Emotions and their reasons

ABSTRACT

Although it is now commonplace to take emotions to be the sort of phenomena for which there are reasons, the question of how to cash out the reason-responsiveness of emotions remains to a large extent unanswered. I highlight two main ways of thinking about reason-responsiveness, one that takes agential capacities to engage in norm-guided deliberation to underlie reason-responsiveness, and another which instead takes there to be a basic reason-relation between facts and attitudes. I argue that the latter approach should be preferred. Not only does a reasons-basic approach promise to fare better in accounting for cases that its opponent struggles to accommodate, but it promises also to uncover a sui generis relation between emotions and their reasons which is at best obscured and at worst denied by its opponent.

KEYWORDS:

• Reasons
• emotions
• rationality
• practical reasons
• epistemic reasons
• affective reasons

There is the common assumption that there are only epistemic and practical reasons … 

we are talking about the preconceptions of philosophers.

And might it be relevant that they have mostly been male?

– Skorupski, in ‘The Domain of Reasons’

We often take emotions to be experiences for which there are reasons. Victims of sexual harassment have reason to be angry, that your childhood pet has passed away is reason for sorrow, and that you were awarded tenure is reason for joy, pride, and perhaps, relief. We often assess our emotions, and those of others, for appropriateness. Is this merely a shallow feature of our folk psychology though, or does it suggest that emotions are amongst those phenomena that can be had for normative reasons?¹ Normative reasons are reasons that justify. These are reasons that count in favour of having a relevant attitude, but that would also justify that attitude rather than merely explain or motivate it. If your friend stood you up last weekend due to a family emergency rather than carelessness, this is a normative reason not to be mad at them for they have not offended you. That getting mad at your friend is likely to make them never do so again, out of fear of retaliation, for example, might be a motivating or instrumental reason to feign anger, or to try to put oneself in a state of anger, but it does not justify your anger. These types of reasons have been called the ‘wrong sort’ of reasons, because they are not the proper reasons that can be followed in experiencing an emotion, that is, one does not become angry based on such considerations. Anger is sensitive to offences rather than to, for example, instrumental reasons for feigning offence (D’Arms and Jacobson 2000).²

There are two main ways of thinking about reason-responsiveness. What I call ‘rationalist’ positions explain reason-responsiveness in terms of compliance with rationality norms (Smith 1994; Korsgaard 1996). On such accounts, considerations become reasons only through adherence to the requirements of rationality. Rationality requirements typically include: coherence amongst one’s beliefs, enkrasia (that one intended to Φ when one judges that one ought to Φ), and adherence to rules of inference such as modus ponens (Broome 1999; McHugh and Way 2018).³ It is by virtue of reasoning in conformity with the requirements of rationality that features of the world become reasons for us on these accounts. Crucially, on rationalist accounts, it is agents, rather than attitudes that are reason-responsive. An alternative approach takes the reason-relation to be basic or primitive. This involves taking there to be a basic normative relation between a fact and an attitude that doesn’t depend on the agent’s conforming to any norms of rationality (Raz 1999; Scanlon 1998; Kolodny 2005). I call these ‘reasons-basic’ accounts.⁴

The question I am interested in is how we should construe emotions as reason-responsive. Are emotions reason-responsive in virtue of agential adherence to rationality requirements, or in virtue of a basic reason-relation? Answers that seem aligned with each stance have been given in the literature on emotion but their relative merits have not been assessed. I will argue that the reasons-basic approach is superior. I will do so first by highlighting a few problems faced by those existing accounts that have rationalistic tendencies. Reasons-basic approaches will emerge as preferable not only because they seem to better cope with these problems but, crucially, because they may allow a sui generis reason-relation distinctive of the emotional realm to emerge.

1. Agential disposition accounts

The most detailed accounts of emotional reason-responsiveness in the literature take agential virtues or dispositions to be central (Tappolet 2016; Jones 2003; Goldie 2004). What these views have in common is that they take reason-responsiveness to rely on properly functioning dispositions to conform to norms of rationality. Agential disposition accounts take emotions to be amongst the agent’s reason-tracking mechanisms, alongside other mechanisms or ‘subsystems’ such as the perceptual systems (Jones 2003). On such accounts, an agent is reason-responsive in light of their emotions, so long as she manifests well-functioning reflective self-monitoring habits (Jones 2003), agential virtues (Tappolet 2016) or dispositions (Goldie 2004). I follow these authors in using these terms interchangeably. The thought is that many sub-systems might track reasons – perceptual, emotional, perhaps motivational – systems we share to a large extent with other animals, but for reason-responsiveness reserved for human agents, there must be well-functioning agential dispositions in play.⁵ The crucial disposition invoked on such accounts is the following: so long as the agent is disposed to intervene and block treatment of their emotion as reason-tracking when the agent has reason to do so, then the agent’s emotions can be seen as properly reason-responsive.

Jones (2003) writes that agents are reason-responsive when:

the agent’s dispositions to reflective self-monitoring are such that she would not rely on that first order sub-system were it reasonable for her to believe that it failed to reason-track  …  this guidance may remain ‘virtual’ – that is, revealed in how the agent would behave in various counter-factual circumstances. (195–196)

Tappolet (2016) writes that:

what is required for reason-responsiveness is the exercise of well-tuned epistemic and practical habits, i.e., of what can be described, broadly, as agential virtues... What is required... is that the agent be disposed to intervene and take active control if she has reason to distrust her emotion. (176)

Similarly, Goldie (2004) writes that reason-responsiveness involves having ‘the right habits and dispositions of thought, such that doubts will arise when and only when they should’ (251).

These views seem to be committed to the following claim:

Counterfactual Claim: One is reason-responsive in being epistemically or practically guided by one’s emotions, in so far as one would not have treated one’s emotion as reason-tracking had there been reason to believe that the emotion failed to reason-track.

The capacity to comply with this counterfactual claim is dispositional. This means that the capacities that confer reason-responsiveness do not involve conscious reflective deliberation but rather a standby sensitivity to when one should engage reflective deliberation, namely when there are reasons to distrust one’s emotion. Tappolet (2018) writes that ‘when there is no reason to distrust your emotion, you don’t need to deliberate to be reason-responsive’ (157). Reason-responsiveness will therefore typically be ‘unreflective, and not part of conscious deliberation  …  (but) rely on our habits and dispositions, at work in the background of our minds, so to speak’ (Goldie 2004, 151).

The counterfactual claim can be read as a claim about dispositional sensitivity to mental-state defeaters. These come in two main varieties; undercutting defeaters, which give one reason to doubt the truth of the grounds of one’s belief, and, rebutting defeaters, which give one reason to hold the negation of the defeated belief, or for holding some proposition that is incompatible with it (Pollock 1986, 38). For an agent to comply with the counterfactual claim they must be sensitive to when there is reason to believe that the emotion has failed to reason track. In other words, the agent must be sensitive to when undercutting or rebutting defeaters are in play. This means being sensitive to whether there is reason to believe that one’s emotional system is malfunctioning, perhaps due to the influence of a foreign chemical substance, as well as to whether one has reason to hold a belief that conflicts with the emotion.

This standby sensitivity to comply with the counterfactual claim depends on explicit deliberative capacities, however:

having sensitivity to when reasons are defeated and when they are outweighed requires the capacity to reflect on the status of the deliverances of those mechanisms that purport to latch onto reasons such as perception, emotion and desire, but also the capacity to reflect on reasoning itself. (Jones 2003, 190)

The agential dispositions on which reason-responsiveness depend are then dispositions to engage reflective reasoning when the agent has reason to, where this reflective reasoning is presumably guided by the sorts of norms the rationalist is committed to (such as coherence amongst one’s beliefs, enkrasia, i.e. that one intend to Φ when one judges that one ought to Φ, and adherence to rules of inference such as modus ponens). Agential Disposition accounts are at the very least committed to one rationality requirement, the counterfactual claim, i.e. that one not trust one’s emotions when one believes one has reason not to. These accounts are committed to the view that emotions are only reason-responsive in so far as agents have well-tuned dispositions not to violate this requirement. Their reliance on the capacity for robust reflective reasoning, however, suggests that further rationality requirements are in play on these accounts:

an agent requires critical reflective ability, dispositions to bring that ability to bear when needed, and dispositions to have the results of such reflection control their behaviour (Jones 2003, 190).

Agential Disposition accounts then, although not full-blown rationalist views,⁶ have a number of rationalist inheritances. First, on these accounts reason-responsiveness is a feature of agents rather than attitudes, and, secondly, it depends on agential capacities for reflective reasoning governed by rational norms, rather than concerning merely a basic relation between facts and attitudes.

1.1. Problems

Agential Disposition accounts struggle to account for cases of outlaw emotion and emotion-based inverse akrasia. Let me briefly outline the sorts of cases I have in mind. Emotions are considered a main culprit in making agents act against their considered judgements of how they ought to act (Arpaly 2000; Jones 2003). In anger, we often violate our commitment to civility, fear often prevents us from following a desired plan of action, and in pride or jealousy we can compromise relationships that we are dedicated to. As a common source of akratic action then, emotions seem to frequently be at odds with the enkratic requirement (that one intend to Φ when one judges that one ought to Φ). When Huckleberry Finn fails to act in accordance with his judgement that he should turn his friend Jim in, however, and instead follows a sense of respect and love for Jim, he arguably acts in light of a reason despite acting akratically. Such cases have been called cases of inverse akrasia or rational akrasia (Arpaly 2000; Tappolet 2016). The thought is that sometimes it is not irrational to violate the enkratic requirement. Many take it to be a condition on a successful account of reason-responsiveness that room for rational akrasia be made, including proponents of Agential Disposition accounts (Arpaly 2000; Jones 2003; Tappolet 2016).

In related ‘outlaw emotion’ cases, emotions conflict with an agent’s wider set of beliefs (Jaggar 1989; Silva 2021).⁷ An otherwise content housewife experiences an outlaw emotion when she becomes angry about her confinement to the home, despite endorsing a large set of beliefs about the value of being a housewife, and feeling a range of emotions that cohere with these beliefs, such as pride in fulfilling this role. Jaggar coined the term ‘outlaw emotions’ to refer to emotions that are ‘distinguished by their incompatibility with the dominant perceptions and values’ (1989, 166). These emotions can be epistemically valuable in granting insight the agent may have otherwise lacked. Much like with inverse akrasia, we might think that an account of emotional reason-responsiveness would do well to make sense of outlaw emotions as sometimes reason-responsive (Arpaly 2000; Jones 2003; Tappolet 2016).

How do outlaw emotion cases and cases of inverse akrasia relate to each other? Cases where one acts on reason-responsive outlaw emotions will be cases of inverse akrasia, but not all cases of emotion-based inverse akrasia will also be outlaw emotion cases. This is because outlaw emotion cases involve emotions that clash with extensive belief systems while cases of inverse akrasia can in principle occur in agents that merely act against their best judgement, where this judgement does not reflect a wider belief-system. The main reason for teasing these two cases apart is to highlight the practical as well as the epistemic role of emotions as reason-responsive states. Inverse akrasia concerns action, while outlaw emotion cases need not. Indeed, outlaw emotion cases are ones typically stressed for their epistemic value (Jaggar 1989; Silva 2021).

Agential Disposition accounts must presumably either deny that inverse akrasia and outlaw emotion cases involve reason-responsiveness or they must provide a story for why this is not so, seeing as these cases seem to be characterized by violations of rationality requirements. In both cases of inverse akrasia and outlaw emotion, there are arguably rebutting defeaters in play. One plausibly has reason to believe a proposition that conflicts with the emotion in both cases. In inverse akrasia, one has a practical belief about what one ought to do, in Huck’s case that he should hand Jim in, which conflicts with his positive emotions towards Jim, while in outlaw emotion cases one endorses a normative belief about the value of being a housewife, for example, which conflicts with the outlaw emotion of anger about being a housewife. In both cases, the agent has at least some reason to believe that their emotion is failing to reason track, given the presence of these conflicting beliefs. These beliefs presumably give the agent reason to doubt that their emotions are properly tracking reasons.

Both Jones (2003) and Tappolet (2016) take the Agential Disposition account of emotional reason-responsiveness to be able to account for cases of inverse akrasia and outlaw emotion, however. On akrasia they are explicit:

The functioning of such sub-systems does not stop being expressive of our commitment to rational guidance just because there is now an opposing all-things-considered judgment. In some cases that all-things-considered judgment may be such that the agent would distrust it, if her self-monitoring capacities were functioning as they should. Thus, the regulated sub-system can be more expressive of the agent's commitment to rational guidance than the all-things-considered judgment: the incontinent action can display the agent's commitment to rational guidance more fully than does the continent action. (Jones 2003, 196)

The focus will be on cases of akratic actions caused by emotions that conflict with the agent’s better judgment. I shall propose that even in such dysfunctional cases, we are able, under certain conditions, not only to track reasons but also to be reason-responsive when we act on our emotions. (Tappolet 2016)

The claim is that, in cases of inverse akrasia, an emotion’s reason-responsiveness is not undermined by its conflicting with an all things considered judgement of how one ought to act. Jones (2003) adds that akratic actions are rational when:

(1) the action is produced by a sub-system that reason-tracks because the agent reason-responded, and (2) the agent would have distrusted her all-things-considered judgment were her self-monitoring dispositions operating as they should. (196)

Agential Disposition accounts would presumably make sense of outlaw emotion cases in an analogous way.

I think Agential Disposition accounts are far too quick to assume success in accounting for these cases. First, consider Jones’ (2003) two requirements for the possibility of emotion-based rational akrasia. The first involves the agent manifesting well-functioning self-monitoring dispositions, this is what allows the emotion ‘subsystem’ to ‘reason-respond’, while the second suggests that the agent actually has malfunctioning self-monitoring dispositions. Well-functioning agential dispositions would not have allowed the agent to endorse the all-things-considered judgement with which the emotion conflicts. If the agent holds such an all-things-considered judgement then they are not manifesting well-functioning self-monitoring dispositions. Cases of inverse akrasia, according to this view then, seem to involve the unattractive result of an agent both having well-functioning agential dispositions (that grant the emotion reason-responsiveness), while manifesting malfunctioning self-monitoring dispositions (in holding the all-things-considered judgement), simultaneously. In so far as these dispositions are agential, it seems that one either has well-functioning dispositions in a given circumstance, or not.

If no deliberation is triggered in cases like Huck’s then one’s dispositions for reflective self-monitoring have failed to pick up on the conflict between one’s emotion-based intention and one’s judgement of what one ought to do. This means that one’s emotions are not reason-responsive, the counterfactual claim is violated, and the agent is not manifesting well-tuned self-monitoring habits. If, on the other hand, deliberation is triggered, due to well-functioning agential dispositions, then either the akratic situation will be dissolved or one will fail to revise one’s belief and be irrationally akratic. It appears that on Agential Disposition accounts one is either rational or akratic, there seems to be little room to account for rational akrasia.

If we want to grant that there are cases of rational akrasia, then it seems that what makes akratic action rational cannot be its dependence on the agent’s dispositions to intervene when they have reason to doubt the emotion is getting things right.⁸ The fact that there is a reason for the emotional attitude itself, independently of the agent’s capacity to engage in deliberative reasoning when there is reason to do so, seems to be (at least part of) the answer.

Similarly, in the epistemic, outlaw emotion cases, the outlaw emotion seems to be reason-responsive despite it conflicting with the agent’s wider set of beliefs. Can Agential Disposition accounts make sense of this? If the agent displayed well-functioning dispositions, outlaw emotion cases would be flagged as problematic for violating the counterfactual claim (as well as coherence norms) and deliberation would be triggered. This means that if deliberation is not triggered, the outlaw emotion is not reason-responsive on these accounts. Outlaw emotions would arguably only be reason-responsive in cases where the deliberation triggered resulted in an endorsement of the outlaw emotion and/or revision of those beliefs with which it conflicts. This is quite a high bar to place on the reason-responsiveness of outlaw emotions, seeing as these cases involve emotions that conflict with the agent’s widespread belief system and where deliberation is more than likely to side with this belief system rather the outlaw emotion. Agential disposition accounts would at the very least drastically limit the cases in which outlaw emotions can be said to be reason-responsive. Any case where the agent does not trigger deliberation and reflectively endorse the outlaw emotion is a case where the outlaw emotion is not reason-responsive. To make sense of outlaw emotion cases as reason-responsive, it seems intuitive to invoke the fact that there is a reason that stands in support of the emotional attitude itself.

I have argued that despite aiming to accommodate cases of outlaw emotion and inverse akrasia, Agential Disposition accounts run into significant problems in their attempts to so. Proponents of these views must say more on how they intend to account for such cases.⁹ A further worry faced by Agential Disposition accounts is that they seem to obscure intuitive differences between distinct ‘reason-tracking’ systems.¹⁰ On these accounts, emotions, perceptions and desires are all reason-tracking sub-systems made reason-responsive by agential dispositions to reflectively self-monitor. Differences between these sub-systems are concealed or even denied. Emotions, however, are phenomena for which we demand and provide reasons, they seem to admit of normative justification while perceptual experiences typically do not. Agential Disposition accounts are hard pressed to account for this difference.¹¹

I have not claimed that there are no moves available to Agential Disposition accounts in response to the worries just outlined, but I take sufficient reason to have been given to warrant exploration of an alternative account of emotional reason-responsiveness.

2. A reasons-basic approach

Although a number of philosophers of emotion seem to take a reasons-basic approach (Deonna and Teroni 2012; Naar 2022; Müller 2022; Mitchell 2021) they do not provide sustained endorsements of this approach nor reasons why such an approach should be preferred. Crucially, they do not propose that the emotion-reason-relation is distinctive or sui generis.¹² I have highlighted a few reasons why a reasons-basic approach might be preferred, namely that it seems more promising in being able to account for cases that trouble Agential Disposition views. Although detailing and defended this claim is beyond the scope of the current paper, it seems that by invoking a reasons-basic conception of reason-responsiveness we can tease apart attitudinal rationality from agential rationality to better account for these cases. A reasons-basic approach also allows us to make initial sense of the difference between perceptions and emotions that Agential Disposition accounts risk obscuring. On a reasons-basic approach, mental phenomena are either the sort of thing that can be had for reasons, or they are not. There is then room to argue that perceptions are not reason-responsive attitudes. The thought would be that intentionality is not sufficient for reason-responsiveness. For reason-responsiveness, the taking of a stance, that reasons can count for or against, is necessary.

Emotions involve an evaluation, or the taking of an evaluative stance, towards which considerations can count for or against. Emotions, much like beliefs or intentions to act, involve taking a stance on something. In the epistemic case, one takes the stance that a certain proposition is true, while in the practical case one plausibly takes the stance that a certain action would be good. Both involve taking a particular stance towards a proposition, one that goes over and above the mere registering of the contents of that proposition. Emotions may be much the same. In having an emotion, one takes a stance on an object or state of affairs, specifically an affective, rather than epistemic, or practical stance. By ‘stance’, I have in mind a personal level orientation towards a proposition. In the case of emotions (and perhaps also in intention and belief), these affective stances are informed by the agent’s underlying cares and concerns (Müller 2019). Emotions plausibly involve a felt, non-conceptually structured, stance, rather than a conceptually structured stance as in belief, or an intention to act as in the practical stance.

This way of thinking lends itself particularly well to attitudinal theories of emotion where emotions are sui generis evaluative attitudes of some sort, for example, felt attitudes of action readiness (Deonna and Teroni 2012) or (non-bodily) felt attitudes (Mitchell 2021), taken towards intentional contents. That being said, reasons-basic approaches are likely compatible with other conceptions of emotion, so long as they don’t construe emotions as literal perceptions incapable of reason-responsiveness.¹³ In what follows I will be assuming that emotions are some sort of sui generis evaluative attitude, remaining agnostic on what exact type of attitude this might be. So long as emotions are attitudes of some sort, they seem to be candidates for reason-responsiveness on a reasons-basic approach. And if emotions are sui generis attitudes, that is, irreducible to other kinds of attitudes (such as beliefs and intentions), then we might ask whether they involve a sui generis reason-relation. This is because, on a reasons-basic approach, reasons involve a relation between a fact and an attitude, such that we might think that the reason-relation differs depending on what type of attitude is involved.

The thought that emotions are sui generis mental states is widespread in the literature. Emotions involve distinctive phenomenology, attentional patterns and dynamic psychological profiles (that is, they have beginnings, middles and ends and can vary in phenomenology and intensity at different times), as well as a strong (for some even constitutive) link to motivation, that are thought to set them apart from beliefs with similar contents (Deonna and Teroni 2012; de Sousa 1987; Tappolet 2016). In this literature it is also assumed that many of these characteristics make emotions distinct from mere intentions to act. The main benefit of a reasons-basic approaches is, I think, that once we focus on the relation between reasons and emotional attitudes, a sui generis reason-relation may emerge. This will be the focus of the rest of this paper.

2.1. Epistemic and Practical Reasons

Amongst those that support construing the reason-relation as basic, reasons are typically construed as facts and there is widespread endorsement of two types of basic reason-relation: reasons for belief and reasons for action (Raz 1999; Scanlon 2014; Skorupski 2010; Maguire 2018).¹⁴ I will take reasons to involve relations between facts and attitudes. By ‘epistemic reasons’ then I mean the relation at play when facts count in favour of beliefs, the paradigmatic epistemic attitude.¹⁵ The same fact can be an epistemic reason and a practical reason depending on which type of attitude it is counting in favour or against (for example, the fact that it is sunny today may be a reason to believe it is a beautiful day and also reason to intend to go to the beach), but not all facts can act as reasons for both, as we will see. I will take the attitudes that epistemic reasons stand in support of to be beliefs, while taking practical reasons to stand in support of actions or intentions to act.¹⁶ Many speak of epistemic and practical reasons as distinct, and often exhaustive, types of reasons. This dichotomy is supported by distinctive features that characterize each type of relation.

Reasons for belief and reasons for action count in favour of beliefs and actions respectively, but they do so in different ways, given that they support different types of attitude. In taking the first steps towards the provision of an account of sui generis emotional reason-responsiveness, I will begin by outlining the differences between the epistemic and the practical reason-relation that are present in the literature. I will be taking these for granted for the sake of the argument, as a background framework against which to map the emotional reason-relation. Three distinctions are typically made between epistemic and practical reasons, they relate respectively to their: formal object, their relation to value and what I will call ‘binarity’. I outline the relevant distinctions between the epistemic and practical realms in turn, before turning to what to make of reasons for emotions which I call ‘affective reasons’.

Practical and epistemic reasons are thought to relate to distinct formal objects. Reasons for belief all stand in the epistemic reason-relation due to their connection to truth, which is taken to be the formal object of belief. All reasons for belief then relate to one concern, truth. Reasons for action, on the other hand, count in favour of actions in virtue of the value of performing a given action. Raz (1999) writes that ‘reasons are facts in virtue of which (…) actions are good in some respect and to some degree’ (23). The formal object of the practical realm, then, is the value or goodness of the action in question. This can be cashed out in terms of the action’s relation to ‘the Good’, or by citing the specific values, or forms of goodness, that discrete actions hold for the agent. For example, an action might hold aesthetic value for an agent (going to the opera), or moral value (confronting a bully), or indeed religious value (an action’s piousness) or mere hedonic value (having an ice cream). Either way, epistemic reasons and practical reasons stand in support of their respective attitudes in virtue of bearing a relation to distinct formal objects which are often construed as constitutive to attitude types: truth for beliefs and ‘the Good’ or forms of the good for action.

This distinction in type of formal object means that practical reasons can relate to a plurality of values, while epistemic reasons do not. There are many ways of being good. Practical reasons that relate to different specific values can be weighed against each other in support of actions or intentions to act. In the epistemic case, all epistemic reasons count in favour (or against) the same concern: truth. When deciding whether to spend the evening reading a book, one weighs practical considerations that relate to distinct values: for example, the aesthetic value of literature, the epistemic value of immersing oneself in the relevant topic, one’s duty to spend that evening with a friend instead, and so on. In coming to believe that the relevant book was written under a pseudonym, on the other hand, considerations of historical evidence, similarities to the writing style of other work of the same authorship, and expert opinion, all bear on the truth of one’s belief, rather than on any other concern.¹⁷

The difference in formal object between the epistemic and practical realms highlights the distinct relation these realms bear to value. The value of holding a certain belief is typically thought to be independent to one’s reasons for having the belief. It might be psychologically beneficial for me to believe my cat to be alive, but that this is so is no reason, in the normative sense, for me to hold that belief. In the practical case, the value of taking a certain action is exactly what counts as a reason for that action. This is the case even when reasons for action relate to values we think of in non-consequentialist terms as well, such as duties and principles. Epistemic reasons are independent of such values as well. One might be committed to patriotic principles and duties, for example, but these are not adequate reasons to believe one’s country to be innocent of war crimes. Epistemic reasons then, are independent of the consequentialist and non-consequentialist value of holding the relevant belief, while practical reasons are dependent on precisely the value of pursuing given actions.

As for the third difference, the received view is that practical reasons demonstrate what I call binarity while epistemic reasons do not (Skorupski 2010; Raz 2011). This is the thought that, while in the epistemic case one can suspend judgement in light of one’s reasons, in the practical case there is no stance analogous to the suspension of belief, one either acts (or intends to act) or one doesn’t.¹⁸ If you have reason to go to the beach and you do not go, then you have not acted on your reasons for going to the beach. If you have reason to believe an opera particularly sublime, on the other hand, and you do not believe this is the case, you might take either of the following stances: you might suspend judgement or you might believe that the opera is not particularly sublime (due to countervailing testimony for example). In the practical case then, there are only two options regarding how to engage with one’s reasons. In the epistemic realm, there seem to be three options for engaging with one’s reasons: in addition to believing P in light of one’s reasons or believing not-P in light of them, one can suspend judgement in light of one’s reasons. There doesn’t seem to be an equivalent to suspension of belief in the practical realm, making practical reasons ‘binary’ because one either acts in line with them or one does not.

To sum up, epistemic reasons count in favour of the truth of beliefs, considerations that count in favour of the value of holding the relevant epistemic attitude are not adequate reasons for belief, and epistemic reasons do not display binarity, for suspension of belief in light of them is possible. Practical reasons differ on all accounts; they count in favour of a distinct formal object to truth, be it ‘the Good’ of the action or more fine-grained values. On either account of the formal object, practical reasons count in favour of the value of performing a certain action, unlike epistemic reasons, and they display binarity because suspension of action or intention is arguably not possible. The epistemic and practical reason-relations appear then to be irreducible. A reasons-basic account, that takes the relation between reasons and attitudes to be primitive, allows us to appreciate that there are not one but arguably at least two types of primitive reason-relation: the practical and the epistemic. Might there be an additional reason-relation characteristic of the emotional realm?

2.2. Affective reasons

Like reasons for action and belief, reasons for emotions are normative, they make the relevant evaluation appropriate to the object in question. The central difference between epistemic, practical and affective reasons is that the latter are reasons for being in a certain affective state, that is for feeling a certain way. To believe an evaluative proposition, even the proposition that an object merits a certain feeling, is not in itself to respond to affective reasons because it is not itself to feel a certain way. Similarly, acting upon an object in ways that are closely related to evaluative properties the object might bear, need not involve responding to affective reasons. Responding to affective reasons involves feeling a certain way in light of reasons. It involves bearing a particular relation to certain kinds of considerations, a relation distinctive of affective as opposed to epistemic or practical attitudes. We will see what sort of relation this is.

Responding to an affective reason involves feeling a certain way in virtue of certain properties of the object. We will see that affective reasons do not stand in support of emotions in ways that are easily reducible to practical or epistemic reason-relations. Affective reasons will emerge as potentially sui generis. Let’s look at how affective reasons do regarding the differences between the received dichotomy of practical and epistemic reasons, namely regarding; formal object, relation to value and binarity.

Emotions, as a type of attitude, are typically not construed as being related to any one formal object. It is particular emotion types that are thought to have formal objects: danger for fear and offence for anger, for example (Teroni 2007; Mulligan 2007). Emotions have, in this sense, narrow formal objects: those evaluative properties characteristic of particular emotion types. In this sense, emotions are concerned only with a subset of an agent’s evidence: those considerations that count in favour of specific formal objects. If this is the case, then affective reasons differ from epistemic reasons by virtue of not relating to one overarching, or wide, formal object.

If emotions, as a kind, admitted of a common formal object, it would not likely be truth. Notions of fittingness or appropriateness, as opposed to truth, are typically invoked with regards to the emotions.¹⁹ One reason for why fittingness, as opposed to truth, is the relevant standard in the emotional realm is that some think emotions do not involve propositional attitudes deemed necessary for truth-aptness (Salmela 2014, 105). Another reason is that, because emotions often involve conative components with world-to-mind direction of fit, they seem to be characterized by a type of intentionality that may not be reducible to the mind-to-word direction of fit most amenable to truth-aptness. Fittingness is the widely endorsed standard that we might think adequately captures these (and potentially other) distinctions. This makes fittingness, as opposed to truth, the candidate ‘wide’ formal object of emotions. On either construal of the formal object of emotions (specific formal objects or fittingness) then, affective reasons and epistemic reasons relate to distinct concerns as they do not share the same formal object.

In contrast to the practical realm, where practical reasons can be thought of as counting in favour of the Good of pursuing a particular course of action, affective reasons do not count in favour of the Good of feeling a certain emotion. Many emotions are not good to feel, they do not feel pleasant nor do they necessarily have beneficial causal effects for the agent. Perhaps affective reasons are more akin to practical reasons construed as relating to distinct values. After all, we saw above that practical reasons need not be thought to relate to the Good, but can be thought of as relating to the specific values of distinct actions (Raz 2011). Reasons for emotions might be thought to relate to specific evaluative properties much like reasons for action relate to the different values of specific actions. For example, anger relates to offences, fear to dangers, sadness to losses, and we saw that we might think actions relate to different value types, going to the opera may be related to the aesthetic value of the action while confronting a bully or eating an ice cream may relate to the moral and hedonic values of these respective actions. However, there is a crucial difference between the practical and emotional realm here: actions, as we saw, still admit a plurality of values as their reasons, while emotions admit only those reasons that relate to the relevant evaluative property. For example, although you may go to the opera for its aesthetic value, those considerations that can count in favour of going to the opera are not restricted to the aesthetic. Hedonic value, one’s enjoyment, may count in favour of the action, as may any instrumental value in attending the opera (seeing and being seen by a certain crowd). Some might even think there is moral value in attending the opera. Emotion types, on the other hand, are (at least typically) restricted in their support to considerations that relate specifically to one type of evaluative property or formal object: danger for fear, offence for anger. Only considerations bearing on the lion’s dangerousness count in favour of fear, for example. While a single action can derive support from an arguably open-ended range of values, with affective reasons this is not the case. A given emotion derives support from affective reasons that count in favour of only a particular evaluative property.

This isn’t to say that affective reasons that count in favour of distinct evaluations cannot support mixed emotional states. Nostalgia and thrill, for example, seem to involve a mixture of evaluative properties. There is, however, a limit to the range of distinct sources of evaluative support that can count in favour of mixed emotional states (which is set by the formal objects of the respective affective attitudes), whereas the range of distinct values that can count in favour of one particular action seems to be far more open ended. We therefore have reason to think that emotions do not share the formal object of intentions or beliefs. This is not altogether surprising, given that emotions are themselves typically taken to be irreducible to beliefs or intentions.

What about the relationship between affective reasons and the value of the attitudes they stand in support of? We saw that epistemic reasons are not reasons that concern the value of holding certain beliefs, but merely concern the truth of the belief. Epistemic reasons are ones that are followed in forming the relevant belief. Practical reasons for holding beliefs, such as it being beneficial for an agent that they hold a certain belief, are not reasons one can follow in the formation of beliefs. We saw that practical reasons, on the other hand, are precisely concerned with the value of performing, or taking, certain actions. On this point, affective and epistemic reasons are much the same. Proper normative reasons for emotions are justifying reasons for the emotional state itself, as opposed to reasons that might count in favour of bringing a certain emotional state about. We saw above when I introduced the notion of normative reason that prudential considerations are the ‘wrong sorts of reasons’ for emotions. That feigning anger would help you get your way, by intimidating someone for example, is not a reason that can justify your anger or in light of which anger can be felt. Affective reasons then do not concern the value of experiencing certain emotions.

What about binarity? Recall that one can occupy three distinct stances on one’s epistemic reason for (and against) P; belief in P, suspension of belief, and belief in not-P. Contrary to the epistemic case, the practical realm is characterized by binarity. One either acts (or intends to act) in response to one’s reasons, or one does not. We might think affective reasons are again in line with epistemic reasons in not being characterized by binarity, after all, if there is reason for anger there are more than two possible outcomes, one can: feel anger, not feel anger, or, one can feel a different emotion altogether, such as joy, for example. This suggests that affective reasons are analogous to epistemic reasons, as opposed to practical reasons, when it comes to binarity. Raz (1999) and Skorupski (2010) at least seem to think this is the case.

I think this reading does not survive scrutiny, however. The first worry with the view that affective reasons do not display binarity is that it is hard to characterize three distinct stances towards affective reasons for a given emotion. First, it is hard to make sense of what ‘suspension’ of feeling would involve. In suspending belief, one remains uncommitted to a belief in P, and one does not endorse P. By failing to feel anger, when one has reason to, one is not ‘suspending affect’ (if sense can be made of such a notion), one is simply not feeling anger. Second, there is also no clear analogue to believing not-P in the emotional realm. What is the negation of an emotion? Is it feeling an ‘opposing’ emotion or simply not feeling the emotion there is reason for? On the former option, while joy and sadness might seem to be natural opposites, it becomes much less clear whether emotion types come in opposing pairs once we consider other emotions. What would anger’s opposite be? Or guilt’s?

As each emotion type has its own narrow formal object, it seems that experiencing a different emotion, say joy, is not a stance that can be taken towards one’s reasons for anger. When concerned with reasons for a given emotion type, only considerations that count specifically in favour or against a particular narrow formal object are relevant. It is far from clear that reasons for joy count against anger. First, one can imagine an agent taking perverse pleasure in their rage, experiencing joyful anticipation of being able to seek vengeance following an offence, such that reasons for joy and reasons for anger could contribute towards a mixed joyful-angry state. In this case, joy and anger have the same intentional object and do not seem to be opposites. If, on the other hand, reasons for anger and reasons for joy concern different objects or states of affair (say you have just won an award and upon leaving the award ceremony find that someone has vandalized your car), then, even though the emotions seem to be in conflict, your reasons for anger and your reasons for joy do not cancel each other out. You might be overcome with rage when you see your car, but this does mean that there are no longer reasons for joy, and vice versa. When concerned with reasons for anger then, one either becomes angry or one does not. Similarly, when one is concerned with reasons for joy, one either becomes joyful or one does not. Experiencing one emotion when we are concerned with the reasons for another would arguably be more analogous to holding a separate belief, Q, rather than believing not-P or suspending belief. Attempts to make sense of a third stance to one’s affective reasons then seem to all collapse into the not experiencing a relevant emotion type, suggesting that affective reasons are binary. One either feels the emotion or does not, experiencing a second different emotion is not a way of responding to reasons for (or against) the first.²⁰

Provided one has affective reason for anger, one has reason to feel that an offence has occurred. By not feeling anger, one is not responding to a reason to feel that an offence has occurred. There doesn’t seem to be a third space to occupy in response to affective reasons for a given emotion type, one either responds to them, or doesn’t. This seems analogous to the practical case where one either acts, or does not act, on one’s reasons for a particular action. In the practical case, one can perform a distinct act, Θ, for which one has separate reasons without thereby violating binarity regarding one’s reasons to Φ. If one responded to reasons to Φ by Θ-ing, one is still not violating binarity, one is performing an action for which one has no reason. The same is plausibly true if one responds to reasons for anger with joy, in the absence of reasons for joy. Affective reasons, then, seem to demonstrate binarity.

Affective reasons seem to be irreducible to either practical or epistemic reasons. Emotions involve distinct formal objects to those involved in the epistemic or practical realm, such that affective reasons count in favour of distinct concerns from epistemic or practical reasons. Affective reasons are like epistemic reasons in that they do not concern the value of holding the relevant attitude, while they share with practical reasons the feature of binarity, as there seems to be no third stance one can take towards affective reasons. Reasons to feel then are unlike reasons to believe or reasons to act.

3. Conclusion

I have argued that a reasons-basic approach to emotional reason-responsiveness allows the sui generis reason-relation between emotions and their reasons to emerge. Agential Disposition accounts at best obscure (and at worst deny) any particularities regarding emotional reason-responsiveness as they cast emotions as mere reason-trackers (alongside perceptual, and other, systems) and construe reason-responsiveness as a domain general agential capacity that applies equally to the practical, epistemic and affective realm.²¹ A reasons-basic account of the sort I’ve sketched here allows us to investigate the normativity of distinct attitudes. It may also help provide more detailed explanations of the cases that Agential Disposition accounts struggled to accommodate – cases of outlaw emotion and inverse akrasia. Although this is a topic for future work, if emotional reason-responsiveness is sui generis, it is at least possible that sometimes there will be a reason for an emotion in cases where corresponding beliefs lack sufficient reason (perhaps because in aiming for truth epistemic reasons are outweighed more easily than affective reasons which count in favour of specific narrow formal objects) . This might help explain why reason-responsive emotions arise in cases of outlaw emotion and rational akrasia while corresponding beliefs often do not. If the details of my sketched account prove contestable, grappling with its inadequacies will, I hope, nonetheless prove fruitful in disentangling affective normativity. Delivering a compelling account of sui generis affective normativity may help vindicate, and, crucially, flesh-out, claims that emotions have a ‘logic of their own’, which although being rife in the emotion literature remain largely programmatic (de Sousa 1987; Greenspan 1988; Deonna and Teroni 2012).

A reasons-basic approach may also have costs that future work should weigh against its benefits. As far as I can tell, the main purported costs of a reasons-basic approach are the loss of those features that we might think underlie the virtue of Agential Disposition views, namely: that appeal to one’s rational or reflective capacities, even if dispositional, has alleged epistemic payoffs, as well as payoffs for our self-conception as rational agents. The epistemic payoff seems to be that while emotions and other tracking systems may go astray, on Agential Disposition accounts we have our reflective capacities to fall back on to provide some epistemic safety. Regarding our self-conception, Agential Disposition accounts manage to preserve what rationalist accounts delivered, what Jones (2003) calls ‘the normative conception of agency’. That is, on these accounts one’s all-things-considered judgement has normative authority, as reflective examination is the gold standard for reason-responsiveness and rationality. To say just a few words on these purported costs: first, appeal to reflective capacities is just one option available to secure epistemic safety, it is not the only option (see Korcz 2021, for example); second, the reasons-basic view might dispute that the normative conception of agency is really threatened on their view, or they might reject that it constitutes a loss to do away this conception of human agency in the first place.²² Indeed at its most revisionary, a reasons-basic approach might even cast the rejection of the normative conception of agency as an additional virtue of their view.

Acknowledgements

For valuable discussion and feedback, I would like to thank the audience at the Ethics of Emotion Conference in Bern, including Justin D’Arms, Roberto Keller, Sabine Döring, Julien Deonna and Fabrice Teroni (I am further grateful to the latter two for commenting on a previous version of the paper). I would also like to thank the editors of this special issue as well as an anonymous reviewer at this journal.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Recently, Maguire (2018) has argued that emotions are not had for reasons. He thinks emotion-supporting facts fall short of reason-hood because they cannot combine in support of emotions, nor provide stronger support for some emotions than others, nor affect each other. Maguire thinks emotions are subject to fittingness conditions, rather than reason-relations. His argument faces a number of problems (for a reply to Maguire see Faraci 2020), not least the underdetermination of the alternative normative standard that emotions are meant to be governed by. It is worth noting that Maguire’s working conception of reason-hood seems to exclude beliefs from being reason-responsive as well, which suggests he is starting from too narrow a conception of the reason-relation. While I disagree with Maguire’s claims that emotion-supporting facts cannot play any contributory or competition roles, I agree with his contention that facts that support one emotion do not cancel out facts that support distinct emotions. This is because, as we will see in Section 2, affective reasons count in favour of very narrow formal objects. Despite disagreeing on some crucial points, I believe Maguire’s argument is to a significant extent in line with my own as he can be read as characterizing a distinctive relation between emotions and the facts that support them. Unlike him, however, I do not deny that this is a reason-relation but rather take it to be sui generis one.

2 I restrict the reasons under discussion to normative reasons, by which I mean justifying or ‘the right sort’ of reasons. My notion ‘appropriateness’ is meant to track cases where emotions are responsive to the right sorts of reasons. By appropriate then, I mean what D’Arms and Jacobson (2000) mean by ‘fittingness’. These are notions meant to tease the right sorts of reasons apart from moral and prudential considerations that might count in favour of holding the relevant attitude, but do not speak in favour of the correctness of the attitude.

3 Cohen (2009) lists nine rules of rationality that include: conforming to the laws of deductive reasoning; properly forming theories from inductive cases; making inferences licensed by an accepted factual generalization; performing actions that further the purposes or interests of the agent; choosing the appropriate kinds of ends. See McHugh and Way (2018) for a critique of the rational requirements view of rationality.

4 I use the label ‘rationalist’ to refer to those accounts Kolodny (2005) calls ‘reductionist’, and ‘reasons-basic’ to refer to those accounts he calls ‘non-reductionist’. His nomenclature captures the idea that rationalist accounts explain reason-responsiveness in terms of rationality while reasons-basic accounts do not explain reason-responsiveness in terms of anything else, construing the reason-relation as primitive.

5 Note that it isn’t entirely clear what ‘reason-tracking’ involves on these accounts. As I understand them, Agential disposition accounts take reason-tracking to involve merely the detection of information, while reason-responding involves being guided, in practical and theoretical reasoning, by reasons. It is background capacities for reflective self-monitoring that, when well-functioning, turn mere information, or features of the environment, into reasons for agents.

6 Note that full-blown rationalist views may exclude emotions from the realm of reason. By tying reason-responsiveness constitutively to reflective norm-governed reasoning, emotions, which are not typically thought to be products or aids to such reasoning, are likely excluded. Agential disposition accounts then resemble what we might consider rationalist accounts that have been modified so as to account for emotional reason-responsiveness: by moving the relevant capacity to the dispositional level.

7 Outlaw emotions are recalcitrant emotions, as they conflict with the agent’s evaluative judgements, but they are not merely recalcitrant. In typical cases of recalcitrance emotion, such as fear of a dog that one believes is not dangerous, there need be only one belief that the emotion conflicts with. In outlaw emotion cases, although there is typically a belief with which the emotion conflicts, making the emotion recalcitrant, the emotion also stands in tension with a large set of further beliefs, often clashing with an agent’s wider belief system (See Silva 2021).

8 Arpaly (2018) makes a similar point in response to Tappolet’s (2016) account.

9 Tappolet (2018) provides a response to the sort of objection I have raise here. Tappolet takes the conflicting judgements in cases of outlaw emotion and rational akrasia to not be proper reasons to doubt one’s emotions are reason-tracking. This is because, Tappolet claims, these conflicting judgements are not justified. Elsewhere (Silva 2021), I have argued that Tappolet’s moves to deny that outlaw emotion cases involve defeaters fail. For now, it is sufficient to note that it isn’t clear that the conflicting judgements typical of outlaw emotion and rational akrasia cases are unjustified, nor that beliefs must be justified to act as defeaters.

10 Differences between the perceptual and the emotional ‘reason-tracking systems’ include the fact that emotions but not perceptions are thought to be rationally evaluable. Further differences include the dynamic nature of emotions as compared to perceptions, that is, occurrent emotions have a beginning, a middle and an end, they can wax and wane in intensity and they can bleed into emotional states of different types. We might think that these dynamic features of emotion affect their rational accessibility, which might suggest further differences between the emotional and the perceptual domain.

11 It is worth noting that this may well be a feature rather than a bug of such views, seeing as agential disposition accounts have often been proposed by philosophers who adhere to a perceptual theory of emotions (Tappolet being the clearest case). On perceptual accounts, emotions are thought to be, in some sense, perceptions of value, where a number of similarities between emotions and perceptions are thought to hold, including at the metaphysical, epistemic and phenomenological level. For emotions and perceptions to come out as similar with respect to their ‘reason-tracking’ roles then would arguably be desirable on a perceptual account. Perceptual accounts have been subject to a number of critiques (Berit and Elijah 2016; Brady 2010; Deonna, Tappolet, and Teroni 2015) as well as defences (Cowan 2016; Pelser 2014; Tappolet 2018) and remain a dominant view in the literature. Whether adherence to a perceptual view rules out a reasons-basic approach to emotional reason-responsiveness is orthogonal to the concerns of the current paper and a topic for future work. I am inclined to think that it does not, however, so long as emotions are not construed as literal perceptions (and perceptions are construed as non-reason-responsive states).

12 Skorupski (2010) is an exception as he argues for a tripartite account of irreducible reason types: practical, epistemic and evaluative. The latter of which is characteristic of the emotional realm.

13 See footnote 11.

14 In the emotion literature, the nature of reasons is typically not spelt out, yet they seem to more often than not be construed as mental states as opposed to facts. Here, I will follow reasons-basic accounts in construing reasons as facts for consistency and I am optimistic that my claims could be adequately modified to incorporate a mental state conception of reasons. It is worth noting that even on mental state conceptions of reasons, it is typically the content of mental states (often construed as propositions) that is thought to act as a reason, rather than the mental state itself. Similarly, factive accounts of reasons do not deny that mental states provide reasons.

15 I set aside other types of epistemic attitude for the purposes of the current paper, such that I am always concerned with beliefs when I refer to epistemic attitudes throughout.

16 I follow Raz (2011) in taking reasons for actions to be reasons for corresponding intentions. Additionally, I follow him (and a main line of thinking) in taking intentions to be distinct from judgements about what one ought to do. The latter are epistemic attitudes with practical content while intentions might be called practical attitudes.

17 Although some have argued that practical concerns sometimes also matter, see Schroeder (2012), I am leaving such cases aside to bring out the differences in epistemic and practical reason-responsiveness.

18 To be sure one can suspend judgement on what one ought to do, but this is an epistemic attitude, rather than an intention.

19 Salmela (2014) and de Sousa (2011) are exceptions to this trend. See Salmela (2014, Ch. 5) in particular for a defence of truth-aptness being the relevant assessment for emotions. Note that he makes distinctions between the type of truth that emotions aim for and the truth beliefs aim for. If an account of emotions as truth-apt can capture the relevant distinctions I highlight, then I am happy to grant that emotions aim for truth. This would, however, make the issue mostly a terminological one, as emotional truth-aptness would still bear differences to truth-aptness in the epistemic realm, and hence beliefs and emotions would not share the same formal object.

20 The suggestion is that reasons for distinct emotion types do not typically count against experiencing a given emotion type, not that there are no considerations that count against experiencing a given emotion type. One might have reason to think one is hallucinating a threat, or hear testimony that an offence was committed accidentally, for example, and these sorts of considerations will typically count against feeling fear or anger, respectively. This is because these considerations concern specifically whether the relevant formal object of fear or anger hold. Reasons that count in favour of anger (evidence of offences), however, do not count against feeling fear (which is concerned with dangers) nor vice versa, because the existence of an offence is no reason to think a danger does not also hold.

21 This is a claim about existing Agential Disposition accounts. Perhaps these views can be amended or supplemented to try to account for sui generis affective normativity. This is a task for my opponent, which would come in addition to the task of having to address the problem cases I outlined in Section 1.1.

22 See, for example, Kornblith (2012) who argues that our reflective self-examinations hold little to no normative or epistemic authority.