The matter of motivating reasons

Abstract

It is now standard in the literature on reasons and rationality to distinguish normative reasons from motivating reasons. Two issues have dominated philosophical theorising concerning the latter: (i) whether we should think of them as certain (nonfactive) psychological states of the agent—the dispute over Psychologism; and (ii) whether we should say that the agent can ϕ for the reason that p only if p—the dispute over Factivism. This paper first introduces a puzzle: these disputes look very much like merely verbal disputes about the meaning of phrases like ‘S’s reason’ in motivating reasons ascriptions, and yet charity requires us to think that something substantive is afoot. But what? The second aim of the paper is to extract substantive theses from certain natural argument for Psychologism and Anti-Factivism—theses which are versions of a Cartesian view of the nature and normative structure of rationality. The paper ends by arguing against these substantive theses on phenomenological and ethical grounds. The upshot is that proponents of Psychologism and Anti-Factivism are either engaged in the project of defending merely verbal theses or they’re engaged in the project of defending false substantive ones.

It is now very common in the literature on reasons and rationality to distinguish between normative and motivating reasons. The former are, to a first approximation, facts which count in favour of the agent performing certain actions or adopting certain attitudes. The latter are, to a first approximation, the reasons for which the agent acts or maintains the relevant attitude.¹ That there is a distinction worth marking here is difficult to question, if only because the fact that p might be a normative reason for one to $\phi$ even if the reason for which one $\phi$s is not that p.²

A debate which has long vexed philosophers is that over Psychologism: the debate over whether our motivating reasons should be identified with certain causally operative psychological states. What, exactly, is at stake in this debate? Isn’t the debate merely over what terms like ‘the reason’ refer to in ascriptions like ‘S $\phi$s for the reason that p’? And why should that be a philosophically interesting debate? The answer isn’t obvious, which is surprising given all the attention this debate has garnered since Davidson (1963) triggered it. In fact, this constitutes a puzzle: surely this debate is substantive, and yet on the face of it it’s merely verbal.

A similar puzzle arises concerning a slightly younger debate which has vexed an overlapping set of philosophers: the debate over Factivism—the debate, roughly speaking, over whether S can $\phi$ for the reason that p even if not-p. Again, isn’t this a merely verbal debate over what phrases like ‘for the reason’ refer to in motivating reasons ascriptions? And why should we find that debate particularly interesting? The answer isn’t easy to determine, and here we find a second puzzle, analogous to the first: surely there must be some substantive issue here, but what?

This paper aims to do three things, the first of which is to motivate the two puzzles just sketched. Section 1 focuses on the debate over Psychologism: I outline the debate; elucidate what it is for a debate to be verbal; and make a prima facie case for thinking that the debate is merely verbal. This raises a puzzle, for charity requires us to think that there is something substantive afoot in this debate. Section 2 motivates the analogous puzzle concerning Factivism.

The paper then attempts to solve these two puzzles by identifying substantive theses in the area. I argue that given certain natural ways of motivating Psychologism and Anti-Factivism, the proponents of those positions are committed to different versions of a substantive view about the normative structure of rationality which I call the Anaemic View. Identifying these substantive theses is the task of Sect. 3.

Finally, Sect. 4 aims to show that the substantive theses with which I associate Psychologism and Anti-Factivism are false, on the grounds that they falsify the deliberative phenomenology and ethical dimension of being a rational agent. Thus, my overall message to proponents of Psychologism and Anti-Factivism is this: your theories can be saved from the charge of being merely verbal, but only at the expense of incurring a commitment to false substantive theses about rationality.

1 Psychologism: a puzzle

In this section I’ll outline the debate over Psychologism (Sect. 1.1). I’ll then specify a framework, largely borrowed from Chalmers (2011), for understanding verbal disputes (Sect. 1.2). Finally, I’ll motivate my puzzle by showing that the debate is prima facie verbal (Sect. 1.3).

1.1 Psychologism versus anti-psychologism

According to Psychologism motivating reasons are to be identified with certain non-factive psychological states the agent is in, which ground their treating p as a normative reason for them to $\phi$ and which cause their $\phi$-ing.³ Typically, in the case of action and practical attitudes the psychological states in question are identified with a suitably related belief/pro-attitude pair. And typically, in the case of doxastic attitudes the relevant states are identified with a set of suitably related beliefs.⁴

According to Anti-Psychologism motivating reasons should be identified instead with what is treated as a normative reason to $\phi$ by the agent—items which are poised to play a certain role in the agent’s deliberation about whether to $\phi$. What is treated as a normative reason to $\phi$ by the agent might in turn get identified with a Fregean proposition: a truth evaluable item individuated at the level of Fregean sense, or alternatively with a state of affairs: an item individuated at the level of Fregean reference to which Fregean propositions are supposed, in some sense, to correspond. In order to avoid taking a stand on this issue I will say that Anti-Psychologism identifies motivating reasons with considerations treated as normative reasons by agents—whatever those are. The relevant psychological states are then treated by the Anti-Psychologist as conditions which enable the consideration to play the role of the agent’s motivating reason, not as constituting that reason.⁵

Suppose Cassidy promises to meet an acquaintance for dinner tomorrow evening. She doesn’t really have the inclination, but she intends to go anyway, because she promised. Cassidy treats the consideration that she promised as a normative reason for her to intend to go. What grounds her doing so, and also what cause her intention, is certain psychological states she’s in: her belief that she promised plus, we can take it, a pro-attitude towards acts of promise-keeping. According to Psychologism, Cassidy’s motivating reason for her intention should be identified with the states of mind just mentioned. According to Anti-Psychologism, it should be identified with the consideration treated by Cassidy as a normative reason: that I promised, where Cassidy’s belief and pro-attitude function merely to enable that consideration to function as Cassidy’s motivating reason.

What gets treated as a normative reason to $\phi$ by the agent is typically some consideration about the way the world is, as in the example just given. The Anti-Psychologist will thus typically identify motivating reasons with worldly considerations, whereas the Psychologist will typically treat them as first-order psychological states directed towards the relevant worldly considerations. But on occasion, what gets treated as a normative reason by the agent is a consideration concerning their own mind. For example, a religious ascetic might treat the consideration that they have a certain sexual desire as a normative reason to engage in self-punitive behaviour, and indeed engage in that behaviour for the reason that they have that sexual desire. Psychologism would say that we should identify the ascetic’s motivating reason with certain second-order states of mind they are in: a belief that they have a certain sexual desire, plus a second-order desire not to have such desires, say. Anti-Psychologism would say that we should identify the ascetic’s motivating reason with the psychological consideration that they have a certain sexual desire.⁶

1.2 What is a verbal dispute?

Is the debate over Psychologism a verbal dispute? We first need to know what it is for a dispute to be verbal. Take an example. Suppose Ted, a Conservative, says: ‘marriage is between a male and a female’ and Matt, a liberal, disagrees by saying: ‘marriage can be between two members of the same sex’. But suppose further that Ted and Matt agree full-well that the law should treat same-sex relationships just the same as heterosexual relationships; that same-sex relationships are in no sense morally inferior to heterosexual ones; that our cultural norms should shift to reflect these moral truths...and so on. In that case, one would have the impression that Ted and Matt aren’t really disagreeing about anything substantial but are merely disagreeing about what the word ‘marriage’ means: Ted thinks it picks out one kind of relation; Matt thinks it picks out another—and that’s it.

We have a rough idea of what a verbal dispute is. Now let’s sharpen it. Here’s Chalmers’s (2011) characterisation:⁷

Intuitively, a dispute between two parties is verbal when the two parties agree on the relevant facts about the domain of concern, and just disagree about the language used to describe that domain. In such a case, one has the sense that the two parties are “not really disagreeing”: that is, they are not really disagreeing about the domain of concern, and are only disagreeing over linguistic matters. (Chalmers 2011, p. 515)

More precisely:

A dispute over S is...verbal when, for some expression T in S, the parties disagree about the meaning of T, and the dispute over S arises wholly in virtue of this disagreement...(Ibid.: 522)

The debate between Ted and Matt fits Chalmers’s characterisation: their dispute arises solely because of a disagreement over the meaning of ‘marriage’.

How are we to go about determining whether a given debate is verbal? Chalmers offers a plausible method for testing whether a dispute is verbal—the ‘Method of Elimination’:

To apply this method to a dispute over a sentence S that is potentially verbal with respect to term T, one proceeds as follows. First: one bars the use of term T. Second: one tries to find a sentence S’ in the newly restricted vocabulary such that the parties disagree non-verbally over S’ and such that the disagreement over S’ is part of the dispute over S...If there is no such S’, then the dispute over S is wholly verbal... (Chalmers 2011, pp. 526–527)

Applied to Ted and Matt, the idea would be this: if we ban the word ‘marriage’ can we re-formulate the disagreement between them so that it comes out as substantive? It looks like we can’t: if the debate is substantive then it would presumably concern some ethical feature possessed by heterosexual relations which is supposedly lacked by same-sex relations. But, banning the word ‘marriage’, we cannot formulate a sentence which captures such an idea, about which Ted and Matt disagree.

1.3 Is the debate over psychologism a verbal dispute?

Is the dispute over Psychologism merely verbal? We ascribe motivating reasons to agents using a range of expressions. Here are some paradigms:

• (1) S $\phi$s for the reason that p

• (2) The reason for which S $\phi$s is that p

• (3) S’s reason for $\phi$-ing is that p.

Psychologists and Anti-Psychologists disagree about what the terms ‘the reason’/‘S’s reason’ mean in these contexts. Psychologists think that such terms refer to certain psychological states of the agent; Anti-Psychologists think that they refer to what the agent treats as a normative reason. But does this debate arise solely because of this linguistic disagreement? If so, the debate is merely verbal.

Let’s apply Chalmers’s Method of Elimination. First, we ban ‘the reason’ and ‘S’s reason’. We now need to find a sentence in which those terms don’t appear but which expresses a substantive thesis about which the two parties disagree. It’s hard to find such a sentence.

The most obvious suggestion is that the parties are in substantive disagreement over the nature of the psychological states the agent counts as being in insofar as (1)–(3) is true of them. On this suggestion, the Psychologist thinks that the truth of (1), for example, involves the agent being in a certain kind of psychological state, whereas the Ant-Psychologist denies this. On the assumption that we could characterise this type of state without using our banned terms, we’d have isolated the desired substantive disagreement.

But this suggestion cannot be right. After all, the following is accepted by both parties:

• ${\textbf{Distinction}}_{\psi}$ Insofar as the agent $\phi$s for the reason that p, we should distinguish between: (i) the psychological states which are causally operative with respect to the agent’s $\phi$-ing and which ground their treating p as a normative reason for them to $\phi$; and (ii) what the agent treats as a normative reason for them to $\phi$, namely: p—something which plays a certain role in their first-person deliberation about whether to $\phi$.

Thus, both Psychologists and Anti-Psychologists agree that when (1) is true, the agent has the following psychological properties: they treat p as a normative reason for them to $\phi$; they are in certain psychological states which ground them treating p in that way; these states cause their $\phi$-ing. But if they agree full-well on that, they agree full-well on what the salient psychological facts are. So there can be no substantive disagreement between them about the nature of the psychological states which underpin (1)–(3).

An alternative suggestion is that Psychologists and Ant-Psychologists disagree on something explanatory. On this suggestion, the Psychologist says that what grounds the truth of (1) is a state of affairs which enables us to explain why the agent $\phi$s by appeal to their non-factive states: S $\phi$s because S believes that p. The Anti-Psychologist, by contrast, would say that the state of affairs which grounds (1) enables us to rationalise S’s $\phi$-ing both in terms their non-factive states and in terms of the fact that p itself: ‘S $\phi$s because p’. Since we’ve specified these explanations without using ‘the reason’ and ‘S’s reason’, we have a candidate substantive disagreement.

But this can’t be right either. For one thing, not all Anti-Psychologists agree that if (1) is true of an agent, we can rationalise their action in terms of the fact that p—Anti-Factivists deny precisely that, as we’ll see. So this can’t be the correct way of characterising the debate between the two parties. Also: it’s not obvious that we have in fact succeeded in characterising a debate without using our banned terms—after all, what makes a rationalising explanation rationalising is that it’s one which specifies the agent’s reasons in $\phi$-ing.

Since these are apparently the only options for reconstructing a substantive debate here, we should conclude that the debate over Psychologism is, apparently, merely verbal: it arises solely because of a disagreement about what ‘the reason’ and ‘S’s reason’ mean in (1)–(3). Such a debate is of little interest to philosophers aiming to theorise about the role of reasons in our mental lives. Those with that interest could simply draw $\hbox {Distinction}_\psi$; introduce technical terms to refer to each term of the distinction—for example, ‘primary reasons’ and ‘deliberative reasons’ to refer to (i) and (ii) respectively; and thence carry out their theoretical aims, making use of the technical vocabulary as necessary, whilst remaining neutral on the linguistic dispute concerning whether our ordinary motivating reasons ascriptions pick out (i) or (ii), or are polysemous.⁸

But this gives rise to a puzzle. It would be very uncharitable to interpret the philosophers who’ve engaged in this long dispute as engaging in a mere back-and-forth about linguistic matters. This means our default view should be that there is some substance to this dispute somewhere: at least one of the parties to it is intending to defend a substantive thesis. But what? I return to this question in Sect. 3.

2 Factivism: a puzzle

I now want to turn my attention to a second debate about motivating reasons: the debate over Factivism. I outline the debate in Sect. 2.1 and then make a case for thinking that it’s merely verbal in Sect. 2.2.

2.1 Factivism versus anti-factivism

The debate over Factivism is premised on the truth of Anti-Psychologism: both the Factivist and Anti-Factivist are essentially committed to the claim that motivating reasons are the considerations one treats as normative reasons for one to $\phi$, typically considerations concerning the way the world is. Factivism combines Anti-Psychologism with the view that S $\phi$s for the reason that p only if p is the case.⁹ Anti-Factivism combines Anti-Psychologism with the view that one can $\phi$ for the reason that p even if not-p.

Here’s an example. My flight to Edinburgh is due to leave at noon and, looking at my watch, I see that I only have fifteen minutes to get to the terminal. I start running, and I do so for the reason that I only have fifteen minutes until my flight leaves. My reason here is identified with the consideration: I only have fifteen minutes until my flight leaves, and not any psychological state(s) I am in. But now suppose I’m mistaken: suppose I’ve misread my watch and, in fact, I have an entire hour until my flight leaves. Do I still count as running to the terminal for the reason that I only have fifteen minutes until my flight leaves? Factivists say no.¹⁰ Anti-Factivists say yes.¹¹

It’ll be helpful to express the difference between the two views in the following way. Let’s call good cases cases in which the agent $\phi$s motivated by a fact they treat as a normative reason for them to $\phi$, and cases of factual error cases in which the agent $\phi$s on the false belief that some consideration obtains, which seems to them to be a normative reason for them to $\phi$. The agent in the latter situation cannot tell, just by reflection, that they are not in a good case, but their belief in their reason is false nevertheless. The Factivist says that S $\phi$s for the reason that p only if they are in the good case. The Anti-Factivist says that S $\phi$s for the reason that p in good cases and cases of factual error.

2.2 Is the debate over factivism a verbal dispute?

Is the dispute over Factivism merely verbal? It’s obvious that there are semantic disagreements between the Factivist and Anti-Factivist. Consider again:

• (1) S $\phi$s for the reason that p

• (2) The reason for which S $\phi$s is that p

• (3) S’s reason for $\phi$-ing is that p.

The Anti-Factivist will want to say that each of (1)–(3) can be true even if p is false and the Factivist denies this. The Factivist might want to say, for example, that ‘for the reason’ in (1) or ‘the reason for which’ in (2) or ‘S’s reason’ in (3) are phrases which are factive: it’s built-into their meaning that the sentences in which they’re embedded require p to be true. The Anti-Factivist will want to deny all of that. Our question is: does the dispute over Factivism arise merely because of this linguistic dispute? If so, the debate is verbal.

Let’s apply Chalmers’s Method of Elimination. First, we ban the phrases ‘for the reason’, ‘the reason for which’, and ‘S’s reason’. Next, we go on the search for a sentence expressing a substantive thesis, in which those phrases don’t appear, and about which our two parties disagree. Again, it’s hard to find such a sentence.

It seems that if there is a substantive disagreement in this area, that’ll be because the Anti-Factivist wants to say that there’s some philosophically interesting condition which obtains across both good cases and cases of factual error, whereas the Factivist wants to say instead that this condition obtains only in good cases. If we could characterise this condition without using our banned phrases, we’d have got what we want.

But this suggestion can’t be right because the following is accepted by everyone:

• ${\textbf{Distinction}}_{f}$ We should distinguish between (i) there being a fact, treated by the agent as a normative reason to $\phi$, which the agent $\phi$s in the light of; and (ii) S $\phi$-ing in a way that manifests a belief in something they treat as a normative reason for them to $\phi$. (i) obtains only in the good case; (ii) obtains across both good cases and cases of factual error.

Both Factivists and Anti-Factivists agree that only when the agent is in the good case do they count as $\phi$-ing in the light of a fact. And both agree that across both good cases and cases of factual error, they $\phi$ because of what appears to them to be a normative reason for them to $\phi$. But if both agree about this, they both agree on what conditions obtain just in good cases and what are neutral across good cases and cases of factual error. And if that’s so, it appears that the only disagreement left between them is the mere verbal disagreement about the truth-conditions of (1)–(3).

Again, such a verbal debate appears to be of little interest to philosophers aiming to theorise about the role of reasons in our mental lives. That’s because philosophers with that interest could simply draw $\hbox {Distinction}_f$; introduce technical terms to refer to each term of the distinction—for example: ‘being motivated by a fact’ and ‘being motivated by a belief’ for (i) and (ii) respectively; and thence carry out their theoretical aims, making use of the technical vocabulary as necessary, whilst remaining neutral on the linguistic dispute concerning whether our ordinary motivating reasons ascriptions pick out (i) or (ii), or are polysemous.¹²

But this gives rise to a second puzzle, analogous to the first: surely there is something philosophically important afoot here—it’d be very uncharitable to read the relevant philosophers as merely engaged in debate about the meanings of certain English words. Our default assumption should therefore be that there is at least some substantive thesis in the vicinity of at least one of the parties to the debate. But what is it? It takes work to answer this, to which I now turn.

3 The anaemic view of rationality

We have two puzzles on our hands. The debates over Psychologism and Factivism look very much like merely verbal disputes, but our default assumption should be that there is at least some substantive thesis in the vicinity. This section solves the puzzles by identifying certain substantive theses about the normative structure of rationality—different versions of the Anaemic View of rationality—in the vicinity of Psychologism and Anti-Factivism. Section 3.1 explains the Anaemic View; Sect. 3.2 distinguishes two version of it; Sect. 3.3 justifies the claims that Psychologism can be associated with one version, Anti-Factivism the other.

3.1 What the Anaemic View Is

We judge agents rational and irrational—these are certain kinds of positive and negative appraisal. Following Comesaña and McGrath (2014, p. 61), we can distinguish ex ante rationality from ex post rationality. $\phi$-ing is ex ante rational for an agent just if $\phi$-ing is a rational thing for them to do, whether or not they actually do it. $\phi$-ing is ex post rational for an agent just if they $\phi$ in a way that warrants praise as rational. This distinction is analogous to the epistemologist’s distinction between propositional and doxastic justification. Our focus in the remainder is ex post rationality—this is what ‘rationality’ will refer to.

The Anaemic View says that $\phi$-ing rationally consists in $\phi$-ing correctly in response to an ought of rationality, but where this status, in turn, is to be thought of as supervening on one’s non-factive psychological states. We are to conceive rationality as a status which is insulated from one’s external environment. But we’re also to think of it as a matter of $\phi$-ing motivated by a rational ought. So: there’s a way of successfully responding to a certain kind of obligation one is under which is available to one no matter what one’s environment is like, and it’s this sort of response which is constitutive of rationality properly so-called.

More precisely, the Anaemic View is the conjunction of:

• Normativive-Guidance Necessarily, S $\phi$s rationally iff (i) S rationally ought to $\phi$; and (ii) S’s $\phi$-ing is a response to what makes it the case that they rationally ought to $\phi$.¹³

• Psychological-Narrowness Necessarily, for any two individuals, S and S*, and for any type of $\phi$-ing both S and S* are engaged in: if S and S* are in the same non-factive psychological states then S and S* are exactly alike with respect to whether they $\phi$ rationally.

Normative-Guidance expresses a kind of deontic view of the nature of rationality. It says that when one $\phi$s rationally, one ought to $\phi$, from the rational point of view. One’s being subject to this ought will be grounded in certain factors, and we are to conceive of $\phi$-ing rationally as $\phi$-ing guided or directed or motivated by these factors. These terms are vague. But they should serve to provide at least a preliminary indication of what the proponent of Normative-Guidance has in mind: the intuitive idea is that rationally $\phi$-ing involves responding correctly to factors that rationally bind one to $\phi$-ing. According to Normative-Guidance, then, $\phi$-ing rationality is a kind of normative achievement: it’s a matter of $\phi$-ing as one ought, where that is a status for which one deserves credit.

Psychological-Narrowness is simply the thesis that rationality is a narrowly supervening status: pairs of individuals who are both $\phi$-ing and who are internal psychological duplicates must also be exactly alike with respect to whether they count as $\phi$-ing rationally. No rational difference without a difference in non-factive states of mind.

Normative-Guidance and Psychological-Narrowness are logically separable: neither entails the other. But if we put them together, the following picture emerges. By Psychological-Narrowness, rationality is a status one enjoys just because of the way things appear to one. But by Normative-Guidance that status consists in responding correctly to an ought of rationality which applies to one. It follows that rational oughts are oughts which apply to one regardless of how one’s environment is, and one is in a position to successfully respond to them regardless of how one’s environment is, too.

I call the view the Anaemic View because it is a way of capturing Davidson’s well-known remark that: ‘...there is a certain irreducible—though somewhat anaemic— sense in which every rationalization justifies...’ (Davidson 1963, p. 9).

3.2 Two versions of the anaemic view

The Anaemic View is a highly abstract thesis: it just says that rationality is a matter of responding correctly to oughts of rationality and that this is an environment-independent status. In order to get a more concrete grip of what the view amounts to, we’d need to know how oughts of rationality are to be conceived by the proponent of the view. Here I’ll distinguish two answers to that question, corresponding to two different versions of the Anaemic View.

According to Structuralism:

• Structuralism about Rationality The Anaemic View is true and oughts of rationality are to be identified with so-called structural requirements of rationality.

Structural requirements of rationality place restrictions on what combinations of attitudes agents are to have. In particular, they function to ban sets of attitudes which are jointly incoherent. They are most naturally stated in the form of material conditionals which state that rationality requires the agent to hold some attitude (the consequent attitude), if they hold some set of further attitudes (the antecedent attitudes). Here are two examples of rational requirements for the practical and epistemic domains respectively:

• Instrumental Requirement Rationality requires S to intend to $\psi$, if she intends to $\phi$ and believes $\psi$-ing to be necessary for $\phi$-ing;

• Modus Ponens Requirement Rationality requires S to believe that q, if she believes that p and believes that p $\rightarrow$ q.

The Structuralist says that we should elaborate the Anaemic View in terms of structural requirements of rationality. On this view, when one rationally $\phi$s one is in certain non-factive psychological states which ground a structural requirement of rationality that applies to one—those states being the antecedent attitudes which appear in the structural requirement.¹⁴ Rationality is then thought to consist in successfully responding to that structural requirement by being guided or motivated by the psychological states which ground its application to one. These structural requirements apply to one just by dint of the way things appear to one: they apply to one independently of the way one’s environment is, and one’s being able to successfully $\phi$ guided by them is also an ability one enjoys independently of one’s environment. Rationality is all about ensuring coherence amongst one’s attitudes and actions, and the power to ensure that is something one cannot lose just by dint of changes to one’s environment.

A different version of the Anaemic View is Subjectivism:

• Subjectivism about Rationality The Anaemic View is true and oughts of rationality are to be identified with subjective oughts.

According to Subjectivism we’re to combine the Anaemic View with the thought that the oughts of rationality are so-called subjective oughts. On this view, we should distinguish between two kinds of ought: the objective ought, which is determined by facts about one’s situation, and the subjective ought, which is determined by what one believes about one’s situation. These are two fundamentally different kinds of normativity. On the one hand there is the familiar kind of normativity associated with normative reasons—oughts which apply to one by dint of objective features of one’s situation. On the other hand there is a subjective surrogate for that kind of normativity—oughts which apply to one by dint of facts about what one believes about the situation, not by dint of the way the situation is, objectively speaking.¹⁵ The two kinds of ought are independent, but related: to a first approximation, the belief that p counts as grounding a subjective ought that applies to the agent just when, and indeed because, it satisfies the following counterfactual: were p true, it would be the case that S objectively ought to $\phi$.¹⁶ Rationality, as opposed to morality or prudence, say, concerns itself only with the latter sort of ought.

Subjectivism has it that when one rationally $\phi$s, one has certain beliefs about one’s environment which make it the case that one subjectively ought to $\phi$. Rationality is then to be thought of as consisting in successfully responding to such a subjective ought by being guided by what one believes. Rationality is all about ensuring that the subjective surrogates of bone fide normative reasons are successfully responded to, and one’s ability to do so is one which cannot be lost just by changes to one’s environment.

3.3 Our two puzzles solved

Let’s now return to the debates over Psychologism and Factivism. I argued earlier that these debates look merely verbal. We’re now in a position to tease out the substantive issues associated with each. In short, there’s a case for associating Psychologism with Structuralism and a comparative case for associating Anti-Factivism with Subjectivism. Since these are substantive theses about the normative structure of rationality, there are substantive disputes in these areas, after all. Our two initial puzzles are thus solved.

Let’s start with Psychologism: the view that one’s motivating reasons are the psychological states one is in which ground one’s treating p as a reason to $\phi$ and which cause one’s $\phi$-ing. Here’s a question I didn’t ask earlier: why should we endorse Psychologism? How should it be argued for? One natural strategy is to appeal to the intuitive idea that agents in factual error cases count as $\phi$-ing rationally. But if an agent $\phi$s rationally, they must $\phi$ for a reason. So, agents in factual error cases count as $\phi$-ing for a reason. But what reason could that be? Whatever the agent’s reasons are, they must be things in terms of which we can provide a rationalising explanation of why they $\phi$ and we can’t appeal to what the agent believes in the factual error case to do that: after all, what they believe is false. The only other option is to identify the agent’s reason in the factual error case with the psychological states that cause their $\phi$-ing and which ground their treating p as a normative reason for them to $\phi$. This gets us the result that Psychologism holds in the factual error case. The further thought is then going to be that whatever holds of the factual error case also holds of the good case: what motivates the agent doesn’t shift across the two cases. And this takes us to the truth of Psychologism in general.

An analogous argument is available for Anti-Factivism: the view that motivating reasons are what one believes about one’s situation, but that one can $\phi$ motivated by the reason that p even if not-p. This argument proceeds to the claim that the agent in the factual error case $\phi$s for a reason just as before: via the idea that they $\phi$ rationally and that this involves $\phi$-ing for a reason. The question again arises: what could that reason be? At this point, the Anti-Factivist will deny that motivating reasons are the things in terms of which we can provide rationalising explanations of why the agent $\phi$s. They’ll instead say that motivating reasons play a deliberative role: they are the things which the agent treats as favouring their $\phi$-ing in episodes of deliberation which lead to the conclusion to $\phi$. The agent’s motivating reason in the factual error case is therefore to be identified with what the agent (erroneously) believes about their situation—not with any psychological state they are in. But since all this holds mutatis mutandis for the good case, it must be that in general the agent’s motivating reasons are to be identified with what they believe about their situation, where such things can play the role of motivating the agent whether or not they are true—and this is just Anti-Factivism.¹⁷

These arguments have a common structure. They each start by showing that the agent in the factual error case $\phi$s for a reason. They then appeal to different principles about what motivating reasons are (an explanatory principle in the case of Psychologism; a deliberative principle in the case of Anti-Factivism) to get us the result that the favoured theory is true of the factual error case. There is then a ‘spreading step’ which enables us to generalise that conclusion to good cases.

I now want to show that if the proponents of Psychologism and Anti-Factivism motivate their positions in these ways, they’ll have to commit to Structuralism and Subjectivism, respectively. Since these are substantive theses about rationality, if this suggestion is right we’ll have succeeded in isolating substantive theses in the vicinity of the two disputes we started out with after all.

Each argument shares a key premise: if the agent $\phi$s rationally, they $\phi$ for a reason. Why should we think this? Why not say, instead, that if S $\phi$s rationally then S $\phi$s in a way that is positively evaluable—for example, that S’s $\phi$-ing manifests the virtue of rationality,¹⁸ or that S’s $\phi$-ing manifests their functioning properly as a rational agent?¹⁹ Each of these evaluative conceptions of rationality is consistent with denying that rationality involves being motivated by a reason, it seems. Opponents of the two ‘arguments from factual error’ could, it seems, perfectly well say that the agent in the factual error case manifests proper functioning as a rational agent; manifests the virtue of rationality; and that they $\phi$ in response to a merely apparent reason, whilst denying that the agent counts as $\phi$-ing for a reason simpliciter. What would be incoherent about that suggestion?

In order to resist this move, the proponent of the idea that $\phi$-ing rationally involves $\phi$-ing for a reason is going to have to reject the evaluative conception of rationality just broached. But if they deny an evaluative view, how are they to account for the positive status they want to ascribe to the agent in the factual error case and which they label ‘rationality’? The only further option is to view rationality along the lines of normative-guidance: they’ll have to say that rationality consists in $\phi$-ing guided by a factor which makes it the case that one rationally ought to $\phi$. They’ll then account for what’s going on in the factual error case by saying that the agent in it is subject to an ought of rationality, and they count as $\phi$-ing rationally because they count as correctly $\phi$-ing in response to that ought.

But Normative-Guidance only gets us to the idea that $\phi$-ing rationally involves $\phi$-ing for a reason if the following principle is true:

• MR-Ought Necessarily, if S $\phi$s rationally and this consists in their $\phi$-ing in response to a factor which makes it the case that they rationally ought to $\phi$, then S $\phi$s motivated by a reason which is to be identified with that factor.

According to MR-Ought, if we’re to model rationality in terms of Normative-Guidance, we’ll have to say that rational agents $\phi$ for reasons, and these are just the things which ground the ought to which the agent counts as successfully responding. Only if one commits to this idea in conjunction with Normative-Guidance will one be able to defend the premise that $\phi$-ing rationally requires $\phi$-ing for a reason, against the attack from the evaluative view of rationality described above.

What this means is that the proponents of Psychologism and Anti-Factivism who motivate their positions in the above ways will have to commit themselves to Normative-Guidance and to MR-Ought. But if they do this, they end up committed in turn to one of the versions of the Anaemic View described above: the Psychologist ends up committed to Structuralism, the Anti-Factivist to Subjectivism. Let me explain.

Consider first Psychologism. What happens when we combine Psychologism, Normative-Guidance, and MR-Ought, as any proponent of Psychologism who motivates their position by the above argument from factual error will have to? What we get is the following claim about rationality: when the agent $\phi$s rationally, this consists in them $\phi$-ing in response to a factor which grounds an ought of rationality which applies to them, and these factors are certain non-factive psychological states of the agent. On this view, rational oughts are grounded in what non-facive psychological states the agent is in, and rationality consists in responding correctly to these oughts, guided by the non-factive psychological states at issue. What sorts of rational oughts are grounded in the non-factive psychological states of the agent? The only option here is to say that they are oughts of structural rationality. So if the Psychologist motivates their position by appeal to the above argument, they end up committed to Structuralism.

Notice that since the Anaemic View and Structuralism were characterised without using the phrases ‘the reason’ and ‘S’s reason’, we’ve succeeded in isolating a sentence—‘The Anaemic View in its Structuralist Version is true’—which we have a grip on independently of the verbal disagreement between Psychologists and Anti-Psychologists and which is surely substantive. Since I’ve shown that every Psychologist who motivates their position by appeal to the above argument from factual error is committed to the truth of that sentence, I’ve therefore shown that there is a bona fide substantive thesis in the vicinity of Psychologism: Structuralism. This solves the first puzzle.

Let me know try to show that the Anti-Factivist who motivates their position by appeal to the analogous argument from the factual error thereby commits themselves to the other version of the Anaemic View: Subjectivism. What happens when we combine Anti-Factivism, Normative-Guidance, and MR-Ought, as any proponent of Anti-Factivism who motivates their position via the argument from factual error needs to? We get the following conception of rationality: rationally $\phi$-ing consists in $\phi$-ing in response to factors which make it the case that one rationally ought to $\phi$, and these factors are to be identified with what the agent believes about their situation, as opposed to any psychological state they are in. On this view, rationality consists in $\phi$-ing guided by the way things appear from one’s perspective, so that one ends up $\phi$-ing in accordance with an ought that applies to one just by dint of things appearing to one in that way. But what sorts of rational oughts are grounded in what one believes about the situation? The only plausible answer here is: subjective oughts. So if the Anti-Factivist motivates their position by appeal to the above argument, they end up committed to Subjectivism.

Again, the Anaemic View and Subjectivism were characterised without using the phrases ‘for the reason’, ‘the reason for which’, and ‘S’s reason’. So we’ve succeeded in isolating a sentence—‘The Anaemic View in its Subjectivist Version is true’—which we grasp independently of the verbal disagreement between Factivists and Anti-Factivists and which is surely substantive. I’ve shown that every Anti-Factivist who motivates their position by appeal to the above argument from factual error is committed to the truth of that sentence. So I’ve therefore shown that there is a bona fide substantive thesis in the vicinity Anti-Factivism: Subjectivism. This solves the second puzzle.

4 The perspective of rationality

I’ve now isolated two substantive theses in the vicinity of Psychologism and Anti-Factivism. Psychologists, at least who motivate their position by appeal to the argument from factual error, are committed to Structuralism about Rationality. Likewise, Anti-Factivists who motivate their position in the analogous way are committed to Subjectivism about Rationality.

I want to finish the paper by critically assessing these substantive positions about the normative structure of rationality. In particular, I want to demonstrate that these views falsify the phenomenology of rationality—they imply that the agent’s own perspective on their $\phi$-ing, when they $\phi$ rationally, is in error. They also imply that rational agency intrinsically involves something like Williams’s (1976) one-thought-too-many syndrome. The overall upshot is that Psychologism and Anti-Factivism can be saved from the charge of being merely verbal positions, but only at the cost of being committed to false substantive ones.

In Sect. 4.1 I argue for a phenomenological constraint on $\phi$-ing rationally. In Sect. 4.2 I argue that neither Structuralism nor Subjectivism meet this phenomenological constraint and raise the Williamsian worry, too. Finally, in Sect. 4.3 I offer, as a coda to the paper, a brief critical discussion of how the availability of a certain kind of disjunctive view of rationality might be thought to undermine the argument of the paper.

4.1 Normative-guidance, rational agency, and the first-person perspective

Earlier we encountered Normative-Guidance, a component of the Anaemic View:

• Normativive-Guidance Necessarily, S $\phi$s rationally iff (i) S rationally ought to $\phi$; and (ii) S’s $\phi$-ing is a response to what makes it the case that they ought to $\phi$.

I left it hanging what it is to $\phi$ in response to what makes it the case that one ought to $\phi$. I now want to argue that if Normative-Guidance is to be plausible, we will have to introduce the following constraint on our understanding of this in response to relation:

• Perspectival Constraint Necessarily, S $\phi$s in response to an ought-making feature, in the sense of ‘in response to’ pertinent to Normative-Guidance, only if the ought-making feature figures as such from S’s first-person perspective.

According to the Perspectival Constraint, whatever grounds the normatively significant factor which moves the agent to their rational response must be something which is an aspect of the agent’s perspective on the world. Moreover, the factor must appear as grounding an ought of rationality from that point of view. Let me clarify.

To say that the factor in question must be present from the agent’s own point of view involves saying that the factor is the object of some intentional state the agent is in, where the intentional state is of a kind which, to a first approximation, constitutes the presentation of the factor to the agent. This latter caveat is intended to rule it out that the relevant mental state is one with a world-to-mind direction of fit: it is intended to rule it out that the state is a desire, wish, hope, emotion, or the like. But that leaves it open what kind, or kinds, of states with the opposite direction of fit can play the role laid down by the Perspectival Constraint, and there are competing views it is possible to have about this. For example, one may wish to identify the relevant sort of intentional state with a belief or knowledge of some truth which is, or correctly represents, the normatively significant factor in question. More liberally, one might wish to leave it open that sensory experiences can play the role of constituting the right sort of presence to mind.²⁰ We needn’t take a stand on this issue: I will think of the Perspectival Constraint neutrally in terms of the idea that responding to an ought-making feature requires one to be aware of it.

The Perspectival Constraint doesn’t just require the rational agent’s awareness of the rationally significant factor. It also requires that the factor must be present as rationally significant to the agent—that the agent in some way registers or manifests sensitivity to the ought-grounding status of the factor in question. Again, there are many ways of cashing this out, some of which are less plausible than others. For example, one might wish to cash out this condition in doxastic terms: as the view that when the agent $\phi$s rationally, they believe or know that the factor in question makes it the case that they ought to $\phi$. Alternatively, and in order to avoid over-intellectualising rationality, one might wish to cash out the condition in dispositionalist terms: as the view that when the agent $\phi$s rationally, this must manifest them correctly treating the factor they’re aware of as a factor which grounds a rational ought which applies to them. This treating condition will then be further identified with having a complex of dispositions towards the factor in question, such as the disposition to use the factor as a premise in reasoning. And, in the practical case at least, these dispositions might in turn be thought grounded in certain non-cognitive states of the agent. Again, we needn’t take a stand on this issue: I will think of the Perspectival Constrain neutrally in terms of the idea that responding to an rational ought-making feature requires the feature to be registered as ought-making.²¹

The Perspectival Constraint requires the ought-grounding factors to be present as such from the agent’s own point of view. This rules out brute externalist forms of rationality, which would classify agents who have no view about whether there is anything to be said for their $\phi$-ing as agents who $\phi$ rationally. For example, the Perspectival Constraint would rule out BonJour’s (1980) Norman, at least in BonJour’s fourth version of his case, as believing rationally.

Why think the Perspectival Constraint plausible? I myself find the verdicts the Perspectival Constraint yields concerning brute externalist cases intuitive. But it’d be better not to rest content with that thought, for it is unlikely to convince the opponent of the Constraint—either they simply won’t share the intuition or they would argue that their alternative is equally intuitive.²² What else can we say?

According to a venerable philosophical tradition, the exercise of rationality on the subject’s part is essentially linked to their capacity for agency. In particular: if an agent $\phi$s rationally, they must have the ability to engage in certain mental acts of conscious reasoning, where these acts constitute them taking control of whether they $\phi$, and in response to what ought-making factor they do so (if any). I think this Agential View of rationality is plausible.²³ But if it’s plausibility is to be brought into focus, two clarifications should be made.

First, we should distinguish between having the general ability to $\phi$ from having the option to $\phi$.²⁴ Consider an agent who speaks German and contrast them with a monolingual English speaker. There is a sense in which the former is able to speak German and the latter isn’t: the former has the general ability to speak German but the latter doesn’t. But now contrast the German speaker with another German speaker who has had their lips glued together in a practical joke. Although both have the general ability to speak German, there remains a sense in which the former, but not the latter, is able to speak German. We can say that the former has the option to speak German whereas the latter doesn’t. Intuitively: having the general ability to $\phi$ in a certain context doesn’t require being able to successfully exercise this ability there-and-then, whereas having the option to $\phi$ does. The Agential View of rationality should be understood as saying that an agent $\phi$s rationally only if they have the general ability to engage in certain acts of conscious reasoning. The Agential View therefore allows that an agent $\phi$s rationally even though—through tiredness, a state of heightened emotion, a state of ill mental health, or because they are unconscious—they aren’t in a position to engage in control-through-conscious-reasoning of their $\phi$-ing there-and-then.

Second, the sort of control at issue is direct. Consider an agent who successfully gets themselves to drop their urge to smoke by engaging in an act of self-hypnosis, or by ingesting a new wonder-drug which functions to suppress that urge. In these sorts of cases, the agent has control over their urge just insofar as they are able to engage in certain acts of self-manipulation. Contrast these cases with a case in which the agent moves a limb or engages in an episode of imagination: in these sorts of cases, the agent exerts control over their body or mind in an intuitively more direct manner. According to the Agential View of rationality, $\phi$-ing rationally requires possessing the general ability to engage in an act of conscious reasoning which constitutes one’s taking direct control over whether one $\phi$s and in response to what.²⁵

This plausible Agential View of rationality requires the proponent of Normative-Guidance to commit themselves to the Perspectival Constraint. For suppose that the Agential View is true. Now take an agent who $\phi$s rationally and suppose further that there is nothing from that agent’s perspective which counts as making it the case that they ought to $\phi$. Is that agent able to exert direct control over whether they $\phi$ through conscious reasoning? I don’t see how they could: after all, there’s nothing for the episode of conscious reasoning in question to operate on. What this means is that if the Agential View is true, and the agent $\phi$s rationally, there will have to be something within the agent’s point of view which appears to them to count in favour of $\phi$-ing and which figures as a premise in the episode of conscious deliberation they are able to engage in.

But now suppose, in addition, that Normative Guidance is true, so that the agent in question counts as successfully responding to a factor that makes it the case that they rationally ought to $\phi$. What is the relationship between that factor and the factor which, by the lights of the Agential View, figures as a premise in the episode of conscious reasoning the agent is able to engage in? From the agent’s own point of view, these two factors are the same. That is, from the agent’s own point of view, the factor(s) which figure in the episode of conscious reasoning as favouring their $\phi$-ing just are the factors that make it rational for them to $\phi$. So, on pain of endorsing an error theory of this aspect of the agent’s perspective on themselves, anyone wishing to endorse both the Agential View and Normative Guidance will have to say that it’s a necessary condition of the agent $\phi$-ing rationally that the factor which makes it the case that they rationally ought to $\phi$ figures as such from their own point of view. But that’s just the Perspectival Constraint.

4.2 The mental life of the rational agent

Neither Structuralism nor Subjectivism meet the Perspectival Constraint. For that reason, they should be rejected. And for that reason, neither Psychologism nor Anti-Factivism, motivated in the ways described in Sect. 3.3, can be associated with plausible substantive theses.

Let’s take Structuralism first. According to the Structuralist, rationality consists in $\phi$-ing in a way that counts as a successful response to a structural requirement of rationality which applies to one. The idea is that if one is in certain non-factive psychological states, that imposes a requirement on one that one acts in certain ways, or holds certain other states, on pain of incoherence. Rationality is achieved when one is in certain non-factive states which impose such a requirement on one to $\phi$, and one $\phi$s in response to these states, so that one counts as discharging one’s structural obligation.

Since the ‘in response to’ relation is subject to the Perspectival Constraint, what the proponent of Structuralism will have to say, further, is this: whenever the agent $\phi$s rationally, they must be aware of certain of their non-factive psychological states which are in turn registered by them to make it the case that they structurally ought to $\phi$.

But the basic phenomenological objection to this idea is that it is not the case, in general, that it seems to the rational agent that certain aspects of their psychology function to make it the case that they ought to $\phi$. Rather, in general, it seems to the agent who $\phi$s rationally that what makes it the case that they ought to $\phi$ are facts about the way the world is which speak in favour of them $\phi$-ing. Consider again the case of Cassidy, who rationally intends to meet her friend for dinner because she promised to do so. Were Cassidy to reflect on the question of why she ought to meet her friend for dinner, that I promised to do so would be consideration that comes to mind but that I believe I promised would not.

Of course, as was noted in Sect. 1.1 we sometimes treat aspects of our psychology as speaking in favour of our performing certain actions or holding certain attitudes. I believe I’ve been abducted by aliens can perfectly well be treated by someone as making it the case that they ought to request an appointment with a psychiatrist, for example. In these sorts of cases it really is the fact that one believes something which the agent treats as grounding the ought that applies to them, not what is believed. But notice that even in these sorts of cases, it does not seem to the agent that they ought to $\phi$ because their belief is part of what makes it the case that it would be incoherent for them not to $\phi$. That is to say: it doesn’t seem to the agent in these sorts of cases that the relevant psychological states ground a structural ought that applies to them. Rather, it seems to the agent even in these sorts of cases that the fact that they are in the relevant psychological states functions to ground an ought in the same sort of way as facts about the world function to ground such oughts—for example, the psychological fact seems to them to ground yet more moral or prudential oughts. So the basic phenomenological objection to Structuralism stands.²⁶

Now let’s turn our attention to Subjectivism. According to Subjectivism, rationality consists in $\phi$-ing in a way that counts as a successful response to a subjective ought that applies to one. The idea is that one has certain beliefs about the way the world is which are such that (to a first approximation): were they true, one would have normative reason to $\phi$, and this suffices for one to be under a special kind of obligation to $\phi$: one subjectively ought to $\phi$. $\phi$-ing rationally is all about successfully comporting oneself to these subjective oughts, by responding to what one believes about the situation.

Again, though, this ‘in response to’ relation is subject to the Perspectival Constraint. That means the proponent of Subjectivism will have to say that whenever an agent $\phi$s rationally, the agent must be aware of items they believe, items which can be there whether or not what they believe is true, and they must register these items as making it the case that they subjectively ought to $\phi$.

The phenomenological objection to this picture should be obvious. For consider what the deliberative life of the agent would be like if the considerations they believe did appear to them as grounding subjective oughts during conscious reasoning about what they should do or think. Suppose it appears to the agent as if the fact that p grounds an objective ought for them to $\phi$. If subjective oughts appeared from the point of view of the agent it would also have to be the case that p appears to them not just as a fact but also as a consideration they believe and as, in its guise as a thing believed, making it the case that the agent subjectively ought to $\phi$ in addition to the objective ought which appears to apply to them. This would mean that for any objective ought that appears to the agent to apply to them, there would be a second ought of a different type which appears to the agent to apply to them. I take it to be phenomenologically obvious that this is not the way things are from the agent’s own point view.

So neither Structuralism nor Subjectivism are plausible theses: given the Perspectival Constraint, both run straight into phenomenological problems. We can add that each theory runs into a kind of ethical objection, too. Each thesis, given the Perspectival Constraint, implies that the rational agent must $\phi$ in recognition of certain subjective features of their situation: either certain of their psychological states qua grounding structural requirements, or the items they believe qua grounding subjective oughts. But since, from the point of view of the agent, there will also be plenty of moral, prudential, or epistemic oughts which apply to them, grounded in objective features of their situation, to which they respond in cases of rationally $\phi$-ing, Structuralism and Subjectivism have the rational agent as responding jointly to certain features of themselves and to certain features of the world (although they will have it that only a response to the former is what constitutes rationality, properly so-called).

But an agent who is doubly-motivated in this way is perverse. Imagine someone who goes to visit their ailing relative both because it’s their relative, so that they have moral reason to go, and because it would be incoherent for them not to go, given their beliefs. Or consider an agent who sits down to play the piano both because they regard mastering the relevant piece as intrinsically valuable and because they recognise that were what they believe about their situation true they would have reason to sit and play, so that they subjectively ought to sit and play. Such agents suffer from something akin to, although not exactly like, Williams’s (1976) one-thought-too-many syndrome. I take it to be an obvious constraint on any successful theory of rationality that rational agents as such don’t display this syndrome.

This concludes my argument. Before ending the paper with some reflections on how the dialectic is affected by the possibility of a Disjunctive View of rationality, let me summarise the argument so far. The debates over Psychologism and Factivism appear to be merely verbal disputes: this I tried to show in Sects. 1–2. But that’s puzzling: presumably philosophers engaged in those debates are debating about something more than just what phrases like ‘the reason’ and ‘S’s reason’ mean in motivating reason ascriptions. There are substantive issues concerning the nature and normative structure of rationality at least in the vicinity of these debates: this is what I tried to show in Sect. 3. Those substantive issues are whether Structuralism, which I associate with Psychologism, and Subjectivism, which I associate with Anti-Factivism, are the correct theories of rationality: two versions of the Anaemic View. In light of the Perspectival Constraint, however, Structuralism and Subjecitvism are untenable on phenomenological and ethical grounds. The upshot for Psychologism and Anti-Factivism is therefore that they can be saved from the charge of being merely verbal theses, but only at the expense of being committed to false substantive ones.

4.3 Coda: disjunctive views of rationality

I want to finish with a comment on how the availability of a Disjunctive View of rationality might have an impact on the argument of this paper. According to Disjunctive Views of rationality: the agent $\phi$s rationally in both the good case and the case of factual error. But there is a difference between the two cases, rationality-wise: in the good case, the agent’s $\phi$-ing rationally consists in them successfully responding to an objective normative reason—whether a moral reason, a prudential reason, an epistemic reason...; in the case of factual error, it consists in them achieving a non-accidental fit between their response and what appears to them to be objective normative reasons to do so. Moreover, these two statuses of rationality are intimately related: successfully responding to what appears to one to be an objective normative reason for one to $\phi$ is to be understood, to a first approximation, as consisting in responding in a way that would be a successful response to an objective normative reason, were one to know the relevant facts. Because the bad kind of rationality is parasitic on the good kind in this way, there’s no danger of the bad kind of rationality rendering the good kind of rationality explanatorily redundant.

I am sympathetic to an account of rationality with the disjunctive character just broached. What I want to do now, however, is describe a version of that disjunctive picture, the availability of which appears to undermine the argument of this paper. I will argue in turn, however, that the argument of the paper extends to this version of the Disjunctive View. The upshot is that if we’re to opt for a Disjunctive View of rationality at all, and in the end I think that we should,²⁷ we shouldn’t opt for this particular version of it.

According to the version of the disjunctive view in question:

• Subjective Ought Disjunctivism In the good case, the agent $\phi$s rationally and this consists in them successfully responding to an objective normative reason. In the case of factual error the agent $\phi$s rationally and this consists in them successfully responding to a subjective ought, grounded in what they believe about the situation. The latter condition is parasitic on the former condition in the following way: What it is for the agent to be under a subjective ought is, to a first approximation, for what they believe to be such that were it true, it would be an objective normative reason for them to $\phi$.

This position is a form of Disjunctivism about rationality because it treats rationality as constituted by different kinds of conditions across good cases and cases of factual error, but treats the condition which constitutes rationality in the case of factual error as parasitic on the condition which constitutes rationality in the good case. What marks this kind of Disjunctivism out as a distinctive form of the view is the way it cashes out rationality in the case of factual error and, correspondingly, how this is supposed to be parasitic on rationality in the good case. According to Subjective Ought Disjunctivism, the rationality of the agent’s $\phi$-ing in the factual error case consists in them successfully responding to a subjective ought, in the sense associated with the Subjectivist version of the Anaemic View described above. And this condition of successfully responding to a subjective ought is parasitic on the condition of responding to an objective ought in the following way: subjective oughts just are oughts which would be objective oughts, were the beliefs on which they’re grounded true.²⁸

The availability of Subjective Ought Disjunctivism appears to undermine the argument of this paper at two crucial points.²⁹ First, the phenomenological and ethical arguments offered in Sect. 4.2 appear to fail to extend to Subjective Ought Disjunctivism. After all, given the way the subjective ought is parasitic on the objective ought, there isn’t an extra ought present in the good case which would be one ought too many, from the ethical and phenomenological points of view. And second, isn’t Subjective Ought Disjunctivism a position which is a form of Anti-Factivism about motivating reasons? If so, since Subjective Ought Disjunctivism is a substantive view and one which, as per the first point, isn’t undermined by the arguments of Sect. 4.2, it turns out there is a version of Anti-Factivism which is substantive and potentially true, after all.

In fact, however, I think the argument of the paper extends to Subjective Ought Disjunctivism. Let me take the second concern first. It’s certainly true that Subjective Ought Disjunctivism is consistent with the letter of Anti-Factivism. Anti-Factivism says that S can $\phi$ for the reason that p even if not-p, where S’s reason is what they believe instead of any psychological state they are in. This looks perfectly consistent with the sort of disjunctive picture advertised above. It would just have to be that this Anti-Factivist identifies $\phi$- ing motivated by a reason with the rationality constituting condition which is common to the good case and cases of factual error.

But it’s not clear to me that this sort of Anti-Factivism is a substantive position. To see this, we should ask: What substantive claim is added to Subjective Ought Disjunctivism insofar as we combine it with the Anti-Factivist view that one $\phi$s for a reason across good cases and cases of factual error? So far as I can tell, we just add the merely linguistic thesis that we can call the kind of rationality constituting condition which is common to each of these cases ‘$\phi$-ing for a reason’. But this is not a philosophically substantive thesis.

Now let me turn my attention to the worry that Subjective Ought Disjunctivism avoids the phenomenological and ethical arguments of Sect. 4.2. Again, I doubt that the view does avoid these arguments. Take the phenomenological datum that it does not in general seem to the rational agent as if they respond to a subjective ought grounded in what they believe. Given Normative Guidance and the Perspectival Constraint, and assuming that the way things seem to rational agents isn’t systematically misleading, it follows that it is not in general true that the rational agent counts as successfully responding to a subjective ought.

Now, the Subjective Ought Disjunctivist does want to say that when one $\phi$s rationally in the factual error case, this consists in one’s successfully responding a subjective ought. So, it appears that they will have to endorse Normative Guidance about the rationality-constituting condition which is present in the case of factual error. But I’ve already argued that the Perspectival Constraint is something any proponent of Normative Guidance will have to commit themselves to. So, the Subjective Ought Disjunctivist will have to say that, in the case of factual error, the rational agent treats what they believe about their situation as grounding a subjective obligation to $\phi$ that they are under. But the phenomenological datum holds of the case of factual error just as it does in the good case: It doesn’t seem to the agent as if there is a subjective ought to which they are responding.

The upshot is that, even if the danger posed by the phenomenological datum to the Subjective Ought Disjunctivist’s account of the good case can be mitigated, that datum undermines the Subjective Ought Disjunctivist’s account of the factual error case any way. Mutatis mutandis for the ethical datum that there is something objectionable about responding to what one treats as grounding a subjective ought in addition to responding to what one treats as grounding an objective ought.

Where does this leave us? What this means is that if we’re to endorse a theory of rationality with a disjunctive character at all, then we should avoid one according to which rationality-in-the-factual-error-case consists in successfully responding to any kind of ought. We should drop Normative Guidance (and hence the Perspectival Constraint) from our account of that kind of rationality. And we should therefore say that the only kind of rationality to which Normative Guidance (and hence the Perspectival Constraint) applies is the good kind of rationality, which is also the most fundamental kind of rationality. How we are to work out the details of this form of Disjunctivism is a task I leave to future work.