Reference Material:
The Myth of Morality by Richard Joyce Ethics: Inventing Right and Wrong by J. L. Mackie
This post is a direct follow-up to my previous one, an introduction to moral error theory. Here I’ll present and comment on Richard Joyce’s argument for error theory while drawing on some of J. L. Mackie’s original work on the topic. Let’s dive right in…
Joyce’s argument has the following structure (77):
- If x morally ought to φ, then x ought to φ regardless of what his desires and interests are.
- If x morally ought to φ, then x has a reason for φing.
- Therefore, if x morally ought to φ, then x can have a reason for φing regardless of what his desires and interests are.
- But there is no sense to be made of such reasons.
- Therefore, x is never under a moral obligation.
I see premise 1 as relatively uncontroversial. To prime our intuition, Joyce uses the example of Nazi war criminals (43). In judging such people, at a war crimes trial for example, we do not take into account what their desires or interests were at the time of their crimes. We simply say (risking considerable understatement) “you shouldn’t have done what you did – it was wrong”. If the Nazis were to defend themselves – “but we were trying to take over the world and create a global Aryan nation” – we would not reason that, given those goals, perhaps what they did wasn’t actually wrong. After all, starting a world war and perpetrating genocide seem like perfectly sensible ways of realizing global domination and racial purity. On the contrary, we judge their actions to be wrong (evil, despicable, etc.) regardless of what their goals were or how they felt about them. That is, we morally assess their behavior against a categorical imperative, and find it wanting in the extreme.
It’s worth elaborating a bit to flesh out this concept of a categorical imperative. It’s a requirement of action or inaction on the part of someone, completely independent of their desires, purposes, or ends. Premise 1 is basically claiming that we express moral requirements as categorical imperatives. Joyce has another nice example to help substantiate this claim (62):
If you are told that somebody named “Jack” broke into a stranger’s house, attacked the inhabitants, and all with the intention of taking their money for idle purposes, then from a moral point of view you have all the information needed to condemn Jack’s action. There is no need to investigate further his desires (or ends); the judgement is not made under the assumption that his desires were a certain way. To agree with this is to agree that morality employs categorical imperatives.
Again, any defense from Jack is going to fall flat. We simply don’t need to know why he was doing what he was doing or what he was trying to accomplish. We morally condemn his actions regardless.
So categorical imperatives have this property, expressed in premise 1, that they apply always to everyone, regardless of desires, motives, purposes, etc. When we morally condemn someone – a Nazi, or Jack – there is a sense in which we say to them, “These are the moral rules. They apply to you, whether you like it or not.” But moral categorical imperatives also have another interesting property, a kind of inescapability or bindingness. This property can be brought out by considering a class of categorical imperatives that share morality’s scope and reach, but not it’s inescapability. These are institutional imperatives.
I know this feels like a detour but bear with me – these ideas are important later in the argument. Institutional imperatives are requirements for or restrictions on action that originate in and derive their force from a particular institution. An institution here, according to Mackie, is “constituted by many people behaving in fairly regular ways, with relations between them which transmit and encourage and perhaps enforce those ways of behaving” (80). An institution can be something as specific as the game of chess or as general as the practice of making and keeping promises. A specific example will help us think about how institutional imperatives differ from moral ones. Joyce invites us to consider Celadus the Thracian, a man forcibly taken from his previous life and, against his will, made a gladiator (34). In a certain fight, Celadus is being overpowered by a superior opponent. He’s lost his sword and is pinned to the ground, and the only way he’ll live to fight on is by throwing sand in his opponent’s eyes. As it happens, throwing sand in the eyes of other fighters is prohibited by the rules of gladiatorial combat. What should Celadus do?
Well, there’s a sense in which Celadus shouldn’t throw the sand. After all, sand throwing is prohibited. But there seems to also be a stronger sense in which Celadus should throw the sand. He never asked to be a gladiator. He certainly doesn’t have much respect for the rules. If throwing sand is the only way he’s making it out of this fight alive, then throwing sand it is. So it seems that Celadus should do (throw the sand to save himself) what he shouldn’t do (throwing sand is prohibited). What’s going on here? What we’re seeing is an institutional imperative in action. The rules of gladiatorial combat are what they are, and they apply to Celadus whether he likes it or not, regardless of how he feels about them. They don’t cease to be the rules just because Celadus has other ideas in mind. In this non-evaporability, they are like moral categorical imperatives. But unlike a moral categorical imperative, an institutional imperative can be superseded by more pressing concerns. It seems perfectly reasonable for Celadus to throw the sand, live to fight another day, and defend his actions – “I know it was against the rules, but it was my only chance. Plus, I never wanted to be a gladiator in the first place. Why should I care that much about the rules?” For this reason, Joyce classes institutional imperatives as weak categorical imperatives (36). Now consider the analogous defense from a Nazi war criminal. We wouldn’t accept them stepping outside the “institution of morality”, so to speak, and claiming that the moral rules were superseded in their case by more important matters. In this way, moral categorical imperatives have a bindingness or inescapability that institutional imperatives just don’t – they are strong categorical imperatives.
Now on to premise 2. Here again the case is, in my opinion, not too difficult to make. Joyce refers to premise 2 as “Mackie’s Platitude” – that if an agent ought to do something, they necessarily have a reason for doing that something (38). Think about the Nazi war criminal example. It would be extremely odd, perhaps nonsensical even, “for their moral jury to admit ‘you ought not to have done it, but we accept that you had no reason to refrain’” (44). That just makes no sense. Of course the Nazis had a reason to refrain from what they did. It was wrong. Put this way, it certainly seems that moral obligations are inherently reason-bringing. However, as with premise 1, there is more nuance that should be unpacked here.
What more can be said about these reasons that necessarily follow from moral obligations? Again, it is helpful to contrast moral reasons with institutional reasons. Institutional reasons are exactly what they sound like – reasons that follow from institutional imperatives. Joyce nicely characterizes these reasons (40-1):
So categorical imperatives have this property, expressed in premise 1, that they apply always to everyone, regardless of desires, motives, purposes, etc. When we morally condemn someone – a Nazi, or Jack – there is a sense in which we say to them, “These are the moral rules. They apply to you, whether you like it or not.” But moral categorical imperatives also have another interesting property, a kind of inescapability or bindingness. This property can be brought out by considering a class of categorical imperatives that share morality’s scope and reach, but not it’s inescapability. These are institutional imperatives.
I know this feels like a detour but bear with me – these ideas are important later in the argument. Institutional imperatives are requirements for or restrictions on action that originate in and derive their force from a particular institution. An institution here, according to Mackie, is “constituted by many people behaving in fairly regular ways, with relations between them which transmit and encourage and perhaps enforce those ways of behaving” (80). An institution can be something as specific as the game of chess or as general as the practice of making and keeping promises. A specific example will help us think about how institutional imperatives differ from moral ones. Joyce invites us to consider Celadus the Thracian, a man forcibly taken from his previous life and, against his will, made a gladiator (34). In a certain fight, Celadus is being overpowered by a superior opponent. He’s lost his sword and is pinned to the ground, and the only way he’ll live to fight on is by throwing sand in his opponent’s eyes. As it happens, throwing sand in the eyes of other fighters is prohibited by the rules of gladiatorial combat. What should Celadus do?
Well, there’s a sense in which Celadus shouldn’t throw the sand. After all, sand throwing is prohibited. But there seems to also be a stronger sense in which Celadus should throw the sand. He never asked to be a gladiator. He certainly doesn’t have much respect for the rules. If throwing sand is the only way he’s making it out of this fight alive, then throwing sand it is. So it seems that Celadus should do (throw the sand to save himself) what he shouldn’t do (throwing sand is prohibited). What’s going on here? What we’re seeing is an institutional imperative in action. The rules of gladiatorial combat are what they are, and they apply to Celadus whether he likes it or not, regardless of how he feels about them. They don’t cease to be the rules just because Celadus has other ideas in mind. In this non-evaporability, they are like moral categorical imperatives. But unlike a moral categorical imperative, an institutional imperative can be superseded by more pressing concerns. It seems perfectly reasonable for Celadus to throw the sand, live to fight another day, and defend his actions – “I know it was against the rules, but it was my only chance. Plus, I never wanted to be a gladiator in the first place. Why should I care that much about the rules?” For this reason, Joyce classes institutional imperatives as weak categorical imperatives (36). Now consider the analogous defense from a Nazi war criminal. We wouldn’t accept them stepping outside the “institution of morality”, so to speak, and claiming that the moral rules were superseded in their case by more important matters. In this way, moral categorical imperatives have a bindingness or inescapability that institutional imperatives just don’t – they are strong categorical imperatives.
Now on to premise 2. Here again the case is, in my opinion, not too difficult to make. Joyce refers to premise 2 as “Mackie’s Platitude” – that if an agent ought to do something, they necessarily have a reason for doing that something (38). Think about the Nazi war criminal example. It would be extremely odd, perhaps nonsensical even, “for their moral jury to admit ‘you ought not to have done it, but we accept that you had no reason to refrain’” (44). That just makes no sense. Of course the Nazis had a reason to refrain from what they did. It was wrong. Put this way, it certainly seems that moral obligations are inherently reason-bringing. However, as with premise 1, there is more nuance that should be unpacked here.
What more can be said about these reasons that necessarily follow from moral obligations? Again, it is helpful to contrast moral reasons with institutional reasons. Institutional reasons are exactly what they sound like – reasons that follow from institutional imperatives. Joyce nicely characterizes these reasons (40-1):
Let us return to the unfortunate Celadus. An enthusiastic Roman onlooker may say “Celadus ought not throw sand” or “Celadus has a reason not to throw sand.” The onlooker isn’t making a mistake – she is merely expressing endorsement of a normative system. But all of this only concerns what an onlooker may legitimately say without violating any rules of linguistic conduct. It doesn’t affect Celadus. He is in the same position as we left him earlier: it is still the case that what he ought to do, all things considered – what he really has a reason to do – is throw the sand. We accepted earlier that the weak categorical imperative of gladiatorial combat doesn’t bind him – that it was something that he could reasonably ignore – and the introduction of the permissibility of the onlooker expressing much the same prescription by reference to reasons does not alter that. We merely fall back on the same rhetoric as was employed earlier: it may be legitimate for an onlooker to say “Celadus has a reason not to throw sand,” but this doesn’t provide Celadus with a real reason – it’s not his reason, it is something he may legitimately ignore.
What we see here is that the lack of inescapability or bindingness of institutional imperatives carries over to the reasons they produce. Institutional reasons are reasons that can be legitimately ignored. Contrast this with the moral case. We think Jack ought not to have broken into the stranger’s home, that he had a reason to not do so. Moreover, we believe this reason to be somehow unavoidable. Imagine Jack putting up a Celadus-style defense: “Well yes, according to the moral rules I should not have broken into the stranger’s house, but what are those rules to me? What reason do I have to follow them? They’re not my rules!”. We would of course not accept this as a legitimate defense. We don’t view morality as just another institution, one that can be opted into or out of based on personal desires or interests. Moral obligations, and the reasons they bring, are everyone’s.
We now come to the crux of the argument. Premise 2 implies that moral discourse is committed to certain kinds of reason for action. These reasons must provide the inescapable and binding force characteristic of moral obligations. Because of this, they must also be non-institutional reasons. What class of reasons could fulfill these requirements? Where could such reasons come from? Joyce makes the case that practical rationality, and perhaps practical rationality alone, can provide such inescapable, non-institutional reasons. The most important point to make here is that practical rationality cannot be legitimately ignored. Joyce shows this by considering 4 skeptical attempts to evade purported reasons for action (50):
We now come to the crux of the argument. Premise 2 implies that moral discourse is committed to certain kinds of reason for action. These reasons must provide the inescapable and binding force characteristic of moral obligations. Because of this, they must also be non-institutional reasons. What class of reasons could fulfill these requirements? Where could such reasons come from? Joyce makes the case that practical rationality, and perhaps practical rationality alone, can provide such inescapable, non-institutional reasons. The most important point to make here is that practical rationality cannot be legitimately ignored. Joyce shows this by considering 4 skeptical attempts to evade purported reasons for action (50):
i. I know that morality requires that I φ, but why should I, according to practical rationality, care about morality?ii. I know that morality requires that I φ, but why should I, according to morality, care about morality?iii. I know that practical rationality requires that I φ, but why should I, according to practical rationality, care about practical rationality?iv. I know that practical rationality requires that I φ, but why should I, according to morality, care about practical rationality?
Of course, ii. and iii. are obviously self-undermining. Asking “why should I, according to x” implies your acceptance of x-type reasons, so you cannot then question their legitimacy. But i. is coherent – indeed this is just moral skepticism. However, iv. is not similarly coherent. It is self-undermining because to ask for any reason at all is to assent to practical rationality. Practical rationality cannot be legitimately questioned – to question it is to accept it.
The question then becomes – what kinds of reasons does practical rationality provide? In particular, can practical rationality supply the categorical feature of moral discourse expressed in premise 1? To address these questions, we need a working theory of practical rationality. Joyce advocates for what he dubs non-Humean instrumentalism, a view similar to that of Michael Smith (108). On this view, you’re practically rational if your decisions are guided or constrained by your subjective reasons. You have a subjective reason to φ if you are justified in believing that a rationally improved version of yourself (you with perfect access to relevant information and ideal reflective capabilities) would advise you to φ. I won’t defend this view of practical rationality in detail here – no doubt there are other conceptions about what it means to be practically rational and there is room for considerable disagreement here. The important point is that, on this view, practical rationality yields reasons that depend on the desires and interests of the agent in question. Your rationally upgraded self knows about your desires, interests and long-term goals, and will advise you based on them. The reasons for action you will have will then depend, at least in part, on those desires and interests. Practical rationality then delivers only hypothetical imperatives, not categorical ones. It cannot supply the categorical imperatives described by premise 1, and premise 4 beckons.
This part of the argument may seem convoluted to some, so it may be useful to restate it more succinctly. Moral obligations imply reasons for action (premise 2). These reasons must have a kind of inescapability or bindingness. That is, they must be non-institutional reasons that cannot be legitimately questioned with a “so what?” response. Perhaps the only source for such inescapable, non-institutional reasons is practical rationality. This is because practical rationality, by its very nature, cannot be legitimately ignored. But practical rationality cannot provide the kind of reasons required by moral obligations. This is because moral obligations are categorical imperatives (premise 1). They apply equally to everyone, independent of people’s diverse interests and desires. The reasons of practical rationality, by contrast, are mere hypothetical imperatives – they depend on one’s desires, interests, and goals. Now, there may well be other kinds of reasons beyond those of practical rationality that can both underwrite categorical imperatives and supply the inescapability of moral obligations. But absent a coherent account of such reasons, it seems the burden of proof is on the moral realist to show that premise 4 does not then follow.
Well suppose we accept this argument. Now what? In Joyce’s words, “if there’s nothing that we ought to do, then what ought we to do?” (175). Of course, addressing this question could be a separate post (or a life’s work) unto itself. And it’s important to note that the question is not self-undermining – the first ought is moral, whereas the second ought is prudential or practical. Let’s also remember that we can, without logical contradiction, still hold first order moral positions (or something like them) while endorsing error theory, a second order position (see my introduction to moral error theory). I’ll just close with part of Mackie’s response to this question, which I found oddly comforting (148):
The question then becomes – what kinds of reasons does practical rationality provide? In particular, can practical rationality supply the categorical feature of moral discourse expressed in premise 1? To address these questions, we need a working theory of practical rationality. Joyce advocates for what he dubs non-Humean instrumentalism, a view similar to that of Michael Smith (108). On this view, you’re practically rational if your decisions are guided or constrained by your subjective reasons. You have a subjective reason to φ if you are justified in believing that a rationally improved version of yourself (you with perfect access to relevant information and ideal reflective capabilities) would advise you to φ. I won’t defend this view of practical rationality in detail here – no doubt there are other conceptions about what it means to be practically rational and there is room for considerable disagreement here. The important point is that, on this view, practical rationality yields reasons that depend on the desires and interests of the agent in question. Your rationally upgraded self knows about your desires, interests and long-term goals, and will advise you based on them. The reasons for action you will have will then depend, at least in part, on those desires and interests. Practical rationality then delivers only hypothetical imperatives, not categorical ones. It cannot supply the categorical imperatives described by premise 1, and premise 4 beckons.
This part of the argument may seem convoluted to some, so it may be useful to restate it more succinctly. Moral obligations imply reasons for action (premise 2). These reasons must have a kind of inescapability or bindingness. That is, they must be non-institutional reasons that cannot be legitimately questioned with a “so what?” response. Perhaps the only source for such inescapable, non-institutional reasons is practical rationality. This is because practical rationality, by its very nature, cannot be legitimately ignored. But practical rationality cannot provide the kind of reasons required by moral obligations. This is because moral obligations are categorical imperatives (premise 1). They apply equally to everyone, independent of people’s diverse interests and desires. The reasons of practical rationality, by contrast, are mere hypothetical imperatives – they depend on one’s desires, interests, and goals. Now, there may well be other kinds of reasons beyond those of practical rationality that can both underwrite categorical imperatives and supply the inescapability of moral obligations. But absent a coherent account of such reasons, it seems the burden of proof is on the moral realist to show that premise 4 does not then follow.
Well suppose we accept this argument. Now what? In Joyce’s words, “if there’s nothing that we ought to do, then what ought we to do?” (175). Of course, addressing this question could be a separate post (or a life’s work) unto itself. And it’s important to note that the question is not self-undermining – the first ought is moral, whereas the second ought is prudential or practical. Let’s also remember that we can, without logical contradiction, still hold first order moral positions (or something like them) while endorsing error theory, a second order position (see my introduction to moral error theory). I’ll just close with part of Mackie’s response to this question, which I found oddly comforting (148):
What the individual can do is to remember that there are, in the different circles of relationship with which he is concerned, various fragments of a moral system which already contributes very considerably to countering specifiable evils which he, like others, will see as evils; that he can at once take advantage of this system and contribute to its upkeep; but that he may be able, with others, to put pressure on some fragments of the system, so that they come gradually to be more favourable to what he sees as valuable or worthwhile.