Appropriateness as correctness
Another prospective solution to the wrong kind of reasons problem has been inspired by Franz Brentano's distinction between truth and correctness as analogous epistemic standards for beliefs and pro-attitudes, respectively. Sven Danielsson and Jonas Olson (2007) have invoked this distinction in suggesting that pro-attitudes can be correct or incorrect in the same way that beliefs can be true or false. In the same way as the right kind of reasons for beliefs bear on the truth of beliefs, rather than on the beneficiality of having the belief, so too the right kind of reasons for pro-attitudes bear on the correctness of those attitudes rather than on the beneficiality of having the attitude. Accordingly, Danielsson and Olson introduce a distinction between two types of reasons, content-reasons and holding-reasons. Content-reasons provide arguments for the correctness of a pro-attitude, whereas holding-reasons are reasons for having the relevant attitude. Content-reasons also give rise to defeasible holding-reasons for attitudes. "Just as we ought in most cases, to have true beliefs or at least avoid having false beliefs, we ought in most cases to have correct conative attitudes or at least avoid having incorrect attitudes" (ibid., p. 519). Such holding-reasons derived from content-reasons also qualify as reasons of the right kind. For instance, if we have a content-reason to admire a talented artist, we also have a holding-reason to have that correct attitude. However, the opposite is not the case: holding-reasons do not give rise to content-reasons. We have a holding-reason to admire the evil demon in the problem case because he threatens to punish us if we don't admire him, but this reason does not render the demon admirable even if it may warrant this emotional attitude from an all-things-considered perspective in spite of its incorrectness in such an abnormal and bizarre case.
The distinction between content-reasons and holding-reasons seems capable of providing plausible responses to hard cases. We have holding-reasons to admire the evil demon, but these reasons do not render his malevolence admirable because they do not provide reasons for the correctness of this attitude. However, the distinction remains insufficient insofar as we attempt to identify value properties in terms of sentiments that are warranted by content-reasons. The problem is that the analogy between the truth of beliefs and the correctness of pro-attitudes breaks down on a closer analysis as there is an important distinction in their respective truth and correctness conditions. To quote Jennie Louise (2009, p. 352),
Truth predicated of belief is not primitive, but rather rests upon the truth of the proposition that is believed. But the point of the normative-priority account is to deny the analogous claim about correct pro-attitudes: the truth-analogues for conative attitudes would be evaluative predicates, so that the correctness of admiration would depend on the admirability of the thing that is admired. Normative priority says that there are no analogues to propositional truth in the case of pro-attitudes. The normative-priority account of correct attitudes therefore has a different structure to that of true belief.
Since pro-attitudes are correct if and only if there are reasons for their correctness, the content of pro-attitudes cannot provide a correctness condition for them in the same way as the content of beliefs provides the truth condition for beliefs. Consequently, Danielsson and Olson and other normative neosentimentalists cannot appeal to evaluative properties as the basis for correct attitudes. They can appeal only to the descriptive properties of objects that provide content-reasons for the pro-attitudes in terms of which the relevant evaluative properties are explicated. Thus, for instance, 'x is admirable if and only if it has properties that provide content-reasons for admiring x'; which amounts to the same as 'x is admirable if and only if it has properties which make admiration of x correct'.
The main problem with this account is that it is uninformative. It provides formal criteria for those emotional responses that are relevant for the identification of corresponding evaluative properties. Even so, we should know more about the correctness conditions of emotions and other pro-attitudes. To quote Deonna and Teroni (2012, p. 48):
Without further specification of the nature of these correctness conditions, we are left with the hardly informative claim that the reasons relevant for the F[itting] A[ttitude]-analysis of evaluative properties are those reasons that make the content of the relevant response correct. This looks more like a way of restating the problem than a way of resolving it.
Therefore, Danielsson and Olson's notion of content-reasons for the correctness of pro-attitudes does not make any headway from D'Arms and Jacobson's proposal in which reasons of fit speak to the characteristic concerns of emotions, defined in terms of formal objects of emotion. In both cases we know that a particular object of emotion has the relevant response-dependent value property if the emotional response is warranted by either content-reasons or reasons of fit, respectively. Yet both accounts are formal as they do not aim at specifying those descriptive properties of particular objects that provide content-reasons or reasons of fit for the relevant emotional responses to those objects. This problem leaves the task of fixing response-dependent value properties by means of corresponding appropriate emotions seriously incomplete.
-  Danielsson and Olson point out that Brentano originally made the distinction between judgments ('Urteile') and conative attitudes ('Gemütsbewegungen, 'Lieben und Hassen'), conceived as psychological attitudes. Beliefs are indisputable examples of judgments that are capable of being true and false, whereas the contemporary term of pro-attitudes refers to conative and evaluative attitudes for which correctness may qualify as an analogue of truth. However, Danielsson and Olson do not aspire to historical accuracy in their interpretation of Brentano,
-  For the distinction between normative and descriptive neosentimentalist theories of value, see Tappolet (2011). Descriptive theories suggest that emotions are perceptual experiences of value and correct when they represent objects accurately as these are. Accordingly, for these theories, the correctness of an emotion is distinct from its justification and therefore does not depend on its warranting reasons but on the epistemic accuracy of its evaluative content. Yet the problem of distinguishing value-tracking emotional responses from incorrect ones remains for these theories as well, even if the problem is framed in different terms that highlight the purported analogy between emotion and perception. However, there are reasons to suspect the solvability of this problem, as for instance Brady (2011) and Salmela (2011) have argued. Therefore, I do not discuss this version of neosentimentalism here.
-  The former formulation is adapted from Danielsson and Olson's (2007, p. 520) claim that x is good means that x has properties that provide content-reasons to favor x, whereas the latter is adapted from Louise's (2009, p. 353) rendering of the same claim as x is good means that x has properties which make favouring x correct.