Emotion and Rationality
By Mark Lance and Alessandra Tanesini
This paper is concerned with the roles played by emotions in rationality, a topic which has been generally, but unjustifiably, ignored by epistemologists. Silence on this matter is, we believe, indicative of the overly narrow view that epistemologists have had of their field. Whatever else we might accomplish by considering the rational role of emotions, we hope to motivate a number of questions and philosophical contexts not commonly considered by epistemologists.
Everyone knows that rationality depends on the doxastic state of the individual. Thus, whether an action, decision, inference, or belief is rational depends on what other things the individual believes, or is justified in believing in the given situation. This holds not just for rationality, but for epistemic norms in general. Many such normative statuses apply directly to beliefs, and all depend at least indirectly on the background doxastic status of the agent. In this paper, it is our purpose to argue that rationality, and epistemic norms more generally, depend as well on the emotional states of the agent. We argue that emotions play a crucial, particular, and essential role in rational thought, that the epistemic status of an agent's doxastic states depends on her background emotions and those she is epistemically entitled to, and that the only cognitive state an agent is justified in having, in many situations, is precisely an emotion. Generally we argue that one cannot characterize crucial features of the rational epistemic agent without recourse to the language of emotions.
In developing this view, we say little about what emotions are. That is, we offer nothing like a complete theory of the emotions and, in particular, do not claim that the epistemic role we attribute to them exhausts the identifying features of emotions. Given this, we cannot claim to have argued that any possible rational agent must have states with all the important characteristics of human emotions, but we do argue that there is a crucial epistemic job performed in us humans by emotions, and that any rational agent would have to have some state or other which did this job. Since the job is quite different from any for which standard propositional attitudes are suitable, any rational agent must have at least emotion-like states. Further, we offer some speculation regarding why various other features of human emotions coexist in a state performing this epistemic function.
~1: Some observations about rational thought
Case 1: A scene from Buffy The Vampire Slayer:
Buffy, Xander, Willow, and Cordelia have just gathered around a table to discuss how to deal with a ghost who is haunting their High School, possessing students, and influencing them to kill teachers.
Buffy: Well, what do we know?
Xander: We know that dog saliva is cleaner than human saliva.
Buffy: Right. Besides that.
The point is that we all know many things. Intelligence is always in part, and often primarily, not a matter of what one knows, but of one's ability to organize that knowledge, and to determine which known propositions are salient or relevant to the issue at hand. The most rational person is quite typically not the one with the largest number of justified claims in her possession, nor the one capable of the most complex reasoning with the fewest inferential errors, but rather she who is best able to marshal those facts as fodder for reasoning which will be most useful in arriving at answers to the questions most salient in the current context. What was wanted in the scene from Buffy was not a list of things known, but candidates for information the inferential deployment of which would provide justification for a course of action vis a vis the haunting spirit.
The globally useful, but locally inconvenient breadth of human knowledge, and the consequent importance to rationality of selection among the range of facts known, applies equally well to inference. Any proposition entails infinitely many others, and stands as good reason to believe an infinite set o others that is far harder to generate. Inferring well is, thus, as much a matter of knowing which of the rationally acceptable inferences it will be intellectually fruitful to pursue in the current context as it is of knowing how to tell a rationally acceptable inference from an unacceptable one. Everyone who teaches logic has noted precisely this point, that it is far easier to see that a proof is valid than to see how to proceed in a proof in the first place, and the point is even more striking in such contexts as theoretical explanation. Any decent undergraduate science student can identify (at least in most cases) which purported explanations of phenomena are genuine, but to find the hypothesis which, through inferential manipulation, is likely to evolve into such an explanation is quite another matter.
We illustrate with an example from chess.
Case 2: a typical scene in a chess lesson: The teacher and his student have reached an interesting position in playing through the student's recent tournament game.
Teacher: Now in this position, I'd be worried about my king's safety and the possibility of a sacrifice to break up my pawn position in front of my king.
Student: No, if he plays Rg8, I'll play .....
Teacher: Slow down, I don't want to analyze yet. I'm worried about, maybe even scared of, a kingside attack.
Student: But I can attack the knight, and then win a pawn on b7.
Teacher: Look, in this position, you need first to be concerned about certain dangers, then to understand why they are threatening, then to calculate. You don't calculate how to win a knight pawn when you should be worrying about your king.
In a typical chess position, there are many things one could calculate (typically on the order of 1020 or 1030 distinct continuations). Successful brute force calculation of typical positions in chess is physically impossible, precluded by basic constraints involving the speed of light, the minimal distance between particles, and the time lapse of fundamental quantum transitions. The present point is that good players do not get worried about threats, or excited about opportunities, as a result of calculation, even in a partial and non-exhaustive manner. Rather, the worry, fear, excitement, etc. come first, and one chooses which moves to consider, and which lines to calculate, on the basis of this prior reaction to the "demands of the position". The best players, we are suggesting, are guided in what sorts of things to apply their calculative methods to in virtue of their well honed emotional competence.
There is a stage in the investigation of many problems - whether intellectual or practical - in which the mark of a rational (competent, wise, experienced, knowledgeable) agent is simply that she knows what issues to investigate, that she is able to take the right - in the sense of reasonable - issues to be salient. The competent criminal investigator has a well developed sense of what leads to follow, what people to suspect, and what features of the crime scene to treat as potential clues. The competent mathematician is drawn to certain conjectures and potential strategies of proof. A good philosopher knows that certain ideas are worth investigating for several years and others not, and that certain ideas are likely to cohere and others not. In each case this understanding - appreciation of salience, sense of where to look - is prior to the justification of beliefs, sometimes even to having beliefs, concerning the facts in question. In the case of chess, this point was summed up, in admittedly hyperbolic terms, by Jose Raul Capablanca, one of the top players in history, who was asked by a reporter how many moves ahead he looks in a game. Capablanca, annoyed by the cliché question replied: "Just one, but it is a very good move."
Now precisely the fact that this feature of intellectual competence occurs prior to the formation of beliefs, or even of explicit hypotheses to test, may suggest to some that it has little bearing on epistemology proper, since the latter is concerned not with rational behavior within the context of discovery, but only with the context of justification. Apart from our general sense - i.e. our intellectual suspicion - that any program which pursues a radical separation between the justification of hypotheses and the rationality of investigators studying them is itself a prime example of a misguided orientation to the issues, the current separation cannot work.
Any epistemology which rejects the quest for Cartesian certainty will concede in one way or another that a belief can be justified even though one has failed to rule out certain logically possible alternative views. In some versions, this point shows up in a requirement that we rule out "relevant potential defeators," or that our belief forming mechanisms be reliable across the range of "relevant alternatives to the actual situation". Whether it is recognized explicitly or not, then, any non-Cartesian epistemology will have to recognize that one's normative relation to various alternative possibilities is a crucial determinant of one's justificatory status. Whether one is justified in believing P will depend on whether one was justified in ignoring possibility Q which is incompatible with P.
For illustration, let us put the point in reliabilist terms. (Of course we don't for a minute wish to endorse reliabilism as a general theory of epistemology.) Suppose that both S and S' have failed to investigate possibility Q, and that in both cases they have done so because of some internal state, emotional or otherwise of the sort discussed above, which led them to take seriously P, P', P'', but not Q. In both cases, the agent then comes to believe P, and for the same reasons, but suppose that the mechanism which led S to ignore Q was highly unreliable, a pre-doxastic sense which regularly leads her to ignore true and important hypotheses, and often leads her to pursue frivolous and false ones. S's pre-doxastic sense, on the other hand, is reliable in the same manner. We take it that in this case we would have reason to think S' justified in believing P and S unjustified, even though the narrowly justificatory mechanisms employed to support P - i.e. those other than the attention focussing mechanisms in question, are the same. And we take it that similar points could be made from the perspective of any plausible epistemology. Whether one is justified in believing P on the basis of R, can depend upon whether one was rational in never bothering to look into the possibility of Q. So if there is a sort of mechanism which determines whether one does look into Q, justification will depend upon having such a mechanism in good rational working order.
Thus, it is not merely the rationality of actions which can depend on the having of the appropriate pre-doxastic salience generators. Jones's belief that Nf6 is the best move in a given situation may well derive its justification, in part, from the fact that Jones has a reasonable overriding concern for the safety of her king in the given position and, hence, no lust for that pawn on the Queenside. This concern does not flow from any beliefs about, or inferences involving, the merits of each course of action, but rather takes the form of a failure ever to form hypotheses about the pawn-grabbing strategy.
When we move away from a precisely constrained and well defined inferential environment like chess, to the sort of messy and unbounded context either of ordinary life or of scientific inquiry, the plausibility that we could do without any salience generator is even less. Can we so much as imagine a rational Buffy who addressed issues concerning the bacteriological content of canine saliva, and every other such issue, on the way to formulating a plan of what to do regarding the ghost? We have already noted that there are limits which make it physically impossible to consider every possible continuation even in a game of chess. Given that one possibility, and often the most rational one, in any inferential situation is to augment the language by the introduction of new conceptual resources, the impossibility in ordinary situations of a non-directed strategy of inference checking seems logical. It isn't too Buddhist a way to put the point, we hope, to say that competent failure to think is a large and essential component of competent thought.
The picture we are urging, then, is one according to which there is a sort of modular division of epistemic labor. Some kind of pre-doxastic sense - apparently one which typically makes use of emotions - leads us to investigate a certain range of hypotheses, issues, conjectures, etc. After having been so led to the right bushes, we beat around in them by marshaling various inferential competencies in an attempt to secure entitlement to the assertion or denial of claims we find there. Here too, crucially, our choice of strategies for justification, of what inferences to pursue, what hypotheses to test, is guided by similar salience generators.
The upshot of our overly brief discussion of the importance of ignoring most potential objections suggests that no strategy for justification of P, short of deductive proof from indubitable premises, will constitute a status of justification in itself. It must always be assessed qua justification by an agent who is already warranted in considering those claims in the vicinity of P and ignoring many others. It is argumentative strategies in the context of salience attunements which fall within the proper scope of justificatory norms.
This is not to say that epistemic entitlement to such pre-doxastic salience generators is categorically prior to justified belief, that one could be justified in strategies of investigation without already being justified in lots of particular beliefs the consideration of which would result from the employment of such strategies. Quite the contrary. Though it may well be that in each case the directing of attention to some subset of the space of propositions precedes the believing or justifying of any propositions within it, one who did not count as justified in many beliefs could not count as entitled to any focus of attention.
Obviously highly theoretical beliefs about, for example, the supervenience of thought on material facts about the world make it rational to turn our investigatory focus away from certain sorts of paranormal explanations of various phenomena. But the fact that the epistemic status of salience generators depends upon the agent's having justified beliefs and vice versa, does not generate a pernicious sort of circularity, for neither sort of dependence is specifically justificatory. It is not as if we need first supply as a premise that it is reasonable to consider P, and then start the justification of P, nor that we must offer up a justification of P as prior grounds for our inclination to pursue the truth value of Q.
The inclination to consider certain claims is non-inferential, but as Sellars argues famously in "Empiricism and the Philosophy of Mind," non-inferential justification is both a normative and a contextual matter. To say that one is non-inferentially justified in believing or doing something is a species of saying that they are entitled to it, the species which says that they are entitled without deriving that entitlement inferentially from something else. But to say that the entitlement does not derive from a process of inference is not to deny other forms of dependence upon background beliefs. In particular, it is perfectly consistent to say that only an agent who also holds a range of justified beliefs may be so entitled.
Thus, the point is simply that only when one's cognitive state reaches a level of complexity such that one is entitled to many salience inclinations and to many beliefs is one entitled to either. There is a division of epistemic labor within the cognitive structure of the knowing subject, but only someone who is competent at both jobs can count as competent at either.
~2: Some elementary facts about emotions
Though we make no claim to offer a complete account of the emotions, we need to note certain facts about them if we are to make plausible our claims concerning their epistemic role. First a minor and uncontroversial point: emotions are not (or at least not only) affective states. Even feelings are probably not identifiable with purely affective states, but in any event emotions clearly involve much more.
One reason for this conclusion is that there are far more emotions than there are affective states, and in many cases two quite distinct emotions typically involve exactly the same affective states. More important, however, emotions are typically conceptually structured. We are afraid of the tiger, worried about the state of the economy, or lustfully attracted to that boy in A Death In Venice. We hasten to emphasize that this is not to say that emotions have beliefs as parts, or are propositionally structured. Typical emotional states do not entail the having of any belief. One can be afraid without believing that the object of fear is dangerous, or be in love without believing any particular good fact about the individual. And we don't even know what candidate would be put forward as the belief entailed by titillation, disgust, or solicitude. Nonetheless, emotions are individuated conceptually, typically in terms of a nominative object. Of course there is a large literature arguing that emotions are, or entail, beliefs. Rather than dealing directly with these arguments, we focus on the role emotions play in rationality, and then argue that the intuitive fact that they don't necessarily involve beliefs is explained by the fact that no belief state could perform their function in rationality.
To say that emotions are not equivalent to beliefs is not to deny that they can play a role in justification. Clearly they have a role in practical inference: one can run out of the room because one is scared of the tiger, and not merely in the causal sense of 'because'. The emotion can be the justification for the action and we have been building the case that they can play a role in theoretical inference.
The intentional object of an emotion can even be a proposition: one is afraid that the grease will catch fire. Indeed, the grammar of emotion language is such that one can meaningfully insert any nominalization into the scope of most emotion terms. Thus, tokens of particular emotional attitudes are potentially as conceptually rich as any cognitive attitude. But even in the case of those emotions which take a propositional object, they do not consist in a belief. To be afraid that the grease will catch fire is not necessarily to believe that it will and emotions, though conceptual, are not propositional attitudes in the sense of requiring a propositional content.
Finally, a point which will be of significance at a later stage of our argument: to have a particular emotion often requires that one be involved in particular social practices. Cross-cultural and historical psychologists have argued for many examples of emotions which exist only within particular cultures, and which seem to require the particular practices of that culture in order to make sense. To pick one example almost at random, Rom Harre discusses the emotion of acedia, which is, he says, something like a particular sort of alienation from, and lack of enthusiasm for, the doing of routine labor to the glory of god. Such an explanation is rather misleading, however, for it suggests that it is an emotion we are familiar with - a combination of alienation and boredom - with an unusual object tacked on. In fact, though, it was experienced as a unified and sui generis emotion.
The hearing of a Tarzan yell provides a good analogy to our point. The yell is, in fact, a combination of three common sounds -a train whistle, a human voice, and some kind of horn. Nevertheless, we hear it simply as a Tarzan yell, with a sensorious quality all its own. Acedia was not only common in medieval society, but much discussed by moral philosophers, theologians, and psychologists, who devoted long discourses to the evaluation, diagnosis, and treatment of those suffering from it. Our point is that without the existence of particular theological beliefs as well as the practice of assigning particular routine tasks to people, especially monks, as part of a ritualized routine of spiritual devotion through labor, the very cognitive space for this emotion evaporates. The point here is not that one cannot have the idea of the emotion except in the context of the relevant practice. We can have the idea of the emotion from reading Harre's discussion. The claim is that the emotion itself is a particular kind of rational comportment towards the practice, an intellectual orientation within it. Thus, while we can have an idea of the emotion, we cannot have the emotion.
The dependency of emotions on social context is also illustrated by the fact that, in many cases, the distinction between various categories of emotions has to do with the kind of claim that is likely to be in the air in the presence of that emotion, or with of the sort of context in which the emotion is appropriately attributed. Thus, the distinction between jealousy and envy has to do with the fact that jealousy typically involves some sort of allegation of moral violation of one's entitlements. This is not to say that to be jealous is, necessarily, to believe that one has been morally wronged, one can be jealous and yet recognize that there are no grounds, but it is to say that jealousy can only arise in contexts in which the jealous person is able to postulate standards of morally correct behavior, especially vis a vis her rights over things or people.
Our suggestion is that an emotion of jealousy is precisely an attunement toward claims involving moral wrong to oneself. Those with the emotion share an epistemic orientation which assigns importance to the settling of issues of moral entitlement, and as such must exist within a social context within which such issues make sense, but they need not share any specific propositional commitments.
~3: Why emotions?
Supposing that one is willing to grant that a competent sense of focus or attunement is essential to rationality, the question remains as to why the states which generate this focus should be emotions. So far, we have merely pointed to some typical examples, with an eye to suggesting that it typically is emotions which serve this purpose. It seems hard to deny that emotions can be rational or irrational, and that actions, inferences, and beliefs are themselves often rational as a result of their relations to justified emotional states. But we would like to understand whether it is mere chance that emotions serve as the rational salience generators in humans, or whether there is some more significant unity to the various features of these mental states.
We consider three features of emotions: their intentional focus, their social embedding, and their affective quality. As for intentional focus, it is already clear that this characteristic is of a piece with their rational utility. As for social embedding, this fits naturally with the role of emotions in rationality insofar as one conceives of theoretical and practical rationality as themselves essentially social. For ourselves this is quite far indeed. In §5, we address certain key social elements of rationality which seem deeply tied to and only explicable in terms of the emotions, and in a fuller treatment of rationality, we would emphasize many systematic ways in which it is essentially embedded in society.
Finally, there is the matter of the affective quality of emotions. Here, plausibly, is an aspect of emotion tied to the particulars of human embodiment, but to say this is not, on our view, to disparage its importance. Indeed, among the reorienting morals we would like epistemologists to draw from the investigation of emotions, the need to see rationality not purely in formal-computational abstractions, but as an embodied practice essentially dependent upon the mechanisms of the body and their rationally choreographed integration with environment and society figures prominently.
Insofar as affective states are involved with behavior - and there is a great deal of psychological literature tying many of the affective states of emotions, moods, and feelings, to particular brain chemicals which themselves stimulate or facilitate particular sorts of bodily reaction - and insofar as the rational upshot of emotional attunements and the consequent beliefs is a rational demand for action, we will have an explanation of the importance of associating an emotion with an affective feel. For example, if flight is a practical consequence of typical beliefs which would be rationally arrived at after a reaction of fear, and if we humans flee better in the presence of brain chemicals which also generate a given affective state, then we would understand the association of fear with that affect.
Is there anything more general and essential to say in this connection? We are inclined to think there is, but to discuss this matter would take us rather far afield, and our thought on the matter is far from settled. What difference would it make to the points we are arguing here? If there is nothing more essential, then we could not claim to have argued that the salience generating states essential to rational agency must necessarily have the sorts of affective quality of many human emotions. Thus, if one wanted to think of these affective states as essential to something's being an emotion, she would not think of emotions as necessary for rationality. For ourselves, we would be more inclined to say that this showed that affect is inessential to emotion, since the role in the regulation of rational attention seems so much more significant, but little hangs on this. In any event, there is a particular task of rationality performed not by belief-like states, but by another sort of which human emotions are a species.
Before leaving this section, we offer the following important case described by the neurobiologist Antonio Damasio, whose work we became acquainted with after writing most of this paper.
Case 3: This case involves a patient "Elliot" who suffered from a tumor, specifically a meningioma, which "had begun growing in the midling area, just above the nasal cavities, above the plane formed by the roof of the eye sockets. As the tumor grew bigger, it compressed both frontal lobes." Surgery was required resulting in significant frontal lobe damage, specifically to areas known to govern emotional response. The description of the resulting impairment is telling:
"Not only was Elliot coherent and smart, but clearly he knew what was occurring in the world around him. Dates, names, details in the news were all at his fingertips. He discussed political affairs with the humor they often deserve and seemed to grasp the situation of the economy. ... Elliot's smarts and his ability to move about and use language were unscathed. In many ways, however, Elliot was no longer Elliot.
"Consider the beginning of his day: he needed prompting to get started in the morning and prepare to go to work. Once at work he was unable to manage his time properly; he could not be trusted with a schedule. When the job called for interrupting an activity and turning to another, he might persist nonetheless, seemingly losing sight of his main goal. Or he might interrupt the activity he had engaged, to turn to something he found more captivating at that particular moment. Imagine a task involving reading and classifying documents of a given client. Elliot would read and fully understand the significance of the material, and he certainly knew how to sort out the documents according to the similarity or disparity of their content. The problem was that he was likely, all of a sudden to turn from the sorting task he had initiated to reading one of those papers, carefully and intelligently, ... Or he might spend a whole afternoon deliberating on which principle of categorization should be applied ... One might say that Elliot had become irrational concerning the larger frame of behavior, which pertained to his main priority ... His knowledge base seemed to survive, and he could perform many separate actions as well as before. But he could not be counted on to perform the appropriate action when it was expected.
"To date we have studied twelve patients with prefrontal damage of the type seen in Elliot, and in none have we failed to encounter a combination of decision-making defect and flat emotion and feeling. The powers of reason and the experience of emotion decline together, and their impairment stands out in a neuropsychological profile within which basic attention, memory, intelligence, and language appear so intact that they could never be invoked to explain the patients' failures in judgment.
It is nice when one's a priori arguments receive empirical verification, but it may be thought that though these results are welcome, they do not support our account of the essential role of emotions in rationality. On our account, patients without emotional faculties should be even more impaired, not merely at large scale focus, but small scale as well. They should not calculate, reason, even take tests well, because they should fail to focus on particular inferences as much as on particular topics.
We have a suggestion. Note that the standard methodology for determining a connection between a site in the brain and a cognitive ability is essentially correlational. We determine that people with damage to a given area, and not to others, lack particular functions. That suggests that any reason to make such an association depends upon a prior understanding of what is involved in the cognitive or mental function. Not surprisingly, given the received opinion, emotions are understood in the psychological literature in a way far more limited than the one we are urging. As we emphasize in the next section, the fundamental emotions turn out, on our account to be things like interest in, attention to, etc. And these are precisely the intellectual emotions which are essential to all forms of cognitive competence.
They are also the emotions least likely to be recognized as such by psychologists and neuropsychologists. These scientists, following on most of the western intellectual tradition consider only more specific and obviously affective emotions such as love, anger, fear, etc. Why this focus would obtain is an interesting matter of historical speculation. We tend to think it is deeply tied into the history of patriarchy within which there have been regular attempts at least as far back as Aristotle to associate women with emotion and men with reason. Now if obviously intellectual emotions like interest, curiosity, and theoretical focus, were taken to be emotions, this dichotomy would have been absurd on its face. Thus there was a need to draw the line at a point which excluded these from the category.
Our unargued speculative suggestion, then, is that ideology pushed categorization, that categorization led correlational neuropsychology, and as a result, tests like those of Damasio actually concern people who have only lost certain specific emotional functions while maintaining more basic generic ones. That even they are so severely cognitively impaired is strong support for our orientation to the issues.
~4: Why emotions are not beliefs
Emotion concepts, as we have noted, need not have a propositional object. In many cases, the object of an emotion is prepositional. One is envious of Lyle Lovitt, or excited about the upcoming superbowl. But some emotion words allow for propositional objects as well, as when one is worried that capitalist commodification is not amenable to piecemeal reform. In any case, there are clearly (at least) three semantically significant elements of each emotion sentence, the agent, the particular emotion, and the object of the emotion.
We suggested above that the epistemic point of an emotion is to direct attention toward the object, in the sense of making salient issues regarding that object. If we are right that emotions play this epistemic role of directing the attention, this suggests that something like interest is the most generic emotion. That is, we want to claim that all emotions involve some specific sort of interest in the object of that emotion.
We mean interest in a thin way, as implying nothing but a direction of the attention. Thus, our claim should not be seen as incompatible with the fact that boredom is an emotion. Of course generalized and undirected boredom is not an emotion on our account, but merely a feeling, but boredom with government sex scandals is an emotion and our point is that to be bored specifically with government sex scandals is a way of being interested in them as we are using the term 'interest'. It is, as Heidegger would say, a comportment toward them. As we put it earlier, it is a way of taking them to be salient in one's intellectual life, though in this case with the goal of making them go away, at least from one's consciousness.
Our present need, in order to argue for the centrality of emotions to rationality, is to clarify the way in which any state playing the salience-generating epistemic role of emotions differs from the beliefs and other propositional attitudes which are the usual concern of epistemology. The point is quite simple. To direct attention to, say the Tiger in the way that fear does is to treat as equally salient both "the tiger is dangerous" and "the tiger is not dangerous". Part of what it is for the tiger to have been thrust upon one as salient to possible danger is for the issue of whether it is actually dangerous to become central to one's investigative activity. One is concerned to see whether it is securely chained up, whether one can successfully run away, etc. But if the relation is to two incompatible propositions - and many others besides - then it is not a belief. An emotion is not, in itself, an undertaking of justificatory responsibility regarding a content, but is an attending to a range of possible contents one might, upon examination of the evidence, come to undertake a commitment to, a set that will not even purport to internal consistency.
If one were to try to assimilate this sort of salience-generating mental state to the category of belief, the best attempt would be as the belief that "one ought to attend to sentences in class C," where C is some class of relevant propositions, or "one ought to attend to topic T". There are several problems. The first - also the deepest and the one we most obviously cannot pursue within the bounds of this paper - is that this suggestion presupposes that normative judgments are actually beliefs. This is a questionable presupposition, having to do both with the proper understanding of normative judgment and with how we ought to understand the category of belief. For now, suffice it to say that one can think of normative assertions as a kind of speech act which has as much in common with an imperative as with a declarative, and that to assimilate its role in reason to that of a belief raises several subtle issues.
A second objection, though more shallow, is easier to pursue within the context of the present discussion. This has to do with the possibility of specifying the class of claims we have labeled as C above. If the only way to specify the class is as the class of sentences which the emotion makes salient, then the suggestion that we understand the role of the emotion in terms of the role of a belief will be circular. In fact, it isn't even clear to us that there is, in reality, a fixed and specific set of sentences which the emotion directs us toward. Our conception of rationality is as ultimately reliant on skill. What rational emotional attunement gives one is a skill at navigating the conceptual world of investigation. Just as one can know how to ride a bicycle, or to assess the significant dangers of a chess position, without being able to state a set of principles determinative of that skill, so our claim is that emotional attunements are ways of being directed toward a set of issues and concerns which operate on the level of skillful behavior. To have such an attunement is to be able to attend to the right things, not to be able to state what these are, much less to capture theoretically how it is that one attends to them.
Nor does it help to say that the belief is that one ought to attend to the subject-matter of the belief, for part of the skill generated by the emotion is that it direct one to particular features of the subject matter, specific claims that call for investigation. When one is afraid of the tiger, not all aspects of tiger-studies become salient. One is not concerned with the tiger's blood type - unless one has a blood-type specific poison at the ready. Thus, again, the epistemic role of the emotion is not captured by any belief, for the emotion makes salient the subject matter in a specific way, one that directs our attention to certain sentences and not others.
So our first claim about the function of emotions is that in us humans they serve the role of salience generators. They direct our attention toward a loosely bounded class of sentences and recommend them as properly worthy of the sorts of epistemic attention we are used to considering: as worth gathering evidence for or against, as worth considering as sources of evidence for other claims, as worth acting upon. Not only do they perform this job, but the job is different from, and prior to, the sorts of jobs done by belief and inference.
Further, not only rationality, but justification and knowledge depend upon being well-attuned emotionally. Of course it can be reasonable to run away from the tiger, because it is reasonable to be afraid of it. It is also reasonable to think that someone will change the subject, because it is reasonable to be bored with the discussion. More importantly, one can be unjustified in believing in his friend's innocence because he should have been suspicious of her story, or rationally inclined to doubt that an accepted theory is defensible because he is interested in an apparent anomaly. In short, what one is justified in believing depends upon what possibilities, what issues, one is justified in ignoring. The choice to consider or to ignore a given issue is one that is based upon a tacit skill at navigating the space of possible issues. Thus, rational navigation of that space is a precondition for the holding of justified beliefs. Put another way, the significant difference between two people, one of whom is justified in believing P and the other of whom is not, can be that the former has and the latter lacks the relevant emotional attunement.
~6: Dialogical emotions and rational communities
We are particularly interested in certain interpersonal emotions which serve a crucial role in constituting rational discourse. We discussed earlier how social contexts are required for the existence of certain emotions, how those emotions only make sense and serve their particular cognitive role against the background of particular social practices. What the present examples will illustrate is the converse of this, that certain emotions serve to constitute practices.
This idea is not new. Indeed it is a familiar view in ethics and political theory to suppose that various sorts of emotional engagement are constitutive of moral community. What we want to emphasize here is the way that they are constitutive of rational dialogical social practice. To illustrate this is to take issue with a particular sort of Kantian ethicist, namely the sort who actually follows the views of Kant. This sort of Kantian denies that emotions are essential to moral life on the grounds that what it is to be a moral agent is merely to be a rational agent, and that morally correct behavior is reducible to some sort of practical rationality accessible by any rational agent. The current discussion undercuts this move by attempting to show that rational agency itself requires emotional involvements of a particular sort.
Fallibilism is generally thought to be an intellectual virtue. We take this not only to be true, but a basic fact of rationality. That is, rather than following by analogy a consequentialist account of virtue and saying that to be an intellectual virtue is to be the sort of state which leads to truth, we are inclined to say that rationality is inherently and fundamentally a matter of virtue, and that our grip on likelihood of reaching truth is always dependent upon a prior assessment of intellectual virtue. But for now, this dispute need not detain us, since our goal is only to consider what the virtue is, not that in virtue of which it is a virtue.
Attempts to capture the intuitive virtue of fallibilism while producing an account which leaves it recognizably virtuous have always run into difficulty. Intuitively, fallibilism is the thesis that epistemic agents ought to exhibit an appropriate level of intellectual humility, ought to recognize that their epistemic practice is fallible, that in some sense they might be wrong.
The most natural readings of the claim that one might be wrong, however, clearly fail to capture the virtue in question. Certainly the ideal is not that we ought to take none of our beliefs to be necessary truths. We, to speak only for ourselves, have many beliefs about modal matters, and take many such claims to be necessarily true. But surely those of us who embrace the virtue of fallibilism think it a virtue even when we are discussing necessitives.
Nor does the obvious epistemic reading seem to work. On this reading, fallibilism would come to the view that one's evidence was never sufficient to guarantee any claim, or that one's evidence never makes it rational to assign subjective probability of 1 to any claim. This is a principle which is somewhat more defensible than the "nothing is necessary" principle, but still seems to miss the point. We can think, quite reasonably, that we do have sufficient evidence, argument, whatever, to make a claim certain, without abandoning my epistemic humility in the relevant sense. Conversely, merely thinking that my claim is not certain is no mark of humility as is evidenced by any manner of many actually existing Bayesians.
What we want to suggest is that we think of fallibilism - epistemic humility - as a kind of emotional state directed toward other agents in the dialogue, or toward positions incompatible with one's own. To be epistemically humble is not to believe that it is metaphysically or epistemically possible that one's view be false, but to be open in the right way to alternative views. It is to take as salient - specifically as placing upon one a demand to listen attentively, attempt to understand, and assess fairly - views put forward as challenges to one's own.
We are all familiar with the phenomenon - sadly at least as common in philosophical circles as it is in the wider world - of a person who is perfectly rational in all those ways capturable as logical principles, a person who keeps their views consistent, gives reasons which relate in the right ways to their conclusions, marshals evidence, makes distinctions etc., and who nonetheless there is simply no point whatsoever in talking to. The familiarity of this phenomenon, the importance to the progress of rational dialogue of our not being this way and recognizing those who are, and the resistance of such dialogical irrationalities to principle based cures suggests that epistemology must take account of the distinction between those who really listen and attempt to understand another's position and those who are unwilling to listen and dogmatic, yet able to be so with logical facility. This distinction is constituted by interpersonal emotions, emotions which thereby serve to constitute us collectively as an intellectual community.
~6.2: Other interpersonal emotions
Similar remarks apply to many other intellectual emotions. Since we cannot begin to discuss them all here, or any in sufficient detail, we confine ourselves to imploring epistemologists to take up the issue of the proper understanding and regulation of such crucial rational emotions as openness (to new ideas), suspicion (of certain explanations), critical distance (from one's own and one's teachers views), interest (in specific topics and leads), sympathy for (unusual opinions), hostility to (foolish approaches to a problem), recognitions (of salience, as in for example certain medical results to questions in epistemology), curiosity (for new data, as in the Popperian imperative), discomfort (with theoretical tensions and anomalies), etc.
Our discussion of emotions, then, leads us to epistemic virtues, but not virtues understood in a derivative sense as that which reliably leads to truth, justification, or some other epistemic good. Rather, we seem to be heading towards a robust sense of epistemic virtue, toward a virtue epistemology which must grapple with such issues as the unity of the rational virtues, the nature of skills and embodiment, with particularism vs. regulism, and with the nature of normativity. To so focus is to depart radically from an epistemological enterprise captivated by internalism and externalism, foundations, or the fourth condition on knowledge.
So much the better.
(1) In a recent presentation one of us was claiming a certain virtue for an account of mind that involved connecting a Heiddeggerian notion of embodied tacit skill and a version of Sellarsian inferentialism. A questioner in the audience asked whether it wouldn't be possible to combine Brandom's account of language with a Cartesian dualism about mind, and whether there was an argument against such an approach. Though lacking any knock-down argument, we take it that being able to distinguish non-inferentially between the wisdom of pursuing the first connection and the foolishness of the second is a symptom of philosophical competence.
(2) Much has been made of the case of chess competence by Hubert and Stuart Dreyfuss. They use this case to build an argument against a purely computational conception of understanding, in favor of a notion in terms of tacit skills. Much of what we have to say here draws on and develops this work. D and D also develop their criticisms of AI from an explicitly Heideggerian perspective, one developed in most detail in Dreyfus's Being in the World, MIT Press, 1991. Though we don't pursue the historical points here, the idea that such epistemic competence is deeply indebted to emotional states which serve to align our comportment toward various issues is, at least on the surface, quite along the lines of Heidegger's discussion.
(3) We take the account of the historical case on faith from Rom Harre. See The Social Construction of Emotions, Blackwell, 1986.
(4) In this regard, Rom Harre reports an experiment in which people were asked to state whether people in a situation were jealous or envious. See The Social Construction of Emotions.
(5) And we have each done some work in this direction. See Tanesini: : An Introduction to Feminist Epistemologies, Oxford: Blackwell, 1998, and Lance: " Some Reflections on the Sport of Language," Philosophical Perspectives, 12: Language, Mind, and Ontology, Jim Tomberlin editor, Blackwell, pp. 219 - 240, 1999, and "The Word Made Flesh: toward a neo-Sellarsian view of concepts, analysis, and understanding," Acta Analytica. Volume 15, Issue 25, 2000, pp. 117-135.
(6) One idea is that the affective quality of emotions allows us to track, or to keep in order, what our bodies are doing. This idea figures in the neurobiological discussion of Antonio Damasio which we discuss below. Damasio's discussion here is difficult to understand and evaluate, however, because he seems to treat as far too similar emotions and feelings, and to ignore the intentional, conceptual aspect of emotions altogether. Despite these problems, we feel this line of thought worth pursuing.
(7) One of us has written in some detail on each of these topics. On normative claims, see The Grammar of Meaning, Cambridge University Press, 1998, esp. chapter 3. On belief, see "Some Reflections on the Sport of Language".
(8) A number of people have been helpful to us in the development of the ideas in this paper. In particular, Rebecca Kukla, Richard Manning, and Maggie Little discussed these issues in depth with us after reading an earlier draft of the paper. Rob Stainton made many helpful comments on the penultimate version. We should also mention that the view here has some features in common with that of Catherine Z. Elgin, who also talks of emotions as salience generators in Chapter 5 of Considered Judgment, Princeton, Princeton University Press.