Cognitive process models of emotion
Cognitive process models argue that it is not possible to restrict the role of appraisal to two separate stages, elicitation and regulation, in emotional processing. Instead, appraisal continues during the entire process of emotion generation and regulation, shaping the emotion in light of constant feedback from other response domains and perceived changes in the emotion-eliciting situation. "The interpretation develops over time, and so does the feeling in a continuously interactive sequence, often a very rapid one. Neither interpretation, nor bodily feedback, nor subjective experience comes first; at the very most, one can talk about which of these complex temporal processes starts first", as Phoebe Ellsworth (1994, p. 227) points out. The continuity of emotional processing makes it difficult to draw a sharp distinction between emotion generation and regulation and to maintain -pace Robinson - that affective and cognitive appraisals map neatly onto these two facets of continuous emotion processing. By virtue of this processing, human emotions may have a strong cognitive imprint even when they are triggered by affective appraisals.
The basic idea of cognitive process models of emotion is that thematically coherent appraisal themes such as "an irrevocable loss" (sadness) or "a demeaning offence against me or mine" (anger) do not cause emotions but are constituted in the process of emotion generation together with other elements of emotional response. All process models emphasize the dynamic, parallel, nonlinear, and recursive character of emotional processing. Appraisal theories of this variant maintain that initial phases of stimulus appraisal trigger initial changes in the various response modalities such as autonomic physiology, action tendencies, motor, vocal and facial expression, and feeling state, whose feedback then influences both subsequent stages of appraisal and reappraisals of previous appraisals. In this way, appraisal and response modalities modify and update each other through parallel distributed processing and recursive feedback, proceeding through amplification to a stabilization phase in which parallel appraisal and response activities begin to constrain each other in a self-organizing manner. Along with increasing stability emerges a coherent appraisal in which different appraisal dimensions consolidate into a superordinate appraisal, which in turn holds the subordinate appraisals in place. "Emergent appraisals (or appraisal-emotion amalgams) are construed as globally coherent states arising and stabilizing through nonlinear causal transactions among appraisal and emotion constituents", as Marc Lewis (2005, p. 174) states. This kind of dependence between wholes and their constituent parts has been characterized as circular or bidirectional causality, and it has been argued to constitute the other main form of nonlinear causation along with feedback in emotional processing (Lewis, 2005; Scherer, 2009).
Cognitive process models maintain that the initiating appraisals in emotion generation process are often implicit, subpersonal, perceptual, and stimulus-driven, and if a satisfactory match to an existing appraisal schema is detected, an emotion ensues effortlessly. However, if further processing is required for determining the meaning of the stimulus to the subject, the level of processing tends to rise in tandem with the progress of the appraisal process. The continuous, parallel, and recursive nature of emotional processing ensures that the outcome of 'slow and careful' semantic appraisals becomes integrated with the result of 'quick and dirty' automatic appraisals that have initiated the emotion. After all, perceptual information reaches the subcortical domains of emotional processing in the centromedial nuclei of the amygdala and its subcortical circuit that includes the striatum, the brainstem, the hypothalamus, and the basal forebrain only a few milliseconds before arriving, through the basolateral nuclei of the amygdala whose function is to integrate sensory information, at the cortical areas of orbitofrontal, anterior cingulate, and ventromedial prefrontal cortex (e.g. Pessoa, 2010; Rolls, 2013; Bzdok et al., 2013). The minor lag in the cortical processing allows the subcortical processing to trigger endocrine, autonomic, and somatomotor responses as well as rudimentary behaviours such as attentional reallocation, orienting, avoidance, and approach before conceptual appraisals begin to influence and control the various response modalities. Indeed, recent neuroscientific evidence suggests that the divide between a subcortical 'low road' and a cortical 'high road' with their distinct types of appraisal, advocated by LeDoux (1995), is misleading (Pessoa & Adolphs, 2010; Rolls, 2013; see also Furtak, 2010). Instead of functioning automatically without top-down influences, the amygdala's responses are strongly dependent on task context and attention, a paradigmatic cognitive process. This suggests that instead of being a hub of the 'low road', the amygdala is a convergence zone for highly processed sensory information from both the subcortical and cortical structures with which it is connected (Pessoa, 2008; Pessoa & Adolphs, 2010; Bzdok et al., 2013). Therefore, its contribution to stimulus evaluation is not unambiguously subcortical. Moreover, the orbitofrontal cortex together with the anterior cingulate cortex engages in more complex processing of stimulus value and its rapid changes than the amygdala especially in primates such as humans, allowing greater flexibility in their emotional behaviour (Rolls, 2013). Thus the different pathways of emotional processing become strongly connected in the normal development of the human prefrontal cortex, allowing us significant control over even those emotions that are elicited through the subcortical circuit of the centromedial amygdala (Bluhm, 2007; Thompson, 2011; Rolls, 2013).
The main theoretical approaches that support the dynamic, parallel, nonlinear and holistic character of emotional processing are process-oriented appraisal theories (e.g. Smith & Lazarus, 1990; Ellsworth, 1991; Lazarus, 1991; Frijda, 1993; Frijda & Zeelenberg, 2001; Smith & Kirby, 2001; Scherer, 2001, 2009) and psychological constructionism, whose main advocates have been James Russell (e.g. 2003, 2009, 2012a,b) and Lisa Feldman Barrett (e.g. 2006a,b, 2009). Both approaches emphasize parallel distributed processing of appraisal and other response components that together constitute emotions. They also share a default-interventionist view about the relation of noncognitive and cognitive processes in emotion generation in maintaining that nonconceptual processes trigger affective responses but conceptual appraisals are required to shape these responses into emotions. Nevertheless, contrary to appraisal theorists, psychological constructionists claim that an actual appraisal process is only typically but not necessarily needed for determining the type and object of emotion. Instead, they suggest that emotions are constructed by categorizing states of core affect on the basis of salient conceptual knowledge about emotions and their objects. Appraisals are conceptually constitutive of emotions, but they do not cause emotions in all cases, as appraisal theorists maintain.
These theoretical disputes are serious to appraisal theorists and psychological constructionists who frame each other as their main theoretical opponents. Even so, I suggest that those disputes are nevertheless relatively minor in comparison to their equally important consensus on the nonlinear and parallel character of emotional processing, which ensures the constitutive role of cognitive processes in human emotions. In what follows, I first discuss the merits and flaws of these two approaches in more detail. Then I outline a novel argument for the constitutive role of cognition in the dynamics of emotion. This argument suggests that the dynamic character of emotional processing blurs the distinction between emotion generation and regulation.