Bounded emotionality and our doxastic norms

ABSTRACT

In this paper I have two main aims. My first aim is to introduce the notion of bounded emotionality. This notion is the analogue of that of bounded rationality in behavioural economics. Bounded rationality says, roughly, that human beings are cognitively limited with respect to their processing and memory capacities. Bounded emotionality says that we are limited in our emotional capabilities, notably in the intensity, duration, and possible combinations of our emotional states. Bounded rationality is a foundational notion in behavioural economics and has played an important role in various fields of philosophy as well. In particular, in epistemology bounded rationality has been thought to have implications with respect to our doxastic norms. My second aim in this paper is to suggest that bounded emotionality also has an important role to play in philosophy. As a case in point, I discuss the implications that bounded emotionality has for our doxastic norms. In particular, I argue that one is not obligated to believe something or to stop holding a belief if one could not believe/fail to believe it without having an emotional response that exceeds one's bounded emotional capacities.

KEYWORDS:

• Bounded rationality
• bounded emotionality
• evidentialism
• epistemic normativity
• emotions
• information overload

1. Introduction

In this paper I have two main aims. My first aim is to introduce the notion of bounded emotionality. This notion is the analogue of that of bounded rationality in behavioural economics. Bounded rationality says, roughly, that human beings are cognitively limited with respect to their processing and memory capacities.¹ Bounded emotionality says that we are limited in our emotional capabilities, notably in the intensity, duration, and possible combinations of our emotional states. Bounded rationality is a foundational notion in behavioural economics and has played an important role in various fields of philosophy as well. In particular, in epistemology bounded rationality has been thought to have implications with respect to our doxastic norms. My second aim in this paper is to suggest that bounded emotionality also has an important role to play in philosophy. As a case in point, I discuss the implications that bounded emotionality has for our doxastic norms. In particular, I argue that one is not obligated to believe something or to stop holding a belief if one could not believe/fail to believe it without having an emotional response that exceeds one's bounded emotional capacities.

The plan of the paper is as follows. In §2, I introduce the notion of bounded emotionality by analogy with the well-known notion of bounded rationality from behavioural economics. In §3, I present two arguments for the claim that our doxastic norms are affected by our bounded rationality. In particular, I argue that bounded rationality requires us to reject two evidentialist norms, Obligation Evidentialism and Permission Evidentialism, according to which we ought or are permitted, respectively, to believe all and only those propositions that fit our evidence. In §4 and §5, I argue that doxastic norms are also affected by bounded emotionality. While the arguments, again against an evidentialist norm, parallel those in §3 that appeal to bounded rationality, they also differ in important respects. The argument in §4 additionally relies on the Could Closure principle, according to which emotional limitations entail limitations on what one could believe or could fail to believe. The argument in §5 relies the Emotional Normativity principle, which says that we are obligated to feel certain emotions in response to certain beliefs we hold. The argument in §5 furthermore faces an objection – the Epistemic vs. All-Things-Considered Normativity Objection – which I address there as well. I conclude by outlining the importance of taking bounded emotionality into account in philosophy and beyond.

2. Bounded emotionality

Information overload – that is, as I’ll understand it, having more information than we can process to form beliefs – is a common experience in the digital era in which we live. And this is largely due to the degree and ease of access we have to staggering amounts of information and misinformation. Information overload, however, is a condition suffered only by boundedly rational agents. Here by ‘bounded rationality’ I mean the descriptive claim originating in behavioural economics that we are cognitively limited with respect to our processing and memory capacities. Behavioural economists contrast boundedly rational agents, like Homo sapiens, with so-called ‘ideally rational’ agents, i.e. Homo economicus, who do not suffer from limited processing or memory capacities, and hence are not subject to information overload.²

In this paper I argue that we should recognize a parallel phenomenon involving limitations in our capabilities to emote, which I call bounded emotionality. Whereas bounded rationality is a claim concerning human cognitive limitations, bounded emotionality is a claim concerning human emotional limitations:

Bounded Emotionality. Human beings are limited in their emotional capabilities, notably in the intensity, duration, and possible combinations of their emotional states.³

Thus, just as bounded rationality limits the information we can process, bounded emotionality limits the information to which we can emote in response.⁴

Consider, for example, one prolific source of information that is available to many of us: news feeds. Digital news feeds are able to deliver an endless supply of information. And, given the negative bias of the news, the majority of this information will provoke negative emotions such as anger, anxiety, sadness, fear, and so on. Given our bounded rationality we cannot even process – that is, read, absorb, fact-check, etc. – all of the news in our news feeds, let alone all of the information from any and all available sources. But suppose that one of the news stories you end up attending to and processing is one about the surge in domestic abuse that is taking place during coronavirus lockdowns. By directing your attention to and reading this news story, you form beliefs about what happened. And then in response to these beliefs, you have certain emotional reactions – you might feel a mixture of anger, horror, and frustration. But, as you read more and more stories, the beliefs you form may quickly outstrip your capacity to form emotions. It may be that you could not achieve certain levels of emotional intensity, or feel certain emotions together simultaneously, or sustain certain emotions for extended periods of time. That is, it will generally be the case that our belief-forming capacities will outstrip our capacities to process and emotionally respond to our beliefs.⁵

In short, the point with respect to bounded rationality is that we generally cannot process all of the information available to us in the digital age in which we live. The point with respect to bounded emotionality is that we cannot emote in response to all of the beliefs we form in response to our information.

To further illustrate the point that our emotional capacities generally cannot keep up with our beliefs given the amount of information available to us, consider two rough but plausible constraints on our emotional responses. First, a set of emotions E should be fitting with respect to one's beliefs, where, roughly:

E is fitting with respect to a belief that p iff: p, if true, would be a sufficient reason to feel E.⁶

Secondly, E should be proportional with respect to one's beliefs, where, roughly:

E felt with degree of intensity i is proportional with respect to a belief that p iff: p, if true, would be a sufficient reason to feel E to degree i.

Some clarifications are in order here. Firstly, which emotions are fitting and proportional with respect to a given set of beliefs may depend on individual agents’ specific circumstances and vary from agent to agent. Secondly, it is open for debate whether there is some specific intensity of emotion which ought to be felt in response a given belief, as opposed to a range of permissible intensities. It's also debatable what shape proportionality takes. We can plausibly say that sadness will be fitting in response to, for example, the belief that a shelter dog was euthanised today. And a greater degree of sadness will be proportional if one holds the belief that two shelter dogs were euthanised today. Likewise, happiness will generally be fitting with respect to the belief that one has won $1,000,000,000, and a higher degree of happiness will generally be fitting with respect to the belief that one has won $2,000,000,000. By analogy with the principle of decreasing marginal utility, some might think that one's happiness (resp., sadness) should not be as much as twice higher in the latter case. It is debatable, however, whether an analogue of diminishing marginal utility applies to all emotions. For example, one might argue that we really ought to feel twice as sad about the fact that two children are going to bed hungry tonight as we ought to feel about one child going to bed hungry tonight. For the purposes of this paper, I do not need to take a stand on these questions. It is enough that some norms roughly like fittingness and proportionality hold.

Our emotions may also be subject additional constraints related to our beliefs beyond fittingness and proportionality. I focus on these because consideration of these constraints is helpful in illustrating why bounded emotionality has implications for doxastic normativity, as we will see.

In sum, bounded emotionality, like bounded rationality, is really a very simple notion. That human capacities to emote are not unlimited is a patent descriptive fact, just as it is manifestly true that human cognitive capacities are not unlimited. What will be of interest to philosophers are the consequences, and in particular the normative consequences, of the fact of bounded emotionality. Thus, I spend the majority and the remainder of this paper discussing the implications of our emotional boundedness for our doxastic norms. In the next section, I rehearse how one can reject certain evidentialist doxastic norms in light of bounded rationality. This will set the stage for me to introduce in the following sections two parallel, but interestingly distinct, arguments against an evidentialist doxastic norm in light of bounded emotionality.

3. Bounded rationality and our doxastic norms

In this section, to lay the groundwork for my arguments from bounded emotionality in §§4–5, I argue that the fact that we are boundedly rational – that is, that we are subject to cognitive limitations – has implications for our doxastic norms. I focus on two prima facie plausible evidentialist norms according to which one ought or is permitted to believe a proposition if and only if the proposition fits one's evidence. I present arguments that appeal to bounded rationality against both norms. The first argument relies on the claim that the agent could not hold certain beliefs, and the second relies on the claim that the agent could not fail to hold certain beliefs. I show that, given bounded rationality, these evidentialist norms conflict with even more intuitive claims.

3.1. Bounded rationality against obligation evidentialism

Here I present argument (I) against Obligation Evidentialism, according to which an agent A ought to believe a proposition p if and only if p fits A's evidence, from bounded rationality.

Consider the following scenario:

Solving Go

Alex is a novice Go player. She knows, however, that according to Zermelo's theorem, two-person, zero-sum games with perfect information like Go, chess and tic-tac-toe have solutions. That is, Alex knows that there is a theoretically determinable optimal strategy in Go consisting of an extensive form strategy formulated in terms of a conjunctive proposition – call this conjunctive proposition g – where for every stage in each theoretically possible game there is a conditional conjunct stating the optimal move for a given player. Alex has also read that the lower bound on the number of theoretically possible games of Go has been estimated to be more than a googolplex (${10}^{{10}^{100}}$) (Walraet and Tromp 2016). She therefore also knows that the complexity of g is such that no computer in existence let alone human could believe g. Likewise, Alex knows that it is not currently possible for any computer (or indeed human) in existence to determine g. In sum, Alex knows that g exists, but, given its length and complexity, it cannot be believed/cognized nor determined in practice by even the most powerful computers let alone human beings.⁷

Let A stand for Alex, and p for the proposition that strategy g is the optimal one. The argument then goes as follows:

(I)

1. Obligation Evidentialism: A ought to believe p iff p fits A's evidence.⁸

2. p fits A's evidence.

3. A could not believe p.⁹

4. Ought-Implies-Can (OIC): for every X, if A ought to X, then A could X.

Claims (1)-(4) are inconsistent. From (3) and (4) it follows that it's not the case that A is obligated to believe p. From that and (2) it follows that Obligation Evidentialism (1) is false.

In this case, (2) follows from the widespread assumption that entailment is a special case of fit. (3) is true in virtue of Alex's bounded rationality, and in particular because Alex could not determine g due to the complexity of the game and her cognitive limitations. And (4) is the familiar Ought-Implies-Can principle.

A proponent of Obligation Evidentialism might consider rejecting OIC in lieu of Obligation Evidentialism. Philosophers have offered various reasons to reject OIC not only in the doxastic context, but also in ethical and other contexts.¹⁰ But by and large, it is still widely accepted. Conee and Feldman have also argued with respect to epistemic justification that ‘[t]here is no basis for the premise that what is epistemically justified must be restricted to feasible doxastic alternatives’ (Conee and Feldman 2004, 87). Their evidentialist norm, however, is different from the evidentialist norms – Obligation Evidentialism and, in the next section, Permission Evidentialism – which we are discussing here. This is because the evidentialist norms we discuss here pick out the beliefs that we ought, are permitted, or ought not hold, and in particular claim that we either ought or are permitted to hold all and only those beliefs that fit our evidence. Conee and Feldman's evidentialist norm of justification, on the other hand, does not instruct anyone to believe anything, as they explicitly note (Conee and Feldman 2004, 86). Rather, it simply states a necessary and sufficient condition for doxastic justification. Thus, that a belief is justified does not necessarily mean that we ought to hold it.

I assume here that OIC is a more plausible principle than are the evidentialist norms under consideration. Much more could be said, of course, about the plausibility of OIC; but henceforth I put objections that OIC is less plausible than the evidentialist norms we consider aside.

Alternatively, one might object that Obligation Evidentialism wasn't really a plausible view in the first place, and so it's no real loss if we decide to reject it. Nelson (2010), for example, suggests that we have no positive obligations to hold any beliefs on purely epistemic grounds, although we are subject to negative obligations not to hold certain beliefs. In fact, even prominent defenders of ideal rationality norms, such as the norm of logical omniscience, grant that it's not the case that we actually ought to be ideally rational.¹¹ All that they claim is that ideal rationality norms are important because they are part of an account of what ordinary norms are. I have no quarrel with them on this point. I remain neutral here with respect to whether any account of doxastic norms of rationality for bounded agents needs to appeal to ideal rationality norms. All that matters for my purposes here is the thesis that bounded rationality constrains what we actually ought to believe.

In light of the above, however, I suggest we consider a much weaker doxastic norm than Obligation Evidentialism that I’ll call Permission Evidentialism. According to Permission Evidentialism, an agent is permitted to believe a proposition if and only if the proposition fits the agent's evidence.¹² Assuming that not being permitted to do something and being obligated not to do it are the same thing, Permission Evidentialism entails that we have negative obligations not to hold particular beliefs, but not that we have positive obligations to hold beliefs. I argue in the next section that we also ought to reject Permission Evidentialism in light of bounded rationality. Furthermore, since Permission Evidentialism is a weaker norm than Obligation Evidentialism, any arguments against Permission Evidentialism will a fortiori be arguments against Obligation Evidentialism. Therefore, in the sections that follow I focus on arguments against Permission Evidentialism – but parallel arguments can be constructed against Obligation Evidentialism in each case.

3.2. Bounded rationality against permission evidentialism

In this section, I present argument (II) against Permission Evidentialism from bounded rationality. Argument (II) also differs from argument (I) in that the relevant claim about bounded rationality is that we could not fail to hold certain beliefs.

Consider the following scenario:

Professor X's Trick Question

Professor X has decided to set a very difficult trick question on her next mathematics exam that will appear to all but the sharpest and most knowledgeable of her students to clearly entail a solution p, which is in fact false. The correct solution that is actually entailed by the evidence is ¬p. Nevertheless, Professor X knows that the majority of her students won't have the capacity to recognize that the question is a trick question, nor stop themselves from believing p upon considering the evidence. Scott, an average student, reads the question, breathes a sigh of relief at its apparent facileness, and comes to believe p.

Here, let A stand for Scott, who can't help but believe the wrong answer p to the trick question. The argument against Permission Evidentialism then goes as follows:

(II)

1. Permission Evidentialism: A is permitted to believe p iff p fits A's evidence.

2. p does not fit A's evidence.

3. A could not fail to believe p.

4. OIC: for every X, if A ought to X, then A could X.

Claims (1)-(4) are inconsistent. From (3) and (4) it follows that it's not the case that A is obligated not to believe p. From that and (2) it follows that Permission Evidentialism (1) is false.

An objector might question the truth of (2) on the ground that p fits A's evidence after all, even though the evidence actually entails ¬p. The evidence is misleading – it was a trick question, and Professor X intended for the majority of her students to be misled by their evidence into believing p. Both the objector and I can agree on the misleading nature of the evidence. According to the objector, however, the evidence is misleading in the following specific way: it fits a false belief in p.¹³

While it is not the place to adjudicate for or against various conceptions of evidence and evidential fit here, I can note that denying (2), that p does not fit A's evidence, is at least somewhat at odds with the ordinary use of these notions. For instance, it would be in order for Professor X to say ‘Look, your evidence doesn't show that p is true’. And in particular, the objector would have to reject the widely held assumption that if A's evidence entails p, then p fits A's evidence, as we have stipulated that the evidence provided by Professor X entails the correct answer.

An objector might also consider rejecting (3) in lieu of Permission Evidentialism in this case. (3), which relies on the thesis that we are boundedly rational, implies that there are some beliefs which some agents cannot help but hold, even in the face of contradictory evidence, due to their cognitive limitations. These beliefs may be recalcitrantly held in this way for various reasons – in §4.2 and §5.2, we’ll see cases in which an agent's emotions prevent them from abandoning particular beliefs they hold. In the case of Professor's X's Trick Question, however, what prevents Scott from not holding his belief, even though he is in possession of evidence which entails its falsity, are his cognitive limitations. I take it that it is plausible then that there are some beliefs that some agents could not abandon due to their bounded rationality, even when they are in fact in possession of contradictory evidence. And in particular, I take premise (3) to be more plausible than Permission Evidentialism.¹⁴

We now conclude this section on the implications of bounded rationality for our doxastic norms. These arguments are meant to be simpler analogues of the arguments in the next sections against Permission Evidentialism from bounded emotionality. Accordingly, I now turn to §4 where I argue that doxastic norms concerning what we are permitted to believe are affected not just by our cognitive limitations, but by our emotional limitations as well.

4. Bounded emotionality and our doxastic norms

In this section, I present an argument that is analogous to argument (II) from §3. There, I argued from bounded rationality against Permission Evidentialism: in particular, I argued that one is not obligated to abandon beliefs in propositions which don't fit one's evidence if one cognitively could not do so. The argument I present in this section, on the other hand, essentially relies on the thesis of bounded emotionality. More precisely, it relies on the claim that there are emotional bounds on our doxastic capabilities. The idea is that there are beliefs that we cannot have or cannot fail to have as a result of our emotional limitations. The argument also relies on what I call the Could Closure principle, according to which emotional limitations entail limitations on what one could believe or could fail to believe. I first discuss and defend the Could Closure principle in §4.1 before presenting the argument against Permission Evidentialism in §4.2.

4.1. The could closure principle

Argument (III) (as well as a similar argument that can be constructed against Obligation Evidentialism) relies on a pair of premises of the form (i), (ii) to infer a lemma of the form (iii), where A is an agent, p a proposition, and E a set of emotions:

1. A could not believe/fail to believe p without feeling E.

2. A could not feel E.

3. So, A could not believe/fail to believe p.

There are several candidates for the kinds of proposition p that an agent could not believe/fail to believe without having an emotional reaction exceeding their bounded emotional capacities. First, p could be such that an agent could not believe/fail to believe it alone without evoking an emotional response that exceeds their emotional capacities to feel. It's clear that there are propositions of this kind in general: propositions about personal experiences of torture, rape, murder, hate crimes, or even the genocide of one's ethnic community; or about extended physical, sexual, or emotional childhood abuse by once-trusted family members; or about egregious crimes that one has committed with severe and abiding repercussions, such as driving while intoxicated and consequently being involved in accidents leading to the deaths of beloved family members and other individuals; and so on.¹⁵ The foregoing propositions are all plausibly such that someone could be unable to believe/fail to believe them because doing so would require emotions beyond what they are able to feel.

Second, p could be such that, relative to a background emotional or doxastic state, the agent could not believe/fail to believe it without experiencing an additional emotional response in excess of the agent's emotional capacities. That is, it could be that the agent has formed or abandoned a number of beliefs, and in doing so exhausted their emotional resources. It is now the case that the agent could neither acquire nor abandon any further belief p without prompting an emotional response that the agent no longer has the emotional capacity to feel.

Third, p could be a conjunctive belief. For example, p could be of the form, ‘emotionally distressing event 1 occurred, and emotionally distressing event 2 occurred, and emotionally distressing event 3 occurred, and so on … ’, where the agent could not believe this conjunctive proposition without evoking an emotional reaction in excess of the agent's emotional capacities. Or, p could be of the form, ‘emotionally comforting event 1 occurred, and emotionally comforting event 2 occurred, and so on … ’, where the agent could not abandon belief in this conjunctive proposition without evoking an emotional reaction exceeding the agent's emotional capacities.

Alternatively, we could formulate our argument with weaker premises considering sets of beliefs instead: that is, there could be a set of beliefs p₁, … , p_n such that the agent could not bear holding or abandoning all of these beliefs. The inference would then take the form:

1. A could not believe/fail to believe all of p₁, … , p_n without feeling E.

2. A couldn't feel E.

3. So, A could not believe/fail to believe all of p₁, … , p_n.

For ease of exposition, I will assume that some instance of (i)-(iii) is true in each of the cases below, but the arguments could be recast with some instance of the weaker premises (i’)-(iii’) instead.

The inference (i)-(iii) involves claims about what A could do. The modality involved is restricted. The idea is not that it is metaphysically impossible for the agent to feel or believe certain things, but that it is not possible for them to do so given their abilities and circumstances. Similarly, when we say that an agent could not run 200 m in less than 10 seconds, we do not mean that it is metaphysically impossible for the agent to do so, but that it is not possible given their (current) physical constitution and circumstances. These are sometimes called ‘circumstantial’, ‘root’, or ‘ability modals’.¹⁶ They are vague and context-sensitive: it is not clear how much of an agent's current make-up is held fixed across the relevant possibilities, and perhaps what remains fixed can vary with context. But it is clear enough that for a suitable choice of E and p, these claims will hold of our scenario relative to any ordinary context. (A worry of equivocation arises when we introduce OIC as a further premise. I address this worry in the next section.)

Inference (i)-(iii) might be thought to face the following objection, however: (iii) does not follow from (i) and (ii). That is, the reasoning from (i) and (ii) to (iii) relies on something like the following principle:

Could Closure (CC): If A could not F without G-ing, and if A could not G, then A could not F.

Could Closure is a basic principle of modal logic. It relies on the intuitiveness of the following kind of claim: if Li could not enrol in a philosophy degree without paying some large sum of money m for tuition, and Li could not pay m, then Li could not enrol in a philosophy degree.¹⁷

With this preliminary in place, we now turn to the argument from bounded emotionality against Permission Evidentialism, with Could Closure.

4.2. Bounded emotionality against permission evidentialism, with could closure

Consider the following scenario:

Faith's Faith

Faith lives in a war-torn country wracked by injustice, disease and famine. Life is very hard for Faith, and she relies on her religious faith in order to survive. In particular, she clings to belief in p, that there is an afterlife with no more strife and with justice for those who have suffered injustice in this life. Unfortunately, perhaps in particular in light of the problem from evil, p does not fit Faith’s evidence. However, Faith could not abandon belief in p without feeling an extreme level of anguish and anger that she could not feel.¹⁸

Here let A be Faith, p be the proposition that there is an afterlife, and E be an extreme level of anguish and anger that Faith could not feel. Intuitively, it's not the case that Faith is not permitted to believe p – that is, that she's obligated to not believe in an afterlife. Our intuition in this case likely arises out of variety of factors, including a reluctance to interfere with personal religious belief, especially given the shielding role this belief plays with respect to Faith's mental and emotional wellbeing. Faith could not abandon her faith without feeling a certain extreme level of anguish and anger, etc., that Faith could not feel. The case suggests that even weak evidentialist norms like Permission Evidentialism are too demanding. More precisely:

(III)

1. Permission Evidentialism: A is permitted to believe p iff p fits A's evidence.

2. p does not fit A's evidence.

3. A could not fail to believe p without feeling E.

4. A could not feel E.

5. CC: If A could not F without G-ing, and if A could not G, then A could not F.

6. OIC: for every X, if A ought to X, then A could X.

Claims (1)-(6) are inconsistent. (4) is true in virtue of Faith's bounded emotionality. From (3), (4) and (5) it follows that A, Faith, couldn't abandon her belief in p. From that and (6) it follows that it's not the case that A is obligated to not believe p. From that and (2) it follows that Permission Evidentialism (1) is false.¹⁹

Argument (III) presented in this section relies on counterfactual links between one's emotions and one's beliefs. In the next section I present a different kind of argument that relies instead on normative connections between beliefs and emotions.

5. Bounded emotionality and our doxastic norms, with emotional normativity

In this section I present an additional argument against Permission Evidentialism. However, in lieu of Could Closure, I use the Emotional Normativity principle, according to which one ought to feel certain emotions in response to certain beliefs that one holds or fails to hold. I discuss and defend this principle briefly in §5.1, before presenting argument (IV) against Permission Evidentialism in §5.2.

5.1. The emotional normativity principle

The principle on which we’ll be relying in lieu of Could Closure in argument (IV) below is:

Emotional Normativity. One ought to feel certain emotions in response to either holding or failing to hold certain beliefs.

For example, one might be obligated to feel anger in response to the belief that one's friend was profiled and verbally abused by ignorant bystanders as part of the rise in anti-Chinese racism due to the COVID-19 pandemic. Or, one might be obligated to feel sadness and regret if one is forced to abandon one's belief in an esteemed authority figure's trustworthiness or moral character.

To be clear, I’m not claiming that all beliefs (or lack thereof) generate emotional obligations, i.e. obligations to feel certain emotions. The belief that ‘typewriter’ is the longest word that can be made using letters from a QWERTY keyboard's top row arguably does not, for instance. Nor do I wish to make detailed claims about the emotional responses that are obligated by beliefs (that is, beyond the fittingness and proportionality constraints I suggested in §2). I’m only claiming that some beliefs or lack thereof generate emotional obligations, and that some such beliefs can be used in the argument against Permission Evidentialism below.

There is relatively little discussion of emotional obligations in the philosophical literature. Lemaire (2014) considers normative principles along the lines of Emotional Normativity. In particular, he considers and rejects the possibility that there might be intrinsic, as opposed to extrinsic, norms for emotion. By ‘intrinsic’ norms here Lemaire means ‘correctness conditions’ of emotions qua emotions. And by ‘extrinsic’ norms for emotion he means those normative constraints that don't apply specifically to emotions qua emotions, such as prudential, moral, and aesthetic norms which apply just as much to actions and beliefs.

However, even if there has not been much explicit discussion of obligations to feel emotions and in particular to feel emotions in response to certain beliefs one holds or fails to hold, there has been much discussion of the rationality of emotions. The rationality of emotions has been cashed out in terms of correctness (as in Lemaire's discussion from above), as well as justification, fittingness, appropriateness, and so on. In particular, there is much discussion of the strategic or instrumental rationality of emotions, and more specifically of emoting or refraining from emoting in certain ways in given circumstances. For example, in a confrontation with a tiger, assuming one has as an end escaping and surviving, one might be thought to be subject to a means-end obligation to refrain from feeling paralyzing fear.²⁰ These instrumental norms support Emotional Normativity, as they suggest that certain beliefs make certain emotions rationally required, permitted or prohibited.

Some might nevertheless find Emotional Normativity to be problematic. In particular, some might think that the language of obligation is inappropriate with respect to emotion, just as one might think that it is inappropriate with respect to individual preferences.²¹ I will not undertake to convince them here. It is sufficient for my purposes to establish the conditional claim that, if Emotional Normativity is true, then bounded emotionality affects our doxastic norms in a further way than the one laid out in section §4. Let us now turn to this point.

5.2. Bounded emotionality against permission evidentialism, with emotional normativity

Consider the following scenario:

Sophie's Choice

Sophie believes that her son Jan survived the children's camp to which he was allegedly sent. Unfortunately, this belief is not supported by Sophie's evidence – she knows that most of the children sent to the camps during the war perished, and she has not heard from or about her son in all of the years since the war ended. But were Sophie to abandon her belief that her son survived, she would be obligated to feel a degree of grief, anger, and despair that she could not feel.

Intuitively, it's not the case that Sophie is not permitted to believe that her son survived – that is, it's not the case that she's obligated to abandon her belief that her son is alive. And this is plausibly at least in part because she could not feel what she ought to feel without this belief. This conflicts with Permission Evidentialism. More precisely, letting A stand for Sophie, p for the proposition that Jan survived the children's camp, and E for an extreme level of grief, anger, and despair that Sophie could not feel:

(IV)

1. Permission Evidentialism: A is permitted to believe p iff p fits A's evidence.

2. p does not fit A's evidence.

3. A is obligated to feel E if A does not believe p (by Emotional Normativity).

4. A could not feel E.

5. Deontic Detachment (DD): for every X and every Y, if A is obligated to X if Y, and if A is obligated to Y, A is obligated to X.

6. OIC: for every X, if A ought to X, then A could X.

Claims (1)-(6) are again inconsistent. From (3), (4) and (5) it follows that if A does not believe p, A is also obligated to do something that A could not do – namely, feel E. From that and (6) it follows that it's not the case that A is not permitted to – that is, is obligated not to – believe p. From that and (2) it follows that (1) Permission Evidentialism is false.

One might consider rejecting Emotional Normativity (3) in lieu of Permission Evidentialism. Again, I take it to be quite plausible that with respect to some belief-emotion pairings, we are subject to emotional obligations. But as I said, I will not attempt a full defence of Emotional Normativity here. It is philosophically significant already to establish that Emotional Normativity and Permission Evidentialism are incompatible. I note, however, that if one rejects Emotional Normativity in lieu of Permission Evidentialism, one will still have to contend with argument (III) against Permission Evidentialism from bounded emotionality in §4, which relies on perhaps less controversial assumptions about counterfactual conditionals and CC.

Deontic Detachment (5) is also subject to putative counterexamples. One counterexample posed by Broome (2013, 120) goes as follows: suppose you have entered a marathon. In this case, prudence requires you to train hard every day, and if you train hard every day, prudence also requires you to eat heartily. DD says that you are therefore required to eat heartily. But suppose now that, despite what prudence requires, you don't train hard every day. Broome then argues that you are not obligated to eat heartily – thus, DD fails.

This putative counterexample does not threaten our argument, however. And this is because in Broome's counterexample the antecedent of the conditional is not satisfied – i.e. you do not train hard every day. The present argument, however, only needs deontic detachment to hold in cases in which the relevant antecedent – in particular, believing a proposition or not believing a proposition – is fulfilled. If we restrict the principle to such cases it is much less controversial: assuming that you are obligated to A, and that you are obligated to B if you A, then, if you A, you are obligated to B. In Broome's case, you are obligated to train hard every day and to eat heartily if you train hard. If you do in fact train hard daily, it is clear that you are obligated to eat heartily.

In sum, I consider that the restricted version of DD, OIC and Emotional Normativity are each more plausible than the Permission Evidentialism principle, and I therefore suggest that we reject the latter.

5.3. The epistemic vs. All-things-considered normativity objection

In this section, I consider an objection to argument (IV) that I call the Epistemic vs. All-Things-Considered Normativity Objection. Here, the objector acknowledges that, all-things-considered, we can't consistently hold both Permission Evidentialism and the principle of Emotional Normativity. These two norms are inconsistent in that Sophie can't adhere to both together: Permission Evidentialism entails that Sophie is not permitted to believe p, and Emotional Normativity entails that Sophie ought to feel some set of emotions in response to her lack of belief in p that she in fact couldn't feel. Sophie could adhere to just Permission Evidentialism and abandon belief in p. And, when Sophie doesn't adhere to Permission Evidentialism and believes p, she could obey just the Emotional Normativity principle and feel what she ought to feel in response to the beliefs she holds. So, Sophie can obey one or the other norm by itself, but she can't obey both together, because obeying each norm prevents her from obeying the other.

But, the objector says, I am only interested in what we ought to believe epistemically speaking. And Emotional Normativity is not relevant with respect to what Sophie ought to believe epistemically speaking. Rather, it is relevant with respect to what Sophie ought to believe all-things-considered. There are no inconsistent norms in argument (IV) concerning what we ought to believe from an epistemic point of view – that is, there are no inconsistent epistemic norms. Clearly, Sophie is epistemically obligated not to believe p given her evidence. Nevertheless, the objector continues, I am happy to concede that all-things-considered, under some broader conception of doxastic permission, obligation, normativity, etc., Sophie is permitted to believe p.

More specifically, from this all-things-considered perspective on these norms, we can recognize that Permission Evidentialism and Emotional Normativity are inconsistent in that we can't obey both together. But upon recognizing this inconsistency, we could claim that one norm has priority over the other – i.e. Permission Evidentialism over Emotional Normativity or vice versa. That is, we could prioritise the ‘epistemic’ evidentialist norm over the non-epistemic Emotional Normativity principle. In this case, in those scenarios in which one is obligated not to believe a proposition that doesn't fit one's evidence, and where not believing that proposition obligates one to feel more than one could feel, the epistemic prioritiser would say that the norm to believe according to one's evidence takes precedence over the norm to emote. Or, we could prioritise the non-epistemic Emotional Normativity norm over the epistemic evidentialist norm. Then, in those scenarios in which one is obligated to feel more than one could feel in response to a lack of belief that one is epistemically obligated to abandon, one ought to prioritise the obligation to feel and maintain the belief. Or, the objector might remain agnostic about whether we ought in general prioritise epistemic over non-epistemic norms, or vice versa.

The point is, however, that there is no normative inconsistency in the sense in which our objector is interested – i.e. in the epistemic sense. So at most what argument (IV) shows is that there could be inconsistent norms concerning what we ought not/are permitted to believe all-things-considered. But this is an uncontroversial and uninteresting thesis, says the objector. Epistemologists have long distinguished between epistemic and non-epistemic forms of justification, rationality, reasons, etc. And many will be happy to say in cases like Sophie's that what one may be epistemically justified, rational, etc., in believing may differ from what one may be pragmatically, or morally, or all-things-considered justified, rational, etc., in believing.²² Thus, in a case in which an eccentric billionaire offers you $1 million to believe some prudentially innocuous proposition q which is disconfirmed by your evidence, most epistemologists would be happy to say that you epistemically ought not believe q, but that in some non-epistemic or non-purely epistemic sense – e.g. prudentially or all-things-considered, etc. – you ought to believe q.²³

5.4. Response

Let me first reiterate that The Epistemic vs. All-Things-Considered Normativity Objection only applies to argument (IV), which relies on the Emotional Normativity principle, and not (III) which relies instead on Could Closure.

That being said, the objection raises broader questions that cannot be addressed fully here. But let me sketch one line of response.

The objection relies on various assumptions about epistemic and all-things-considered normativity to which various philosophers have presented serious objections, and many of which I would reject (again, for reasons I don't have the space to fully lay out here). Firstly, the objection assumes that epistemic and all-things-considered normativity can diverge, so that, for example, one may not be epistemically permitted to believe some proposition, but be all-things-considered obligated to believe that proposition. Mills (1998), however, argues that we should reject this ‘divergence thesis’ on the basis that our acceptance of it is due to a conflation of the assessment of acts which cause one to believe with the assessment of believing itself.²⁴

Secondly, in contrasting epistemic normativity with all-things-considered normativity, the objection assumes that epistemic and non-epistemic norms are commensurable, that is, that they can be weighed against one another to determine what one all-things-considered ought/is permitted to do. In doing so, they assume the falsity of the Incommensurability Thesis, according to which there is no meaningful question about whether epistemic oughts trump or are trumped by other oughts (Feldman 2000, 694–95; Kelly 2003, 619). Kelly and Feldman suggest, however, that the view that epistemic and non-epistemic norms of belief are incommensurable is the natural view. While I don't share this intuition, and don't have a settled opinion about the truth or falsity of the Incommensurability Thesis, I think it is at least worthwhile noting that the objector makes the non-trivial assumption that the Incommensurability Thesis is false, when it is taken by thoughtful philosophers like Kelly and Feldman to be both true and the natural view.

Thirdly, varieties of pragmatic and moral encroachment have been proposed that suggest that it is the epistemic status of beliefs that is encroached upon by moral and pragmatic considerations. Moss (2018, 177–9), for example, defines ‘moral encroachment’ as the thesis that the ‘epistemic status of [a belief] can depend on its moral features’, specifying that a belief's ‘epistemic status’ includes but is not limited to whether a belief is justified, constitutes knowledge, and so on. Worsnip (2020, 3) likewise discusses varieties of pragmatic encroachment in which ‘pragmatic considerations make a difference to what one ought to believe tout court via making a difference to what one epistemically ought to believe, or is justified in believing’. And so, those friendly to encroachment might think that there is an additional, analogous form of ‘affective encroachment’, where the epistemic status of beliefs can depend on considerations related to our bounded emotional capacities.

Fourthly and finally, the objection assumes that there is such a thing as epistemic normativity in the first place, where epistemic normativity is understood as a kind of normativity distinctive to beliefs. Although this assumption is a widely held one, it is not without opponents. Papineau (2013) argues against precisely this claim, and for the thesis that the only kinds of normativity to which our beliefs are subject are kinds with which we are otherwise familiar – moral, prudential, aesthetic, etc. While it is beyond the scope of this paper to offer a full defense of this kind of broadly pragmatist view, I am sympathetic to the notion that we ought to discard any notion of the ‘epistemic’ as a descriptive or normative dimension distinctive of belief; and that we ought instead to consider and assess beliefs in the same way in which we assess non-doxastic attitudes, states, actions, etc.²⁵ Relatedly, Cohen (2016) argues that we have good reason to question the status of the notion of the ‘epistemic’. In particular, he argues that ‘epistemic’ functions largely as an undefined technical term, and more specifically that its meaning is not clear enough to make sense of many disputes concerning so-called ‘epistemic’ notions.²⁶

6. Conclusion

In this paper, I had two aims: the first was to introduce the notion of bounded emotionality, and the second was to suggest that bounded emotionality has an important role to play in philosophy, and in particular epistemology, and beyond. More specifically, I argued that our doxastic norms ought to take into account our emotional limitations as well as our cognitive limitations. Thus, we are not obligated to hold or abandon beliefs when we could not do so without exceeding our emotional capacities.

I suggest that we should continue to explore the relevance of bounded emotionality in epistemology and in other philosophical fields where the notion of bounded rationality has proved fruitful, such as ethics. For example, philosophers have used bounded rationality to defend satisficing doxastic norms over optimizing ones.²⁷ Satisficing doxastic norms could likewise be defended using bounded emotionality. With respect to moral norms, ethicists have considered whether cognitive limitations can excuse immoral acts or change the ethical standards against which we judge persons and their actions.²⁸ We might consider the ethical implications of emotional limitations along the same lines. For example, one might prima facie think that one is subject to moral obligations to engage in myriad altruistic activities for innumerable worthy causes – contributing financially to various charities, donating to food banks, volunteering one's time and expertise to look after vulnerable, underserved and disadvantaged communities. But one might also know that doing so will in some cases involve placing oneself in emotionally overwhelming circumstances. It would be reasonable to think that one's bounded emotionality ought to be taken into account with respect to one's apparent moral obligations. The prospect of exceeding one's emotional capacities in carrying out some moral act may be relevant in determining whether or not that act is in fact obligatory or even permitted.

Beyond philosophy, I suggest that bounded emotionality may have significant implications for economics, psychology, and other social sciences. For example, bounded rationality has enabled behavioural economists, who work at the intersection of economics and psychology, to more accurately describe, predict, and explain human decision-making. It has also better enabled them to intervene to foster more optimal decision-making – i.e. to ‘nudge’ people.²⁹ Behavioural economists would also do well to take into account our bounded emotionality in predicting, describing, and intervening in human decision-making, because our emotions and our emotional limitations affect our behaviour and restrict our choice options just as our intellect and cognitive limitations do. And some decisions that may be optimal or rational when we don't take emotional limitations into account will be suboptimal and even self-destructive if we take agents’ emotional limitations into account.

For example, turnover rates in child welfare and social services are extremely high due to phenomena like compassion fatigue, which Newell and MacNeil (2010) define as ‘the overall experience of emotional and physical fatigue that social service professionals experience due to chronic use of empathy when treating patients who are suffering in some way’; and burnout, which Maslach and Jackson (1981) define as ‘a syndrome of emotional exhaustion and cynicism that occurs frequently among individuals who do “people-work” of some kind’.³⁰

This is all to suggest that it would be well-worth further exploring the normative and descriptive implications of bounded emotionality in philosophy, economics, psychology, and beyond, as well as in everyday life. I have argued in this paper that bounded emotionality has important implications for philosophy. But we might also see bounded emotionality as a bridge between normative philosophy and these various empirical social sciences. It is becoming increasingly obvious, especially in this era of information overload, that bounded emotionality will have to play an important role in future debates.

Disclosure statement

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

Notes

1 Herbert Simon introduced the term ‘bounded rationality’ (see Simon 1957).

2 This terminology – ‘ideal rationality’, ‘ideally rational’, etc. – is the terminology used by economists, and in particular behavioural economists, to describe the cognitive capabilities of Homo economicus (a.k.a. ‘Econs’) rather than Homo sapiens (Thaler 2015). My use of these terms here, and in particular my use of the term ‘ideal’, should not be taken to imply that Homo sapiens ought to be ideally rational, or that we are necessarily subject to the same doxastic standards as are Homo economicus.

3 Note that the term ‘bounded emotionality’ is used with another meaning in communication studies, namely to denote the use of a range of emotions in an organization within bounded limits in terms of what is acceptable (see, e.g., Jeanes 2019). Thanks to an anonymous referee for pointing out this alternative usage of the term.

4 The emotional capacities of Homo economicus are not considered in the relevant economic and philosophical literature, as far as I’m aware. And so it wouldn't be strictly correct to say that Homo economicus are ideally emotional just as they are ideally rational, and compare this with the bounded emotionality and bounded rationality of Homo sapiens. In principle, however, it could be helpful to think of Homo economicus as having ideal emotional capacities in contrast to our bounded emotional capacities.

5 This emotional overload that piggybacks on information overload is referred to nowadays by such terms as ‘feed fatigue’, ‘social media fatigue’, ‘doomscrolling’, and ‘doomsurfing’.

6 This is a rough formulation. A refined fittingness principle would need to keep fixed relevant background factors. Take the case in which p is the true proposition that Florian has broken his promise to Ai. Prima facie, p is sufficient reason for Ai to feel E, anger and disappointment. But let's say that Florian was prevented from keeping his promise because he was in an incapacitating car accident. In this scenario, p no longer seems like sufficient reason for Ai to feel E. See, e.g., Schroeder (2009) for recent work on objective and subjective reasons, which may be helpful here in filling out this fittingness principle. These issues are orthogonal to my main argument, however, so we can leave them aside. Thanks to Clayton Littlejohn for discussion of this point.

7 Many thanks to Julien Dutant for helpful discussion of this case.

8 I use the terminology of ‘ought’ and ‘obligation’ interchangeably in this paper – i.e., I take it that one is obligated to X iff one ought to X.

9 Or at least could not believe ‘responsibly’ where by this I mean based on or in a manner suitably sensitive to the evidence. That is, presumably our evidentialist obligations and permissions are not merely obligations/permissions to form beliefs that fit the evidence, but to do so based on the evidence. In some cases, as in the Solving Go case, the cognitive complexity of the believed proposition p will be the reason an individual could not believe p. In other cases, p may not be too cognitively complex for a doxastic agent to believe, but it may be that believing p responsibly is beyond the cognitive capacities of the agent in question. I leave out the caveat that beliefs need to be formed responsibly according to the various evidentialist norms I will consider here for simplicity.

10 See King (2017) for a helpful overview of objections to OIC.

11 See, e.g., Smithies (2015, 2780–1).

12 Note that Permission Evidentialism isn't inconsistent with having obligations to believe – it just doesn't entail that we have them. One could combine Permission Evidentialism with another obligationist evidentialist view according to which, with respect to certain matters, we are in addition obligated to believe some things that fit our evidence.

13 Thanks very much to an anonymous referee for raising this objection.

14 Alternatively, we could replace (3) with the claim that A couldn't believe p responsibly, and (1) with the corresponding refinement of Permission Evidentialism. See footnote 9.

15 An anonymous referee helpfully suggests that we may want to distinguish cases in which our bounded emotional capacities are quantitatively exceeded (exemplified by the doomscrolling case, in which a given agent could not emote in response to believing a series of propositions, etc., even over an extended period of time) from those in which our emotional capacities are more qualitatively exceeded (exemplified by the case ‘Faith's Faith’ in §4.2, in which Faith could not emote in response to abandoning a particular belief due to the intensity of the emotional response involved, or the impossibility of having a certain combination of emotions, etc.). The referee doubts whether there are in fact such qualitative cases, and suggests that whether there are is an empirical question.

For my purposes, it is sufficient for my core arguments if we accept that there could be such quantitative-type cases. But I also take the qualitative cases I’ve proposed to be convincing. For example, if we accept that agents’ emotional capacities can be exceeded in the more quantitative doomscrolling-type cases in virtue of the amount of time that it would take to emote in response to a series of propositions, I take it that we could imagine qualitative cases involving a proposition in response to (the abandoning of) which a given agent could not emote due to the intensity or complexity, etc., of the emotional response involved, even over an extended period of time. But I am happy to concede that whether there are cases of the more qualitative kind is ultimately an empirical question. Many thanks to the anonymous referee for encouraging me to consider this point.

16 See especially Kratzer (2012), as well as Hegarty (2019) and Portner (2010).

17 Thanks very much to both an anonymous referee and Julien Dutant for extensive feedback and helpful discussion of many of the points in this and the following section.

18 Compare this case with a story recounted by philosophy professor and atheist Stephen Asma in his book Why we need religion:

One day, after pompously lecturing a class of undergraduates about the incoherence of monotheism, I was approached by a shy student. He nervously stuttered through a heartbreaking story, one that slowly unraveled my own convictions and assumptions about religion. Five years ago, he explained, his older teenage brother had been brutally stabbed to death, viciously attacked and mutilated by a perpetrator who was never caught. My student, his mother and his sister were shattered. His mother suffered a mental breakdown soon afterward and would have been institutionalized if not for the fact that she expected to see her slain son again, to be reunited with him in the afterlife where she was certain his body would be made whole. These bolstering beliefs, along with the church rituals she engaged in after her son's murder, dragged her back from the brink of debilitating sorrow, and gave her the strength to continue raising her other two children — my student and his sister (Asma 2018, 3).

Given the way Asma's story is framed, it's not clearly the case that the mother could not feel the emotions that she couldn't help but feel if she were to abandon her religious belief. Rather, it seems to be the case that she could not abandon her religious beliefs without having an extreme emotional response that would have debilitated her and essentially caused her to self-destruct. In our case, Faith straightforwardly lacks the emotional capacity to have some emotional response that she could not fail to have if she were to fail to believe p. And therefore, as I argue in the next section, OIC applies. In Asma's story, on the other hand, OIC does not seem to apply because the mother does not straightforwardly lack the emotional capacity to have the emotional reaction that she could not help but have without her faith, but rather could not feel the relevant emotions without self-destructing.Thanks to anonymous referee for encouraging me to distinguish between cases involving straightforward lack of emotional capacity to feel certain emotions (to which the OIC principle will apply), and more complex cases involving a lack of emotional capacity to feel certain emotions without self-destructing or violating some other moral or practical norm of self-preservation (to which the OIC principle as I’ve formulated it does not straightforwardly apply). While the implications of bounded emotionality for our doxastic norms in the latter kind of case are of course well worth exploring further, it is only the former kind of case involving straightforward lack of emotional capacity that I am concerned with in this paper.

19 An objector might suggest an alternative approach in which our bounded emotionality and bounded rationality are not relevant with respect to our doxastic norms – e.g., whether or not Permission Evidentialism or Obligation Evidentialism are true – but with respect to whether or not we are blameworthy when we violate such doxastic norms. So, for example, let's assume that Permission Evidentialism is true, and that Faith violates this doxastic norm by believing p. It's also still the case that Faith could not feel what she could not help but feel if she were to abandon her belief in p, given her bounded emotionality. This fact does not falsify Permission Evidentialism, however, but rather excuses Faith in believing p, so that even though Faith violates a doxastic norm, she is not blameworthy for this violation.

It is common to distinguish between blameworthiness and obligation violation in both ethics and epistemology, such that one can be both morally or doxastically obligated to x but not morally or doxastically blameworthy for failing to x (see, e.g., Alvarez and Littlejohn 2017; Greco 2021; Littlejohn forthcoming; and Biebel 2018). According to this alternative approach, Faith's case is analogous to that of a kleptomaniac who could not keep themselves from stealing. Intuitively, it's true both that the kleptomaniac ought not steal, and also that they are not blameworthy for stealing precisely because they lack the capacity to keep themselves from stealing.

This alternative approach definitely has its merits and relies on the well-established philosophical distinction between blameworthiness and obligation violation. While I do not have a conclusive objection against it, I would note that my approach provides an arguably simpler and more direct answer to the question of what we ought or are permitted to believe. Furthermore, some might share my intuition that there is something more positive to be said for the boundedly appropriate beliefs associated with my approach than that they are merely excusable. For example, some might think that we should say something more positive about Faith's religious belief in p than that it is a kind of excusable doxastic dysfunctioning. We should say, rather, that Faith is permitted or even obligated to believe p given her circumstances. Many thanks to an anonymous referee for suggesting this alternative approach.

20 See, e.g., De Sousa (1987, chap. 6) on the strategic rationality of emotion.

21 Many of us will be familiar with the notion that our feelings are not subject to normative regulation. One common thought, for instance, is that we shouldn't feel guilty if we fail to experience grief after the death of a loved one, and that may suggest that we don't have an obligation to feel so. Likewise, a common thought is that when it comes to individual preferences – what ice-cream flavour, colour palette, species of pet one prefers –, or one's romantic and sexual preferences – whom we love or are attracted to – we are not under any obligation either.

22 See Mills (1998), who says that what he calls the divergence thesis, according to which epistemic justification can diverge from practical or all-things-considered justification, is widely accepted. For example, Pollock writes:

A justified belief is one that it is ‘epistemically permissible’ to hold … .Epistemic permissibility must be distinguished from both moral and prudential permissibility. … .For instance, it is popularly alleged that lobsters do not feel pain when they are dunked alive into boiling water. It is extremely doubtful that anyone has good reason to believe that, but it may be prudentially rational to hold that belief because otherwise one would deprive oneself of the gustatory delight of eating boiled lobsters (Pollock 1986, 7–8).

See also, e.g., Davidson (1982), Heil (1992), and Pears (1984), for examples of authors who accept the divergence thesis and have written extensively about how to solve puzzles to which it gives rise.

23 The Epistemic vs. All-Things-Considered Normativity Objection also applies to a parallel argument against Obligation Evidentialism from bounded emotionality and the Emotional Normativity principle. At the heart of the objection is a distinction between epistemic and all-things-considered normativity. And so the objection will apply to with respect to any ‘epistemic’ norms we care to discuss, whenever there is an inconsistency between the demands of the epistemic and the non-epistemic norms on our beliefs.

24 See also Basu and Schroeder (2019) on a further worry about the relationship between epistemic and non-epistemic normativity which they call ‘the problem of coordination’.

25 Those who find Papineau's outright rejection of the notion of epistemic normativity too radical may find Rinard (2015; 2019) and Maguire and Woods’ (2020) pragmatist but ‘epistemic’-friendly views about doxastic norms more palatable. According to their views, there is (or may be) such a thing as epistemic normativity, but all legitimate or authoritative reasons for belief will be pragmatic ones.

26 I use the term ‘doxastic’ norm rather than ‘epistemic’ norm to describe the norms of belief in this paper because of these sorts of points about the legitimacy of the ‘epistemic’.

Importantly, however, rejecting epistemic normativity does not undermine this paper's discussion of our doxastic norms. That is, even if there is no such thing as a distinctively epistemic normativity, it will still be worth asking what we ought or are permitted to believe.

27 See Stirling, Goodrich, and Packard (2002) and Grüne-Yanoff (2007).

28 See Gigerenzer (2010), Rudy-Hiller (2018), and Zimmerman (1997), among others.

29 See e.g., Thaler and Sunsteinn (2009, 170–4) for a well-known illustration of nudging.

30 A related phenomenon recently discussed in relation to the COVID-19 pandemic is ‘societal burnout’, which Smullens (2020) defines as ‘a state of being in which we are overwhelmed, overburdened, and overloaded by grave, threatening societal problems.’