On the interdependence of cognition and emotion

Abstract

Affect and cognition have long been treated as independent entities, but in the current review we suggest that affect and cognition are in fact highly interdependent. We open the article by discussing three classic views for the independence of affect. These are (i) the affective independence hypothesis, that emotion is processed independently from cognition, (ii) the affective primacy hypothesis, that evaluative processing precedes semantic processing, and (iii) the affective automaticity hypothesis, that affectively potent stimuli commandeer attention and evaluation is automatic. We argue that affect is not independent from cognition, that affect is not primary to cognition, nor is affect automatically elicited. The second half of the paper discusses several instances of how affect influences cognition. We review experiments showing affective involvement in perception, semantic activation, and attitude activation. We conclude that one function of affect is to regulate cognitive processing.

Acknowledgements

Support for this research is acknowledged from National Institute of Mental Health Grant MH 50074 to GLC.

Notes

¹That is, we, along with others (e.g., Barrett, 2006), suggest that there is not a brain centre dedicated to specific emotions such as fear, happiness, etc. But, there are specific areas critically involved in emotion processing. For instance, the amygdala is critically involved in the emotion of fear, but is not specifically dedicated to fear.

²The conception of emotion we raise, affective independence and affective primacy, comes mainly from Zajonc (1980, 2000). The affective automaticity derives from arguments made by Bargh and colleagues (Bargh, 1997; Ferguson & Bargh, 2003).

³In particular, the strongest evidence for such a route comes from affective blindsight individuals. Individuals have damage to area V1 of the visual cortex and as a result have no conscious perception of the world. However, these individuals still demonstrate affective reactions to fear-inducing visual stimuli. In the literature though, this is still a debated issue. First, the pathways involved are unclear. That is, although information may not be visually conscious to blindsight individuals, areas of the visual cortex still receive visual information (area V4 and extrastriate) from subcortical structures such as the pulvinar and superior collicolus. Therefore, although the area V1 is damaged, areas of the visual cortex still receive the same visual information. Storbeck, Robinson, and McCourtt (2006) examine this issue more extensively.

⁴We will use the term “semantic” to describe the meaning analysis that we propose precedes affective analysis. What we have in mind specifically are at least three achievements: (1) the integration of multiple features of the object into a single “object” code; (2) the identification of this object; and (3) the categorisation of the object (e.g., as animate or not). The term semantic, then, refers somewhat more directly to the achievements of area IT (especially invariance, identification, and categorisation) that seem to occur in order for a person to retrieve affective associations.

⁵A host of fMRI studies have demonstrated the activation of the amygdala to masked fear faces and other emotional stimuli. Such studies are interesting because individuals do not have a conscious perception of the image. However, the amygdala only shows enhanced activation to arousing images (e.g., fear faces), but not to non-arousing faces (e.g., houses). Although such evidence suggests that amygdala activation can occur without perceptual awareness, we still suggest that the visual system still codes that image and sends its input forward to the amygdala in the same manner as if the stimulus was presented supraliminally. Moreover, imaging studies have a weakness of comparative activity. Therefore, it is difficult to gage how much processing is done between masked and non-masked fear faces. In addition, there is plenty of evidence to suggest that the visual cortex processes masked and non-masked images in a similar manner. Moreover, evidence from single-cell recording suggests that the visual system can still determine whether a face or a house was presented regardless of whether each image was presented with a mask and subliminally. Therefore, studies demonstrating that the amygdala activates for a subliminal, but not a supraliminal picture does not mean that the visual cortex did not send the same information. There is no reason to believe that the categorisation processes performed by area IT are conscious. Indeed, on the basis of ERP data, we might conclude that unconscious categorisation routinely precedes conscious categorisation. Furthermore, unconscious categorisation by the visual system may occur extremely quickly after stimulus exposure, in as little as 48 ms for “global templates” (Sugase, Yamane, Ueno, & Kawano, 1999) and 70–80 ms for classes of stimuli (Van Rullen & Thorpe, 2001). Interestingly, Van Rullen and Thorpe (2001) also found that the initial (70–80) categorisation-related ERP component was not highly correlated with a participant's response to the task at hand, whereas an ERP component that occurred at 190 ms post-stimulus onset was. Thus, categorisation appears to occur quite rapidly and seems to occur independently of later, possibly more conscious, categorisation processes. Relatedly, people can classify objects on the basis of category membership even with no awareness of the distinct categories guiding their response (e.g., Reed, Squire, Patalano, Smith, & Jonides, 1999). In summary, we conclude that categorisation occurs within later stages of the visual cortex, specifically area IT. Moreover, other data suggest that these same visual areas are not sensitive to the affective significance of objects (Iwai et al., 1990; Nishijo, Ono, & Nishino, 1988a; Rolls, 1999; Rolls, Judge, & Sanghera, 1977). Thus, within area IT and other later stages of the visual cortex we appear to have considerable evidence for categorisation prior to affect retrieval. Recall that studies have found distinct category-related ERPs within 70–80 ms post-stimulus onset (e.g., Van Rullen & Thorpe, 2001). Object identification also appears to occur rapidly, perhaps within 100 ms of stimulus onset (Lehky, 2000; Rolls & Tovee, 1994). These findings suggest that categorisation tends to occur prior to identification. Nevertheless, studies that present masked stimuli have demonstrated that even stimuli presented as briefly as 20–60 ms with pre- and postmasks are still sufficiently processed by area IT to support object identification (Dehaene et al., 2001; Rolls, 1999; Vogels & Orban, 1996). In the latter connection, Rolls, Tovee, Purcell, Stewart, and Azzopardi (1994) argued that such subliminal presentations reduce the amplitude of neural responses to stimuli, but do not change fundamental neural identification processes (see also Kovacs, Vogels, & Orban, 1995, for similar results). Thus, the primary difference between subliminal and optimal viewing conditions pertains to the amplitude of the neuronal responses within area IT, but sufficient processing still occurs to produce an invariant neural code (i.e., identification). From this perspective, demonstrations of “unconscious” cognition or affect are not particularly special from a neurological point of view.

The section title implies that cognition does not modulate emotion. We would suggest, like others have, that in fact cognition does modulate emotion (e.g., Ochsner & Gross, 2005), but such a discussion is beyond the scope of this article.