A Rational Solution to the Problem of Moral Error Theory? Benjamin Scott Harrison In his Ethics, John Mackie (1977) argues for moral error theory, the claim that all moral discourse is false. In this paper, I consider Michael Smith s (1994) 1 moral rationalism as a challenge to the error theory. I begin by discussing moral error theory in more detail and show that it has undesirable consequences ( ). I then show that these undesirable consequences can be avoided if we embrace moral rationalism. Finally, I argue that Mackie s argument from relativity results in a standoff between error theory and moral rationalism I I follow Finlay (2008) in interpreting Mackie s argument for moral error theory as consisting of the following two claims: P Moral judgements presuppose objective values. E There are no objective features of reality. We should approach the first of these claims through Kant s distinction between categorical and hypothetical imperatives; that is, between commands that have authority independent of an agent s desires and commands that are conditional on said desires, where commands are truth-apt statements to the effect that something ought to be done or that it would be good to do it (Foot 1972, 305; Smith 1994, 77). In addition to being commands, we can think of hypothetical imperatives as pieces of advice because they help an agent satisfy her ends (following Joyce 2011a, 521). If an agent wanted to quell her hunger, the command you should eat is true when said of her because eating would satisfy her end. If she did not want to quell her hunger, be it that she were fasting, for example, the very same hypothetical imperative would be false. We could not command her to eat with any authority because the advice, if taken, would not result in the satisfaction of her ends. Moral judgements appear to differ: for example, you should not murder is not mere advice for those seeking to avoid legal consequences. If moral judgements were hypothetical in this way, we could not truly state of an agent, who did not fear the consequences of her actions, that she should not murder, rape or steal and this seems incorrect. In Kant we find the alternative: the categorical imperative presents an action as of itself objectively necessary, without regard to any other end (1959, my emphasis). As categorical imperatives, moral judgements have the authority we require. I believe Kant is slightly mistaken here as some categorical imperatives, namely subjective categorical imperatives, present actions that are nonobjectively necessary. Nevertheless, I will argue that moral judgements are not like this; they do in fact present actions as objectively necessary. We will then be in a position to understand P ; that is, by determining the existence of a relation between the objective necessity of actions presented by moral judgements, that is, objective categorical imperatives and objective values. To that end, we must understand the distinction between objective and non-objective necessity; a distinction I take Smith to advance using the notions of de re and de dicto necessity against Foot s claim that morality and etiquette are analogous. Consider the following rule of etiquette: you should fasten your top button at a dinner party. Foot rightly asserts that this is a categorical imperative because it does not fail to apply to someone who has his own good reasons for ignoring this piece of nonsense (1977, 208). It presents an action as necessary without regard to any other end, but is the action 1 All otherwise unattributed references are to Smith (1994) hereafter.
objectively necessary? Smith rightly argues that we are only externally motivated to follow the rules of etiquette (83). When we judge that fastening our top button is good etiquette, we are motivated by the desire to do the done thing ; we are not directly motivated, that is to say, internally motivated by the content of the judgement itself. Thus, if an open top button were considered smart, we would be no longer be motivated to fasten it. This is because we are motivated to do what etiquette requires, where this is read de dicto and not de re; we are motivated to do whatever is considered good etiquette. This is not so with moral judgements: Good people care non-derivatively about honesty, the weal and woe of their children and friends and the like, not just one thing: doing what they believe to be right, where this is read de dicto and not de re. (75) To see why it is that we are motivated to do what is moral, where this is read de re and not de dicto, we should return to Kant (1959), who states that [i]f the action is thought of as good in itself, and hence as necessary the imperative is categorical. In my view, Kant is wrong to think that an action must be thought of as good in itself for the imperative to be categorical. That the action is at all necessary independent of any other end is sufficient for the imperative s being categorical; for example, fastening your top button is necessary simply because an institution deems it so. However, when the action is good in itself, it is presented as being objectively necessary rather than non-objectively necessary. Thus if moral acts were good in themselves, moral judgements would be apt to motivate us internally in virtue of their content. Then we really would desire to be honest (and the like) as Smith claims; that is, desire to do what is moral, where this is read de re and not de dicto. Foot calls this the automatic reason giving force of moral judgements (1972, 309). This explains why moral judgements appear to have more force than the rules of etiquette. Unlike moral judgements, the rules of etiquette do not present actions that are good in themselves; they present actions that are deemed good by an institution (83 84). We can now understand P claim that the content of moral judgements are objectively necessary moral actions, which are objectively necessary in virtue of their being good in themselves; and as I see things, that is by instantiating the objective value of being morally right. Error theory is then the conjunction of this claim and E which asserts that there are no such objective values. Thus when Mackie states that my thesis that there are no objective values is specifically the denial that any such categorical imperative element is objectively valid but is constituted by our choosing or deciding to think in a certain way (1977, 29 30), he is asserting that all categorical imperatives are subjective categorical imperatives. We can think of the objective validity of categorical imperatives as their having authority in virtue of objective values rather than in virtue of an institution (the latter being our deciding to think in a certain way ). These are our objective and subjective categorical imperatives, respectively. The problem of moral error theory follows and is well explained by Joyce through the character of Gyges with his ring of invisibility, who sought to satisfy his basest passions by stealing and raping at will (2011a, 524). Because our moral framework is categorical we can carry on legitimately saying Gyges, you ought not do that! But if our utterances are merely a verbal output that has been validated by an institution of our own creation, then it all begins to sound rather shrill why should he do otherwise? (Joyce 2011a, 524) Moral judgements continue to apply to Gyges despite his disinterest in morality because they are categorical; that is, just as the rules of etiquette apply to those who think they are
nonsensical. However, like the rules of etiquette, moral judgements would only be subjectively valid if error theory is true they would have authority only if we think in a certain way and, in the case of Gyges, that might mean valuing the love and happiness of others, and so on. Thus while morality applies to Gyges (insofar as we can truly assert that he acts immorally), it fails to grip him because he does not desire the things which underpin it; quite simply, he chooses to ignore it and, if there are no objective values (values that we must all share), no argument could ever convince him otherwise. Moral judgements are the alleged pretenders of such arguments. This we know from our analysis of P In an attempt to stop people like Gyges, we push moral judgements towards objective validity; we presuppose objective values and our moral discourse falls into falsity. II There are three positions in opposition to error theory (as identified by Joyce 2011b, 153). Noncognitivists argue that moral discourse does not aim at the truth (e.g. Ayer 1990). Others grant E, claiming that P (see Joyce 2011b for a discussion of this, the concessive approach). Lastly, some tackle the problem head-on and deny E Moral rationalism falls under this latter head-on approach and it is against moral rationalism that we are to assess the argument from relativity. Smith s moral rationalism is a two-fold view (Copp 1997, 33 34). First, it is the thesis that there are objective values, which are called for by normative reasons. Call this the objectivity of normative reasons. This turns on reducing valuing to believing (147 150), which I understand as the claim that an agent values things that are iff she believes that she has a normative reason to do things that are. Thus she values moral rightness iff she believes she has a normative reason to be honest; believes she has a normative reason to prevent suffering; believes she has a normative reason to promote equality and so on. Second, normative reasons (reasons hereafter) are practical requirements of rationality that justify an action, such that an agent has a reason to just in case, if she were fully rational, she would want herself to. As Smith states, what it is desirable for us to do [what we value] is what our more rational selves, looking down on ourselves as we actually are from their more privileged position would want us to do in our actual circumstances (151 152). Call this the practicality of reasons. We then have a knockdown argument to E. Everyone should value the same thing, an objective value of the sort E denies. To demonstrate this, I will argue that we have a reasonable argument to offer Gyges. We can assert, Gyges, your desires are irrational the requirements of rationality compel you to be moral. It is important to note that this argument need not actually compel him to stop (see Joyce 2011a, 524). I take it that a reasonable argument need only make true the judgement Gyges, you ought not do that!, which I believe it does. Thus this argument is enough to rescue moral discourse from error theory since moral judgements of this sort are in fact true. This argument stands and falls with two claims that underpin Smith s moral rationalism. The first claim is as follows: C There is a convergence in the desires of fully rational agents such that in any circumstances C, fully rational agents always desire to in C. (173) If this were false, the objectivity of reasons would be false. Scattered desires would entail subjective reasons because it would be possible for two agents to have contradicting reasons in the same circumstances C since their respective and more rational selves would desire that they contradict one another in C.
Before we continue, it is important that we specify the content of the desires that rational agents converge upon. Smith only requires for apply to higher level desires and not lower level desires (171 173). To understand this distinction, we might compare an agent s desire to eat cake over pie and her desire to eat ceteris paribus whatever happens to take her fancy. The first desire is conditioned by taste, namely an innocent preference for cake to which rational agents are indifferent (see Smith 1999, 89). The second desire is conditioned by her ideals, and it is on these desires that fully rational agents converge. Moreover, it is rational for an agent to follow her tastes so long as it does not conflict with her ideals. Consider that by eating cake she may satisfy one ideal, of eating whatever takes her fancy, but come into conflict with another, of refraining from gluttony. It follows that requires all fully rational agents to agree on what counts as mere taste and converge upon the same ideals (Sobel 1999, 142). Furthermore, desires conditioned by taste become relevant feature of circumstances C. 2 3 We should now move onto the second claim: The practicality requirement: If an agent judges that it is right for her to in circumstances C, then either she is motivated to in C or she is practically irrational. (61 62) Without the second claim, the practicality of reasons would fail, but taken together, these claims entail that reasons are both objective and motivating (see Smith 1997, 88). They are the foundations of moral rationalism, which make it true that in any circumstance everyone should be motivated to do the same thing, inter alia to be moral. Thus as I see it, Gyges should believe that it is right to stop his abhorrent behaviour because, on proper reflection, he would know that, were he fully rational, he would desire to stop. As Gyges should believe it is right to stop, he should be motivated to stop or else fail to be practically rational. Moreover, if motivation requires a means-end belief and an appropriately related desire, it follows that Gyges must desire to stop or else fail to be motivated to stop. 4 Thus it follows that if someone believes they have a reason to, then rationally she should desire to (148). But Gyges does not desire to stop and so is practically irrational. III The argument is premised on the objectivity of reasons, which depends on C. Moreover, its inferences depend on the practicality requirement. I cannot attempt a discussion of the practicality requirement in this paper. We shall therefore consider the argument from relativity as set against the objectivity of reasons. 2 We can include, along with those of taste, desires conditioned by talents and attachments (173) and indeed any other desire that fully rational agents are at least indifferent to people having and in favour of their acting on them once they have them (Smith 1997, 89). 3 They are a feature of C because what fully rational agents desire to do in C depends on the their personal tastes. Suppose I prefer cake and you prefer pie. Though your desire to eat pie is rational, my desires to eat cake is similarly rational. Were you in my circumstances, namely circumstances in which you were had a preference for cake, you would desire to eat cake too (and vice versa) because, as fully rational agents, we both desire to eat ceteris paribus whatever happens to take our fancy. 4 This derives from the Humean theory of motivation, which I will not attempt to defend in this paper (see 92 129 for a defence of the Humean theory).
Mackie holds that scientific disagreement results from speculative inferences or explanatory hypotheses based on inadequate evidence whereas [d]isagreement about moral codes seems to reflect people s adherence to and participation in different ways of life (1977, 36). I take it that if moral judgements were speculative inferences based on inadequate evidence, they would be the kind made about the desires of fully rational agents; more specifically, about where these desires converge. In my view then, Mackie is denying that we value what our more rational selves would desire, where this is read de re and not de dicto, and asserting that we value social norms instead, where this read de dicto and not de re. And since we would value whatever society expects of us, moral values would not be objective values. We might argue that moral disagreements need only be resolvable in principle (see Brink 1984, 117). After all, the existence of objective values turns on a convergence between our more rational selves and not our actual selves. However, I do not think Mackie wants to argue so directly against objective values on the basis that our actual selves disagree. Instead, it seems that Mackie is merely offering a better explanation for moral judgements over often distorted perceptions of objective values, namely our adherence to subjective social norms (1977, 37). It would be more charitable to suppose that this further work is done by his argument from queerness, which condemns objective values for being entities or qualities or relations of a very strange sort, utterly different from anything else in the universe (Mackie 1977, 38). But if we are to expect more from the argument from relativity, then it had better be because it suggests an inherent problem with objective values as called for by reasons rather than merely pointing out that moral disagreements remain unresolved in actuality. For as we have already seen, reasons are olds even if we are often mistaken in our beliefs about the reasons we have. We might therefore interpret the argument from relativity as an attack on C (201). As I see things, if this argument were successful, then even if we were to value the desires of our more rational selves, these values would still be determined by social norms. Before I explain, we must carefully consider what Smith calls a fully rational agent. When Smith (156) says that an agent is fully rational, he means that she has (i) (ii) (iii) no false beliefs all relevant true beliefs, and deliberates towards a maximally coherent desiderative profile. It is important to realise that fully rational is being used to describe an agent that is immune from reasoned criticism. I follow Sobel in thinking that we must distinguish between two conceptions of rationality if we are to understand what Smith means here, namely rational deliberation and ideally rational deliberation (1999, 138). In the first instance, an agent may deliberate rationally from false beliefs. In other words, she is immune from reasoned criticism so long as she makes proper use of the false content of her beliefs. To use Sobel s example, we deliberate rationally when we leave the building upon hearing a fire alarm even if there is no fire; that is to say, we deliberate rationally from the false belief that the building is on fire by leaving. However, Smith states, the desires [a fully rational agent] has which are based on her false beliefs or ignorance are not immune from reasoned criticism (1997, 91). Clearly then, Smith rationality qua rational deliberation; it requires rationality qua ideally rational deliberation. We can understand ideally rational deliberation as when an agent deliberates rationality in ideal circumstances; that is, circumstances in which she has no false beliefs and all relevant true beliefs as expressed by (i) and (ii).
As I see things, it is the first interpretation of rationality that Copp has in mind when he asserts that, It is certainly not a failure of rationality in any ordinary sense if someone fails to believe certain truths about, say, the hair on my head. An agent may be fully rational without being omniscient (1997, 44). As Copp is talking about rationality qua rational deliberation, he is correct. This conception of rationality does not require omniscience since it admits of rational deliberation based on false beliefs. However, if we choose to understand rationality as ideally rational deliberation, then Copp is simply incorrect. If knowing certain truths about the hair on Copp s head were ever relevant to the circumstances in which a fully rational agent found herself, then she simply must know them on pain of irrationality. Moreover, rationality is almost entirely a philosopher s term of art and so, within reason, it can be framed to suit the philosopher s purpose (Smith 1997, 91). Smith has chosen this conception of rationality because requires fully rational agents to have true beliefs that we would not otherwise expect anyone to know. A second objection to Smith s account of rationality turns on (iii), the claim that fully rational agents must deliberate towards a maximally coherent desiderative profile; that is, a set of desires that could not permit the addition of any further desires without resulting in an inconsistency (see Sayre-McCord 1997, 75 76). Sayre-McCord thinks that deliberation towards a coherent set of desires is not a requirement of rationality. He argues that if an agent were to desire coffee ice cream, she would have a more coherent desiderative profile if she also desired ice cream simpliciter. But unless she had a desire fetishism, that is, a desire to have a more coherent desiderative profile, it would not be a failure of her rationality if she did not desire ice cream simpliciter. She could be utterly indifferent to what she would desire if she had a more coherent desiderative profile. This is because a desire s belonging to a coherent desiderative profile does not present an ipso facto reason for having it; there must be an independent reason for her to desire ice cream simpliciter (Sayre-McCord 1996, 154). So as Sayre-McCord sees it, working towards a coherent desiderative profile is not a necessary requirement of rationality. However, Smith thinks that generating new desires only makes for a more coherent desiderative profile if it allows us to make more sense of that desiderative profile (1997, 93). And desires that would give us a greater understanding are higher level desires, for example, the desire to eat ceteris paribus whatever happens to take our fancy (see Smith 1997, 94). To illustrate this, consider again the agent that desires to eat coffee ice cream. What could we say about her rationale here? That she often desires to eat ice cream simpliciter? Perhaps, but a better explanation would point to a more general, higher level desire, namely that she desires to eat whatever she fancies. It is apparent that Copp and Sayre-McCord have different conceptions of rationality from Smith s. However, if we are clear that Smith s fully rational agents are agents that participate only in ideally rational deliberation towards a maximally coherent desiderate profile then, if there is an objection to be made, it must be that even rationality qua ideally rational deliberation towards a maximally coherent desiderative profile does not entail a convergence amongst higher level desires (see Sobel 1999, 143 147 for such objections). Furthermore, if the higher level desires of fully rational agents do not converge; if is false, the argument from relativity would provide a better, if not the only explanation for our moral judgements, namely our adherence to social norms. Suppose that the desires of fully rational agents do not converge. As I see it, the question to ask is this: what could cause two agents with all relevant true beliefs, who deliberate rationally qua ideally rational deliberation towards a maximally coherent desiderative profile to disagree on a moral issue? And one possibility is the desire to do what society expects. Fully rational agents would converge on the desire to do ceteris paribus whatever society expects, which would lead them in different directions. Thus even if moral judgements were inferences
made about the desires of our more rational selves, they would still reflect our adherence to social norms. Of course, this possibility is not enough to establish the error theory; it remains possible that fully rational agents will converge on moral issues. Further consideration of fully rational agents and their nature is therefore necessary (see Smith 2002, 343). Only in this way can we hope to break this standoff between the moral error theory and moral rationalism. References Ayer, A. J. 1990. Language, Truth and Logic. London: Penguin Books. Brink, D. O. 1984. Moral Realism and the Sceptical Arguments from Disagreement and Queerness, in Australasian Journal of Philosophy, 62, 111 125. Copp, D. 1997. Belief, Reason, and Motivation: Michael Smith s The Moral Problem, in Ethics, 108(1), 33 54. Finlay, S. 2008. The Error in the Error Theory, in Australasian Journal of Philosophy, 86, 347 369. Foot, P. 1972. Morality as a System of Hypothetical Imperatives, in The Philosophical Review, 83(3), 305 316. Joyce, R. 2011a. The Error in The Error in the Error Theory, in Australasian Journal of Philosophy, 89, 519 534. Joyce, R. 2011b. The Accidental Error Theorist, in R. Shafer-Landau (ed.) Oxford Studies in Metaethics, vol. 6, 153 180. Oxford: Oxford University Press. Kant, I. 1959. Foundations of the Metaphysics of Morals. New York: Liberal Arts Press. Mackie, J. L. 1977. Ethics: Inventing Right and Wrong. Harmondsworth: Penguin Books. Sayre-McCord, G. 1996. Coherentist Epistemology and Moral Theory, in W. Sinnott-Armstrong and M. Timmons (eds.). Moral Knowledge?: New Readings in Moral Epistemology, 137 189. Oxford: Oxford University Press. Sayre-McCord, G. 1997. The Metaethical Problem, in Ethics, 108(1), 55 83. Sobel, D. 1999. Do the Desires of Rational Agents Converge?, in Analysis, 59(3), 137 147. Smith, M. 1994. The Moral Problem. Oxford: Blackwell Publishers. Smith, M. 1997. In Defense of The Moral Problem : A Reply to Brink, Copp, and Sayre-McCord, in Ethics, 108(1), 84 119. Smith, M. 2002. Exploring the Implications of the Dispositional Theory of Values, in Philosophical Issues, 12(1), 329 347.