Please cite the published version of this paper. This version may contain typographical and typesetting errors. Published as Cox, Ryan. 2019. “Only Reflect.” Philosophical Topics. 47 (2): 183–204. https://doi.org/10.5840/philtopics201947222
Is normative reflection a means of controlling our emotions? If so, what exactly does it give us control over? Does it give us control over whether we have the emotion in the first place? Does it give us control over whether we continue to have an emotion we already have? Does it give us control over the strength of the emotion? Does it give us control over whether our emotions are under the control of reason? Does it merely give us control over how we behave in light of our emotions?
It is widely assumed, both on the basis of ordinary experience, and on the basis of empirical research, that normative reflection is a means of controlling our emotions, and that, most importantly, it gives us control over whether we continue to have an emotion we already have. A common piece of advice in ordinary life and self-help books (e.g. Laurent and Menzies 2013) is to “step back” and “reflect” on one’s anger, one’s jealousy, one’s envy, or one’s fear, for instance, and to ask whether one has any reason to be angry, jealous, envious, or afraid. So, the claim that normative reflection is a means of controlling our emotions is part of common wisdom and, we might assume, has some empirical backing from ordinary experience. The psychological literature on emotion regulation (Gross 1998a, 1998b, 1999) has investigated such “cognitive strategies” for controlling emotions and provides further empirical support for the hypothesis that normative reflection is a means of controlling our emotions (Augustine and Hemenover 2009; Koole 2009; Webb, Miles, and Sheeran 2012).
While it is widely held that normative reflection is an effective means of controlling our emotions, it has proven to be notoriously difficult to provide a plausible model of such control (Hieronymi 2009, 156). Prior to testing the hypothesis empirically we face the following question: how could normative reflection be a means of controlling our emotions? How is it meant to work? What would normative reflection have to be, what powers would it have to have, in order for it to be a means of controlling our emotions? What kind of control could normative reflection give us over our emotions? What limits are there on the control that normative reflection gives us over our emotions? These are questions we might hope to get some traction on from the a posteriori armchair (Nolan 2015), drawing on a priori reflection, common wisdom, and the evidence of the empirical sciences.
This essay addresses some of these more philosophical questions about reflective control in this spirit. It will do so while keeping a close eye on the empirical research that has been undertaken to date on emotion regulation. I thus hope to provide both the philosophical foundations for a theory of reflective control and a theory which can ultimately be subjected to empirical investigation in order to answer the opening questions of this essay. My main aim is to examine and reject two models of reflective control, the higher-order model and the first-order model, and to defend a model which avoids the twin pitfalls of each of these models. Let me briefly describe each model and my alternative before getting started.
The higher-order model of reflective control—which has its source in certain philosophical views about the relationship between normative thought, rationality, responsibility, and freedom (e.g. Burge 1996; Korsgaard 1996; McDowell 1998) —gives a special role to beliefs and judgements about the normative status of our emotions in controlling our emotions, but in doing so, I will argue, gives higher-order beliefs and judgements more power than they in fact have, and fails to explain some of the central features of reflective control.1 The first-order model of reflective control—which can also be found in the philosophical literature (Hieronymi 2009), but has its source in the psychologists’ notion of “re-appraisal” (Lazarus 1991; Gross 1998a, 1998b)—gives a special role to first-order evaluative beliefs and perceptions about the objects of our emotions in controlling our emotions, but in doing so, I will argue, fails to explain how normative reflection could be a means of controlling our emotions at all.
I defend a model of reflective control which avoids the twin pitfalls of the higher-order and first-order models of reflective control, while learning from them both. I argue that normative reflection is a means of bringing our emotions under the control of reflective reason, where an emotion’s being under the control of reflective reason is to be understood in terms of its being under the control of one’s first-order evaluative beliefs and perceptions in accordance with one’s reflective commitments. Unlike the higher-order model, my alternative view gives first-order evaluative beliefs and perceptions a central role to play, thus capturing a central insight of work in psychology on reappraisal. But unlike the first-order model it finds a place for higher-order, normative reflection in the model which vindicates the thought that normative reflection is a means of controlling our emotions.
Here’s how I will proceed. I begin, in Section 1, with some very brief remarks about emotions and associated evaluative beliefs and perceptions. Then, in Section 2, I make some initial remarks about normative reflection and control. With these preliminaries out of the way, in Section 3, I discuss the higher-order model of reflective control. In Section 4, I discuss the first-order model of reflective control. In Section 5, I explicate and defend my alternative model of reflective control.
In what follows, I will assume that a kind of broadly cognitive theory of emotions is correct. A cognitive theory of emotion can be characterised as a theory which “makes some aspect of thought, usually a belief, central to the concept of emotion and, at least in some cognitive theories, essential to distinguishing the different emotions from one another” (Lyons 1980, 33). This is a very permissive characterisation of a theory of emotion. To be a little bit more committal, in what follows, I will be assuming that emotions are caused and maintained by broadly evaluative beliefs and perceptions (Arnold 1960; Armstrong 1968; Lyons 1980; Gordon 1987; Lazarus 1991).
In speaking of evaluative beliefs and perceptions, I am speaking of what psychologists call “appraisals” (Arnold 1960; Frijda 1986, 468–69; Lazarus 1991). These are beliefs and perceptions which involve abstract, or high-level, evaluations of one’s situation, such as the perception of someone’s remark as an insult or some such, or of the goat’s brains as disgusting, or of the bear as dangerous.2 It is perceptions like these which cause and maintain our emotional responses, but these, in turn, are caused and maintained by more particular and descriptive beliefs, judgements, and perceptions, like the judgement that someone meant what they said, or that the goat’s brain is slimy, or that the bear is large and approaching fast.
It follows almost trivially from such an account that our emotions are under the control of reason, in at least one sense of the expression ‘being under the control of reason’. If being under the control of reason just means being under the control of our evaluative beliefs, judgements, and perceptions in accordance with rationality, then our emotions are, to a large extent, under the control of reason. As George Pitcher notes, on a cognitive theory of emotions “it becomes a little easier to understand how one’s reason can control one’s emotions” for “we understand fairly well how reason can control evaluations and some kinds of apprehensions” (Pitcher 1965, 345). This is not to say, however, that our emotions are always under the control of reason, that they are never irrational. It is just to say that they are typically caused and maintained by evaluative beliefs and perceptions in accordance with rationality.3
While it follows almost trivially that our emotions are under the control of reason on this view, it does not follow trivially that they are under our control. Contrary to what some philosophers have claimed (Solomon 1973, 31–32; 1976, 186), it doesn’t immediately follow from the fact that our emotions are under the control of reason, that they are under our control, that they are under the control of the person whose emotions they are. If our evaluative beliefs and perceptions aren’t under our control, then how can our emotions be under our control in virtue of being under the control of our evaluative beliefs and perceptions? Indeed psychologists have reasoned in the opposite direction claiming that “[our] inability to control the motivational and perceptual bases of the appraisal process may explain [our] inability to control the emotions that result from that process” (Roseman and Smith 2001, 9). The cognitive theory of emotions may increase the prospects of our having control over our own emotions, but it doesn’t trivially entail that we do.
When philosophers discuss reflection as a means of controlling our emotions they usually have in mind normative reflection, the kind of reflection initiated by questions like “Do I have any reason to be ashamed?” or “Is this a reason to be ashamed?” In calling this normative reflection what is meant is that it involves the application of normative concepts like the concepts of reasons, justification, and appropriateness. These questions differ from purely factual, theoretical, or descriptive questions about our emotions like “Am I angry?” or “Am I ashamed?”.
Normative reflection may involve more than just reflecting on such normative questions. It may involve reflection on non-normative questions or on normative or evaluative questions about things other than my emotions themselves. So, I may begin by asking whether I have any reason to be ashamed of my body, say, and this question might lead me to take up questions about my body, “Am I really too tall” in the course of trying to answer this question.
Now, we should be careful to distinguish between reflection in this sense and what psychologists call “appraisal” and some philosophers call “evaluation”. When psychologists focus on “reflection” as a means to control, they speak of “reappraisal” and “cognitive change” (Gross 1999, 560–61). Gross and Thompson refer to cognitive change as “changing how we appraise the situation we are in to alter its emotional significance, either by changing how we think about the situation or our capacity to manage the demands it poses” (Gross and Thompson 2007, 14). So, psychologists tend to be concerned with changes in evaluative beliefs and judgements about the objects of our emotions. Only recently has the psychological literature attended to the distinction between reflection on our emotions themselves and reflection on their objects. In a recent meta-survey, Webb, Miles, and Sheeran, distinguish between reappraising one’s emotional response and reappraising the emotional stimulus (Webb, Miles, and Sheeran 2012, 778). We will keep a clear distinction between higher-order normative beliefs and judgements about our emotions themselves and first-order beliefs and perceptions about the objects of our emotions and our external circumstances in what follows.
For the purposes of this essay, I intend to use the word ‘control’ with its ordinary meaning. This is because I think that the this is the meaning that it is used with when we first ask “Is reflection a means to controlling our emotions?” We want an answer to this question, with this sense of ‘control’ for the purposes of ordinary life.
In order to better understand claims about control we should note that attributions of control come in two main forms. We speak, on the one hand, of having control over our emotions. But when we have control over our emotions, what exactly is it that we have control over? So we speak, on the other hand, of having control of whether he have a particular emotion, whether we continue to have a particular emotion, to what degree we have a particular emotion. Attributions of control using these interrogative forms are often a lot more informative than attributions using nominal forms. It will be important throughout to ask exactly what it is that we have control over, where the answer to this is given in the interrogative form.
The ordinary notion of control is a causal concept with wide application. We say, for instance, that the temperature of the room is under the control of the thermostat. The temperature of a room is under the control of a thermostat just when there is a kind of causal dependence between various states of the thermostat and the temperature of the room. What is it for something to be under the control of a person? Well, it seems, it is just for there to be a kind of causal dependence between various states of the person and the that thing. To put it how it is usually put: it is for there to be a kind of causal dependence between a person’s will and the thing. So the ordinary meaning of control is connected to a kind of causal dependence between states of a thing and something else. In the case of a person, it is a causal dependence between things and a person’s will. Thus, we speak here of voluntary control. Within the ordinary notion of control there is an important distinction to be made between direct and indirect control. The distinction can be made as follows: one has direct control over something insofar as one does not control is by means of controlling something else. A man’s control of his weight is not direct. He controls it by means of controlling how much he eats and exercises. These latter things may be under his direct control.
If our question is whether we have direct control over our emotions, then the answer seems to be obvious: we do not. Our emotions are not under our direct voluntary control. But this shouldn’t disappoint us. Our question is whether normative reflection is a means of controlling our emotions. If it is, it would be an indirect means: the control we would have would be indirect in any case. Here it is helpful to think about another indirect means we have of controlling our emotions: controlling our emotional behaviour. Psychologists have long noted that we can indirectly control our emotions by controlling our behaviour. As Willam James writes: “In rage, it is notorious how we ‘work ourselves up’ to a climax by repeated outbreaks of expression. Refuse to express a passion and it dies” (James 1884, 197). This gives us a paradigm example of an indirect means of controlling our emotions. We will want to know how normative reflection departs from this paradigm if it is also an indirect means of control.
Some philosophers will prefer to argue that it is a mistake to think that the question is a question about control in the ordinary sense, that is, about voluntary control. If we are concerned with broader questions about rationality, freedom, and responsibility, then we need to look for a sense of control other than voluntary control. Perhaps voluntary control is just one species of the broader genus “control” and the genus, rather than just the species, is what has this connection with rationality, freedom, and responsibility. This view is usually motived by the thought that unlike things like headaches, things which merely happen to us, our emotions are not things which merely happen to us but are, rather, things we are in some sense responsible for. Since we do not have direct voluntary control over our emotions, it is argued that there must be some sense in which we have control over them which marks the distinction between things which merely happen to us and things which we are in some sense responsible for. I think that these philosophers are right to think that there is something important here. But the failure of these philosophers to provide an account of this special kind of direct control, despite their continued insistence that there must be something here, suggests that there is no special sense in which our emotions are under our direct control. What makes the distinction here is that our emotions but not our headaches are under the direct control of reason. But it is one thing for them to be under the control of reason and another for them to be under our control.
The only kind of control we care about is voluntary control, where that is understood as a kind of causal dependence on our wills. The only question which remains then is to say exactly what we could hope to have voluntary control over with respect to our emotions. I will argue that there is a sense in which we can have voluntary control over whether our emotions are under the control of reflective reason. This is all the control we need to vindicate the usual claims about rationality, freedom, and responsibility.
According to the higher-order model of reflective control, in addition to being under the control of our first-order evaluative beliefs and perceptions, our emotions are under the control of our higher-order normative beliefs and judgements. On this model, if I am angry, say, I may arrive at the conclusion that I have no reason to be angry and my belief or judgement that I have no reason to be angry will, if the case is typical, cause me to no longer be angry. The defining feature of this model is that it is my normative belief or judgement that I have no reason to be angry which causes my anger to dissipate or lessen. As David Owens puts it “the instrument of reflective control” on such views “is a normative higher order judgement” (Owens 2000, 11). On this model, normative reflection is merely a means of arriving at a higher-order normative belief or judgement.
This model is usually supported by appeal to a certain picture of rationality. On this picture, our mental states typically conform to the principle of rationality. The principles of rationality require us to have certain mental states or combinations of mental states under certain conditions. And it is plausible to think that the principles of rationality require you not to have a particular emotion if you believe you have no reason to have it. Thus, insofar as you are disposed to conform to the principles of rationality, you will be disposed to cease having some emotion if you believe that you have no reason to have it.4
We should be careful not to misunderstand the higher-order model. The view is not that one reaches some conclusion about the normative status about one’s emotion and then does something to control one’s emotion accordingly. There is a possible picture on which we engage in normative reflection in order to work out what the normative status of our emotion is and then act to bring our emotion into line with our beliefs about their normative status. So, if I conclude that I have no reason to be angry I might then take steps to bring it about that I am not angry, thereby ensuring that I do not have an emotion I believe that I have no reason to have. On this view, rational reflection is a means to working out the normative status of one’s emotion which, in turn, is a means to working out how to act in order not to have unjustified emotions. The higher-order view, by contrast, holds that one’s emotional responses are under the direct control of one’s higher-order normative thought (insofar as one is rational). While our control over our emotions is indirect, the control of higher-order thought over our emotions is direct on this model.
With the higher-order model clearly in view we now turn to some challenges for it.
The first challenge is that it doesn’t explain why reflection itself has consequences for one’s emotional life, prior to, or independent of, one’s reaching a conclusion about the normative status of one’s emotion. Consider the following widely quoted passage from Christine Korsgaard:
[O]ur capacity to turn our attention onto our own mental activities is also a capacity to distance ourselves from them, and to call them into question. I perceive, and I find myself with a powerful impulse to believe. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I believe? Is this perception really a reason to believe? I desire and find myself with a power impulse to act. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I act? Is this desire really a reason to act? The reflective mind cannot settle for perception and desire, not just as such. It needs a reason. Otherwise, at least as long as it reflects, it cannot commit itself or go forward. (Korsgaard 1996, 92–93)5
This is a particularly apt characterisation of reflection. It sees reflection as having consequences itself for one’s emotional life. It is as if the emotion itself is “suspended”, “bracketed”, or “called into question”, while one is engaged in reflection. As it stands, the higher-order model has no explanation of this. On the higher order model, we shouldn’t expect any changes in one’s emotional life until one reaches a conclusion about the normative status about one’s emotion.
The higher-order model might, however, be modified in order to account for this. The simplest modification would hold that uncertainty about the normative status of one’s emotion has the consequence of suspending, bracketing, or calling into question, the emotion itself. But it is hard to see how this modification isn’t just ad hoc. Unless one’s emotion was caused and maintained by a higher-order normative belief or judgement, it is hard to see how becoming uncertain about the normative status of one’s emotion could have the consequences of suspending it. Of course, the higher-order model might claim that our emotions are caused and maintained by normative beliefs and judgements, but such a view is not particularly plausible.
The second challenge comes from the possibility of dispassionate, detached, or disengaged normative reflection. At this stage I want to take this distinction to be an intuitive one. Here’s how to understand it. The thought is that I might engage in higher-order normative reflection on my emotions in just the same way that I might engage in higher-order normative reflection on your emotions. When I engage in normative reflection on your emotions, I am just trying to work out, in the first instance, whether you have any reason for your emotional response. I may have the further aim of advising you or criticising you, or perhaps even taking some means to changing your emotion. Perhaps I want to ensure that you don’t have any emotional responses you have no reason for having. Now, I can undertake this kind of reflection on my own emotions. When I do I might be said to be taking a third-personal, dispassionate, detached, or disengaged stance towards my own emotions. Of course, this just raises the question of what it would be to take a first-personal, passionate, attached, and engaged stance towards my own emotions, and this is a question we will come to. But for now, I hope that it is intuitively clear enough that this third-personal stance is not the ordinary stance we take towards our own emotions in reflection. All I need for the argument of this section is the claim that there is a distinction here. I don’t need an account of it.
With this distinction in hand I want to argue that the higher-order model loses much of its plausibility when we consider normative beliefs and judgements arrived at by means of this kind of third-personal, dispassionate, detached, reflection. The proponent of this view is committed to the claim that it doesn’t matter how these beliefs or judgements are arrived at, since it is believing that you have no reason for some emotion and continuing to have the emotion which rationality does not permit. So no matter how you arrive at the belief, we should expect your disposition to conform to the principles of rationality to manifest itself. Now, I want to claim that as a matter of psychological fact, normative beliefs arrived at in this way have no immediate consequences for our psychological lives. My normative belief about my own emotions arrived at in this way has no more consequences for my emotional life than my normative belief about your emotions has for your psychological life. Of course, the proponent of the higher-order model will just deny this since their theory predicts otherwise. But I think that we intuitively recognise the possibility of such detached reflection and that a feature of it is that it doesn’t have the consequences for our psychological lives that engaged reflection does. And this suggests that the higher-order model is motivated by a restricted diet of examples.
Let me make this point carefully since it is central to my criticism of the model. If we begin from ordinary cases of normative reflection, we can agree with the proponent of the higher-order model that one is less than fully rational if one concludes that one has no reason to have a particular emotion but continues to have it. The proponent of the higher-order model draws two lessons from this. One is that the principles of rationality require one not to have a particular emotion if one believes one has no reason to have it and that it is the normative belief or judgement which causes one’s emotion to dissipate in accordance with rationality. But this is a hasty conclusion drawn from a restricted diet of examples. If we consider the possibility of dispassionate, disengaged normative reflection, we see, in the first instance, that one’s normative belief or judgement may have no consequences at all for one’s emotional life. Now, what does this mean for the higher-order model? Well, the proponent of the model might think that this does nothing to undermine their story in the ordinary cases. Cases of dispassionate reflection are cases where one fails to be as one rationally should be. Somehow cases of dispassionate reflection involve a masking of this disposition to give up the emotion upon believing that one has no reason to have it. Now, I don’t want to get embroiled in a fruitless clash of intuitions about whether in such cases one is less than fully rational because the principles of rationality prohibit believing you have no reason to have an emotion while continuing to have it. What I want us to do is squarely consider the hypothesis of the higher-order model and the alternative concerning the relevant causal mechanism, once we have cases of dispassionate reflection clearly in mind. The alternative hypothesis to the hypothesis offered by the higher order model takes these cases to show that it isn’t the higher-order normative belief or judgement which has the consequences for one’s emotional life. The purported cause is present in this case but the effect isn’t. With a nod to J.S. Mill we conclude that the normative belief or judgement is not the cause in the normal cases and we hypothesise that there is a common cause of the consequences for one’s emotional life and the higher-order belief or judgement in the normal cases. But the proponent of the higher-order model, in order to hold that the belief or judgement is the cause in the normal cases, must hold that cases of dispassionate reflection are cases of prevention: there must be something which prevents the normative belief or judgement from having its ordinary effect in such cases. Since it isn’t at all clear what this could be, the higher-order model is left without a plausible explanation.
The third challenge for the higher-order model is that, at least as it has so far been stated, it fails to make sense of the idea that I have control over my emotions. At best, on this view, my emotions are under the control of my higher-order normative beliefs and judgements. Now, why is this a problem? Well, it seems that unless my higher-order normative beliefs and judgements were under my control, my emotions wouldn’t be either.6 Of course, I have control over whether or not I engage in normative reflection. But I don’t have control over the outcome of such reflection.
Now, it might be argued at this point that this is to misunderstand the higher-order view. The argument is not of the familiar kind: since your emotions are under the control of your normative thoughts and your normative thoughts are under your control, your emotions are under your control. Rather, someone might argue, one way for your emotions to be under your control is for them to be under the control of your normative beliefs and judgements.7 But it isn’t clear what is gained by speaking this way. Sure, you can use words this way. But our question wasn’t whether there is a way of using the word ‘control’ on which our emotions are under our own control. The question was whether reflection was a means of controlling our emotions. And even if we grant that this is a way for our emotions to be under our control, it is hard to see how normative reflection could be a means to such control. We would just have this control over our emotions. Normative reflection wouldn’t be a means to it at all.
I suggested above, however, that there is an alternative option open. The proponent of the higher-order view could distinguish between two senses in which one’s emotions might be under the control of reason. They can distinguish between a minimal sense, understood in terms of one’s emotions being under the control of one’s first-order evaluative thoughts and perceptions, and a stronger sense of being under the control of one’s higher-order normative beliefs and judgements. Then they could argue that normative reflection is a means to bringing one’s emotions under the control of reason in this stronger sense, since it is a way of arriving at such higher-order normative beliefs and judgements. As I will argue below, I think that this is the correct general model, but, give the other problems with the higher-order model, I think that it gets the details wrong—it doesn’t properly characterise what it is for an emotion to be under the control of reflective reason.
Many of the challenges raised for the higher-order model point towards a first-order model of reflective control. As I pointed out earlier, normative reflection often involves raising and answering first-order questions about the objects of our emotions. I reflect on whether I have any reason to be angry with Jones and then find myself reflecting on what Jones did, and asking whether he wronged me. My initial perception of his behaviour may have been that it was meant as an insult. But I am now questioning my initial perception. In engaged normative reflection these questions are not bracketed as they are in dispassionate reflection. I am led by the normative question to open and consider these questions. In asking whether I have any reason to be angry with someone, I might consider whether they really meant to insult me. These questions, and their answers, are, of course, relevant to the question of whether I have any reason to be angry. Reflective control, on this model, involves what psychologists call “reappraisal” and as such is a well established mechanism for emotion regulation. Moreover, the consequences of changes in first-order thoughts for our emotional lives has often been noted by philosophers. Solomon calls it “a feature of emotions that has been pointed out by many analysts” and give an example: “A change in what I am angry”about” demands a change in my anger; if I no longer feel wronged by John, who only bought a car that looks like mine, I cannot be angry at John (for stealing my car) any longer” (Solomon 1973, 23).8
Nobody doubts that changes in one’s beliefs, judgements, and perceptions on these matters have consequences for one’s emotional life. After all, these are the beliefs, judgements, and perceptions which are agreed to cause and maintain one’s emotion. So it isn’t any surprise when the belief, judgement, or perception maintaining an emotion is removed that the emotion itself is removed. So the first-order view just exploits the explanatory resources already provided by cognitive theories of emotions. In this sense the first-order theory is very economical. It also provides plausible explanations of the facts about suspension and dispassionate reflection.
Consider first the facts about suspension. If one raises the question of whether someone’s behaviour constituted an insult, one may suspend one’s belief or judgement that they insulted one, and this has the immediate consequence of one’s emotion being suspended. As we saw in the criticism of the higher-order view, we expect the explanation of suspension to involve a model like this. The first-order view connects suspension of the emotion to suspension of the first-order evaluative belief or perception during reflection.
Now consider the facts about detached, disengaged, and dispassionate reflection. The first-order theory can claim that engaged reflection involves taking up and considering first-order evaluative questions, questions about the objects of one’s emotions. This, according to the first-order theory is just what such reflection is. Detached reflection brackets these questions. And now it is clear why detached reflection does not have consequences for one’s mental life while engaged reflection does.
So if reflection results in a change in one’s emotional life, it is due to a change in these object-directed evaluative questions, and not in the answer one reaches to the normative question. And it isn’t very surprising that if one’s answer to the object-directed question was arrived at by means of “simple nonconcious, lower level cognitive processing” then it might change when one engages in “complex, conscious, high-level cognitive processing” (Roseman and Smith 2001, 9). Now, of course, the answer one reaches to the normative question will accord with one’s answer to the object-directed question. Answering negatively the question of whether he meant to insult you may lead you to believe that you have no reason to be angry. But the change in emotion is due to your change in the answer to the former question and not to the belief that you have no reason to be angry. If this is right, then it is no surprise that it appears as though reaching conclusions to questions about the normative status of your emotions causes a change in your emotional life. But the appearance is deceptive. There is a common cause of the change in your emotional life and your conclusion about the normative status of your emotions, namely, your judgement about the object of your emotion. As Pamela Hieronymi puts it:
[I]f the alternative picture is correct, the powerful picture according to which one’s lower-order attitudes conform or are sensitive to one’s higher-order judgements (insofar as one is rational) is misleading…. First-order attitudes could not properly be thought of as sensitive to the higher-order judgement, because, insofar as one remains of one mind, the first-order attitude will be revised or suspended in the process of arriving at the higher-order judgement. When things go well, the attitude and the judgement do not cohabit the mind. At best, the attitude is sensitive to a stretch of the reasoning that supports or generates the higher-order judgement. (Hieronymi 2009, 160–61).
There is much to be said for the first-order model then. It is economical in that it exploits the explanatory resources made available by cognitive theories of emotion. It avoids the problems of the higher-order theory in this regard. It also provide an explanation of the facts about suspension and disengaged reflection. However, it faces some challenges of its own.
The first challenge is analogous to the third challenge to the higher-order view. Suppose we grant that our emotions are under the control of our first-order beliefs, judgements, and perceptions. How does that amount to our having control over our emotions? I do not have control over my first-order beliefs, judgements, and perceptions. So how can I have control over my emotions? At best, again, it seems that my emotions are under the control of my first-order beliefs, judgements, and perceptions.
Now, at this point there are various responses available to the proponent of the first-order view. The first of which is also available to the higher-order view, but I have delayed discussing it until here. The response is to deny that we do not have control over our judgements or evaluations. Notice my careful change to ‘judgements or evaluations’. The claim would not be nearly as plausible if it were about ‘beliefs and perceptions’ since beliefs and perceptions are things we do no have voluntary control over. Now, as we have already seen, some proponents of cognitive theories believe that if such a theory is correct, and our emotions are caused and maintained by our evaluative judgements, then they are under our control, since our evaluative judgements are under our control. But this view gets most of its plausibility from appealing to notions like ‘judgement’, ‘evaluation’, ‘answer’, ‘interpretation’ and ‘appraisal’ which have corresponding verbs ‘judging’, ‘evaluating’, ‘answering’, and ‘appraising’. Now, each of the latter are clearly things we do and have control over. We judge, evaluate, interpret, answer, and appraise. But the claim is not that our emotions are under the control of these activities which we do have control over. The claim is that they are under the control of judgements, evaluations, answers, interpretations, and appraisals, considered as states, and we simply do not have voluntary control over whether we are in these states.
The proponent of the first-order view might at this point argue that what we have control over is whether our emotions are under the control of reason. Since our emotions are under the control of our first-order evaluative beliefs and perceptions, and reflection is a means of arriving at or modifying such beliefs and perceptions, it might be thought that engaging in reflection is means of bringing our emotions under the control of reason. The problem with this response is that, on this minimal understanding of ‘being under the control of reason’ our emotions either are or are not under the control of reason, and engaging in normative reflection does nothing to bring our emotions under the control of reason. At best, it exploits the fact that our emotions are under the control of reason. So this style of explanation is not available to the proponent of the first-order model.
Now, one might try to argue that there is a sense in which our judgements, beliefs, and perceptions are under our control, it just isn’t the kind of control we have been imagining so far. And we can all agree that if there is a sense in which they are under our control, they are under control in this sense in a way that our sensations are not. I have no control over whether I feel a pain in my foot. But I am meant to have some kind of control over whether I think that you wronged me. Now, it is true that there is a distinction here. But it isn’t clear that it is a distinction in kinds of control I have over things. What we might say here is that yes, my beliefs, judgements, and perceptions, are, if I am lucky, under the control of reason. And this certainly is a difference. But are they thereby under my control? Well, they are under the control of my capacity for reason and nobody else’s. Someone might want to say that one way for our beliefs, judgements, and perceptions to be under our control is for them to be under the control of reason. Well, yes. But, again this seems to be just another verbal trick. And again, how is it that reflection is meant to be a means of control if our emotions simply are under our control in virtue of being under the control of reason?
So although the first-order view can make sense of the idea of our emotions being under the control of our first-order evaluative beliefs and perceptions, it cannot make sense of our emotions being under our control or of normative reflection being a means of controlling our emotions. In the next section I argue that this shouldn’t be surprising given the downgraded status the first-order model gives to normative reflection.
What, according to the first-order theory, is the relation between normative reflection and first-order reflection? Let’s agree that normative reflection often involves first-order reflection on the objects of our emotions. But what is the relation between the two? Why should someone take up a first-order evaluative question about the object of their emotion upon engaging in normative reflection? One answer to this is that the normative question “Do I have any reason to be angry?” is a question about reasons and the first-order evaluative question “Did Jones wrong me?” is a question about something which is a candidate to be a reason. But then we ask: well, why not take one’s answer to the first-order question for granted, and conclude that, yes, indeed, you have a reason to be angry with Jones? Why must you take up the first-order question? If normative reflection could take place in a way that bracketed the first-order questions, then the first-order theory’s explanation would not work. Now, the passage quoted from Hieronymi earlier makes it sound more or less accidental as to whether someone engaged in normative reflection will take up the first-order question rather than taking their answer to it for granted. And a proponent of the first-order theory might claim that this is all there is to say. Normative reflection has consequences for one’s emotional life only when it leads one to take up the first-order question and perhaps only when doing so leads one to suspend one’s answer to it or change one’s mind. But this seem to significantly downgrade normative reflection. Moreover, might normative reflection sometimes have consequences for one’s emotional life without one changing one’s first-order evaluations?
These problems point to a deeper one. On the first-order theory the capacity for normative reflection, for normative thought, brings with it no new powers. A creature with the capacity for normative reflection has no new powers of control over their emotions that a creature lacking the capacity does not already have. A creature with the capacity for normative reflection may more often be led to take up first-order questions which have consequences for their emotional life. But that is the extent of it. This seems wrong. And this is something the higher-order theory seems to have right, for on the higher-order theory, the capacity for normative reflection brings with it new powers of control. We need a way of taking what is right from both the higher-order and first-order model. We need to make sense of the consequences of first-order thought for our emotions while giving normative reflection powers of their own. In the next section I will offer a model which does just this.
Let’s take stock. We have seen that the higher-order model of reflective control mistakenly assumes that the higher-order beliefs and judgements about the normative status of our emotions play a causal role in our emotional lives which they do not in fact play. Yet, the higher-order model does provide us a plausible model of how we might be said to be able to control our emotions by means of bringing them under the control of reason. If the higher-order model had a plausible account of what it is for us to bring our emotions under the control of reason, then we would have every reason to accept it. But the higher-order model is based on a mistaken assumption about the power of normative thought. The first-order model, by contrast, rests on a plausible assumption about the causal role played by first-order evaluative beliefs and perceptions about the objects of our emotions. But it fails to explain the connection between normative reflection and first-order evaluative beliefs. Sure, in the course of reflecting on the normative status of our emotions, we might take up first-order evaluative questions about the objects of our emotions. And if we do, then we have a nice explanation of the consequences of normative reflection on our emotional lives. But, then again, we might not take up these questions. The first-order theory offers no explanation of the connection between the two. We have also seen along the way that there is not much to be said for attempts to define a sense of control distinct from the ordinary sense of control and to claim that we have this kind of control over our emotional lives. We are interested in whether reflection is a means to controlling our emotions on a perfectly ordinary sense of control. The question is what exactly we have control over in a perfectly ordinary sense of control.
In order to provide an alternative model to both the higher-order and first-order theory, I propose that we begin by distinguishing, as the higher-order theory does, between two senses in which our emotions may be under the control of reason.
The first sense, as we have seen, is the minimal sense in which our emotions are under the control of our first-order evaluative thoughts and perceptions about the objects of our emotions in accordance with the principles of rationality. As I argued earlier, once we accept a cognitive theory of emotions it becomes an almost trivial matter to claim that our emotions our under the control of reason in this sense. Of course, there will be exceptions, and the extent of the control of our evaluative thoughts and perceptions over our emotions may be limited. But, when it is in place, it amounts to the control of reason over our emotional lives. What we need is a distinct sense in which our emotional lives might be under the control of reason.
We saw that the higher-order theory fails to provide this. Our emotional lives being under the control of reason, in this stronger sense, is not a matter of them being under the control of higher-order normative thoughts. This view is, on the one hand, psychologically implausible, and on the other hand, rationally implausible, positing as it does, a conflict between first-order control of reason and higher-order control of reason. But I nonetheless think that the higher-order model points us in the right direction. Where the higher-order theory goes wrong is in its appeal to beliefs and judgements about the normative status of our emotions. We need some other way of marking the distinction.
What we need to appeal to here is the notion of being committed to some consideration’s being a reason for some reaction and a corresponding notion of withholding commitment to some consideration’s being a reason for some response, where being so committed is distinct from merely believing that something is a reason. In order to make good on this claim, we are going to need to have a better understanding of such reflective commitments, of what I mean by being committed to some consideration’s being a reason for some response and what I mean by withholding commitment to some consideration’s being a reason for some response. Now, before proceeding, I want to be upfront about the challenge I now face. I want to explicate the notion of reflective commitment. But I want to do so in a way that isn’t utterly obscure. Many philosophers have felt the need to appeal to something like commitment or endorsement in this context.9 But these notions are very hard to make sense of. The words sound appropriate, but without further explication the view being gestured at remains obscure. This is difficult terrain. I want to see if I can do better.
To begin, I want to return to the metaphors of stepping back, gaining some distance, calling into question, suspending, and bracketing. What do we step back from, or gain distance from? What is called into question? What is suspended or bracketed? Korsgaard’s passage suggests that it is both the mental state in question and one’s reasons for it. But the question cannot merely be the theoretical question ‘Is this consideration a reason for me to do so and so?’ which would have equal application in the third-person ‘Is this a reason for him to do so and so?’ The question is, in part, whether some consideration is to be my reason for doing so and so. Merely asking whether something is a reason for doing so and so, doesn’t call it into question, suspend it, or bracket it. But asking whether something is to be my reason for doing so and so immediately calls its status as my reason for doing so into question, brackets its role, suspends it qua reason. Now, the question of whether something is to be my reason for doing something calls for decision and the resulting state is one of reflective commitment: one will either end up being committed to the consideration’s being one’s reason for doing something or will end up withholding commitment from the consideration’s being one’s reason for doing something. This is neither a matter of believing that something is a reason, or it to be one’s reason, or a matter of* intending or desiring it to be one’s reason*.
So far so good, but the threat of obscurity is looming large. It is one thing to give a negative characterisation distinguishing reflective commitments from beliefs, desires, and intentions. It is another thing to give a positive characterisation. So let me do something more to explicate the notion of reflective commitment. I will do so in terms of its typical causes and effects, thereby analysing it in the general spirit of functionalist theories of mind. Doing so will reveal why commitment is distinct from both believing and intending. The typical effect of the state of being committed to some consideration’s being a reason for having some emotion is for the typical effects of one’s beliefs and perceptions concerning that consideration in causing and maintaining one’s emotion to be reinforced. The typical effect of the state of withholding commitment from some consideration’s being a reason for having some emotion is for the typical effects of one’s belief and perceptions concerning that consideration in causing and maintaining one’s emotion to be inhibited. Reflective commitment, then, is to be understood in terms of its typical effects of reinforcing and inhibiting the typical causal roles of first-order evaluative beliefs and perceptions. These are distinctive functional roles of reflective commitments and it is these which distinguish them from beliefs, desires, and intentions. We may also look to a typical cause of reflective commitment, and that is, unsurprisingly, normative reflection. Normative reflection on whether various considerations are reasons for or against will typically lead to commitment one way or the other.
Reflective commitment also has a role to play in explaining our ability to answer ‘why’-questions about our emotions in a way that is independent of observation and inference from evidence. Someone who is committed to some consideration’s being a reason for being angry will be disposed to answer the question ‘Why are you angry?’ by offering that consideration, and someone who is withholding commitment from some consideration’s being a reason for being angry will be disposed to deny that that consideration is their reason for being angry. In this way, we distinguish reflective commitment from belief and intention, since neither of these are understood in terms of their consequences in reinforcing or inhibiting the causal roles of other mental states nor in terms of their grounding our capacity to speak for our reasons without relying on observation or inference.
By understanding reflective commitment in terms of its typical effects of reinforcing and inhibiting the typical causal roles of first-order beliefs and perceptions we have avoided utter obscurity and staked out a claim which may or may not be true. If there are no mental states the typical effect of which is to reinforce or inhibit the causal roles of first-order beliefs and perceptions, then there are no states of reflective commitment. We can argue for such states on the basis that they best explain the facts about reflective control. That will be my strategy for the remainder of this essay.
With the notion of reflective commitment now firmly in hand, we can make sense of two kinds of normative reflection one might engage in. Disengaged, dispassionate, or theoretical reflection is normative reflection which merely aims at belief or knowledge. One engages in such reflection merely with the aim of working out what the normative status of one’s emotion is. In contrast, there is a kind of engaged, passionate, or deliberative reflection, which is normative reflection which aims at commitment. Normative reflection is engaged insofar as the question is not merely one of whether some consideration is a reason but, rather, is whether the consideration is to be one’s reason.10
We are now in a position to define a stronger sense of an emotion’s being under the control of reason. It is under the control of reason in this stronger sense when the role played by one’s evaluative beliefs or perceptions about its object is grounded in one’s being committed to the content of that belief or perception as a reason for having the emotion in question. We can also see how the absence of such an emotion can be under the control of reason in this stronger sense, if its absence is due to the typical effects of a particular belief or perception being inhibited by one’s withholding commitment. And with this stronger sense of an emotion’s being under the control of reason in hand, we can explain how normative reflection can be a means of bringing our emotions under the control of reason in this stronger sense. We engage in such reflection with the aim of arriving at a commitment concerning reasons for and against having that emotional response. Call this stronger sense of being under the control of reason being under the control of reflective reason. What is under our control, then, in the perfectly ordinary sense of control, is whether our emotions are under the control of reflective reason.
We can now explain why merely by engaging in such reflection our emotion may be suspended. Once the normative question has been raised with a view to determining whether to commit to the reason as a reason, the default is to withhold commitment. Now, withholding commitment, unlike simply not being committed or being uncommitted, is a positive state. One counts as withholding commitment only if one has considered the question of whether something is to be one’s reason. Thus, insofar as one remains uncommitted, upon engaging in reflection, one withholds commitment. And if such withholding is effective, then one’s first-order evaluative beliefs and perceptions do not play their typical role in causing and maintaining one’s emotional response. Thus, we can explain how the emotion may be suspended merely upon engaging in reflection. We can also explain why reflection may have consequences for one’s emotional life even though one hasn’t changed one’s views at all about the first-order facts. There may be mere differences in commitment. So we do better than the first-order view here. But we incorporate the first-order view’s important focus on first-order evaluative questions about the object of our emotions, for what we commit ourselves to, and what we withhold our commitment from, are considerations about the objects of our emotions. The power of normative reflection is limited to reinforcing the rational dispositions which are already in place or inhibiting the rational dispositions which are already there. It cannot create ex nihilo such a disposition, so it cannot bring about an emotional response ex nihilo (in the way that a mere normative belief or judgement could on the higher-order view). So we avoid the kind of conflicts in control by unreflective and reflective reason the higher-order view is committed to.
In appealing to the notion of commitment and withholding commitment, I am not claiming that these are things we have direct control over. Just like higher-order beliefs and judgements, and first-order beliefs and perceptions, we do not have control over our commitments. We engage in normative reflection, aimed at arriving at commitment, considering reasons for and against getting angry, and we either arrive at a commitment or end up withhold commitment. Normative reflection aimed at commitment is, in this sense, just like normative reflection aimed at belief, and first-order evaluative reflection. If someone wants to say that we have control over our own emotional lives merely in virtue of our emotional lives being under the control of reason, in the stronger sense just characterised, then they are welcome to. But I don’t see what could be gained by insisting on this way of talking. I do, however, think that when our emotional lives are under the control of reason in this sense, then our emotional lives have a deeper claim to being truly our own. There is a sense in which by engaging in normative reflection aimed at arriving at a commitment, we more fully face up to, and take responsibility for, our emotional lives. This is a sense that goes beyond the sense in play when our emotional lives are under the control of reason in the weaker, minimal sense. They have a deeper claim to being truly our own because there’s nowhere to step back to, outside the point of view of our commitments, to distance ourselves from our emotional responses, in the way that normative reflection provides a way of stepping back from our first-order beliefs and evaluations and the role they play in causing and maintaining our emotions. So I think it would be better to say that by engaging in normative reflection I more fully make my emotions mine, than it would by to say that by engaging in normative reflection I have control, in some special sense, over my emotions. There’s just no special sense in which I have control over whether I have the emotions in question. The control I have over my emotions is just the ordinary kind of voluntary control. But it isn’t, at least in the first instance, voluntary control over whether I have some emotion or continue to have it. I can’t pick and choose. Rather, what I have voluntary control over, in the first instance, is whether my emotion is under the control of reflective reason. And normative reflection aimed at commitment is a means of bringing my emotion under the control of reflective reason.
I have argued that the kind of control we have over our emotions is the same kind of control we have over the position of our limbs, or the position of the phone on our desk. But I have argued that what we have control over is not, in the first instance, whether we have a particular emotions or to what degree we have them. Rather, what we have control over is whether our emotions are under the control of reflective reason. I have characterised being under the control of reflective reason in terms of being under the control of our first-order evaluative beliefs and perceptions in accordance with our reflective commitments. And I have given an account of reflective commitments which supports this view. On this view, normative reflection is a means of arriving at reflective commitments and is thus a means of bringing our emotions under the control of reflective reason. Now, I want to end by pointing out that in having this kind of control we will often have control over whether we continue to have particular emotions. This is a consequence of this view, since many of our emotions cannot survive the scrutiny of normative reflection aimed at commitment. We find ourselves unable to arrive at a positive commitment to many of the things we pre-reflectively take to be reasons for particular responses. Upon reflection, many of these considerations appear ridiculous to us. It isn’t as the first-order view would hold, that we change our initial evaluative beliefs or perceptions about the world. We rather find that we can’t commit to treating the various considerations as reasons. And so our emotions dissipate. Knowing all of this we can see that normative reflection is a means of controlling our emotions in the additional sense that we can often extinguish problematic emotions merely by engaging in normative reflection. The advice of common life and self-help books isn’t so bad after all: only reflect! (But let that not be their only sermon).