounds. Emotivism, Stevenson claims, understands moral judgments primarily to be expressing, rather than reporting, the appraiser’s noncognitive emotions and desires and attempting to arouse similar attitudes in the appraiser’s audience (18-­‐19). Notice that there are two distinguishable functions alluded to here: expressing one’s own attitudes and arousing similar attitudes in others or, as Stevenson sometimes says, influencing others. One could apparently aim for either goal without the other. Though Stevenson sometimes suggests that the interpersonal aim is primary (19), it’s not clear why either wouldn’t do. If so, we might represent the emotivist claim about the primary function of moral judgments disjunctively: to express the appraiser’s own attitudes or to influence the attitudes of others. In “The Emotive Meaning of Ethical Terms” Stevenson suggests that subjectivism is incomplete, leaving out the primary emotive meaning of ethical terms. On one interpretation, emotivism supplements subjectivism with a distinctively emotive element. This conception of emotivism is roughly what Stevenson will later identify as the First Pattern of Analysis: To say “X is good” is roughly synonymous with “I approve of X; do so as well” [EL 81]. So understood, emotivism is superior to simple subjectivism along all three dimensions. Like Ayer, Stevenson recognizes that subjectivism cannot represent moral disagreement. If moral judgments are just reports of the appraiser’s attitudes, then it’s not clear why Axel and Bert disagree; each is reporting his own attitudes toward the conduct in question. Unlike Ayer, Stevenson takes this to be a problem for subjectivism. Also unlike Ayer, Stevenson thinks that the emotivist can and should have an account of moral disagreement. The emotivist understands moral disagreement, not as the assertion of incompatible propositions, but as the expression of conflicting attitudes (19-­‐20, 26-­‐27). According to the emotivist, Axel and Bert are not asserting anything, much less inconsistent things. They are merely expressing contrary attitudes. Their disagreement is like the one at a sushi bar when two friends look at the offerings and one says “Yum!” and the other says “Gross!”. Stevenson 7 also says the disagreement might be understood as like the practical disagreement between Axel and Bert if Axel wants to go to the beach and Bert wants to go to the mountains for a hike. Subjectivism also seems ill-­‐suited to explain the dynamic aspects of moral judgments, because it’s not clear why your liking something is a reason for me to like something. But if moral judgments not only report the appraiser’s attitudes but also instruct others to join in, then we can see how they have a dynamic interpersonal function (27). Subjectivism also fails to represent the autonomy of ethics insofar as it implies that an appraiser can support her moral judgment simply by pointing out the psychological fact that she approves of the conduct in question. But that is not how we try to support moral judgments. We don’t cite autobiographical facts about ourselves to explain why apartheid is unjust. By contrast, emotivism maintains the autonomy of ethics. While reasoning about the non-­‐moral facts may help resolve disagreements when we share fundamental attitudes and disagree only about what further attitudes the non-­‐moral facts recommend, not all disagreements are disagreements about the non-­‐moral facts. There are also disagreements in fundamental attitudes, and these disagreements won’t be resolved by the evidence of the sciences (27-­‐29). According to Stevenson, moral terms have both descriptive meaning and emotive meaning, but their emotive meaning is primary. This is reflected in the First Pattern of Analysis. But the descriptive element of the FPA is the autobiographical element that it shares with the subjectivist analysis. But the subjectivist analysis goes wrong by making moral judgments autobiographical. When I judge apartheid to be unjust, I don’t seem to be making any claim about myself or my attitudes. I seem to be making a claim about apartheid on which my attitudes are perhaps consequential. Considerations such as these, led Stevenson to introduce in Ethics and Language a Second Pattern of Analysis. “This is good” has the meaning of “This has qualities or relations X, Y, Z …,” except that “good” has as well a laudatory emotive meaning which permits it to express the speaker’s approval, and tends to evoke the approval of the hearer [207]. Adapting ideas from Hare, we might interpret Stevenson’s SPA as distinguishing descriptive criteria for emotive meaning and emotive meaning. On this view, any moral judgment picks out certain features of a person, action, or situation in virtue of which the appraiser expresses her attitudes (e.g. approval) toward that person, action or situation and/or encourages others to adopt those attitudes. Notice that SPA drops the subjectivist element in FPA. Emotivism, according to SPA, is not a supplement to subjectivism, however important. QUESTIONS Notice that emotivism analyzes moral judgments into the attitudes of appraisers. The standard analysans of emotivism are approval/disapproval. But there are a great many apparently distinct kinds of moral judgments possible: judging good, judging right, judging virtuous, judging admirable, etc. It would seem that emotivism is committed to pairing distinct moral judgments with distinct attitudes. It’s not clear that we can identify sufficiently many distinct attitudes for this purpose without invoking the very moral properties to individuate attitudes that we are trying to analyze. I don’t say that this is an insurmountable challenge for expressivism, but it is a challenge. Emotivists seem to subscribe to internalism about moral motivation, claiming that motivational states, such as pro-­‐attitudes, are essential parts of moral judgments. But this seems to render impossible disengaged moral judgments. We often represent the possibility of amoralist skepticism -­‐-­‐ skepticism about the authority of moral requirements -­‐-­‐ by the amoralist -­‐-­‐ someone who professes indifference to the moral requirements that she recognizes. This is a philosophical amoralist, and we might think that it is a problem with emotivism or any other form of internalism that denies this possibility. Moreover, we might want to recognize non-­‐philosophical amoralists who are indifferent to what they judge morally required, not out of philosophical principle, but out of apathy, depression, or 8 the sort of volitional impairment that we find in those with damage to the prefrontal cortex of the brain (e.g. Phineas Gage). Stevenson thinks emotivism superior to subjectivism, because it has a better account of moral disagreement. We might concede this comparative claim, but wonder if the emotivist account of disagreement is as good as the realist account. The realist interprets Axel as ascribing the property of impermissibility to the act type <torturing babies for fun>. By contrast, Bert denies this claim. They assert inconsistent propositions, which can’t both be true. By contrast, the emotivist sees this as a conflict of attitudes, like the conflict between wanting to go to the beach and wanting to go to the mountains. It’s not clear if this practical conflict is all that’s involved in a moral disagreement. Both impulses can be understandable, even reasonable. By contrast, it’s hard to believe that both Axel’s and Bert’s claims about torturing babies for fun can be understandable and reasonable. It seems that one of them must be right and the other wrong. The emotivist must also count moral disagreements in unusual ways. The emotivist identifies moral agreement with agreement in noncognitive attitude and moral disagreement with disagreement in noncognitive attitude. Consider the moralist and the amoralist who intuitively share the same moral views. God believes that compliance with the Decalogue is right and approves those who are compliant. Satan also thinks that compliance with the Decalogue is morally right and disapproves of those who are compliant for that very reason. The emotivist must say that God and Satan have a moral disagreement, because they have conflicting attitudes toward the same actions. But it seems that they agree about what morality requires and disagree only in attitude. There are also the worries about revisionary expressivism and the problems it has with the semantics and logic of moral assertion and inference. Most of these problems would seem to arise for Stevenson since he treats moral judgments, for the most part, as being not truth-­‐apt. In Facts & Values Stevenson does help himself to a deflationary or disquotational account of ascriptions of truth (216-­‐
19). But while this may allow him to make emotivist sense of ascriptions of truth to moral judgments made by oneself or others, it is not clear that this entitles him to the property of truth, necessary, for instance, for the truth-­‐theoretic explanation of validity. I don’t mean to suggest that there are no emotivist resources to reconstruct the relevant semantic and logical structures and relations. But this is the project of accommodationist expressivism, pursued by later expressivists, such as Gibbard and Blackburn, not by Stevenson himself. RATIONALIST MORAL MOTIVATION We have looked at worries about noncognitivism. The externalist solution may also seem to be a solution of last resort, because it may seem to deny the platitude that moral judgments are motivation-­‐
ally efficacious. For this reason, we might look seriously at rationalist theories of moral motivation, because they promise to represent moral judgments as intrinsically motivational without giving up cognitivism. Some philosophers, such as Nagel and McDowell, maintain cognitivism and internalism about motivation by rejecting the assumption that motivation requires a desire or pro-­‐attitude; they insist that purely cognitive states -­‐-­‐ beliefs -­‐-­‐ can motivate. For instance, both Nagel and McDowell motivate this rationalist view about moral motivation by appeal to an analogy between prudential and moral motivation. Seeing what's in one's interest or what morality requires is often all that's needed to motivate and explain action. In such cases, ascription of the desire to perform the act in question is "merely consequential." However, even if ascription of proximate desires is consequential, motivation seems to depend on more ultimate desires. Proximate or foreground belief can be sufficient for motivation only when more ultimate desires are part of the background. If we count appeal to my belief that it is raining as explaining my action, this is only because we take my desires to go out and to stay dry for granted as part of my psychological background. Likewise, we can agree that moral beliefs are sufficient to motivate the virtuous person, but this is only because a virtuous person is someone with a certain well-­‐developed psychological profile that is structured by various cognitive and conative states. 9 In support of the need for prior conative structure in order for normative belief to motivate, we might make several observations. First, this picture is supported by the fact that acquiring the same belief about what morality requires can produce different behavior and motivation in the virtuous and the non-­‐virtuous. Same input, different output. By itself, this just means that there must be some difference in internal psychological architecture between the virtuous and the non-­‐virtuous. The differences might be cognitive, not conative. However, we can presumably find cases of differential output (motivation) with common input (moral beliefs), despite common beliefs. Compare the virtuous person, who tracks the morally good, with the vicious person, who tracks the morally bad. Same input, same cognitive architecture, different output. If so, then there must be a difference in conative architecture. The implausibility of the purely cognitive picture of moral motivation is a consequence of the plausibility of belief-­‐desire psychology. Proponents of belief-­‐desire psychology often characterize the difference between beliefs and desires in terms of their different directions of fit with the world (e.g. Aristotle, Green, Anscombe, Stampe, Humberstone, Velleman). The difference between beliefs and desires is a special case of a more general difference between representations and pro-­‐attitudes. On this view, representations, such as beliefs, are states of the agent whose content she adjusts to conform to information she receives about the state of the world. By contrast, pro-­‐attitudes, such as desires, are states of the agent on the basis of which she acts to make the world conform to them. We can think of the difference in terms of the response to a perceived mismatch between the content of the intentional state and information about the way the world is. If the state is a belief, the agent tends to respond to such a mismatch by modifying the content of the intentional state to match the way the world is or appears. If the state is a desire, the agent tends to respond to such a mismatch by acting so as to modify the world to conform to the content of the state. On this sort of belief-­‐desire psychology, agents act in order to satisfy their desires based on their beliefs about the world, in particular, their beliefs about the causal means to and necessary conditions of satisfying their desires. MORAL MOTIVATION BY RATIONALIZED DESIRE A different form of rationalism denies that motivation is possible without a prior pro-­‐attitude but insists that certain beliefs, in particular, normative beliefs entail pro-­‐attitudes. On this view, the belief that I have a moral reason generates a desire to perform the action in question. This solution to the puzzle has two component parts. • Normative Internalism: beliefs about one's practical reason entail (pro tanto) desire. • Conceptual Rationalism: belief that an action is morally required entails belief that it is rationally authoritative. Nagel and Smith both seem to endorse this sort of moral motivation by rational desire. Smith's version: • Normative Internalism: To judge that I have reason to φ is to judge that I would desire to φ if I was fully rational, that is, deliberating correctly (e.g. with full information, coherently, and employing means-­‐ends reasoning) (155-­‐61). (This is, I think, a conceptual claim about normative judgment, not just a claim about the truth-­‐conditions of normative judgment.) • Conceptual Rationalism: To judge that I am morally required to φ is to judge that I would desire to φ if I was fully rational and that φ-­‐ing "is an act of the appropriate substantive kind" (184). (This is also a conceptual claim.) However, I'm not sure the details of Smith's version of motivation by rationalized desire need concern us. Though there is an important grain of truth in normative internalism, both theses are open to question. Normative internalism implies that there is an important connection between judgments of practical reason and the will. This idea might be defended by appeal to belief-­‐desire psychology. On this view, as we have seen, intentional action is viewed as the product of representational states, such as 10 belief, and pro-­‐attitudes, such as desire, which display different directions of fit with the world. On this sort of belief-­‐desire psychology, agents act in order to satisfy their desires based on their beliefs about the world, in particular, their beliefs about the causal means to and necessary conditions of satisfying their desires. But on this sort of psychology, we can also understand how normative beliefs would tend to influence desire. For normative beliefs are beliefs about how the world should be. But if desires are precisely states that tend to make agents modify the world in accordance with their content, then we should expect normative beliefs normally to affect desires. If we accept belief-­‐desire psychology, because of their different directions of fit, we can claim that desire can be responsive to ought judgments. This shows how one can accept the Humean dictum that action depends on desire without accepting the Humean dictum that reason can only be the slave of the passions. However, to say that normative beliefs can and normally do influence desire is not to say that normative beliefs have such influence necessarily. Other things being equal, normative beliefs have conative influence. But other things need not be equal if there is some relevant form of psychological interference. In some cases of weakness of will, normative beliefs apparently motivate but provide insufficient motivation. In other cases of weakness of will, normative beliefs may not motivate at all. This second sort of weakness of will is selective if the interference is intermittent; it is global if the interference is systematic. Depression might produce selective weakness of will (cf. Stocker, Mele), but damage to the prefontal lobe of the cerebral cortex (as in the famous case of Phineas Gage) might produce systematic weakness of will. This is already to recognize one kind of amoralism -­‐-­‐ someone who recognizes moral requirements yet remains indifferent. The kind of amoralism whose possibility Smith concedes is a kind of unprincipled amoralism. It is unprincipled, not in the sense that it is random or lacks a psychological explanation, but in the sense that it is due to psychological interference with the normal process by which results of practical deliberation affect an agent's motivational set; indifference does not reflect principles the agent accepts. The possibility of this sort of amoralism undermines the strong internalist assumption about moral motivation. If, as I believe, the puzzle about moral motivation is best construed as resting on this strong internalist assumption, then recognition of the possibility of this sort of amoralism is sufficient to vindicate the externalist solution. But there is another problem for this form of rationalism about moral motivation, even if we accepted normative internalism. Conceptual rationalism seems problematic. Doubts about conceptual rationalism reflect the possibility of principled amoralism in which one is indifferent to moral requirements, either globally or selectively, because one has doubts about the rational authority of moral requirements. In this way doubts about conceptual rationalism actually presuppose (weak) normative internalism. It is possible to remain unmoved by what one judges morally required as long as one can hold conceptions of morality and practical reason according to which moral requirements, globally or selectively, lack rational authority. There are many conceptions of morality and practical reason according to which their coincidence is at best imperfect. One familiar set of conceptions involves recognizing impartial or other-­‐
regarding duties and an instrumental or prudential conception of practical reason. For even if it is often true that fulfilling other-­‐regarding duties would satisfy one's desires or promote one's interests, this coincidence need not always take place. Fulfilling one's other-­‐regarding duties need not always satisfy one's desires or promote one's interests. Notice that principled amoralism does not require the truth of anti-­‐rationalism. It requires only the denial of conceptual rationalism and the possibility that one be able to form conceptions of morality and practical reason according to which their demands can diverge. EXTERNALISM ABOUT MORAL MOTIVATION Doubts about the plausibility of moral motivation by rationalized desire tend also to support externalism about moral motivation. 11 •
•
Unprincipled Amoralism undermines strong internalism about moral motivation. Principled Amoralism undermines both strong and weak internalism about moral motivation. As long as anti-­‐rationalist beliefs are possible, whether or not they are true, internalism about moral motivation must be false. But we might doubt the coherence of externalism, since it depends on the coherence of anti-­‐
rationalism. Consider this worry about anti-­‐rationalism. 1. If I am under a moral requirement to φ, there is a moral reason for me to φ. 2. If there is a moral reason for me to φ, there is a reason for me to φ. 3. If there is a reason for me to φ, it would be pro tanto irrational for me to fail to φ. 4. Hence, if I am under a moral requirement to φ, it would be pro tanto irrational for me to fail to φ. This argument would show weak internalism to be a conceptual truth. But it is not compelling. Sometimes when we say that I have a reason to φ, we mean • (a) There is a behavioral norm that enjoins φ-­‐ing and applies to me. In this sense of reason, moral norms do imply reasons. There are as many kinds of reasons as there are norms, including moral reasons, legal reasons, reasons of etiquette. But we often have something more in mind in ascribing reasons. • (b) There is a behavioral norm that enjoins φ-­‐ing, it applies to me, and it would be pro tanto irrational for me not to φ. If there is reason, in this sense, to act on a norm, then practical reason endorses this norm. But not all reasons for action in the first sense are reasons for action in the second sense (cf. Foot, "Morality as a System of Hypothetical Imperatives"). For instance, it is arguable that failure to conform to requirements or reasons of etiquette or law need not be pro tanto irrational. It is clear that moral requirements are moral reasons and that moral reasons are (a)-­‐reasons. It is not clear that they are (b)-­‐
reasons. So the sense of reason for action in which (3) is true need not be the same sense of reason for action in which (2) is true. If so, it's arguable that it is only by failing to distinguish these two senses of reason for action that the rational authority of morality could fail to seem an open question. SMITH'S ANTI-­‐EXTERNALISM Smith rejects the possibility of principled amoralism. He thinks that putative amoralists should always be interpreted as making moral judgments that employ moral language in "inverted commas" – they are expressing indifference not to the moral standards they accept but to the moral standards that others around them accept. So he thinks that externalists who appeal to the possibility of amoralism are begging the question. Smith defends internalism by appeal to an analogy between colors and morals: motivation is essential to moral judgment in the way that qualitative sensory experience is essential to color judgment. Finally, Smith rejects what he takes to be the externalist explanation of the way in which the motivation of the "good and strong-­‐willed person" normally tracks changes in her moral beliefs. He believes that this requires a de dicto, rather than a de re, concern to do the morally correct thing. This commitment, Smith thinks, is implausible and commits the externalist to "one thought too many" (74-­‐75). The externalist denies that an agent's motives must track changes in her moral beliefs. However, even the externalist should claim that the motives of the "good and strong-­‐willed person" reliably track changes in her moral beliefs. Smith thinks that the concerns of such a person must all be derived from her moral concerns. This, he thinks, is inconsistent with intrinsic concern for one's 12 associates; the externalist must ascribe to the good and strong-­‐willed person "one thought too many" about whom to save. Externalism leads to an over-­‐moralized self. EXTERNALIST REJOINDERS There are several places the externalist may want to put her foot in the door. Inverted commas readings reject a literal reading of the putative amoralist's claims. On the one hand, people's self-­‐
reports should not always be taken at face value (e.g. Huck Finn, maybe). But why should we always reject a literal reading? What if the putative amoralist persists in describing his judgments as moral judgments even after being offered an inverted commas interpretation? Why does Smith think that the appeal to principled amoralists question-­‐begging? The externalist claim is not that the internalist does not have a consistent position; it is that the position is implausible. Internalism makes a very strong generalization about the connection between moral judgment and motivation, and externalism challenges that, not as incoherent, but as implausible. Specifically, it is implausible, because it asserts that something is impossible, which seems to be both possible and actual. In appealing to principled amoralism as a reason for rejecting (weak) internalism, the externalist is reasoning much as Smith himself does in appealing to unprincipled amoralism as a reason for rejecting strong internalism. The strong internalist could preserve consistency by insisting that the apathetic and depressed are using moral language in inverted commas. Moreover, in cases in which the amoralism, whether principled or unprincipled, is acquired, there will be continuity in discrimination and belief across the change in motivational effect. Since we interpreted the earlier judgments as bona fide moral judgments, we should treat the later judgments, which are continuous with the earlier judgments, as bona fide moral judgments. The charge of question-­‐begging may suggest that the internalist and the externalist just make different conceptual claims about morality and the question then becomes which, if either, conceptual claim to accept, and this may seem like a stalemate. But this is not a fair description of the debate. Whereas the internalist assumes quite a bit about the concept of morality – that it excludes the possibility of holding beliefs about morality and beliefs about practical reason according to which they could diverge in their demands – the externalist assumes much less about the concept of morality – in particular, the externalist can recognize a wide variety of possible views about the nature and content of morality and its relation to practical reason. If we have general doubts about the limits of conceptual analysis and wish to be able to entertain very diverse conceptions of philosophical concepts, then we should for that reason prefer externalism to internalism. Now consider Smith’s color analogy. We might wonder if the color analogy begs the question in favor of internalism. Why should we find an analogy between moral and color judgments especially apt unless we already accepted internalism? The internalist strategy assumes a sort of dispositional conception of colors, which is itself potentially controversial. But even if we make this assumption about colors, there are disanalogies. The sort of sensory experience that might be required for color judgment is not the sort of conative state that is required by the internalist for moral judgment. In a dispositional analysis of color, color judgments are true in virtue of the object being such as to elicit a certain sort of visual experience in a standard observer. But it won't imply that anyone making a color judgment, as such, must have this visual experience. So, it's not clear that the moral analog of this could deliver the desired internalist connection between moral judgment and actual motivation. Also, it is not clear that the capacity for visual experience is a condition for making color judgments. Can't the blind make color judgments, perhaps by inheriting meaning/reference from linguistic participation with the sighted community? Here, we may need to distinguish cases of congenital/acquired blindness. Now consider the externalist explanation of the good and strong willed person. Where is the incompatibility between intrinsic concern for associates and moralized concern? If friendship is a virtue, then won't moralized concerns include concerns for friends? Perhaps the externalist will think that expressions of friendship should be regulated by beliefs about the permissibility of those expressions. Is that a problem? Don't we think that acting on concerns for friends should be regulated 13 by our beliefs about what's permissible? It's okay to favor one's spouse in a rescue situation when that is permissible, but not when it's not (e.g. when you are an EMT). It's okay to favor friends, but not by poisoning their rivals. Smith’s worry about the externalist seems relevantly like Williamsesque worries about Kant's insistence that virtuous agents act from the motive of duty. Critics often complain about the good will that it treats sympathetic or compassionate aid as morally tainted. But this misunderstands the doctrine of the good will. The good will does not require contramoral sentiments; it requires only that they be regulated by a sense of duty such that the agent would not have had the special concern or would not have acted on it had she believed the action in question was impermissible. Why can’t the externalist claim about the motivation of the good and strong-­‐willed be understood in a similar way?