De Sousa on emotional truth
Ronald de Sousa has put forward an argument that emotions can be literally true and false. De Sousa rejects strong cognitivism that equates emotions with evaluative judgments. His basic reason is the evidence of groundless emotions, such as fear of flying, that we may experience in spite of contrary, well-founded beliefs. However, de Sousa suggests that perceptions are a form of cognition that is capable of accommodating groundless emotions. The argument is based on an analogy between groundless emotions and visual illusions that may persist in the face of the knowledge that they are illusory. De Sousa admits that emotions differ from sense perceptions by being relatively opaque. When you try to describe your visual experience, you end up depicting what you see, whereas the emotional case is quite the reverse: an attempt to describe an emotionally arousing situation may succeed only in portraying the subject's own affective state. Moreover, emotions encompass objects that are accessed in different ways. They often relate to some specific target in the world, but they also represent the subject's bodily state that the former, exteroceptive perception has initiated. Yet de Sousa proposes that the liability of perceptions to assessment in terms of veridicality and illusoriness indicates that they amount to a kind of cognition.
De Sousa argues that emotional truth refers not to semantic satisfaction but to success, which is tied to the correctness of the emotional evaluation. A further distinction can be made between emotions with a propositional object and emotions with a direct object. In the former case, "E(p) is satisfied iff p is true, [while] E(p) is successful ifp actually fits E's formal object," whereas in the latter case, "E(t) is satisfied iff t exists, [but] E(t) is successful if t actually fits E's formal object" (de Sousa, 2002, p. 72). For instance, the formal object of fear is the property of dangerous. A phobic person's fear of spiders is satisfied if spiders exist but it is not successful if spiders are not dangerous. In contrast, if someone is afraid of monsters, his or her emotion is not satisfied if there are no monsters. Yet the emotion may be successful, since monsters would be dangerous if they existed. Emotional truth is thus a matter of fittingness of the particular emotional object with the relevant formal object.
The next question is how to evaluate the correctness of emotional evaluations. Here de Sousa introduces his axiological hypothesis of emotions as perceptions of value. Values are out there in the world but they are not there independently of our emotional responses, but only by virtue of them. De Sousa calls his view "axiological holism", because it stipulates that we apprehend value only in light of a complex set of factors that include biological facts, social norms, and 'paradigm scenarios' of individual biography. "It is the totality of all these factors - biological, social, personal, and more - that may properly be confronted with one another in the hope of arriving at something like reflective equilibrium" (de Sousa, 2002, p. 74).
De Sousa's account of emotional truth has two main weaknesses. The first problem is that the truth of an emotion cannot be defined in terms of its success alone. This would entail that my fear of monsters is true insofar as monsters are dangerous, whether or not they exist, which is absurd. The propositional content of one's emotion must also be semantically satisfied or the target of one's emotion must exist. For example, my fear that terrorists will attack my hometown is true only if a terrorist attack would actually be dangerous, and such an attack is actually underway or in preparation. In general, an emotion is true if and only if its actual object fits the formal object of the relevant emotion type, and the propositional content of the emotion is semantically satisfied or the target of the emotion exists or did exist.
I can obviously fear that terrorists will attack my hometown for good as well as bad reasons. Intelligence reports generally provide more reliable reasons than an astrologer's prophecy. Yet if the astrologer's prophecy by chance comes true, my fear that was supported by this prophecy may have been true, even if it was not warranted in the first place. Here is an analogy with unwarranted beliefs that can turn out to be true by virtue of epistemic luck. The truth of a belief or an emotion is then independent of its warrant. Nevertheless, the distinction between good and bad reasons is relevant in another respect. For we say that a person knows that p if and only if his or her belief or emotion that p is both warranted and true.
The notion of emotional knowledge may sound awkward. After all, we do not have a concept for such an epistemic attitude as emotional knowing as distinct from feeling an emotion. On the other hand, we talk about perceptually knowing that p by sensorially perceiving that p. Insofar as emotions involve evaluative perceptions of actual objects in terms of formal properties, an analogous account may be available for emotions. Thus, roughly speaking, a subject S has emotional knowledge that p is v if and only if (1) S feels e about p (or is disposed to feel e about p in the relevant situations), (2) p is true, (3) v is the formal property of e, and (4) S is warranted in feeling e about p. One reason why we do not have a distinct concept for this epistemic attitude is that emotional knowledge is not about emotions but about values. True and warranted emotions give us axiological knowledge about those values that emotions represent in their content, namely formal objects, as Kevin Mulligan (2010, p. 485) points out, without subscribing to this view, however. We often express axiological knowledge by the means of evaluative judgments, without feeling the relevant emotion at the time. Even so, emotions representative functions. "On this view, emotions do not merely represent formal objects as obtaining, they also represent what is to be made to obtain when formal objects obtain... An emotion is satisfied when what it successfully represents as to be made to obtain fits what it successfully represents as obtaining" (p. 762). Thus, for instance, "we can think of fear as being satisfied when it motivates the emoter to avoid dangers, of anger as being satisfied when it motivates the emoter to get back at slights, of shame as being satisfied when it motivates the emoter to repair failures to live up to an ego ideal, or disgust as being satisfied when it motivates the emoter to expel noxious substances, and so on" (ibid). The main problem with this proposal is that the proper emotionally motivated responses depend so much on changing situations that it is impossible to define them even in terms of generic actions. For instance, getting back is not always the proper response to slights, nor is expelling noxious substances to disgust as this emotion can emerge in interpersonal and moral contexts that have nothing to do with bodily contamination. Therefore, I suggest that we limit the satisfaction criteria of emotions to their cognitive representations, while recognizing their motivational function as well may provide primary access to those evaluative properties whose true and warranted ascription amounts to emotional knowledge.
Another problem with my proposed account of emotional truth appears to be that we can never know whether our present fears or hopes - as well as other future-oriented emotions - are true or false because in order to know this, we should know if the feared or hoped-for state of affairs will or has come true. Yet the moment we come to know this, our fear turns either to sadness or terror, and hope to joy, as Robert Gordon (1987) points out. Gordon argues that fear and hope are epistemic emotions, because by definition they involve an uncertain belief that p, whereas terror, sadness, and joy are factive emotions, since their subject must believe or rather know that p in order to feel the emotion. Therefore, it seems that fears, hopes, and other epistemic emotions can never be true except ex post facto. This is indeed a disturbing conclusion. Fortunately, it can be avoided, for there is a flaw in the reasoning. It lies in the erroneous equation of epistemic emotions with beliefs having an analogous content.
To believe that p is to assert p as true. Thus, when I believe that terrorists will attack my hometown, I assert the proposition that terrorists will attack my hometown as true, and this proposition is true if and only if terrorists actually attack my hometown. In like manner, a factive emotion, such as sadness about having my article rejected, asserts that my article has been rejected and that this failing entails a significant loss for me. Both beliefs and factive emotions are thus semantically satisfied if their propositional content is true. In contrast, in feeling an epistemic emotion like fear that terrorists will attack my hometown, I do not assert this proposition as true. Instead, my fear asserts that this event will - more or less -probably happen and if it does, it will be frightening. Accordingly, the fittingness of my fear depends on whether a terrorist attack on my hometown would actually be frightening, and whether such an attack is actually as probable as my fear takes it to be. The question about the semantic satisfaction of my fear then turns into a question about objective and subjective probability of an eventual terrorist attack.
Subjective probability refers to the probability that my fear and its underlying evidence together attribute to the possibility of a terrorist attack, whereas objective probability refers to the probability of such an attack in the light of all relevant evidence. My subjective probability assessment depends on my information about recent terrorist activities in my home country and its status on a proclaimed terrorist target list as well as on my reliance on the national and the international police, among other things. However, vivid and tormenting memories of the 9/11 terrorist attacks in New York City and Washington DC may raise my subjective probability estimate for a terrorist attack to the point that I become afraid of a possible attack every time I encounter terrorism-related news in the media. Yet if there is no evidence indicating that terrorists are preparing an attack on my hometown, the objective probability for such an attack is not in all likelihood high enough to render my fear fitting.
The determination of the objective probabilities of future events is obviously a difficult if not an insurmountable task, for it requires acquaintance with all relevant evidence concerning the actualization of those events, and we rarely have such acquaintance. This is because such evidence generally surpasses the evidence that is available for an individual subject at a particular moment. For instance, unbeknownst to me, the CIA or other intelligence agencies may possess information that is relevant for determining the probability of a terrorist attack on my hometown. In this case, my subjective probability assessment does not overlap with the objective probability of the event, even if the assessment is rational by my own standards. But rational warrant is not equal to truth. Indeed, objective probability may be an ideal construct, since no real person may be in the position to assess the probability of a given event in the light of all relevant evidence. Yet it is the closest approximation of truth for the semantic satisfaction of epistemic emotions. Therefore, I propose that a subject Ss epistemic emotion E is semantically satisfied if the subjective probability that the emotion ascribes to the actualization of the future state of affairs x corresponds to the objective probability of the actualization of x, determined at the moment of Ss experiencing the emotion.
It is important to emphasise that there is no fixed degree of objective probability that would render an epistemic emotion satisfied, because the satisfaction depends on the concern involved in the situation. If stakes are high, even a low objective probability may make the emotion fitting. If one takes a medical test for cancer after having discovered a lump in one's body, fear can be a fitting emotion, even if the objective probability for a positive test result were much below 0.5. In contrast, even a high probability of a future event does not make an emotion fitting if the event is not important enough. Such an emotion might be fear of missing one's customary train if the next train leaves within 20 minutes and one's schedule for the day will not be ruined by missing the train. Obviously, the threat of missing one's customary train may merit some fear or rather worry, but this must be in proportion to the significance of the event. In general, the probability and the significance of an anticipated event are inversely related when it comes to fittingness and intensity: the more significant the event, the lower objective probability warrants an intense fitting emotion about it, and vice versa. A highly probable but relatively insignificant event warrants only a weak emotion about it, if any. This concludes my discussion on semantic satisfaction as a criterion of emotional truth.
The second main problem of de Sousa's notion of emotional truth is its vagueness. The idea of a wide reflective equilibrium of biological facts, social norms, and individual experience as the standard of fittingness remains overly sketchy. Unfortunately, de Sousa does not give any example of such an equilibrium. Moreover, the notion of a reflective equilibrium does not give us any grounds for adjudicating conflicts between biological, social, and personal factors. Take, for instance, a believer in voodoo who is paralyzed on learning that he or she has been cursed. The paralysis is self-inflicted through self-suggestion, but the result still constitutes a significant harm for the subject. Should we conclude that fear of a voodoo curse is true for the believer, even if it is false for a non-believer for whom a similar curse does not constitute a danger?
There is a lively debate about the truth-aptness of moral as well as aesthetic discourses in the contemporary epistemology. Crispin Wright has introduced the property of superassertibility as the truth predicate for these discourses. He proposes that superassertibility is "the property of being justified by some (in principle accessible) state of information and then remaining justified no matter how that state of information might be enlarged upon or improved" (Wright, 1992, p. 47). Wright applies superassertibility to comic discourse about what is funny. However, I believe that his discussion is applicable to the emotion of amusement whose formal property is the funny, and likewise to other emotions as well. For it seems to me that emotions, properly understood, meet the syntactic and disciplinary requirements of minimally truth-apt states.
-  De Sousa (2011, p. 65) purports to evade my argument for the necessary role of semantic satisfaction in emotional truth. Thus he states that "I can happily say that a delusional fear, such as the fear of monsters or of God, is a false fear, on the ground that what doesn't exist isn't really dangerous." Contrary to what de Sousa assumes, I accept this response but I do not see how it helps him, because by invoking existence as a criterion for true and false fears, de Sousa subscribes to my account of emotional truth in which existence figures as the standard of semantic satisfaction for emotions that take a direct object, such as fear of monsters.
-  Andrea Scarantino (2010) has proposed another rival account on the satisfaction condition of emotions. His account is based on the idea that emotions have both cognitive and conative representative functions. "On this view, emotions do not merely represent formal objects as obtaining, they also represent what is to be made to obtain when formal objects obtain... An emotion is satisfied when what it successfully represents as to be made to obtain fits what it successfully represents as obtaining" (p. 762). Thus, for instance, "we can think of fear as being satisfied when it motivates the emoter to avoid dangers, of anger as being satisfied when it motivates the emoter to get back at slights, of shame as being satisfied when it motivates the emoter to repair failures to live up to an ego ideal, or disgust as being satisfied when it motivates the emoter to expel noxious substances, and so on" (ibid). The main problem with this proposal is that the proper emotionally motivated responses depend so much on changing situations that it is impossible to define them even in terms of generic actions. For instance, getting back is not always the proper response to slights, nor is expelling noxious substances to disgust as this emotion can emerge in interpersonal and moral contexts that have nothing to do with bodily contamination. Therefore, I suggest that we limit the satisfaction criteria of emotions to their cognitive representations, while recognising their motivational function as well.
-  For emotions that take a direct object, the formula is: t is v if and only if (1) S feels e about t (or is disposed to feel e about t in the relevant situations), (2) t exists, (3) v is the formal property of e, and (4) S is warranted in feeling e about t.
-  There is an obvious flaw in Gordon's account of factive emotions. For a person must only believe that he or she knows that p in order to feel a factive emotion that p. If Gordon were right, Romeo could not have mourned Juliet's death because Juliet was not in fact dead at the time Romeo believed this to be true. Other philosophers, such as Roberts (2003, p. 94) and Ben-Ze'ev (2000, p. 543) have made the same point against Gordon.
-  Recall my distinction between emotional authenticity and truth here. If my emotion is warranted by the totality of evidence that is available for me, the emotion is true to me even if it may not fit the world. See my discussion on emotional authenticity in Chapter 4.