In Defence of Moral Error Theory
Department of Philosophy
My aim in this essay is largely defensive. I aim to discuss some problems for moral error theory and to offer plausible solutions. A full positive defence of moral error theory would require substantial investigations of rival metaethical views but that is beyond the scope of this essay. I will however try to motivate moral error theory and to clarify its commitments.
Moral error theorists accept two claims—one conceptual and one ontological—about moral facts. The conceptual claim is that moral facts are or entail facts about categorical reasons (and correspondingly that moral claims are or entail claims about categorical reasons); the ontological claim is that there are no categorical reasons—and consequently no moral facts—in reality. I accept this characterisation of moral error theory and I try to unpack what it amounts to in section 2. In the course of doing so I consider two preliminary objections: that moral error theory is (probably) false because its implications are intuitively unacceptable (what I call the Moorean objection) and that the general motivation for moral error theory is self-undermining in that it rests on a hidden appeal to norms.
The above characterisation seems to entail the standard formulation of moral error theory, according to which first-order moral claims are uniformly false. Critics have argued that the standard formulation is incoherent since—by the law of excluded middle—the negation of a false claim is true. Hence if 'Torture is wrong' is false, 'Torture is not wrong' is true. Contrary to what moral error theorists contend, then, moral error theory seems to carry first-order moral implications that by the theory's own lights are uniformly false. In section 3 I suggest a formulation that is consistent with the standard formulation of moral error theory, free of first-order moral implications, and subject to no logical difficulties.
In section 4 I consider and rebut Stephen Finlay's recent attack on moral error theory. According to Finlay the conceptual claim is false because all moral claims—and indeed all normative claims—are, or should be understood as, relativized to some moral standard or system of ends. Moral error theorists thus attribute to ordinary speakers an error that simply isn't there. Finlay aims firstly to undermine evidence in favour of the conceptual claim and secondly to offer a positive argument for why the conceptual claim is false. I consider Finlay's attempt to undermine what he calls linguistic and disputational evidence for the conceptual claim and argue that it fails. I then argue that Finlay's theory does not avoid commitment to error theory and that it fits less well with ordinary moral though and talk than does Mackie's.
In section 5 I consider the worry that the error theorists' rejection of categorical reasons proves too much; in particular the worry that error theorists' qualms about categorical reasons apply equally to claims about hypothetical reasons, i.e., claims to the effect that there is reason to take the means to one’s ends. In my view error theorists like Mackie and Joyce have failed to pay due consideration to this problem. What the challenge establishes, I submit, is that error theorists cannot just take for granted that hypothetical reasons are metaphysically unproblematic; they must offer an account of hypothetical reasons that shows that they are. I argue that the only plausible account available to error theorists is one according to which claims about hypothetical reasons reduce to non-normative claims about relations between means and ends.
2. Motivating Moral Error Theory
Ever since John Mackie's seminal discussion, standard arguments for moral error theory are routinely lumped together under the label 'arguments from queerness' (Mackie, 1977: Ch. 1). In my view some of these arguments have considerably more force than others. The most acute of Mackie's queerness worries about moral facts is not that moral facts— i.e. facts to the effect that some agent morally ought to do or not to do some action; that there are moral reasons for some agent to do or not to do some action; that some action is morally permissible; that some institution, character trait, or what have you, is morally good or bad; and the like—would be intrinsically motivating in the sense that belief in a moral fact guarantees (some degree of) motivation to behave as the putative moral fact prescribes. This worry is obviously premised on motivational internalism, the view that there is a necessary link between an agent's making a moral judgement and that agent being to some extent motivated to act in accordance with it. But as many critics have pointed out, motivational internalism is after all a highly controversial view (e.g. Brink, 1984; Dworkin 1996). Differently put, it is far from clear that ordinary speakers' notions of moral fact and moral judgement are committed to the strong kind of motivational internalism Mackie takes for granted.
Richard Garner and other commentators have noted that the most acute of Mackie's queerness worries is rather that moral facts would have to be, as Mackie said, objectively prescriptive. What makes moral facts queer is that they make demands from which we cannot escape (Garner, 1990; Joyce, 2001; Finlay, 2008; Robertson, 2008).1
Ronald Dworkin has complained that Mackie's talk about the objective prescriptivity or 'inbuilt to-be-pursuedness' of moral facts is overly metaphorical (1996: 114). I agree that Mackie's discussion is sometimes opaque and I will therefore try to unpack what it is that Mackie and other moral error theorists object to.2
As Mackie and other error theorists have noted, there is a sense in which we are all familiar with objective prescriptivity as instantiated in the real world.3 For instance, it is a familiar fact that chess players ought not to move the rook diagonally and that there are reasons for soccer players not to play the ball to their own goalkeeper when under pressure. But these are not examples of the kind of objective prescriptivity Mackie objected to. Mackie did not deny that there are rules and standards according to which certain agents in certain situations ought or have reason to behave in certain ways (Mackie 1977, pp. 25-7).
The kind of objective prescriptivity Mackie did object to is one that involves categorical reasons. To say that there are categorical reasons for some agent, A, to behave in some way, , is to say that there is reason for A to irrespective of whether A's -ing would promote satisfaction or realization of some of A's desires or aims, or would promote fulfilment of some role A occupies. Suppose for instance that torturing animals for fun is morally wrong and that donating 20% of one's income to charity is morally required. It seems commonsensical that there would then be reasons for any agent not to torture animals for fun and to donate 20% of her income to charity, even if doing so would not satisfy or realize one of her desires or aims, or promote fulfilment of some role she occupies. In other words, moral facts entail facts about categorical reasons and moral claims entail claims about categorical reasons.
Elsewhere I have distinguished between transcendent and immanent norms (Olson, 2009). The former apply to agents categorically; their reason-giving force transcends particular aims, activities, or roles. Immanent norms, by contrast, are those whose reason-giving force depends on agents’ engagement in certain goal-oriented or rule-governed activities or their occupation of certain roles, e.g. institutional or professional roles; the reason-giving force of immanent norms does not transcend rule-governed activities or roles, which is why immanent norms imply merely non-categorical reasons.4 Another way of putting it is to say that while immanent norms determine correct behaviour according to rules or fixed standards, it does not follow that there are categorical reasons to comply with these norms. For transcendent norms, it does follow that there are categorical reasons for compliance.
As mentioned above, it is a plausible conjecture that on the commonsense conceptions of moral norms these are examples of transcendent norms, whereas the norms of e.g. chess, soccer, grammar, and etiquette are prime examples of immanent norms. To say that a norm is a moral norm is to say that there are reasons for any agent to comply with that norm, irrespective of her desires, ends, or roles. To say that some norm is a norm of etiquette or grammar, by contrast, is not to say that there are categorical reasons to comply with it, but rather that some sort of behaviour would be incorrect relative to a certain standard of etiquette or relative to the rules of grammar.5 In my terminology, norms are transcendent or immanent and reasons are categorical or non-categorical.
Error theorists do not object to the existence of immanent norms and non-categorical reasons. There is nothing metaphysically queer about the fact that there is (conclusive) non-categorical reason for chess players not to move the rook diagonally, since this is just the fact that moving the rook diagonally is incorrect according to the rules of chess; there is nothing metaphysically queer about the fact that there is (non-conclusive) reason for soccer players not to play the ball to their own goalkeeper when under pressure, since this is just the fact that such play tends to give the opponent team opportunities to score (and preventing the opponent team from scoring is one of the goals in soccer). Similarly, there is nothing metaphysically queer about the fact that there is non-categorical reason for a soldier to comply with the orders of a general, since this is just the fact that complying with the orders of those superior in military rank is part of the role of being a soldier. Note that a soldier might not desire to comply with the general's order and he might have no ends that would be served by his compliance. That is why I add that error theorists can recognize non-categorical reasons that depend on agents' roles. Agents can occupy roles they have no desire to fulfil.
Moral norms and moral reasons, as we have seen, are a different a matter. The reason-giving force of moral norms transcends agents' desires, aims, and roles. One way of unpacking the popular view that moral facts are non-natural is in terms of categorical reasons. On this interpretation, what non-naturalist realists mean to capture in claiming that moral facts are non-natural is precisely that these facts are or entail categorical reasons.6 By contrast, facts about, e.g., etiquette and rules of grammar are natural since they do not entail categorical reasons.
Following others (e.g. Smith, 1994; Miller, 2003), we can call the claim that moral facts are or entail categorical reasons (and correspondingly that moral claims are or entail claims about categorical reasons) the conceptual claim. Moral error theorists accept the conceptual claim but they also accept the ontological claim that there are no such reasons in reality. Some naturalist realists aim to demystify moral facts by denying the conceptual claim (e.g. Brink, 1984). I shall consider and reject one such recent attempt in section 4 below.
Other realist critics of moral error theory accept the conceptual claim but deny the error theorist's ontological claim.7 The problem for these realists is precisely to explain how there can be facts that in themselves, i.e. irrespectively of the desires, aims, or roles of human beings and other agents, require, or count in favour of, certain forms of behaviour.8 A popular realist rejoinder is to adopt a 'partners in guilt (or innocence)' strategy and claim that moral facts are not metaphysically queerer than, e.g., mathematical and logical facts, or facts about set theory (cf. Scanlon 1998: 62-4).9 The latter kinds of facts about abstracta may be metaphysically problematic in a number of ways but they do not display the feature that moral error theorists find especially queer about purported moral facts—they do not entail categorical reasons.
Someone might object that, e.g., logical facts do entail categorical reasons for belief. An example might be that the fact that p and if p then q entail q, entails that if one believes p and if p then q, there is reason to believe q or give up at least one of the prior beliefs. The error theorist's response is that the reason here is non-categorical, since the claim that if one believes p and if p then q, there is reason to believe q, or give up at least one of the prior beliefs, simply amounts to the claim that according to the modus ponens rule, if one believes p and if p then q, it is correct to believe q, or give up at least one of the prior beliefs (cf. Olson 2009b). By contrast, when we make moral claims we do not merely mean to state or express correct moral rules for behaviour; we mean to say that there are categorical reasons to comply with these rules.
2.1 Two Initial Objections: The Moorean Argument and the Hidden Appeal to Norms
At this point one might object that metaphysical doubts about transcendent norms and categorical reasons are based on pretty advanced, or at least controversial, philosophical theorizing. And are we not comparatively more certain that some actions—e.g. torturing animals or children for fun—really are morally wrong than we are that reality harbours no categorical reasons and consequently no moral truths? Since it marshals commonsense against philosophical theorizing, let us call this argument the Moorean argument against moral error theory.10
But metaphysical qualms about categorical reasons are not the sole cornerstone of the case for moral error theory. Moral error theorists often give debunking explanations of why we humans tend to believe that there are moral facts (Mackie 1977: 105-24; Joyce 2001; 2006). One important ingredient in these debunking explanations is the evolutionary advantages of moral beliefs. For instance, moral norms against stealing, harming, cheating, etc., tend to promote senses of trust and security, which facilitate cooperation, which in turn raise prospects of survival. As Mackie said, in human evolutionary history morality serves as a 'device for counteracting limited sympathies' (1977: 107).11
Belief in transcendent norms and correlative categorical reasons is useful in other respects too: it puts pressure on individual agents and makes them less likely to succumb to temptations to maximise expected short-term egoistic or parochial benefits. In short, morality persists in the world of human thinking partly because of its socially useful coordinating and regulative functions.
In addition there are plausible hypotheses as to how and why belief in moral facts originates in the individual human mind that are congenial to moral error theory. Shaun Nichols (2004) argues that belief in moral norms originates partly because of the linkage to affect. Witnessing suffering in others tends to give rise to intense distress in most human beings and this is at least part of the explanation why most people are strongly motivated to enforce and comply with norms against harming innocents, such as animals and children. Reactive distress causally explains beliefs to the effect that violations of norms against harming are generalizably wrong (Nichols 2004: 180). This clearly echoes Hume's famous dictum that moral judgement stems from a 'productive faculty, [that] gilding or staining all natural objects with the colours, borrowed from internal sentiment, raises, in a manner, a new creation' (1998: 163).
These are rough sketches of attempts at debunking explanations of moral beliefs. Other writers have offered highly impressive and detailed elaborations and I won't delve deeper into the matter here (see e.g. Nichols, 2004; Joyce, 2006). Suffice it to say that these elaborations have enough plausibility to undermine the Moorean argument. For once we take these debunking explanations into consideration it is far from clear that we are more certain that some actions—e.g. torturing animals or children for fun—really are morally wrong, than we are that there are no categorical reasons and consequently no moral truths. Furthermore, debunking theories à la Mackie, Joyce, and Nichols also explain the certainty with which we hold certain moral beliefs. The explanation is simply that the regulative and coordinating functions they facilitate are of such vital importance to us.
It is fairly obvious that the argument against categorical reasons that proceeds via Mackie's queerness worry and debunking explanations of moral beliefs is based on an appeal to Occam's Razor. The gist of the argument, after all, is that error theory offers a theoretically simpler and hence preferable explanation of the phenomenon to be explained (i.e. moral thought and talk) than do competing realist explanations.12 But appeals to Occam's Razor and considerations of theoretical simplicity seem to be appeals to norms. And consequently the moral error theorist's argument against the existence of some norms, e.g. moral norms, seems to involve a hidden appeal to other norms, which makes it smack of self-defeat (cf. Sayre-McCord, 1988: 277f.).
In response the moral error theorist should concede that appeals to Occam's Razor and considerations of theoretical simplicity are indeed appeals to norms. But these are immanent rather than transcendent norms. To say that a theory T offers a theoretically simpler explanation of some phenomenon than a distinct theory T' is not to say that the comparative simplicity of T is a categorical reason to prefer T to T'. It is just to say that T is in one respect preferable to T' according to a standard of theory assessment commonly adopted in many natural and social sciences, to wit that T is preferable to T' if T makes fewer problematic assumptions without loss in explanatory power. This is the case with moral error theory as compared to realism. The greater theoretical simplicity of the former as compared to the latter is therefore a non-categorical reason to prefer moral error theory to realism. Appeals to norms like Occam's Razor are hence unproblematic from the moral error theorist's naturalist perspective.
I hope that what have been said so far makes moral error theory seem if not a promising theory then at least not a dead end in metaethics. That much suffices as a rationale for my defensive project in the remainder of the essay. I shall consider three challenges to moral error theory, starting with the most basic one, according to which the standard formulation of moral error theory is incoherent.
3. Formulating Moral Error Theory
It is routinely said that according to moral error theory, first-order moral claims are uniformly false. A first-order moral claim is a claim that entails that some agent morally ought to do or not to do some action; that there are moral reasons for some agent to do or not to do some action; that some action is morally permissible; that some institution, character trait, or what have you, is morally good or bad; and the like. But this raises the question of what to say about the truth-values of negated first-order moral claims, which leads to two worries: Is the standard formulation of moral error theory coherent?13 Can it be maintained that moral error theory lacks first-order moral implications?
Mackie insisted that his error theory is purely a second-order view and as such logically independent of any first-order moral view (1977: 15-17). But this can be doubted. According to the standard interpretation of Mackie's error theory a first-order moral claim like ’Torture is morally wrong’ is false. According to the law of excluded middle it follows that its negation, ’Torture is not morally wrong’, is true. That torture is not morally wrong would seem to imply that torture is morally permissible. More generally, then, the apparent upshot is that contrary to Mackie's contention moral error theory does have first-order moral implications. And rather vulgar ones at that; if moral error theory is true, any action turns out to be morally permissible!
But it seems that we can also derive an opposite conclusion. According to moral error theory, ’Torture is morally permissible’ is false. According to the law of excluded middle it follows that torture is not morally permissible, which seems to entail that torture is morally impermissible. More generally, then, the apparent upshot is that any action is morally impermissible! This may not be a vulgar first-order moral implication, but it is surely absurd. It also transpires that the standard formulation of moral error theory leads to a straightforward logical contradiction since we have derived that it is true that, e.g., torture is morally permissible (since any action is morally permissible) and that it is false that torture is morally permissible (since any action is morally impermissible). Ronald Dworkin has argued that this demonstrates the impossibility, indeed the incoherence, of being "sceptical about value [...] all the way down" (1996: 91).
Walter Sinnott-Armstrong has suggested the following way out of the predicament: the scope of moral error theory is to be restricted, to the effect that only positive first-order moral claims are deemed uniformly false (2006: 34-6). A positive first-order moral claim is defined as a claim that entails something about what some agent morally ought to do or not to do, what there is moral reasons for some agent to do or not to do, and so on and so forth; or what would be morally good or bad, or morally desirable or undesirable, and so on. It says nothing about mere permissibility.
Restricting moral error theory to positive first-order moral claims only, rids moral error theory from incoherence and from the absurd implication that anything is morally impermissible. But one may object that it remains the case that a negative first-order moral claim such as ‘Torture is not morally wrong’ entails 'Torture is morally permissible' since it seems to be a platitude that any action that is not morally wrong is morally permissible. In other words, moral error theory would still imply vulgar first-order moral nihilism, according to which anything is morally permissible. But Mackie's contention that his error theory is purely a second-order view and as such logically independent of any first-order moral view must be taken to include the first-order moral view that anything is morally permissible. In other words, Mackie's moral error theory holds that no first-order moral claims are true and claims about moral permissibility are no exception.14
A better way out is to deny that the implications from 'not wrong’ to 'permissible’ and from 'not permissible’ to 'wrong’ are conceptual and maintain instead that they are instances of conversational implicature. To illustrate, ‘not wrong’ conversationally implies ‘permissible’ because normally when we claim that something is not wrong we speak from within a system of moral norms, or moral standard for short. According to most moral standards any action that is not wrong according to that standard is permissible according to that standard.15 But the implication from ‘not wrong’ to ‘permissible’ is cancellable. The error theorist can declare that torture is not wrong and go on to signal that she is not speaking from within a moral standard. She might say something like the following: ‘Torture is not wrong. But neither is it permissible. There are no moral properties and facts and consequently no action has moral status.’ This would cancel the implication from ‘not wrong’ to ‘permissible’. (Analogous reasoning of course demonstrates why the error theorist’s claim that torture is not morally permissible does not commit him to the view that torture is morally impermissible and hence morally wrong.) On this view, error theory has neither the vulgar implication that anything is permissible nor the absurd implication that anything is impermissible.
But one might object that the problems remain. The law of excluded middle entails that if ’Torture is wrong’ is false, then ’Torture is not wrong’ is true. If the latter claim is a first-order moral claim, the standard formulation of moral error theory still has first-order moral implications, i.e. implications that by its own lights are false.
In response, recall that according to our above definition first-order moral claims are claims that entail that some agent morally ought to do or not to do some action; that some action is morally permissible; that some institution, character trait, or what have you, is morally good or bad; and so on. Now, according to the view on offer a negated claim like ’Torture is not wrong’ does not entail that torture is permissible; it merely conversationally implies that it is since the implication from ‘not wrong’ to ‘permissible’ is cancellable. Likewise, 'Torture is not morally permissible' does not entail that torture is impermissible and hence wrong; it merely conversationally implies that torture is impermissible and hence wrong. Thus negated atomic claims involving moral terms are not strictly speaking first-order moral claims, but some such claims conversationally imply first-order moral claims.16 Since claims like 'Torture is not wrong' are true we cannot derive that their negations (such as 'Torture is wrong') are true. This saves the standard formulation of moral error theory from the threat of incoherence and from implausible first-order moral implications. I shall continue to say, then, that according to moral error theory first-order moral claims are uniformly false.17
Having defended moral error theory against the most basic challenge I turn in the next section to the challenge that the theory is ill motivated since the error it claims to identify in ordinary moral discourse is a chimera.