Expressivism and Explaining Irrationality: Reply to Baker

Abstract

In a recent paper in this journal, Derek Baker (Erkenntnis 83(4):829–852, 2018) raises an objection to expressivism as it has been developed by Mark Schroeder (Being for, Oxford University Press, Oxford, 2008). Baker argues that Schroeder’s expressivist (1) is committed to certain sentences expressing rationally incoherent states of mind, and he objects (2) that the expressivist cannot explain why these states would be rationally incoherent. The aim of this paper is to show that Baker’s argument for (1) is unsound, and that (1) is unlikely to be true. This obviates the need to explain the alleged rational incoherence, and so Baker’s objection to Schroeder’s expressivism is undermined.

1 Introduction

In his paper ‘Expression and Guidance in Schroeder’s Expressivist Semantics’ (2018) Derek Baker raises an objection to expressivism as it has been developed by Mark Schroeder in his book ‘Being For’ (2008). The objection is based on a particular argument. Baker argues that if the view Schroeder developed on behalf of expressivism is true, then the following sentence “must be expressing inconsistent psychological commitments” or an “inconsistent state of mind” (2018, p. 838):

• (S) ‘Murder is wrong, but not blaming for murder makes sense’.

So Baker’s argument is supposed to establish the claim (henceforth: Baker’s claim) that the mental state expressed by (S) is, as I prefer to say, rationally incoherent,¹ and his objection is based on that claim.

Baker’s objection to Schroeder’s expressivism is not that the above sentence (S) intuitively does not express a rationally incoherent mental state.² Rather Baker accepts the incoherence and objects that Schroeder’s expressivism is unable to explain the rational incoherence of the state expressed by (S), at least without raising serious problems elsewhere in the expressivist theory.

The aim of this paper is to show that Baker’s argument is unsound, and so does not establish Baker’s claim about the incoherence of the state expressed by (S). Moreover, I shall argue that even more promising arguments for that claim also fail and so it is unlikely that the state expressed by (S) is rationally incoherent. But if Baker’s claim is false, this undermines his objection to expressivism: one cannot object to expressivism that it is unable to explain why a state is rationally incoherent, when in fact it is not rationally incoherent.

2 Baker’s Objection and Schroeder’s Expressivism

Let us start by taking a closer look at Baker’s objection and Schroeder’s expressivism. On Baker’s understanding, (S) is a conjunction of two atomic first-order language sentences. The first conjunct, ‘murder is wrong’, contains the normative predicate ‘wrong’ applied to ‘murder’. The second conjunct, ‘not blaming for murder makes sense’, contains the normative predicate ‘makes sense’ applied to ‘not blaming for murder’.³ Baker primarily, though not exclusively, uses ‘makes sense’ in his paper as the normative predicate in the second conjunct, but he suggests that it could be replaced by ‘is the thing to do’, ‘is rational’, ‘is reasonable’, ‘is fitting’, ‘is appropriate’, or ‘one ought to’.⁴

The heart of Schroeder’s metaethical expressivism is a formal semantics for a first-order language containing moral or normative predicates like ‘wrong’. This expressivist semantics provides a recursive mapping from every sentence of the language to the mental state it expresses. Schroeder stipulates that in his particular form of expressivist semantics, all sentences express noncognitive states of being for, even non-normative sentences, since he believes that otherwise expressivists will not be able to provide a constructive solution to the Frege–Geach problem.⁵ Being for is supposed to be a noncognitive, desire-like, motivational state that takes complex mental properties as its content. For atomic sentences, Schroeder offers the following semantic clause:

• Atom For every predicate F and singular term t, sentences of the form F(t) express for$(\psi (t))$, where $\psi (t)$ is a complex property one is for having, namely the property of standing in the $\psi$-relation to t.⁶

Schroeder suggests that ‘t is wrong’ expresses being for blaming t, or slightly more formally: for(blaming(t)). The sentence ‘murdering is wrong’ then expresses the state of being for blaming murdering. Since, as a quasi-realist, Schroeder’s expressivist will hold that to believe what a sentence says is to be in the state expressed by it, it follows that to believe that murder is wrong is to be for blaming murder, and likewise for all other beliefs.

Another important stipulation of Schroeder’s theory is that being for is capable of what he calls A-type inconsistency: A set of mental states is A-type inconsistent iff the states are all of the same genus (e. g. belief, intention, or being for...) and the set of states is rationally incoherent if their contents are (logically) inconsistent. It follows, that a set of states of being for is rationally incoherent if their contents are (logically) inconsistent. This implies that for($\alpha$) is rationally incoherent with for($\lnot \alpha$), where $\alpha$ is a schematic letter standing for an arbitrary metalanguage formula describing a property one is for having. It is natural to reformulate A-type inconsistency as a requirement of rationality: rationality requires that one is not being for (logically) inconsistent contents (cf. No Contradictory Being Fors below). It follows from my definition of rational incoherence as violation of a rationality requirement that being for inconsistent contents is one way of being rationally incoherent, but it need not be the only way.⁷

According to Baker, Schroeder’s semantic clause for atomic sentences (together with his clause for conjunction) determines that (S) expresses the following state:

• (U) for$\big ($blaming(murder) and $\psi$($\lnot$blaming(murder))$\big )$.

Baker here uses $\psi$ as a placeholder, because Schroeder’s semantics does not include a lexical entry for the predicate ‘makes sense’.

Now, if Baker’s claim is correct, which says that the state that is expressed by (S) must be an incoherent or inconsistent state of mind, and Schroeder’s theory says that (U) is the state expressed by (S), then (U) must be an inconsistent state of mind. This raises the question: “does Schroeder’s theory predict an inconsistent state of mind?” (Baker 2018, p. 838).

The answer to this will obviously depend on what $\psi$ is, and the largest part of Baker’s paper is concerned with the question of whether ‘$\psi$’ can be replaced with something that allows Schroeder to predict the rational incoherence of (U). In fact, Baker thinks there are ways to predict the incoherence (for instance, if ‘$\psi$’ is replaced with ‘being for’). But he argues that every replacement that allows Schroeder to predict the incoherence creates problems elsewhere in the expressivist theory (e. g. having to do with the inexpressibility of some states). So his objection is that, ultimately, expressivists are unable to explain the incoherence.

I think that expressivists could say some things in reply to Baker’s various replacements of ‘$\psi$’. But it is unnecessary that I go into the downstream details of Baker’s objection here, since I want to suggest that Baker’s claim about incoherence is false anyway. I do so by showing that Baker’s argument for the claim is unsound (Sect. 3), and that better arguments for it can and should be resisted by Schroeder’s expressivist (Sect. 4). This undermines Baker’s objection to expressivism, because if there is no incoherence to be explained in the first place, then it cannot be objected that expressivism is unable to explain said incoherence.⁸

3 Baker’s Argument for Incoherence

Why does Baker think that the sentence ‘murder is wrong, but not blaming for murder makes sense’ must express a rationally incoherent mental state? He offers the following argument:

the agent who utters the above sentence [(S)] is expressing a commitment [...] to blame for murder, while simultaneously the same agent judges it makes sense not to blame for murder—and since ‘makes sense’ is normative, she must also be expressing a commitment not to blame for murder. So it seems like we should want to say that ‘Murder is wrong but not blaming for murder makes sense’ is semantically incompetent or she is irrational—she’s committing to doing something and simultaneously committing to not doing it. (Baker 2018, p. 837, italics original)

Here is another version of the argument:

We can put the point this way: either the agent does not blame for murder, in which case she fails to have an attitude she is committed to having; or she blames, in which case her emotions don’t conform with her own better judgment, her judgment about which reaction would make the most sense. So she is guilty of inconsistency in her psychological commitments. (Baker 2018, p. 838)

This argument relies on two principles. The first is:

it seems like an implicit commitment about the nature of the being-for attitude is that failing to have the target attitudes that one is for having is a failing of rational coherence. Presumably it was always part of the story that an agent who is for blaming for murder but does not in fact blame for murder is guilty of irrationality (all else being equal). (Baker 2018, p. 843)

I shall reformulate this as a requirement of rationality, and call it:

• Forcia: Rationality requires: if one is for($\alpha$), then one is $\alpha$.

The argument also relies on a second principle which is:

[There] is the widespread intuition that an agent who sincerely judges that something is the thing to do will be motivated to do it, barring weakness of will or other forms of defective agency. Someone who judges that some feeling is the one that makes sense suffers from a case of irrationality (of recalcitrant emotion) if they do not in fact feel it. (Baker 2018, p. 837)

Applied to his second conjunct he says:

If I judge ‘Not blaming for murder makes sense,’ what this judgment should intuitively regulate is my attitude of blame (or my lack of such an attitude). If I continue to blame for murder, I am acting, by my lights, irrationally. ... [that judgment] does rule out the rational coherence of continuing to blame for murder. (Baker 2018, p. 839)

We can also reformulate this as a requirement of rationality, and call it, following Broome (2013):

• Direct Enkrasia: Rationality requires that if one believes one ought to $\lnot \alpha$, then one is $\lnot \alpha$.

Two points on my reformulation of Baker. First, Direct Enkrasia is formulated in terms of ‘ought’, instead of ‘makes sense’ or ‘is the thing to do’, because this is the way Enkrasia (see below) is standardly formulated. This should be fine for Baker, since he explicitly “treat[s] ‘One ought to x’ as equivalent to ‘x-ing makes sense’ or ‘x-ing is reasonable’.” (Baker 2018, p. 831). Second, in the quote above Baker says that someone who jugdes that something is the thing to do will be motivated to do it. My reformulation does not mention motivation because, if instead of Direct Enkrasia we interpret Baker as saying that it is rationally required that if one believes one ought to $\lnot \alpha$, then one is motivated to $\lnot \alpha$, then Baker’s argument will become invalid.

So here is Baker’s argument in a more explicit form:

• Forcia: Rationality requires that if one is being for $\alpha$, then one is $\alpha$.

• Direct Enkrasia: Rationality requires that if one believes one ought to $\lnot \alpha$, then one is $\lnot \alpha$.

• Corollary: one necessarily violates a requirement of rationality (i. e. one is rationally incoherent) if one is for $\alpha$ and believes one ought to $\lnot \alpha$.

• Belief as Being For: believing that t is wrong is being for blaming for t.

• Baker’s Claim: Hence, one necessarily violates some requirement of rationality if one believes that t is wrong and believes that one ought not blame for t. (From Corollary, Belief as Being For, and substitution of $\alpha$ with ‘blaming for t’)

This argument is valid. It is impossible to satisfy both Forcia and Direct Enkrasia, if one believes one ought to $\lnot \alpha$ while also being for $\alpha$, which means that one cannot be in both states and be fully rationally coherent. By replacing $\alpha$ with ‘blaming for t’, we get that it is rationally incoherent to believe one ought not to blame for t while also being for blaming for t. Since, according to Schroeder’s noncognitivist ‘analysis of belief’ (Belief as Being For), being for blaming for t is believing that t is wrong, it follows that the belief that t is wrong is rationally incoherent with the belief that one ought not blame for t. Hence, Baker’s argument entails his claim about incoherence.

Even though Baker’s argument is valid, it is not sound. Contrary to what Baker assumes, Schroeder’s expressivist does not, need not, and must not accept either Forcia or Direct Enkrasia. I shall argue for each of these claims in turn.

Schroeder’s expressivist does not assume Forcia, at least not explicitly. Schroeder nowhere in his book or his papers⁹ assumes “that an agent who is for blaming for murder but does not in fact blame for murder is guilty of irrationality (all else being equal)” (Baker 2018, p. 843). What Schroeder does assume instead is “that being for has the motivational property that someone who is for($\alpha$) will tend to do $\alpha$, all else being equal” (Schroeder 2008, p. 93). This is neither formulated as nor intended to be a requirement of rationality.¹⁰ Since Schroeder’s expressivist does not assume Forcia, she is not committed to accepting Baker’s argument.¹¹

Schroeder’s expressivist does not assume Direct Enkrasia either, at least not explicitly. This again shows that Schroeder’s expressivist, as he currently formulated the view, need not accept Baker’s argument.

Schroeder also does not implicitly assume Forcia or Direct Enkrasia, because his theory does not need these principles. Neither of them plays a theoretical role in his theory. The only principle Schroeder actually needs in order to establish his results especially about logic is A-type inconsistency.

Perhaps Schroeder would have to assume additional requirements besides A-type inconsistency, if he wanted to explain not only logic, but also how it is possible that certain requirements of rationality (e. g. Enkrasia) hold, if normative beliefs really are noncognitive states of being for. But whatever additional requirements expressivists need in order to explain some widely accepted requirements, they must not accept either Forcia or Direct Enkrasia, since they are false, or so I shall argue now.

Direct Enkrasia is false. Rationality does not require that rational agents succeed in doing what they believe they ought to do. Rational agents may be omniscient, but they need not be omnipotent. That one sometimes fails to do what one believes one ought to do, does not make one irrational, because it is sometimes, in fact often, beyond our control to secure that we succeed in doing what we believe we ought to do. This line of argument is rooted in the widely accepted principle that rationality supervenes on the mind.¹²

Direct Enkrasia will be compatible with the supervenience principle, and may in fact sound more plausible, if it is restricted to $\alpha$’s that are not actions or states of the world, but mental states. It has, however, been convincingly argued by John Broome that “we have to reject the Direct Enkratic Condition, even restriced to mental states. Its apparent attraction is spurious, and the counterexamples show it is false” (Broome 2013, p. 96).¹³

Similar points can be made about Forcia. If Forcia is not restricted to mental states, then it violates the supervenience constraint for rational requirements. One cannot be accused of irrationality just because one fails to succeed to do what one is for doing. Bad luck is not the same as bad reason. But even if Forcia is restricted to mental states, it seems false.¹⁴ Toppinen, to whose paper Baker refers in a footnote, argues against ‘Schroeder-style views’ that there is “no appropriate rational connection, at all, between desiring to desire $\varphi$ and the desire to $\varphi$”, and likewise that “it is possible for me to be fully rational, to be for desiring $\varphi$, and yet not to desire to $\varphi$” (Toppinen 2015, p. 160f.; italics original). Moreover, Baker himself should be sceptical about Forcia, given that he says about desire that “it is an open question whether there is any breakdown in rational coherence in failing to have an attitude one desires to have” (Baker 2018, p. 840). I do not see why this should be any different in the case of being for. Hence, Forcia is not plausibly a requirement of rationality.

To summarize: Schroeder’s expressivism does not assume Forcia or Direct Enkrasia. Moreover, Schroeder does not need to assume these principles in order to establish the results of his expressivist semantics and logic. And more importantly, these principles must not play a role in his theory, since they do not appear to be requirements of rationality at all. So I conclude that, since either of the crucial premises of Baker’s argument is false, the argument does not establish the truth of his claim about incoherence.

4 Better Arguments for Incoherence?

In the previous section I showed that Baker’s argument for incoherence is unsound. But showing that an argument is unsound, does not show that its conclusion is false. So might Baker’s claim nevertheless be true? In this section I offer some better arguments for Baker’s claim of incoherence, but show that even these better arguments can and should be resisted by expressivists. Here is another argument:

• No Contradictory Intentions: Rationality requires that if one intends $\alpha$, then one does not intend $\lnot \alpha$.

• Enkrasia: Rationality requires that if one believes one ought $\lnot \alpha$, then one intends $\lnot \alpha$.

• Corollary: One necessarily violates a requirement, if one intends $\alpha$ while believing one ought $\lnot \alpha$.

• Belief as Intention: Believing that t is wrong is intending to blame for t.¹⁵

• Baker’s Claim: it is necessarily rationally incoherent to believe that t is wrong while believing that one ought not blame for t.

This is valid. No Contradictory Intentions and Enkrasia cannot jointly be satisfied, if their antecedents are true, because their consequents are inconsistent (Corollary). What we need in order to derive Baker’s claim from this is the additional noncognitivist premise which says that to hold a moral belief is to intend something (Belief as Intention). This premise is Schroeder’s Belief as Being For, simply with ‘being for’ being replaced by ‘intention’. Baker’s claim then follows straightforwardly from Corollary, Belief as Intention, and by substituting ‘blame for t’ into ‘$\alpha$’.

This argument is better than Baker’s original one, because it relies on the widely accepted requirements No Contradictory Intentions and Enkrasia.¹⁶ I will not debate their truth here, and simply assume that expressivists accept them for the sake of argument. But even when Schroeder’s expressivist accepts these as requirements, the argument has an obvious flaw: Schroeder does not accept Belief as Intention, but Belief as Being For. So, this argument does not apply to Schroeder’s expressivism as it is currently formulated.

So is Schroeder’s expressivism safe from this new argument? One might think that there are two problems, both deriving from the fact that being for and intention share some of their main features.

Here is the first problem. What is important about the being for attitude is that it is a noncognitive, desire-like motivational pro-attitude that is rationally incoherent to hold towards inconsistent contents. But these characteristics seem to be true of intention as well. This suggests that Schroeder could perhaps have formulated his theory in terms of intention, instead of being for. If he did, he would accept Belief as Intention. But then the better argument would apply to his theory and commit him to Baker’s claim about incoherence.

Are there any reasons that speak against replacing being for with intention? I think the fact that identifying moral belief with intention allows one (via the better argument) to derive Baker’s claim (which is intuitively false) is itself a reason not to identify moral belief with intention.

Another reason against the identification with intention is that if the belief that t is wrong were the intention to blame for t, then the belief that one ought to blame for t would require (via Enkrasia) to believe that t is wrong. This does not sound correct to me in general. It seems rationally fine to believe that blaming for t is the thing one ought to do even though one does not believe that t is wrong, for instance, because one thinks t is not wrong.¹⁷

Another consideration that speaks against replacing being for with intention is the following. Many think that moral belief provides motivation that is pro tanto. This means it can be overridden or outweighted by stronger motivation, perhaps provided by some other moral belief, or a belief about what one prudentially ought or ought all things considered to do. Desire, and plausibly being for, also provide pro tanto motivation. In contrast, to desire the motivation that is provided by intentions seems rather pro toto. The motivational force of an intention seems to be the result of weighing the motivational forces of competing desires or other pro-tanto-motivation-providing attitudes. This is also supported by the observation that if one intention is stronger than some other incompatible one, then we will normally drop the weaker one completely, which is less frequently the case with desires: we often stick to our desires even if they are incompatible and of very different strengths. This suggests that moral belief is more desire-like, and hence rather like being for, than like intention.

These considerations suggest that the Schroederian expressivist should not accept Belief as Intention.¹⁸ In any case, Schroeder can escape the better argument by simply sticking to his choice of letting being for be the central attitude in his theory.

This brings us to the second problem mentioned above. If being for and intention are similar in important respects, then perhaps being for is also governed by requirements that are structurally similar to those governing intention. Suppose, for instance, that the following variation of Enkrasia governs the being for attitude:

• Forkrasia: Rationality requires that if one believes one ought $\lnot \alpha$, then one is being for $\lnot \alpha$.

This principle combined with Schroeder’s Belief as Being For and with the following principle (implied by A-type inconsistency)

• No Contradictory Being Fors: Rationality requires that if one is being for $\alpha$, then one is not being for $\lnot \alpha$.

again entails Baker’s claim.

This version of the better argument has two premises that are actually accepted by Schroeder, namely Belief as Being For and No Contradictory Being Fors. However, Forkrasia is not part of Schroeder’s theory, at least not an explicit part, and I do not think anyone has defended this particular principle.

But is Forkrasia implicitly accepted by Schroeder? Perhaps this is so. In a short paragraph of his book Schroeder investigates the idea of explaining the motivational character and inconsistency-transmittingness of intentions on the basis of the motivational character and inconsistency-transmittingness of the attitude of being for. In this context he assumes, though quite hypothetically, that “intending to do A ... involves or commits to being for doing A” (Schroeder 2008, p. 101). Call this assumption: Intention-For. This bridge-principle, however, together with Enkrasia entails Forkrasia. This means that Intention-For, Enkrasia, Belief as Being For, and A-type inconsistency jointly entail Baker’s claim. Should we accept this argument?

Since Schroeder accepts Belief as Being For, and A-type inconsistency, this argument crucially rests on Enkrasia and Intention-For. I have already noted that it is unclear if Schroeder’s expressivist accepts Enkrasia or not, but that I assume it for the sake of argument.

So what about Intention-For, the view that intention involves or commits to being for? Interestingly, there is a very similar thesis accepted by many philosophers of mind, namely the thesis that intention involves desire. Kieran Setiya says that it

is a matter of consensus in the philosophy of intention that intending to do A entails wanting to do A, in the motivational sense for which the ‘primitive sign of wanting is trying to get’ (Anscombe 1963, p. 68). Doubts about this entailment are attributed to ambiguities in ‘desire.’ When I intend to do A reluctantly, from the motive of duty, I may deny that I want to do it, but what I lack is ‘appetite’ not ‘volition’ (Setiya 2018).

Most authors understand the notions of ‘wanting’ or ‘desire’ here rather narrowly, as they figure, for instance, in discussions of the so-called Humean theory of motivation.¹⁹ The entailment thesis is mostly taken to be intuitively true or defended indirectly by showing that apparent counterexamples to it fail. The only direct argument for the entailment I know of is this. If a person has or expresses an intention that she is going to do something, it is always appropriate or legitimate to ask that person “Why do you want to do that?”²⁰ It seems that the question would not always be appropriate, if what it presupposes, namely that one wants something, would not be made true by that person’s having an intention. This suggests that it is impossible to intend something without wanting or desiring it. This is supposed to justify the view that intention entails desire.

In the present context the question is: Does intention also entail being for? Answering this is difficult, because we would have to know more about the being for attitude in order to answer it. For instance, is it a decidedly moral attitude or not? If we think of it as being some sui generis attitude of moral approval, then the entailment does not seem to hold: intending to go swimming does not require that one morally approves of swimming.²¹ If, on the other hand, being for is not a decidedly moral attitude, then it could just be the attitude that goes along with our ordinary notion of being for something. After all, ‘to be for something’ or ‘to be in favor of something’ are familiar English expressions (that contrast with ‘being against’, see OED), and it is natural to think of voting as a way of indicating what we are for.

But this ordinary notion of being for something also does not support Intention-For. Consider the following case. Former Prime Minister Theresa May intended to implement Brexit.²² But even if, in some sense of ‘want’, she wanted or was motivated to implement Brexit (because doing so was her political duty or mandate), it appears incorrect to describe her as being for Brexit. Famously, she voted Remain, and so she was for Britain’s remaining in the EU, not for leaving it. Moreover, if May had on some occasion declared that she intends to implement Brexit (perhaps by saying ‘I shall implement Brexit’), it may have been appropriate to ask why she wants it (she wanted it, because she promised to deliver on the ‘will’ of the British people), but it would have been inappropriate to ask her why she is for Brexit, since everyone knows she was not for it, and perhaps she would even have voted Remain again, if there had been a second referendum.²³

Cases like this one cast doubt on the thesis that intention entails being for. For this reason, Schroeder’s expressivist should not accept it. I conclude that even the better arguments for Baker’s incoherence claim fail. Thereby, Baker’s objection to Schroeder’s expressivism is undermined.

5 Conclusion

The main aim of this paper has been to argue against Baker’s claim that it is rationally incoherent to believe that murdering is wrong while believing that one ought not blame for murdering. I have done so by arguing that Baker’s original argument for the incoherence rests on false premises, and is thus unsound. I showed that there is a better argument available for the incoherence of these beliefs which, however, does not apply to Schroeder’s expressivism as the view is currently formulated, and there are reasons for Schroeder not to reformulate his view in terms of intention or to assume that intention entails being for. I take my arguments to provide inductive evidence for the falsity of Baker’s claim. But if Baker’s claim about incoherence is false, he cannot object that expressivism is unable to explain the incoherence.