Nonconceptual appraisals as constitutive causes of emotion
Robinson's argument for noncognitivism is founded on distinguishing the non-cognitive elicitation of emotions from their cognitive monitoring and treating the former as decisive in determining the identity of emotion. Robinson admits that conceptually complex thoughts and beliefs sometimes participate in the evaluation of emotion-eliciting stimuli at early stages of appraisal. Yet the affective appraisals that ultimately elicit emotions and function as their components are nonconceptual and noncognitive. "Emotions are processes in which a rough-and-ready affective appraisal causes physiological responses, motor changes, action tendencies, changes in facial and vocal expression, and so on, succeeded by cognitive monitoring" (Robinson, 2005, p. 97). If this argument is correct, then all elicitors of emotions qualify as noncognitive at the algorithmic level of analysis whereas the role of cognition is limited to emotion regulation that begins only after emotion elicitation.
Robinson illustrates her view by analysing the fright caused by a significant economic loss. She admits that sophisticated conceptual appraisals may be necessary to realise that the evaporation of 10 000 USD from one's stock portfolio during a recent economic crisis constitutes a significant loss. Yet she suggests that this appraisal only paves the way for a causally efficient, fast and automatic affective appraisal about the loss. "It is only after lengthy cognitive evaluations that I realise how badly I am doing, but once I make this discovery, then all of a sudden I make an affective appraisal that rivets my attention on this discovery and its implications for my well-being," (Robinson, 2005, p. 62). Robinson argues that this subdoxastic and subpersonal affective appraisal is decisive as it gives rise to the other noncognitive components of emotion: physiological changes, changes in facial and vocal expression, action tendencies, and subjective feelings. In contrast, the preceding conceptual appraisal has only a contingent role in emotion elicitation because all affective appraisals do not have conceptual appraisals as their antecedents. More importantly, conceptual appraisals that either precede emotions or participate in their regulation after their elicitation do not influence the identity of emotions but only their duration, intensity, and expression. Therefore, all human emotions have nonconceptual content.
Robinson characterizes affective appraisals of the contents of conceptual thoughts or judgments as "meta-responses" that appraise the cognitively evaluated situation as bad for me (in bold to emphasize the qualitative difference to cognitive appraisals) or good for me in a "rough and ready" manner (ibid., p. 62). The problem with this proposal is that appraisals of this kind are hardly sufficient for determining the type of the emerging emotion as fear, anger, sadness, grief, joy, pride, and so on. Goodness and badness for the self only imply the valence of an emotion as either positive or negative, not its distinct type. Nor does a general appraisal of goodness or badness for the self represent the particular situation that figures as the particular object of emotion. Instead, an affective meta-appraisal appears to implement the higher-level cognitive appraisal whose content - an unexpected economic loss - determines both the identity of the emerging emotion as fright and its particular intentional object. The insistence on the necessary role of affective appraisal in emotion elicitation seems to refer to the activation of amygdala that is found in both bottom-up and top-down processes of emotion generation as neuroscientists have pointed out (e.g. Ochsner, Ray, Hughes, McRae, Cooper, Weber, Gabrieli, & Gross, 2009; Ochsner & Gross, 2007).
Robinson appears to recognize the problem with the content of affective appraisals as she suggests that the components of these appraisals may be the same as those identified by cognitive appraisal theorists. Her examples include This threatens me (or mine)! (fear) and This is an enemy! (hate), or, more concisely, Threat! (fear) or Offence! (anger). The argument is that affective appraisals of this kind always elicit emotions "regardless of how little or how much cognitive work it takes to detect the threat in question" (ibid., p. 94). If this proposal works, then an affective appraisal could specify the identity of the emerging emotion and represent its particular object after all.
Unfortunately, an elaborate affective appraisal appears superfluous in both respects in situations where such an appraisal has been preceded by a cognitive appraisal of threat or offence or some other core relational theme. Robinson correctly argues that cognitive thoughts and evaluations "are never the whole story" (ibid., 91), but her insistence that an affective appraisal that belongs to the emotion must nevertheless reiterate the entire evaluation in noncognitive terms strikes one as an unnecessary overlap that evolution would have eliminated if it had belonged to our emotional repertoire, given that the function of emotions is to facilitate fast responding to situations that bear on the subject's life or well-being. Alternatively, if there is no overlap between cognitive and affective appraisals, then affective appraisals are "meta-responses" that rely on cognitive appraisals in implementing emotions whose type and particular object are determined by cognitive appraisal. In this way, cognitive appraisals figure essentially in the elicitation and constitution of emotions after all.
The crux of Robinson's theory is her view that "affective appraisals can be automatically evoked not only by simple perceptions such as a sudden loud sound but also by complex thoughts and beliefs" (ibid., p. 97). If affective appraisals are capable of taking conceptual appraisals on the personal significance of events as their input, then affective appraisals are not so modular - opaque, cognitively impenetrable, informationally encapsulated, mandatory, domain specific and neutrally hard-wired - as Robinson and affect programme theorists who support similar views about the nature of emotional appraisals suggest. Instead, affective appraisals are flexible as they are capable of mediating emotional responses to both perceptual stimuli and semantic information (Roberts, 2003). In functional terms, affective appraisals qualify as garden-variety cognitions by virtue of their causal role in emotion elicitation. However, these appraisals also qualify as cognitive at the algorithmic level of analysis insofar as they are capable of relying on the outcome of conceptual appraisals in their functioning. This conclusion seriously undermines Robinson's noncognitive interpretation of affective appraisals and her entire theory.
Prinz is then right in arguing that emotions can be elicited by both cognitive appraisals and noncognitive perceptions. However, this upshot does not determine the cognitive or noncognitive status of human emotions because we must look beyond the first fractions of seconds in emotion generation to see whether or not cognition has an essential role in emotion. Noncognitive theorists admit that cognitive appraisals and reappraisals may influence the duration, intensity, and expression of emotions but not their identity. Therefore, these theorists exclude late cognition from emotion proper and associate it with emotion regulation instead. Robinson is an exception as she emphasizes the nature of emotions as processes that involve cognitive monitoring by default rather than contingently. In fact, Robinson admits that cognition may amount to more than merely monitoring as these "secondary appraisals" have the capacity of influencing the identity of emotions. Her example is the distinction between fear and anger. Robinson notes that "both fear and anger may occur when someone has done something to us which we regard as a wrong or a threat. Fear is associated with an appraisal that one is unable to deal with the situation... [whereas] anger is associated with an appraisal that I can deal with the situation" (Robinson, 2005, p. 202).
This is a striking - indeed a theoretically fatal - concession from a noncognitive theorist whose main argument for noncognitivism is that cognitive appraisals have only a monitoring function in regulating noncognitively elicited emotions. If cognitive secondary appraisals are sometimes needed to distinguish between such basic emotions as fear and anger, they do not merely monitor the unfolding of emotional responses but take an active role in shaping those responses as empirical emotion regulation researchers have pointed out (e.g. Shaver, Schwartz, Kirson, & O'Connor, 1987; John & Gross, 2007). Such influences on the duration, intensity, and expression of emotions as well as on their identity allow us to treat cognitive appraisals as necessary components of emotion after all. Indeed, Robinson seeks to present her account as a plausible synthesis of noncognitive and cognitive theories of emotion while remaining subtly on the noncognitive side. Nevertheless, my previous discussion has shown that Robinson is actually a subtle cognitivist as her insistence on the pervasive role of cognitive monitoring in all human emotions associates her view with cognitive process models of emotion.