Stubborn emotions, stubborn beliefs

Abstract

Recalcitrant emotions, such as fear of flying, are emotions that persist even though they are in tension with the emoter’s considered belief. A widely accepted argument against cognitivist theories of emotion holds that recalcitrant emotions show that emotions are more like sensory states than like thoughts or beliefs. I show that this argument does not succeed: Emotions are usually sensitive to our changing beliefs in a way that is more akin to cognitive states than to sensations. Moreover, empirical evidence strongly suggests that beliefs sometimes behave much like recalcitrant emotions do.

1 Introduction

Imagine that you, a person in possession of a normal degree of rationality, see your neighbor walking their dog, Fido, on the other side of the street. You know Fido and believe him to be very gentle. His owner has given assurances that he is quite friendly, and Fido has never demonstrated any aggression in your past meetings. Your considered belief—the belief you could provide reasons for and which you would stand behind—is that Fido is not dangerous. Nevertheless, upon seeing the dog you are frightened. You are so afraid that you begin to tremble and even cross the street to avoid him. This is the familiar phenomenon of stubborn or ‘recalcitrant’ emotion.

Recalcitrant emotions highlight the ways that emotions seem to be beyond our rational control, come unbidden, have us at their mercy. You are afraid of the dog—afraid that the dog will bite you, perhaps—while you simultaneously believe that the dog is gentle. These two mental states, your emotion and your belief, seem to present the world in two incompatible ways. And your thought does not change your fear, which persists even in the face of your belief that the dog is gentle and nonthreatening. Given the rational nature of belief, if you can think that the dog is not dangerous while feeling afraid of the dog, then emotions and beliefs must be two completely different types of mental states.

It is compelling and intuitive to think that recalcitrant emotions drive a wedge between emotion and cognitive states and show that emotions are not even partly composed of thoughts or beliefs. This is important, because it might otherwise seem attractive to explain the intentionality of emotion by saying that emotions at least partly consist in thoughts or beliefs. Like these paradigmatically cognitive states, emotions are about objects and states of affairs that may be distant in both time and place from the emoter: You are happy that you are visiting friends, sad that your favorite bookstore has closed. So, emotions look like they are ways of relating to a structured content, just like mental attitudes such as belief are. Cognitivist views of emotion can account for this easily: According to these views, emotions either are judgements or beliefs, or else have these cognitive states as constituent parts.¹ However, recalcitrant emotions that persist in spite of your considered beliefs seem to reveal that emotions do not even have beliefs as constituent parts, since this recalcitrance seems to many to be incompatible with belief. Some version of this argument is widely endorsed as decisive against cognitivist views of emotion.²

In spite of its intuitive power, the argument from recalcitrance has proven difficult to pin down precisely.³ One important and promising formulation of this argument draws an analogy with perceptual illusions.⁴ However, a more nuanced picture of belief has emerged from empirical work, revealing that many of our intuitions about rationality and belief are likely idealized and mistaken. Drawing on this work, I argue that the formulation of the argument from recalcitrance which draws a close analogy with perceptual illusion fails because it relies on an incorrect picture of belief. Emotions may turn out to be among, or to have as constituent parts, the most paradigmatically cognitive mental states we have, after all.

The paper is organized as follows. Section 2 sets out the first formulation of the argument from recalcitrance via analogy with perceptual illusion. Section 3 draws upon empirical work and recent discussion thereof to show that beliefs can exhibit recalcitrance, too, and hence that the encapsulation formulation of the argument from recalcitrance fails. Section 4 considers and responds to two potential objections to the claim that beliefs are sometimes recalcitrant. In Sect. 5, I consider a different sort of argument that recalcitrance poses a problem for cognitivism, before raising several considerations against that line.

2 Recalcitrance as evidence of encapsulation

Recalcitrant emotions look a good deal like perceptual illusions. When you are driving on the highway and seem to see a large puddle in the distance in the middle of the road—even when you know that it is an optical illusion due to heat rising off the pavement—your belief that the puddle is illusory does not make the visual appearance of the puddle go away. It still looks to you as if there is a puddle there, even though you know better. Recalcitrant emotions seem very much like your visual perception of the illusory puddle: In recalcitrant emotion, as in known perceptual illusions, one of your mental states continues to present the world one way in spite of conflicting with another mental state that presents the world differently. And so, it may seem natural that analogies with recalcitrant perceptual illusions such as the classic Müller-Lyer illusion are everywhere in discussions of emotional recalcitrance.⁵

Persistent perceptual illusions may be a reason for thinking that perception and cognition are distinct mental systems. The illusion above persists even when you believe that the puddle doesn’t exist or that the lines in the figure are the same length, and so the perception must not be able to access the information that belief is drawing upon: your perceptual state is informationally encapsulated from your belief to the contrary. If one processes is encapsulated from another, Process A cannot access information stored in Process B. Informational encapsulation has seemed to some a good reason for positing that two processes belong to distinct mental systems (Fodor, 1983, p. 69).

It appears that recalcitrant emotions likewise show that emotion is or belongs to a distinct mental system from cognition: The persistence of some emotions in the face of our considered beliefs to the contrary looks like evidence that emotions, like perceptions, are encapsulated from cognition. Several authors in the literature appear to have an argument along these lines in mind when they mention persistent perceptual illusions in connection with recalcitrant emotions, even though most do not spell it out.⁶

Accordingly, this formulation would be: Recalcitrant emotions reveal that emotions are informationally encapsulated from cognition; mental state A’s being encapsulated from mental state B provides a good reason for positing that A and B belong to distinct systems; so recalcitrant emotions give us a good reason to think that emotion is a distinct system from cognition, and thus that emotions do not have beliefs as even necessary parts. Each form of cognitivism about emotion holds that belief is at least a necessary part of emotion, and so cognitivism is false.

Let’s spell this formulation out explicitly:

The Encapsulation Formulation of the Argument from Recalcitrance:

• 1. Recalcitrant emotions show that emotions are encapsulated from cognition (information available to cognition is not available to emotion).

• 2. If a mental state or process is encapsulated from cognition, that mental state is not cognitive.

• 3. (2) is incompatible with the claim that beliefs are at least necessary parts of emotions.

• 4. Thus, cognitivism about emotions is false.

This formulation has several advantages over other formulations of the argument that focus on the degree to which having recalcitrant emotions is irrational. First, the encapsulation formulation makes the notion of recalcitrance precise. Moreover, this formulation makes it much clearer what is meant by ‘cognitive:’ It is a minimal criterion for a state to be cognitive that it is not encapsulated from cognition.⁷

This argument faces two problems. First, not all token emotions are recalcitrant; occasional recalcitrance doesn’t show that emotions as a system are encapsulated from cognition. Second, beliefs exhibit stubbornness, too. If stubbornness is evidence of encapsulation, then at least some beliefs are encapsulated. I explain these points in the following section.

3 Two problems for recalcitrance as evidence of encapsulation

3.1 Many instances of emotions are sensitive to changing beliefs

Emotions are sometimes recalcitrant,⁸ but they aren’t always recalcitrant. We would be very different beings if all—or if even most—of our emotions were recalcitrant. Instead, emotions often change in light of changing beliefs in a way that suggests that many instances of emotion are sensitive to belief. If every instance of emotion were recalcitrant, you would not ever transition from fear to relief upon learning that you passed a crucial exam that you had antecedently believed you had failed. As Solomon notes, if I’m angry that you’ve stolen my car but then learn that my car has not been stolen after all, my anger will sometimes persist, but in many cases will dissipate (1976, p. 119). It might even be replaced with relief if, for instance, my discovery of the car’s location is the way I learn that my car hasn’t been stolen. If all occurrences of emotions were recalcitrant, it would be difficult to make sense of the way emotions are often sensitive to changing beliefs.

In any case, it is not obvious that emotions are recalcitrant so often that we should consider them to be encapsulated from cognition.⁹ Occasional instances of recalcitrant emotion are not on their own enough to show that emotions as a system must be encapsulated from cognition. However, even if the advocate of the argument from recalcitrance grants this point, they might persist by pointing out that beliefs do not ever exhibit recalcitrance, and hence that even occasional recalcitrance of emotion is a reason for rejecting cognitivism and perhaps for drilling down into the claim that emotions are encapsulated from cognition. However, as I’ll show in the following section, this response won’t go through. Beliefs are sometimes stubborn in much the way emotions are.

3.2 Empirical evidence for stubborn beliefs

An extensive body of empirical work demonstrates that we do not always update our beliefs in accordance with classical ideals of rationality.¹⁰ People often persist in their standing beliefs in the face of good evidence to the contrary (Anderson, 1983; Anderson & Sechler, 1986; Anderson et al., 1980; Slusher & Anderson, 1989); and equivocal evidence may even strengthen, rather than moderate, preexisting beliefs.,^11,12 Additionally, people tend to selectively expose themselves to congenial over uncongenial data, vigorously argue against evidence that challenges their existing beliefs, and judge evidence that supports their preexisting views as stronger than evidence that challenges them.¹³

More specifically, robust and well replicated empirical work shows that beliefs are sometimes stubborn in much the way that emotions sometimes are: Some beliefs persist in spite of countervailing evidence. Take, for instance, the phenomenon of misinformation persistence. People are evidently troublingly poor at updating their beliefs even once they learn that the information on which those beliefs are based is false.

As Wilkes and Leatherbarrow’s (1988) experimental paradigm for misinformation persistence shows, people who are otherwise typical reasoners often fail to revise their beliefs even once they learn that the information which they had previously accepted is false. For instance, in one iteration of this well replicated paradigm, Johnson and Seifert (1994) gave study participants a series of false news articles about a warehouse fire. The fifth article in the series stated that there were cans of oil paint and gas cylinders stored in the room where the fire started. In one of the experimental conditions, participants were informed immediately after they read this fifth article that there were in fact no cans of oil paint or gas cylinders stored in the room where the fire started. In the other condition, subjects didn’t receive this corrective until they had read the entire sequence of fake news articles about the warehouse fire. The participants completed a brief distractor task and were then given a questionnaire that was designed to determine what they believed was the cause of the fire. Over ninety percent of subjects in both misinformation conditions remembered the correction and reported that they accepted this correction, but nevertheless ninety percent used the misinformation in their reasoning as they answered the questionnaire.¹⁴

The subject in these trials gives us good reason to believe that they hold incompatible beliefs: both the correction they accept, as well as its contradiction, which they use in their reasoning in answering the questionnaire. The subject doesn’t have any other properties by which to pick out the gas cylinders, etc., outside of the details the experimenter has provided, and presumably isn’t thinking of two different sets of gas cylinders. So, the subject looks like they maintain two contradictory beliefs. This study is especially nice because it is not straightforwardly a case of emotionally-laden motivated reasoning. It doesn’t bear on participants’ political or personal beliefs, nor even many of their more mundane beliefs. Study participants don’t come into the study with antecedent beliefs about the case, and they likely don’t have much of an antecedent bias toward the misinformation.

Does this amount to evidence that ordinary reasoners have recalcitrant beliefs? A critic might object that misinformation persistence is importantly weaker than recalcitrant emotion: In cases of misinformation persistence the subject may not be aware of both beliefs. If a subject isn’t aware of two beliefs, it makes sense to think they won’t be aware that the two beliefs are incompatible, either. Meanwhile, on a cognitivist reading of recalcitrant emotion, the subject not only has two incompatible beliefs, but is aware of them both, and is aware of the tension between the two states.

If this critique of the misinformation persistence findings holds up, then a more precise characterization of recalcitrance emerges. In order to count as recalcitrant, a belief will not only need to be (a) incompatible with another belief the subject holds, but it will also have to be the case that (b) the subject is aware that the beliefs are incompatible.¹⁵ Even if this is enough to reject that the misinformation studies are evidence of recalcitrant belief, this potential objection guides us nicely toward a more precise characterization of recalcitrance. The task is now to show that beliefs exhibit (a) and (b).

Indeed, one form of belief perseverance shows just that. In the backfire effect—sometimes termed the backlash or boomerang effect—beliefs get stronger in the face of disconfirming evidence, rather than weakening as one would expect. The effect was recently found by Nyhan and Reifler (2010), among others, and is nicely illustrated in the following experiment.

In Batson’s (1975) experiment, participants were divided into those who believed that Jesus was the son of God and those who believed that he was not. Both groups were then given what they were told was an unpublished article that was “denied publication in the New York Times at the request of the World Council of Churches because of the obvious crushing effect it would have on the entire Christian world” (Batson, 1975, p. 180). According to the unpublished article, archeologists had brought new, decisive evidence to light that proved that the authors of the New Testament themselves knew that Jesus wasn’t the son of God.

After reading the article, the participants reported whether they believed its claims. They then took another test to track whether their belief that Jesus was the son of God had changed. Unsurprisingly, participants who antecedently did not believe that Jesus was the son of God and reported that they believed the claims of the article maintained and strengthened their belief that Jesus was not the son of God. Similarly, those who did antecedently believe that Jesus was the son of God and did not believe the article’s claims about the authors of the New Testament did not change their belief: This group continued to believe that Jesus was the son of God. But there was a group of participants who antecedently believed that Jesus was the son of God who also reported that they believed the article’s report to be true. For all of the participants in this last group, their belief that Jesus was the son of God was strengthened even though they also reported that they accepted the claims of the article stating that Jesus was not the son of God. They reported that they had just received good evidence that not-P and accepted that not-P, and yet the strength of their belief in P increased in spite of the very evidence they accepted and in spite of their new reported belief that not-P.¹⁶

The subjects reported both beliefs in quick succession. Participants in this last group have (a) incompatible beliefs and (b) seem to be aware of that incompatibility. One might even think that what’s happening in the backfire effect is more irrational than what we see in recalcitrant emotion: In the backfire effect, the subject not only has two incompatible beliefs and is aware of their incompatibility, but one of the beliefs is strengthened by the acquisition of a contradictory belief.

This looks to be a compelling case of recalcitrant belief. If this is right, then some token beliefs are recalcitrant. However, this doesn’t seem to be a good reason to think that beliefs are encapsulated as a system from cognition—they are beliefs, after all. The recalcitrance of some token beliefs does not establish that beliefs as a system are encapsulated from cognition. Or else, if stubbornness implies encapsulation, then at least some beliefs are encapsulated and encapsulation on its own is insufficient to establish that the state in question isn’t cognitive.

At this point, a reader might worry that these cases of stubborn belief all look to be ‘hot,’ motivated by emotional concerns, and hence insufficiently strong as evidence of stubborn belief. However, there is reason to think that even ‘cool’ beliefs exhibit this stubbornness. Jern et al. (2014) find that belief polarization can occur even for ‘cool’ rather than only for emotionally charged ‘hot’ beliefs, where the effect is usually observed.¹⁷

Evidence of belief perseverance gives us good reason to accept that the some of the token beliefs of ordinary, actual reasoners are stubborn in a way that’s closely analogous to the way that some episodes of emotions are. If this is right, then the stubbornness of a token mental state is not enough to show that this state cannot have a cognitive state even as a mere constituent part. Beliefs, stubborn or not, are cognitive states if anything is, and so the recalcitrance of token instances of a mental state type cannot show on their own that the state in question is not cognitive.¹⁸

There is compelling evidence that, from time to time, the beliefs of ordinary, actual human reasoners exhibit stubbornness akin to the stubbornness of emotion. If this is right, then the potential objection I considered at the end of the previous section does not go through: Appealing to recalcitrance of some instances of emotion is not enough on its own to show that emotion exhibits a feature that belief never does. Occasional recalcitrance of emotion is not enough on its own to tell against cognitivist views of emotion, and the encapsulation formulation of the argument from recalcitrance fails.¹⁹

4 Recalcitrance, belief, & rationality

4.1 Recalcitrance and the nature of belief

The success of Sect. 3.2 depends on the claim that the relevant mental states in the studies are actually beliefs. Some may want to resist this claim on the grounds that (a) having incompatible beliefs where (b) one is aware of having incompatible beliefs would violate the nature of belief.²⁰ It might intuitively seem that you can’t have incompatible beliefs. If beliefs by their nature just aren’t the sorts of things that can be recalcitrant, then the examples of what I have termed recalcitrant beliefs in the previous sections couldn’t be cases of beliefs.

That line assumes that a token mental state must meet the ideal criteria of belief to count as a token belief, and that those criteria preclude recalcitrance. For instance, Gendler (2008) and Levy (2015) seem to agree that token beliefs must satisfy the following criteria:

• i. Feature in normatively respectable transitions; and

• ii. Be consistent with other beliefs.²¹

Perhaps (i) and (ii) reflect at least some norms of ideal rationality that govern belief and according to which we evaluate belief. And yet, that beliefs are governed by these norms doesn’t require that each token belief satisfies (i) and (ii). Many have conflated the governing norms of belief with definitional criteria for counting as belief.²² This is a mistake.

We rarely make it criterial for each token of a mental state to meet the normative ideals of its type. You can be in a perceptual state that is more or less accurate or precise. Suppose you have poor vision and aren’t wearing any kind of corrective lens, for instance. You’ll still have a visual experience, even if it’s very blurry. Or try imagining you’re sitting at your kitchen table over a cup of coffee. You might imagine some aspects of the scenario in great detail, and others vaguely. You might even just imagine the scenario vaguely, in very little detail at all, because your mind is wandering elsewhere (or perhaps because you’re reading an academic paper). Your mental state need not represent all details of the imagined scenario in great detail in order for the state to be one of imagining. Similarly, when remembering, one can make lots of mistakes that fall short of confabulation before falling short of remembering at all. So even if a token state is subject to norms that govern its state type, the token can fall short of the norm without failing to be a token of the type altogether. Moreover, though many would agree that belief is governed by the norm of aiming at the truth, even the criterialists wouldn’t make arriving at the truth criterial of every token belief. A token belief can be governed by the norm of aiming at the truth even when the belief is itself false.

The question is not whether a token state meets the governing criteria to count as belief, but whether it is part of a system that is answerable to those normative criteria.²³ Accordingly, we ought to allow that some token beliefs may fall short of ideal rationality. I have cited evidence that we have incompatible and even recalcitrant beliefs, in spite of what many intuitions about belief would like to admit.²⁴ If this is right, then there is evidence that some beliefs do in fact fall short of normative ideals such as (i) and (ii).

4.2 Recalcitrance and the nature of the evidence

Some might accept that some beliefs fall short of (i) and (ii), and also accept that we seem to have evidence that beliefs do not behave in the optimally rational way that intuition suggests, but nevertheless deny that the studies cited above show that beliefs rise to the level of recalcitrance. Current critical debate merits some discussion here. There isn’t yet consensus about the extent of the backfire effect.²⁵ Some recent studies have failed to replicate it, and others have found the effect to be highly contextually constrained.²⁶ However, backfire has been observed even in even ‘cool’ and unemotional cognition, suggesting that the effect is in fact not especially constrained.²⁷ Moreover, some of the criticism lodged at the robustness of the backfire effect focuses on the fact that the effect is more likely to be observed in behavior rather than in self-report. But behavior is arguably a more reliable measure of belief in this domain. The cost of reporting one’s beliefs is lower than the cost of changing one’s other behavior, after all, not to mention that self-report is notoriously fraught.

Note, too, that the backfire effect is more extreme than recalcitrant belief would seem to require: The subject’s antecedent belief does not just persist but is strengthened by evidence to the contrary. Beliefs need not meet the backfire threshold in order to count as stubborn. Hence, even if backfire is mooted, we still have good evidence of stubborn beliefs.

Most importantly, the backfire effect is just one species of belief perseverance. Other perseverance effects, such as confirmation bias in the evaluation of evidence, are robustly replicated.²⁸ People are more likely to rate information that confirms, or is consistent with, their prior beliefs as stronger than information that contradicts those prior beliefs.²⁹

Insofar as perseverance is a form of stubbornness—your belief persists in the face of countervailing evidence and so is not evidence-sensitive in the way we expect beliefs to be—the work on belief perseverance still suggests that beliefs are sometimes stubborn in much the way that emotions sometimes are. As long as you accept that the states in these studies are beliefs, you should accept that beliefs turn out to behave in a way that falls far short of the intuitive notion of rational belief. If that is right, then we should rely on empirical study rather than intuitions to ask whether beliefs sometimes also satisfy (a) and (b), above, and rise to the level of recalcitrance.

It is also worth noting that there is also active debate over whether and the degree to which belief perseverance poses a problem for a Bayesian picture of belief.³⁰ Some argue that Bayesian principles are consistent with, or even entail, belief perseverance.³¹ Jern et al. (2014) model, for instance, entails biased assimilation effects.

However, my point does not hang on how the debate over Bayesianism about belief resolves. Both sides of that debate grant that beliefs exhibit perseverance. They differ only on whether a Bayesian picture of belief can satisfactorily account for belief perseverance. Nevertheless, those arguing that belief perseverance is rational in the sense that it is consistent with Bayes actually provide support for my position. Here’s why: Some may think that stubborn beliefs are always the result of ‘hot’ cognition about emotionally-charged issues, so one might worry that belief on its own doesn’t exhibit stubbornness but does so only under the influence of emotion. But if belief perseverance is consistent with Bayesian principles, and if even emotionally ‘cool’ beliefs exhibit stubbornness, then this worry is blocked: Stubborn beliefs are not to be explained exclusively by the interference of emotion.

None of this is to deny that there is rational failure in cases of stubborn belief. Things certainly seem to have gone wrong from a rational point of view when a person has a stubborn belief. But without careful attention to the ways that beliefs actually behave, we won’t have a full grip on the limits on belief, given that beliefs turn out to violate several stated intuitions about how they behave.

5 Norms of rationality, belief, and emotion

A different formulation of the argument from recalcitrance focuses on a more straightforwardly normative concern, beginning with the observation that we usually don’t judge a person in the grip of a recalcitrant emotion to be seriously irrational, while we do ordinarily judge a person who has conflicting beliefs to be very irrational. Many people are familiar with feeling afraid of something that they believe is safe, or guilty about doing something that they do not really believe was wrong, and so it is very understandable to have recalcitrant emotions. And, the argument goes, people typically find it more irrational to have stubborn beliefs than to have recalcitrant emotions: If you have had plenty of evidence that p, but you still believe that -p, or if you have beliefs that are mutually incompatible, then intuitively you are being very irrational. So, the thinking goes, people tend to judge those in the grip of recalcitrant emotions as less irrational than they would judge people who were in the grip of similarly stubborn beliefs. But if cognitivism is right and emotions are beliefs, are reducible to beliefs, or at least have beliefs as constituent parts, then a person who has a recalcitrant emotion would, counterintuitively, be just as irrational as someone who has stubborn beliefs. On this line, the primary problem with cognitivism's handling of recalcitrant emotions is that cognitivism makes the emoter out to be more irrational than intuition says they are.³²

A complete treatment of this formulation is beyond the scope of this paper, but I will point toward some potential ways of responding, here. First, there is at least one instance of conflicting beliefs that it seems eminently rational to hold. Second, at least some instances of recalcitrant emotions resemble this case of rational but conflicting beliefs. Third, I think we should be circumspect about how much everyday judgments of comparative irrationality reveal about the structures of the states in question. There will not be sufficient space to do justice to this final point, but I will give some initial considerations in support of it.

But first, an instance of rationally holding conflicting beliefs. There is one classic case of belief that clearly rises to the level of recalcitrance, but which nevertheless appears eminently reasonable. In the Preface Paradox,³³ it seems perfectly rational to hold two mutually incompatible beliefs, even though one knows that the beliefs are incompatible. Suppose you write a very long book. You are a diligent and conscientious researcher and writer, so you’ve double-checked each reference carefully and cannot find an assertion you don’t believe to be true. At the same time, you are intellectually humble and know that in a book this long, no matter how careful your fact-checking, it is unlikely that every claim is correct. Thus, as you write the preface to your book it seems very rational to say that you believe of every individual assertion that it’s true, but that you do not believe the conjunction of the book’s assertions to be true. This is a case where many would be happy to say that the believer has two conflicting beliefs but is nevertheless quite rational. So, there is at least one instance where most would be happy to say that a person is not irrational in holding conflicting beliefs.

Moreover, at least some cases which we might be inclined to deem recalcitrant emotion resemble the Preface Paradox more closely than they resemble the stubborn beliefs discussed in Sect. 3. Cases of recalcitrant fear tend to dominate the discussion of recalcitrant emotion, and fear tends to be forward-looking in Gordon’s (1994) sense: Fear is usually about states of affairs that have not yet come to pass, or which have passed but about which the emoter is not certain. Even when fear seems to be directed at an entity or person instead—as when someone says they are afraid of dogs, of planes, etc.—usually the fear is about what the entity in question might do or might have done (bite, crash, etc.).³⁴ In some instances, an emoter will have good reason to believe that a fearful outcome is unlikely, but will also have some reason to believe that the fearful outcome might occur. For instance, suppose you’ve just checked that you’ve turned off all the knobs on the stove. This gives you good reason to think that the burners are not emitting gas and that the stove is safe. However, you also believe that over the many occasions you’ve checked the knobs, it’s unlikely that you’ll never err. Suppose that, further, you’ve also gone through the very same procedure in the past only to return home and realize that you left the very last knob turned on and that the stove has been emitting gas.³⁵ In this case, it seems you have reason both to believe that the stove is off and to fear that it is on; there doesn’t seem to be much in the way of rational failure. This parallel is not restricted to fear but should hold instead for any cases where you have considerations counting both for and against the relevant belief.³⁶

Finally, what reasons do we have for judging that a person in the grip of recalcitrant emotion is less irrational than a person who has conflicting beliefs? Perhaps a person who has recalcitrant emotions trusts or endorses their belief but not their emotion, and hence is less irrational: They disavow the content of their emotion while endorsing the content of their belief, and this is less irrational than either endorsing two incompatible beliefs or endorsing a belief which persists in spite of evidence to the contrary.³⁷

The assumption that people usually don't endorse their recalcitrant emotions requires careful treatment. While it may be true that people will usually verbally disavow their recalcitrant emotions and endorse the conflicting beliefs, it is a mistake to take this verbal behavior as decisive. This tendency to report trusting one’s own beliefs over the emotions that conflict with them may itself reveal more about attitudes toward emotion than it does about one’s actual degree of trust in their emotions versus trust in beliefs: Many people generally think reason and considered belief ought to outweigh their emotions, even—or perhaps especially—when the two states are in conflict. One may report endorsing their considered beliefs over their recalcitrant emotions even when their actions suggest that the emotion wins out.

Given that the behavior of some instances of belief turns out to violate many familiar intuitions about belief, as I discussed in Sect. 3.2, theorists should be circumspect about relying on intuitions about the comparative rationality of having recalcitrant emotions compared with holding conflicting beliefs. Moreover, these intuitions about comparative rationality might reflect attitudes toward emotion and belief, as well as practices of privileging belief over emotion, rather than revealing much about either the structure of the states themselves or which norms govern the respective states.

6 Conclusion

The paper does corrective work to show how we should go about investigating the place of emotions in the mind. We want to know which features of emotion can point us toward their place in the mind. Much of the literature has treated recalcitrance as if it is up to the task, especially in light of concerns over encapsulation and intuitions about the rationality of belief. I argue that this is a mistake. Advocates of the argument from recalcitrance have relied on recalcitrance and an analogy with persistent perceptual illusions to drive a wedge between emotions and belief, and to align emotions with perception over belief. I have argued that not all instances of emotion are recalcitrant: Unlike perceptions, many instances of emotion change in response to changing beliefs. Moreover, I have drawn attention to empirical evidence that suggests that some beliefs exhibit the very stubbornness that seems to be characteristic of recalcitrant emotion. Hence, it is unlikely that emotion can be shown not to be cognitive by appeal to recalcitrant emotion.

At the very least, this paper shows that the problem of emotional recalcitrance needs to be spelled out much more carefully and with greater sensitivity to the variety of irrationality we see in actual beliefs than it has been so far. Finally, I have drawn an analogy with a clear case of recalcitrant but rational belief—the preface paradox—to suggest that a second formulation of the argument from recalcitrance is also unlikely to succeed. This suggestion is compatible with Cognitivist views that say emotions are beliefs, are reducible to beliefs, or have beliefs as necessary parts.

If my arguments go through, then emotions may well involve beliefs—the most cognitive sorts mental states we have.