From true emotions to sentimental values
The challenges of neosentimentalism
There is a popular research program in contemporary metaethics that purports to explicate value concepts and value properties in terms of appropriate or fitting emotions and other pro-attitudes. Theories that focus on emotions in this task are known as neosentimentalist theories of value as they follow the classic sentimental-ism that was initiated, among others, by David Hume and Adam Smith and developed by Franz Brentano and A.C. Ewing. Contemporary neosentimentalists include Justin D'Arms and Daniel Jacobson (2006a,b, 2000a, b, 2010; O'Arms, 2005); Simon Blackburn (1998), Allan Gibbard (1990), Julien Deonna (2006), John McDowell (1997a, b); Kevin Mulligan (1998), Ronald de Sousa (2004), Holmer Steinfath (2002), Christine Tappolet (2011), and David Wiggins (1997). In spite of their disagreements on the metaphysical status of value and the cognitive or noncognitive nature of value judgments, these theorists agree on a certain basic idea that justifies their classification into the same theoretical approach. This idea is that values and their concepts depend fundamentally on human sensibilities as they can be analysed or explicated in terms of appropriate (fitting, merited, correct, justified, warranted) sentiments or emotions. Accordingly, neosentimentalism comes in two closely associated versions, conceptual and metaphysical.
The conceptual version of neosentimentalism seeks to analyse or explicate value concepts in terms of appropriate or fitting sentiments. Sentimentalist analyses have typically been proposed for thick value concepts such as admirable, funny, enviable, shameful, disgusting, and offensive that have a one-to-one connection with distinct emotions - admiration, amusement, envy, shame, disgust, and anger, respectively. For concepts of this kind, D'Arms and Jacobson (2000b, p. 729) formulate the following "response-dependency thesis":
RDT: to think that X has some evaluative property Ф is to think it appropriate to feel F in response to X.
Value judgments are then, essentially, judgments about appropriate/correct/warranted emotions.
In contrast, the metaphysical version of neosentimentalism defines or identifies value properties by means of appropriate or fitting sentiments. The responses are the same as in the conceptual version but instead of concepts, we talk about properties. Rephrasing the response-dependency thesis of D'Arms and Jacobson for value properties, we have
RDT*: for X to have some evaluative property Ф is for X to be such that it is appropriate to feel F in response to X. Or, more briefly,
RDT**: X has some evaluative property Ф if feeling F is appropriate in response to X.
In what follows, I shall focus on the metaphysical version of neosentimentalism because I believe that it suffers from less serious problems than its conceptual cousin.
Neosentimentalist theories of value face two major problems. The first problem concerns the relation between evaluative concepts and the associated emotional responses. It seems that in order to avoid a vicious circle, we must deny that emotional responses involve those evaluative concepts that we purport to analyse or explicate in terms of the relevant responses. For instance, if we claim that amusement contains the concept of being funny, we cannot analyse this concept in terms of the emotion of amusement, because grasping the sentiment already requires an understanding of the concept of being funny that is part of the emotion. Many sentimentalists, both traditional and contemporary, have tried to avoid this problem by adopting a noncognitive view of the emotions. Unfortunately, this theory is unappealing on independent grounds, as I have argued in previous chapters of this book. Noncognitivism is particularly implausible in the case of such emotions as pride, shame, guilt, indignation, admiration, and gratitude, which emerge late in ontogenetic development and are not merely caused but felt for reasons.
Neosentimentalists sympathetic to cognitivism about emotions have purported to defend themselves against the circularity objection in two main ways. The first strategy is to admit the circularity but argue that it is not vicious because the elucidation of value concepts requires an epistemologically indispensable detour through sentiments that cannot be identified and understood without the relevant concepts (e.g. Wiggins, 1997; McDowell, 1997a, b). The detour is benign and informative because it allows us to explicate the application criteria of those value concepts in our evaluative practices in spite of remaining circular. The second strategy is to argue that emotional content can be evaluative even if it does not involve concepts. A nonconceptual view of emotions purports to alleviate the circularity worry by suggesting that emotions "involve a pre-judgmental or non-conceptual representation of values" (Tappolet, 2011, p. 129). Even so, the solution is not entirely satisfactory insofar as emotions still are or involve appraisals that -either nonconceptually or functionally - represent their formal objects, such as danger (fear), loss (sadness), or offence (anger). It does not matter if the content of emotional appraisal is conceptual or nonconceptual insofar as this content is identified or explicated in terms of the same evaluative concept that is defined by reference to the emotional content. To quote Michael Brady (2008, p. 471).
If a subject wants to know about the content of the evaluative judgment that something is dangerous, it does not help for her to be told that something is dangerous if it is correct to appraise it as dangerous. That is something that she already knows.... Insofar as evaluative judgments concern the accuracy of such appraisals, then the resulting sentimentalist account of evaluative judgment will be at best circular and at worst vacuous.
Therefore, the combination of emotional cognitivism and metaethical sentimentalism appears to be a dead end right from the outset (see Svensson, 2004, for a particularly thorough discussion of this problem).
However, it is important to observe that the circularity problem focuses on the possibility of analysing value concepts in terms of appropriate emotional responses. If we do not read the response-dependency thesis as a conceptual or semantic claim but instead as a metaphysical claim about the nature of value properties, then the thesis may retain its appeal. The idea is that a particular joke is, for instance, funny if amusement is an appropriate response to hearing the joke for the first time. It may be possible to track value properties by the means of appropriate emotions that ascribe the same properties to particular objects and events even if a definition of value concepts in terms of such responses would be circular or vacuous. The main challenge for this proposal is how to fix the relevant notion of appropriateness. Unfortunately, this task has turned out to be notoriously difficult, amounting to the second main problem for neosentimentalist theories of value.
The conflation problem refers to the difficulty of distinguishing between reasons that are relevant and irrelevant for the appropriateness or warrant of emotions. D'Arms and Jacobson (2000a) coined this problem as the "conflation problem", whereas Rabinowicz and Ronnow-Rasmussen (2004) analysed a similar, "wrong kind of reason" problem in all buck-passing theories of value. The problem is that insofar as neosentimentalist theories of value are incapable of discriminating between reasons that are relevant and irrelevant for the appropriateness or warrant of emotions or other pro-attitudes, they conflate these reasons together. Prudential, strategic, moral, and all-in reasons may often repudiate an emotion even though evidential reasons render it appropriate or warranted. We may recall or imagine situations, for instance, in funerals and other formal ceremonies, where it would be rude and imprudent and therefore all-in unreasonable to be amused by a particular event, even though the event is actually funny. Rabinowicz and Ronnow-Rasmussen (2004) focus on contrary cases, in which we have seemingly good reasons to have an emotion or another pro-attitude even though the object intuitively lacks the evaluative property that our pro-attitude ascribes to it. Their prime example is a demon who wants us to admire him for his malicious set of mind and threatens to punish us if we don't admire him. This seems like a psychologically viable reason for admiration, although we intuitively think that this kind of malicious mind does not merit admiration. However, insofar as we are unable to discriminate between relevant and irrelevant reasons for appropriate emotions, the neosentimentalist program is seriously flawed or at least incomplete.
In what follows, I focus on the conflation problem in neosentimentalist theories that restrict the scope of relevant values to those with a one-to-one relationship with some particular sentiment or emotion, such as fear, anger, shame, pride, embarrassment, and amusement. It is an open question as to whether or not other values allow a sentimentalist explication, but I do not engage in that study here as it has not been the aim of most neosentimentalist theories. Instead, I introduce the conflation problem in existing neosentimentalist theories in more detail. I then proceed to D'Arms and Jacobson's solution to this problem in terms of fitting reasons. I am sympathetic to their way of defining reasons of fit on the basis of each emotion's characteristic concern as well as to their strategy of examining actual emotions piecemeal in order to find the criteria of fitting responses. However, I argue that their project has not yielded plausible results. In general, D'Arms and Jacobson exaggerate the agent-relativity of reasons of fit for such emotions as fear and envy and the objectivity of those reasons for amusement and disgust. This suggests that we need an alternative account of reasons of fit. As such an account I investigate Danielsson and Olson's distinction between content-reasons and holding-reasons in which the former kind of reasons bear on correctness of emotional evaluation. Unfortunately, content-reasons prove to be as uninformative as D'Arms and Jacobson's reasons of fit. Finally, I introduce a proposal that emerges from my account of emotional truth, presented in the previous chapter.
-  The general approach that purports to analyze values in terms of correct or fitting pro-attitudes, whether conative or affective, is known in recent metaethical literature as the 'buck-passing' account of value, thanks to T.M. Scanlon (1998), who coined this view. It maintains that to be good or valuable is to have properties that give us a reason to take up a certain pro-attitude towards the bearer of those properties. Neosentimentalist theories are a subclass of the 'buck-passing' view as they invoke fitting or appropriate emotions in their analysis of value. Accordingly, I will focus on neosentimentalist theories of value and discuss problems of the more general 'buck-passing' account only as far as they concern neosentimentalist theories as well. The most important case is the problem of the wrong kind of reasons, raised by Justin D'Arms and Daniel Jacobson (2000b) as "the conflation problem" for neosentimentalist theories in particular, and by Wlodek Rabinowicz and Toni R0nnow-Rasmussen (2004) for all 'buck-passing' accounts of value.
-  I speak about emotions rather than sentiments in this chapter, even though some adherents of sentimentalism stick to this traditional notion, mainly to widen the scope of affective states that may ground values, as far as I can see. Since I am not a native English speaker, I am not the right person to determine the correct meaning of "sentiment". However, if we assume, with some theorists that employ this concept, that "sentiments" include all affective phenomena - emotions as well as feelings, moods, affective dispositions and traits, pleasures and pains, etc. - emotions are still the most promising candidates among sentiments to ground values. Unlike feelings and moods, emotions are intentional states directed at particular objects that they present in a certain evaluative light. By virtue of these characteristics, emotions can be assessed in terms of their appropriateness or rationality. No doubt, we sometimes talk about irrational feelings (of fear or guilt, for example) or moods (of happiness or anxiety, for instance). However, it seems to me that when we do so, we ascribe to these states a representational content and a direction of fit. Whether or not other affective states besides emotions possess those qualities is a theoretical problem that I cannot address here.
-  Brady also applies this argument against noncognitive theories of emotion such as D'Arms and Jacobson (2003) and Prinz (2004), who maintain that emotions are functionally or involve appraisals. However, their possible circularity, which concerns the semantic project of defining value concepts rather than the metaphysical task of determining value properties, is not my worry here because I have rejected these theories of emotion on independent grounds elsewhere in this book.