Contra Lance Bush

One way to argue for moral realism is simply to rule out all other logical possibilities. If non-cognitivism, error theory, and subjectivism are the only options for the moral anti-realist, then all that is needed is to rule out these three options to show that moral realism is correct.

Lance Bush objects to this way of arguing for moral realism. One reason that he objects is on the grounds that those three positions are not the only three possible positions. I will argue that his objections here are not relevant to the argument, since we can in fact show by a process of logical division that positions relevantly similar to these three positions are in fact the only logical possibilities.

I will continue to go through the process of elimination that this division allows, showing how one can rule out the anti-realist positions. In some places, I will consider objections Lance Bush has raised against some of the arguments against the anti-realist positions.

Finally, I will consider the role empirical evidence has in reasoning about meta-ethics.

Logical division

As said above, the first step in the process of elimination argument for moral realism is showing that the three positions (non-cognitivism, error theory, and subjectivism) are the only three logically possible positions. Once this has been established, it would be sufficient to rule out these three possibilities.

Lance Bush insists that those three views are not the only options for the moral anti-realist:

In short: these three views are not the only antirealist views that one can endorse. Antirealism only requires you to reject moral realism. It does not require you to make a claim about the meaning of moral claims. You could be agnostic about the meaning of moral claims, or endorse indeterminacy, variability, or incoherentism. Fundamentally, the problem with the claim that these are the only three options is that all three positions rely on substantive assumptions about language and meaning that are not an inherent feature of moral antirealism as a position, and one is therefore not obliged to endorse one of these positions to qualify as an antirealist.

As a matter of psychology, I suppose that Lance Bush is right. But what is philosophically relevant is that those are the only three logical possibilities. Then you can derive the disjunctive claim that moral realism, non-cognitivism, error theory, or subjectivism is true. And then if you further establish that the last three are false, you have established the truth of moral realism. The fact that someone could be agnostic or have contradictory views, for example, is not relevant to this point. Of course, there are various ways of dividing up the various meta-ethical positions. We could simply divide it into moral realism and moral anti-realism and by denying moral anti-realism arrive at moral realism, for example. But these three positions are the ones commonly held, and if it is in fact the case that they are logically exhaustive and they can all be refuted, then this would be sufficient.

If Lance Bush was just making a psychological point, then this would not be worth responding to, but he acknowledges the philosophical significance:

One of the strategies you can employ if you insist there’s only three antirealist positions is an argument by process of elimination: if you can demonstrate some deficiency in each of these forms of antirealism, you can declare victory for realism by default. I think it’s dialectically important, as a results, for BB, Huemer, and others, to sideline positions like mine and show they’re not legitimate possibilities. This allows the process of elimination arguments to go through.

He further claims that he has not seen a good argument for these being the only possible positions:

I have yet to see any good argument for why these are the only three options, though. This appears to be, once again, a matter of assertion.

But we can use a process of logical division to show that these (or at least something like these) are in fact the only three logical possibilities.

First, either it is possible that there are some moral sentences that purport to be truth-apt, or it is not the case that it is possible that there are some moral sentences that purport to be truth-apt (i.e., it is necessary that there are no moral sentences that purport to be truth-apt). Let us call the latter claim non-cognitivism. Next, assuming the former claim, it is possible that some moral sentences are in fact true, or it is not possible. Let us call the latter error theory. Next, it is possible that some of the true moral sentences are true in a stance-independent way, or it is not possible. Let us call the latter position relativism and the former one moral realism.

Besides the division above, I think that it is necessary to further divide moral realism. If it is possible that there are true stance-independent moral statements, we can further ask if it is possible that some are true in a way that is irreducible to non-moral statements, or if it is not the case that it is possible that any are irreducible to non-moral statements. Let the former be called moral non-reductionism and the latter be called moral reductionism. Some reductionist views are often classed as being moral realist views, but they are incompatible with my own understanding of morality, so I want to address such views.

Now it is clear from the above that only non-cognitivism, error theory, relativism, moral reductionism, and moral non-reductionism are logically possible. And further it is clear that one (and only one) of them must be true.

The only way to deny this besides denying the law of the excluded middle or the law of non-contradiction, would be to simply deny the coherence of the statements.

Now, Lance Bush rejects the category of ‘moral claim’:

Here’s the problem: I don’t think there is any category like this. I don’t think there is a category of “claim,” a “moral claim,” that shares some specific semantic content. As an aside, I don’t even think there is a moral domain at all, so I’d have considerable objections to there even being a well-defined category of “morality” in the first place.

But this is not a major issue, since the moral realist making this argument can simply stipulate what belongs to the category. The realist can give some examples such as ‘One ought not torture babies on a whim’, ‘One ought to torture babies’, ‘Pleasure is good’, ‘Pleasure is bad’, etc. Then the moral realist and anti-realist can simply discuss the sentences that the realist is stipulating to be moral sentences.

Lance Bush furthers objects to the idea that sentences literally have meanings or purport to be truth-apt, rather thinking that people mean things. But this also is not a problem, since I can simply take his suggestion of treating this as short-hand for what people mean:

When we talk about the “meaning of moral statements,” I take this as an awkward and non-literal characterization of what people mean when they make moral statements, where “moral statements,” would be operationalized as some rough attempt at capturing a subset of ordinary discourse. […]

Besides those, another controversial term might be stance-independent. I would want this to be understood broadly to mean something like being independent of what anyone thinks, feels, says, or does regarding the choice or action or object that moral properties are being attributed to.

It must be admitted that I am simply stipulating the definitions of the terms non-cognitivism, etc. Perhaps some people or even most people use them slightly differently. For example, error theory is sometimes stated as something like the view that moral claims are generally false, but I made sure to define it in such a way that it includes the view that moral claims are incoherent as well. Lance Bush claims in the same article that the view that moral claims are incoherent is not covered in the three possibilities. And perhaps that is true in the way that some people define the positions. However, this is only philosophically significant if the objections to error theory do not also apply to this sort of view. Even if my definitions differ from the normal definitions, as long as the objections given against the views as usually defined also apply to the views as I defined them, then this is not philosophically significant (at least in the context of this particular argument).

The next thing to do is to go through the process of elimination to see if the normal objections apply to my stipulated definitions.

Process of elimination

Non-cognitivism

One minor issue with how I defined non-cognitivism is that you could simply deny the existence of moral sentences or sentences in general. It was defined as the view that necessarily, there are no moral sentences that purport to be truth-apt. So in addition to the view that possibly there are moral sentences but they do not purport to be truth-apt, it also covers the view that it is impossible that there are moral sentences and the view that it is impossible that there are sentences at all. Besides the fact that the view that sentences do not exist is absurd (keep in mind that no specific metaphysical view of sentences needs to be assumed), the standard type of arguments do technically apply to this view. (The view that there cannot be moral sentences specifically, since, for example, morality is incoherent, was effectively covered above by the consideration that I can simply stipulate what sentences I mean when I say ‘moral sentences’.)

The standard sort of objection is an argument along these lines:

(1) If it is grammatically possible to embed some moral sentences in conditionals, then some moral sentences purport to be truth-apt.

(2) It is grammatically possible to embed some moral sentences in conditionals.

(3) Therefore, some moral sentences purport to be truth-apt.

An example of embedding a moral sentence in a conditional is If abortion is wrong, then God is angry. Compare If shut the door, then God is angry. The latter is ungrammatical, but the former is grammatical. Besides conditional statements, we can use similar arguments for similar cases, for example ‘I wonder if it is wrong to eat meat’; it would not make sense to say ‘I wonder if do not eat meat’ or ‘I wonder if Boo eating meat!’, etc.

Now, if you simply deny the existence of sentences, you would deny the second premise, but the conclusion of the argument does rule out non-cognitivism as I defined it, since it concludes that there are actually moral sentences that purport to be truth-apt, but non-cognitivism as I defined it said that it is impossible that there are moral sentences that purport to be truth-apt. Of course, it is true that this response to non-cognitivism does not directly address the view that sentences do not exist. (To address such a view directly, one can point to the many examples of sentences, such as the sentences used in this essay, or the sentences that you speak.)

Anyway, Lance Bush has an objection to this sort of argument against non-cognitivism:

I’ve always found this to be an atrocious and silly objection to noncognitivism. Perhaps a very crude and flat-footed form of noncognitivism is vulnerable to this criticism. According to such a view, the only appropriate use of “X is wrong” amounts to something like “X? Boo!” or “Don’t X.” So this would make certain utterances like the ones BB provides appear incoherent. And it probably does successfully indicate that such a view is incoherent. Only, is a noncognitivist obliged to think that ordinary moral language is as rigid and inflexible as this? No. People may use moral claims primarily to express nonpropositional claims even if one can shift into using them in a propositional way in various contexts. […]

As far as I can tell, Lance Bush is simply suggesting that the non-cognitivist could say that in some cases, moral sentences do purport to be truth-apt. This response does not work for my stipulated definition, but does it work at all? Even if we define non-cognitivism to allow for this, the moral realist can simply focus on the subset of moral sentences that the non-cognitivist acknowledges do in fact purport to be truth-apt. Regarding that subset, we can ask the same questions. Could any of those sentences in fact be true? If not, we have error theory (as I defined it), so any objections to that can be used against the so-called non-cognitivist. If the non-cognitivist says that some can be true, then we can further ask whether they can be true in stance-independent ways or not. If not, then we can use arguments against subjectivism for the subset.

So if all one is claiming is that some moral sentences are non-cognitive, then Lance Bush’s response works, but it does not if one is claiming that all such sentences are non-cognitive. And in the former case, it must be that one of the other positions applies to the subset of cognitive moral sentences, so we can move on.

Error theory

If there is a single example of a true moral sentence, then error theory is false. Bentham’s Bulldog, whom Lance Bush is responding to in the essay that I have been quoting, gives several examples, such as ‘Pleasure is better than pain’ and ‘Torturing people for fun is generally wrong’ (note that the error theorist would also deny the truth of pain is equally good or better than pleasure as well as torture is generally permissible).

Strangely, Lance Bush responds by insisting that for this objection to work the moral sentences must be stance-independently true:

The only way in which this objection would work is if Bb meant that there are obviously some stance-independently true moral statements. So error theory holds that there are no stance-independent moral facts. BB’s objection to this? That it’s obvious that there are such facts. This is not an objection. It is simply an assertion to the contrary. I could simply retort: “No, it’s obvious there are no such facts.” After all, it seems obvious to me. That’s literally true.

But moral error theory (also called moral nihilism) denies that moral statements are true in any sense – stance-independently or otherwise. This is not just the case for how I chose to define error theory. For example, in Ethical Intuitionism, (p. 4) Michael Huemer says moral nihilism (“a.k.a. ‘the error theory’”) ‘holds that evaluative statements are generally false’. He goes on to say ‘Why? Because evaluative statements assert that things have objective value properties, but in reality there are no such properties’.

But even if you define moral nihilism in such a way that it makes a claim about why the statements are false, it is still sufficient to find a case of a true moral statement (other than the kind of true moral statements that the nihilist must account for like it is not the case that murder is immoral and it is not the case that murder is moral, which must be true if you say that murder is immoral and murder is moral are false). If moral nihilism is the conjunction all moral statements are false & moral statements purport to be about stance-independent properties, to show that the conjunction is false, all that is necessary is to show that one of the conjuncts is false, and one of the conjuncts is all moral statements are false, so it is sufficient to show that that is false.

Even Lance Bush appears to agree that there are true moral statements (at least statements that some people would call moral statements). He even accuses Bentham’s Bulldog of making unethical arguments: ‘I rarely say this, but BB’s “arguments” here are not just terrible, they are unethical: even if BB isn’t aware of how misleading his remarks are, he has a minimal moral obligation not to be so negligent and sloppy in his presentation so as to mislead audiences’.

Of course, a committed error theorist is not going to be convinced no matter how many counter-examples you give. If you can convince such a person at all, you would probably need to first argue against objections that the person has to moral realism.

Anyway, if you do accept the counter-examples, then this objection to error theory works even if you define it to include the possibility that moral sentences are incoherent, since incoherent sentences cannot be true. So my stipulated definition did not cause any issue here.

Why think that any moral statements are true?

If the person who is going through this process of elimination already agrees that some moral statement is true, then this is enough to move on to consider relativism. But what reason is there for thinking any of these claims are in fact true?

Here is the sort of reasoning one could use, in the form of a deductive argument:

(1) If some people perceive that other things being equal, one should not cause suffering, then other things being equal, one should not cause suffering.

(2) Some people perceive that other things being equal, one should not cause suffering.

(3) Therefore, other things being equal, one should not cause suffering.

The first premise is obviously true. In general, if you perceive that X, then X. If it was not the case that X, then your supposed perception was not really a perception. Based on this general a priori principle, we can derive the first premise.

Some people just means at least one person, so the second premise can be justified based on introspection (for those of us who introspect that we perceive the relevant fact, anyway).

The argument does not require moral realism. An individual subjectivist would also agree that he perceives certain moral truths, but his account would be that he is simply introspecting on his own preferences (or whatever it is about himself that he thinks makes his moral statements true). A moral naturalist who thinks that ‘one should not do X’, just means that ‘X causes suffering’ would grant that he perceives the truth of the moral claim based on analytic a priori reasoning. A moral non-naturalist would say that he perceives the truth through synthetic a priori reasoning.

That being said, I suppose that certain metaethical theories besides error theory would be incompatible with the second premise. If you are a cultural relativist, then you might say that when you say ‘Causing suffering is generally wrong’ you mean something like ‘My culture generally disapproves of causing suffering’. Depending on how this is interpreted, it is not something that could be directly perceived. If you mean that most people in the region where you live disapprove of causing suffering, then you would not be able to directly perceive that fact. But perhaps you just mean that some people around you show disapproval of causing suffering, then you can perhaps perceive it by relying on memory or other faculties. Or perhaps you mean that you were culturally conditioned to disapprove of causing suffering, in which case you might perceive that through introspection and/or memory.

Of course, if some metaethical theory is incompatible with the second premise, then if upon introspection it does appear that you perceive moral truths, then this would give you evidence against that metaethical theory. So even if the argument ends up being incompatible with certain metaethical theories, it does not require explicitly presupposing that any of them are incorrect.

By the way, my own view is moral non-naturalism. I think that I (and presumably others) are capable of perceiving moral truths through the use of our rational faculties. Similarly to how I can perceive arithmetical, geometric, and logical truths through my rational faculty. I, and people with similar views, think that we can be directly aware of abstract objects and their relations (in this case, moral abstract objects). Anyway, the argument does not require assuming this view.

The committed error theorist will presumably just reject the second premise. If there are no moral facts, then of course you are not perceiving them. But just as non-nihilists might be unable to convince error theorists by appealing to their own introspection or intuition, this response of the error theorist should not be convincing to anyone else.

If the nihilist does want to argue against the second premise, how might the nihilist do so? First, the nihilist might try to argue against subjectivism and moral naturalism. If the nihilist is successful in this, then he might go on to argue that perceiving non-natural moral facts is impossible. He might present arguments for nominalism, or present evolutionary debunking-type arguments, for example.

Relativism

Counter-example thought experiment

Suppose that there is a world where everyone has positive or neutral dispositions towards torture – they approve (or at least do not disapprove of it), they say that there is nothing wrong with it, they do not punish people for engaging in it, etc.

The relativist must say that if someone in that world was, for the first and only time, to say ‘You should not torture people on a whim’ (and has the same sort of meaning as I do when I say it truly), then this statement could not be true. Of course, this applies to any moral statement.

To deny this requires denying moral relativism.

A technical note

I defined relativism broadly enough to include positions such that ‘You should not torture people on a whim’ can mean something like ‘People say disapproving things about torturing people on a whim’ or otherwise mean something about what people say. In which case you might think that since the person is saying ‘You should not torture people on a whim’, it could be true, since now at least one person is saying something disapproving of whimsical torture. But I do not think that this is a serious issue.

You can interpret the sentence in a few ways assuming that it is a statement about what people say.

Either it is about what the speaker says or what others say. If the latter, then the statement is false, since no one else has said anything disapproving about torture.

If the former, how are we to interpret the statement? Either he is saying that he is saying what he is saying or he is saying that he is saying something else. For example, if the latter is the case, perhaps he is saying something like ‘I am saying something disapproving about torturing people on a whim’. But then his statement is false. Saying that you are saying that you disapprove is not the same as saying that you disapprove.

Next, let the former be the case so that ‘You should not torture people on a whim’ means something like ‘I am saying “You should not torture people on a whim”’. In this case, there are two further interpretations. Either the embedded ‘You should not torture people on a whim’ is merely a quote of meaningless English words, or it is meaningful. But if it is meaningful, then his original statement was viciously circular, it means ‘I am saying that I am saying that I am saying that…’, ad infinitum. Such a circular statement cannot be true (so we have error theory, not relativism).

Finally, it could be that ‘You should not torture people on a whim’ means ‘I am saying the English phrase “You should not torture people on a whim”’, where the embedded ‘You should not torture people on a whim’ is being considered independently of its meaning. Finally we have a case where what he is saying could be true. At this point, though, such a view can hardly be called relativism, because it is not dependent on anyone’s stances toward torture. Instead, it might be considered a type of reductionism. But it would be an absurd sort of reductionism, since it suggests that when I say ‘Torture is wrong’, I am not even saying anything about torture, but only the word torture.

Further, if one acknowledges the existence of abstract mind-independent propositions, then I could alter the thought experiment to use those instead. The proposition [you should not torture people on a whim] (let it be whatever proposition is expressed by the anti-realist when he says ‘You should not torture people on a whim’) would be false in the hypothetical world if some sort of relativism is true. (I used a sentence for the thought experiment instead of a proposition originally, because I think that Lance Bush does not believe that there are abstract propositions.)

Ability to reason about morality

Another common sort of objection to relativism is that one can reason about morality, which would not be the case if relativism were true.

If put into a deductive argument, you could say:

(1) If moral relativism is true, it is not the case that it is possible that there are any true moral claims about which one can reason exactly the same way as stance-independent claims.

(2) It is the case that it is possible that there are some true moral claims about which one can reason exactly the same way as stance-independent claims.

(3) Therefore, it is not the case that moral relativism is true.

The first premise is clearly true. Moral relativism as I am defining it states that it is possible that there are true moral claims, but only ones that are not stance-independent. If we accept that in general, it is impossible to reason about two classes, A and B, in exactly the same way if A has some property that B does not have, then it is clear, given relativism, that it is not possible to reason about true moral claims in exactly the same way as stance-independent claims, since the former lack the property of being stance-independent, which the latter has.

Now, obviously, the committed anti-realist will simply deny the second premise. But if someone does in fact accept the second premise, then clearly such a person should reject moral relativism.

Is there any reason to think that the second premise is in fact true?

First, you might have a moral fictionalist account (depending on what you think a fiction is, this view might fall under error theory rather than relativism, or could even fall under moral realism). ‘Torturing babies is wrong’ might just mean ‘According to the moral fiction, torturing babies is wrong’. Compare how ‘Sherlock Holmes lives on Baker Street’ might be considered true when we interpret it as ‘According to the Sherlock Holmes stories, Sherlock Holmes lives on Baker Street’. In a sense, we can reason about fiction. Suppose that a novel says that some character has no siblings, you might conclude that the character has no brothers.

Even in the case of fiction, there is a limit to our ability to reason about it. A fictional work could have inconsistent statements. It might be said that the character has no siblings, but elsewhere be stated that he is a second child (and that the first child is still living). If a sibling is a child by the same parent, and it is the case that there is someone born of the same parent who is still living, then clearly there must be a sibling of the character. At this point, we must simply appeal to the fact that fictional statements are not literally true.

In the case of morality, though, I might reason that because a conjunction of views leads to some sort of inconsistency, then one of the views must be false. But if morality is just fictional, you could just insist that there is nothing wrong with inconsistent statements in fiction.

For example, one might use the argument from marginal cases to argue against the mistreatment of animals. Suppose you give this argument:

(1) If there are no non-physical differences between group X and Y and it is wrong to torture group X, then it is wrong to torture group Y.

(2) There are no non-physical differences between severely mentally retarded humans and normal adult pigs.

(3) It is wrong to torture severely mentally retarded humans.

(4) Therefore, it is wrong to torture normal adult pigs.

(The second premise might not be exactly correct, but that can be safely ignored for this metaethical discussion.)

How is premise 1 to be interpreted? You could say ‘According to the fiction, if there are no non-physical differences…’ or you could say ‘If there are no non-physical differences between group X and Y and according to the fiction, it is wrong to torture group X, then according to the fiction, it is wrong to torture group Y’.

In the first case, the argument is not valid, since premise 1 is not a conditional. Instead, the whole argument must be treated as a fictional argument.

Compare this argument about a fictional character Alice:

(1a) If Alice has no siblings, then Alice has no brothers.

(2a) Alice has no siblings.

(3a) Therefore, Alice has no brothers.

If we interpret 2a as meaning ‘According to the fiction, Alice has no siblings’, and we interpret 1a as ‘According to the fiction, if Alice has no siblings…’, then it is not a valid argument. We must treat the whole thing as a fictional argument. The problem then is that logic and reason need not apply within fiction. Perhaps it is the case that the fiction states that Alice has no siblings but also that Alice has brothers.

Alternatively, 1a could be ‘If according to the fiction, Alice has no siblings, then according to the fiction, Alice has no brothers’. But in this case, what is the justification for believing the premise? We know that if X has no siblings, then X has no brothers is necessarily true because to be a brother is simply to be a male sibling – if there are no siblings at all, then there are no siblings with the property of being male. But if according to the fiction, X has no siblings, then according to the fiction, X has no brothers is not a necessary truth. In fact, there are probably many fictions that claim that a character has no siblings without explicitly claiming that the character has no brothers, and there are presumably some with inconsistent statements about whether a character has siblings.

Now consider the moral case above. If we interpret the whole argument as fictional, then one can simply say that logic does not apply to fiction. One might agree that according to the moral fiction, if there are no non-physical differences, then there are no moral differences, and agree that there are no non-physical differences, but simply say that since logic does not apply to fictions, none of this matters.

The other interpretation is that the argument is not fictional but that it is about a fiction. The first premise says ‘If there are no non-physical differences between X and Y, and according to the fiction, it is wrong to torture X, then according to the fiction, it is wrong to torture Y’. This is not a necessary truth or something that can be known a priori, though. Maybe you can use induction. But if you are familiar with all that the fiction says, then you cannot conclude anything novel. You already know if the fiction states the conclusion. If it does not, then you would simply reject the first premise. And even if you use induction for the first premise, once you arrive at the conclusion, you can simply conclude that the first premise was false after all, since there is a case where there are no non-physical differences between X and Y and the fiction says that it is wrong to torture X, but the fiction says that it is not the case that it is wrong to torture Y.

Next, according to the individual subjectivist who believes that the moral claims in the above argument are true, this is how the argument must be interpreted (or something along these lines):

(1) If there are no non-physical differences between group X and Y and I disapprove of torturing group X, then I disapprove of torturing group Y.

(2) There are no non-physical differences between severely mentally retarded humans and normal adult pigs.

(3) I disapprove of torturing severely mentally retarded humans.

(4) Therefore, I disapprove of torturing pigs.

This has some of the same problems that I mentioned above. What reason could you have for believing premise 1? Maybe for all cases you thought of before considering this argument, you disapproved of torturing both X and Y when there were no non-physical differences (for example, maybe you disapproved of torturing both men and masculine women). But then if you do not currently disapprove of torturing pigs, you would simply discover that premise 1 is wrong. You can only conclude premise 1 based on introspection or a mixture of introspection and induction (according to the individual subjectivist).

In general, with relativism, I suppose that you must deny that there are a priori general moral principles. They would have to be a priori general principles about what you or others think or do regarding some choice or action or object. In general, what people think and do can be completely arbitrary.

I do not expect any of this to be convincing to a committed anti-realist like Lance Bush. For my initial argument, some might simply say that of course the second premise (‘It is the case that it is possible that there are some true moral claims about which one can reason exactly the same way as stance-independent claims’) is false since in fact any true moral claims are stance-dependent. And regarding the inability to use a priori general principles, they will probably simply insist that there is no problem with that (or try to explain it away).

However, for a more naive subjectivist, who upon reflection realizes that he in fact does (at least attempt to) reason about morality just as if it was stance-independent, it might be convincing. He must at least admit that his thinking about morality is inconsistent. And if he wants to maintain that he is properly reasoning about morality, he must abandon moral subjectivism. Speaking from experience, I was once a moral subjectivist, and although it was long enough ago that I do not remember my exact reasons for abandoning it, I do think that a major reason was that it seemed to me that I was attempting to reason about morality in a way that was inconsistent with moral subjectivism.

A note about fictionalism

I treated moral fictionalism as a type of relativism, but depending on how it is conceived, it might not be. If the so-called fiction is mind-dependent, then of course it is a type of relativism. But perhaps the fiction is mind-independent. Then what? Is it a type of moral realism by my definition?

If we imagine the fiction to be something like a set of mind-independent propositions, then what could these propositions be? If ‘Torture is wrong’ means ‘“Torture is wrong” is a member of the set of moral propositions’, then we have the same sort of issue I mentioned in my previous ‘technical note’. If the embedded ‘torture is wrong’ means the same thing as the original ‘torture is wrong’, it will be viciously circular and so incoherent – so it would count as error theory. Alternatively, maybe you think it means something else. In that case, then perhaps you need to say that the sentence ‘torture is wrong’ has two meanings, the literal and non-literal meaning. The literal meaning might try to ascribe some wrongness property to torture, but since no property exists, that meaning is incoherent or false. The non-literal meaning is the according to the fiction meaning. If you want to make a distinction between literal and non-literal meanings, then I could simply change the original division so that instead of dividing into views that say it is possible that there are true moral claims and views that say that it is not possible that there are true moral claims, I could simply say ‘literally true’ instead of ‘true’.

One could also say that it is a set of sentences rather than propositions. So that when you say ‘Torture is wrong’ you are saying ‘There is a set that contains the English phrase “torture is wrong”’. This technically would qualify as moral realism as I defined it. But it would be an absurd view. If the person with the view were to say ‘Torture is wrong’, he must admit that what he is saying is not about torture in any way (though it is partly about the word ‘torture’). Further, my concern about not be able to reason about morality using a priori principles applies here as well, since the moral fiction is just an arbitrary set containing meaningless phrases.

Moral reductionism

Originally I was only going to cover anti-realist views, but my discussion about fictionalism led me to consider that it would be worth discussing moral reductionism as well.

As I said above, besides the division of views I made, one could further divide moral realism into moral reductionism and moral non-reductionism, with the former being the statement that it is possible that there are true stance-independent moral statements, but that they are all reducible to non-moral claims, and the latter the view that it is possible that there are true stance-independent moral statements that are not reducible to non-moral claims.

It is clear that, assuming that moral realism as I defined it is correct, one of these views must be correct. Moral realism as I defined it said that it is possible that there are true stance-independent moral sentences. So either all of the possible true moral statements are reducible, or it is not the case that all of them are.

Technically the view resembling moral fictionalism which I considered above counts as moral reductionism (at least in the definition I stipulated). If ‘You should not torture people’ means there is a set that contains the English phrase ‘You should not torture people’, and all other true moral statements are similar, then moral statements are reducible to non-moral statements. But such a view as this is absurd. The statement ‘Torture is wrong’ has nothing to do with torture on this view, only the word torture. Granted, I cannot be sure what other people mean when they make such statements, but I am quite confident at least that I am saying something about torture if I say torture is wrong. I highly suspect that others will feel similarly about their own use of moral statements upon introspection.

The traditional sort of moral reductionist view is more reasonable than that, though. For example, it might be that ‘I should not torture’ just means something like ‘torture generally causes suffering on net’.

The traditional sort of response to this sort of view is Moore’s open question argument.

Here is a statement of the argument from Wikipedia:

Premise 1: If X is good by definition, then the question “Is it true that X is good?” is meaningless.

Premise 2: The question “Is it true that X is good?” is not meaningless (i.e. it is an open question).

Conclusion: X is not (analytically equivalent to) good.

For example, if you say that ‘good’ just means pleasure, then ‘Is it true that pleasure is good?’ is no more a reasonable question than ‘Is it true that the bachelor is unmarried?’. If you understand what the word ‘good’ means and it just means pleasure, then there could be no reason to ask if pleasure is good.

Moore was focused on the predicate good and thought that ought-statements are reducible to statements about goodness. But you can apply the same sort of argument to ought-statements. You might think that ‘You ought not torture’ means ‘Torture generally causes suffering on net, and more so than alternatives’, but then supposing that you know in fact that torture does generally cause suffering on net, and more so than alternatives, can you still reasonably ask whether you ought not torture?

I cannot be sure what others mean when they say such things, but upon introspection it seems to me that I could reasonably ask whether I should or should not do something, even if I knew what sort of consequences it leads to. I lean towards hedonistic utilitarianism, but I am not certain of it. If my moral statements are given a moral reductionist interpretation, then apparently this uncertainty must simply be a function of my uncertainty about what I mean with my statements.

Alternative logical division of positions

Instead of dividing the positions as I did, it might be better to divide them differently. The reader might have noticed that the objection to error theory also applied to non-cognitivism and the objection to moral reductionism also applied to relativism.

So perhaps it makes more sense to divide meta-ethics into three positions: moral non-reductionism, moral reductionism, and moral nihilism. The first would say that it is possible that there are some true irreducible moral statements; the next, that it is possible that there are some true moral statements, but only if they are reducible; and the last, that it is not possible that there are any true moral statements.

If I designed the argument rather than defending an already existing one, perhaps I would have done it this way. Of course, there is nothing wrong with presenting further arguments against particular versions of these views, for example presenting the arguments against relativism in addition to arguments against reductionism in general.

Even if this is considered an issue, it is an issue with presentation rather than substance.

Empirical concerns

Lance Bush seems to put a lot of emphasis on the need for empirical research. He seems to think that to determine what metaethical position is correct, you might need to do empirical research about how people use language:

1. I take facts about what people mean to be empirical questions

2. I endorse folk metaethical indeterminacy, i.e., I think that with respect to metaethical considerations, ordinary people rarely mean to express claims that determinately fit a realist or antirealist analysis at all

Mainstream analytic philosophers addressing questions about the meaning of moral claims tend to either (1) appear not to treat such questions as empirical at all, and instead see such claims as a priori. I reject this view, or (2) may nominally regard such questions as empirical but believe armchair philosophy is sufficient to settle such questions reasonably well. I disagree with (2) as well. As such, I either fundamentally reject the philosophical presuppositions behind the approach most philosophers take to addressing metaethical questions, or, at best, believe they’re using embarrassingly bad methods for addressing those questions: I am adamantly against generalizing from the armchair. If you want to know what nonphilosophers mean when they make moral claims, you need to engage in empirical research.

Taking into considerations Bush’s concerns about philosophers’ making empirical claims, I made sure to be careful to avoid any empirical claims in this essay. It might be true that if one were to consider non-cognitivism to say that when most people make moral sentences, they are making sentences that do not even purport to be truth-apt, that we would need an empirical investigation to settle the issue. However, if we instead define it as saying that (it is necessary that) there are no moral sentences that purport to be truth-apt, and upon reflection I realize that I use moral language in a way that purports to be truth-apt, then there is no need for any further empirical research. Likewise, when I considered subjectivism and moral reductionism, I made sure to focus on introspection.

In the case of error theory, if you can find a single example of a moral statement that is true, there is no need for empirical research about what others believe about morality. Regarding my perception-based argument for the truth of moral statements, the controversial premise was based on introspection.

However, empirical studies could provide some evidence against the claim that I perceive that some moral statement is true. Consider a non-moral case. For example, if I look outside and believe to be seeing a squirrel in my backyard, but then a bunch of others insist that they see no squirrel, I might reasonably start to think that I might be hallucinating. If there are a bunch of studies where people insist that all moral statements are false, then that provides some reason for thinking that I am mistaken. However, it is important to distinguish between reports about people’s perceptions and simple claims that something is true or false. For example, in the squirrel case, imagine that there was a widespread religion that denies the existence of squirrels and insists that when you seem to be seeing squirrels, you are really experiencing a hallucination caused by demons. If the people who disagreed with you about the squirrel all said that their anti-squirrel religious beliefs were the reason that they believe that there is not a squirrel when you think that you are seeing one, then this would undermine the evidence against your belief that their claims would otherwise provide. If the reason that people deny a moral belief is due to ideological reasons, then it would not be strong evidence against the belief.

Likewise, for studies specifically about metaethics that purport to be about people’s intuitions but merely report on beliefs, this distinction needs to be kept in mind. It could be that they adopt such beliefs despite in fact perceiving the truth of moral realism.

The value of introspection

Someone might take issue with so much reliance on introspection.

There are certainly cases of poor reasoning using introspection. Imagine if I introspect on my sexual preferences and conclude that I am not sexually attracted to other men, and then I further conclude that homosexuals do not exist. This would of course be bad reasoning. Imagine instead that I am considering the claim that there are no heterosexuals, and introspect that I am a heterosexual. This obviously would be strong evidence (from my own perspective) against the claim that there are no heterosexuals. Not only that, but if someone else is claiming that there are no heterosexuals, my report of my introspection would be some evidence against that person’s claim from the other person’s perspective as well.

Further, even if my introspection does not provide strong evidence for others to believe in something, I, and other moral realists using similar arguments, can invite readers to use their own introspection. Imagine that I am a counter-apologist against the hypothetical anti-squirrel religion I mentioned above. The religion claims that there are no squirrels, but in fact I have seen squirrels myself, so clearly there are squirrels. Not only does my report of having seen squirrels provide some evidence against the anti-squirrel religion, but, more importantly, I can also invite members of the religion reading my writing to consider their own experiences. Have they seen squirrels? If so, there must be squirrels after all. Of course, if they have never seen any squirrels, perhaps living in a location where there are not any, then they cannot rely on their own perception, though my report still provides some evidence.

Now, regarding the specific uses of introspection in this essay, are any problematic?

In the non-cognitivism section I relied on introspection regarding my own use of moral language. If non-cognitivism is defined as saying that it is not possible for moral sentences to purport to be truth-apt, but upon introspection, I realize that I in fact do use moral sentences that purport to be truth-apt, this would certainly be enough from my own perspective to refute non-cognitivism as defined (I am not claiming that any self-described non-cognitivists in fact hold such a view, they might not). Further, people who can introspect and arrive at this same conclusion about their own uses of moral language can also rule out such a view. Finally, even if they dispute that they use moral language in such a way, my claim that I do provides them with significant evidence against such a view.

Likewise, when introspecting regarding relativism or moral reductionism, presumably I am capable of having a decent idea of what I mean with my moral language and so can rule out certain interpretations. Further, others should at least be willing to consider that I do in fact correctly introspect about my use of moral language.

The most questionable case is the perception-based argument I used in the error theory section. My report of perceiving certain moral truths is some evidence for others that there are in fact moral truths. However, it need not be conclusive. Compare a case where someone claims to perceive auras. This does provide some evidence that they are perceiving something that they call an aura, but I would not become a realist about auras. Of course, some people might simply be hallucinating. People sometimes think that they see things that are not real. Another possibility is that the person is in fact perceiving things such as facial expressions and body language and then reifying them into a mysterious supernatural aura.

The fact that perception of auras seems to rely on some sort of perceptual faculty that I do not have makes such things especially hard for me to believe. If the people perceiving them claimed to visually see them, then I would simply rely on my own sight to be quite confident that they do not exist. If the people perceiving them claim to rely on some sort of perceptual faculty that I lack, I would find it hard to believe that they have a faculty that I lack, and so would put much less credence in their claim to perceive auras than if they were claiming to rely on a faculty that I think that I have.

Moral anti-realists can use similar strategies to reject moral realism as I would use to reject auras. Still, they should not deny that moral realists who believe themselves to perceive moral truths do in fact have reason to believe in such truths. Further, they should not deny that it provides the anti-realists with some evidence (albeit weak evidence).

I suspect that moral anti-realists (as well as moral realists) are often relying on background metaphysical and epistemic assumptions. If they deny in general that it is possible to perceive non-natural truths, then they will put less credence in claims of specifically perceiving non-natural moral truths. If someone claimed to rationally intuit some metaphysical truth, I would take that more seriously than claims to perceive things like auras, because I am committed to the belief that people (certainly myself at least, but presumably others) are in general capable of perceiving things through a rational faculty, but not that people have an aura-perceiving faculty. Of course, I will still give much more credence to my own intuitions.

So I do not think that my use of introspection is problematic. Some of it is comparable to cases that I would expect are almost entirely uncontroversial (such as using introspection on your own preferences to reject the view that people with such preferences do not exist). The one most questionable case is the argument I considered when discussing error theory. But even here, if I introspect and it appears to me that I am perceiving something, it is surely perfectly reasonable for me to believe it. And there is nothing wrong with appealing to readers to use their own perception – just like the hypothetical case of appealing to the members of the anti-squirrel religion regarding their own sightings of squirrels. Further, my report of my perceptions does provide at least some evidence to other people who do not have similar experiences (but of course I do not think that such second-hand reports should in all cases convince opponents, just as I would not be convinced by the anti-realist who claimed to perceive the truth of anti-realism). Of course, if you just grant in the first place that there are moral truths, then you can skip that argument.

Objection: Perhaps people mean multiple things

One might object to my use of introspection on the grounds that I (and others who are introspecting) might mean multiple things with their moral sentences. For example, perhaps if I was to say ‘Murder is wrong’, I might, at least in some contexts, be simultaneously saying ‘Boo murder!’, ‘I disapprove of murder’, and ‘There is a stance-independent reason not to murder’.

However, I do not think that this causes any serious issues. First, if what I said in the non-cognitivism section is correct, then this would lead me to conclude that I at least intend a purportedly propositional meaning or meanings even if I additionally intend a non-cognitivist meaning. Next, in the error theory section, I can conclude that I perceive that at least one of the meanings is true, if not more than one.

What about for subjectivism? You might think that I can confuse myself into thinking that the objective meaning is true due to my perception of the truth of the subjective meaning. But I think that I can just focus in on the meaning or meanings I was ascribing truth to and then consider the thought experiment. Does it appear that in fact the meaning or meanings I was ascribing truth to are stance-dependent? Further, can I in fact engage in the type of reasoning one would engage in for stance-independent claims for the meaning or meanings that I was ascribing truth to? In fact, it might even be the case that one of the moral statements I was ascribing truth to was something I concluded from some process of reasoning involving a priori general principles of the sort that cannot be intuited regarding stance-dependent things.

Finally, a similar strategy can be employed for moral reductionism. Are the meaning or meanings I was ascribing truth to subject to the open-question argument? If so, the meaning or meanings must not be reductionist.

So I do not think that this objection has much force.

Conclusion

Lance Bush takes issue with the claim that there are exactly three alternatives to moral realism: non-cognitivism, error theory, and relativism. As a matter of psychology, he is right that you could hold different views. Further, if we are not careful in how we define these terms, there might be other logical possibilities. But neither of these points is philosophically significant (from the point of view of making the process of elimination argument, anyway). One can easily use a process of logical division to arrive at four logical possibilities that resemble moral realism plus the three moral anti-realist theories. Perhaps these stipulated definitions differ from the standard ones a little bit, but the same sort of objections used against these positions generally address the stipulated definitions as well, so for the sake of the argument, this is unimportant.

The sort of arguments I presented for or against certain meta-ethical positions do not rely on the sort of empirical evidence that require something like a scientific study. They all either rely on introspection or other types of perception. There is no issue with relying on ones own perception and encouraging readers to do the same. Further, second-hand claims of ones own perception does provide some weak evidence from the perspective of others.