Distinct brain systems underlie the processing of valence and arousal of affective pictures
Introduction
Humans are characterized by a wide variety in emotional experiences. One widely accepted theory of emotion states that emotional experience can be reduced to two almost independent dimensions, namely arousal and valence. Valence refers to the pleasantness of a stimulus, whereas arousal reflects the intensity of an event, varying from calming to exciting (Lang et al., 1993, Russell, 1980).
Over the last decade, a growing number of functional neuroimaging studies have been devoted to the identification of the neural substrates of different human emotions (see for reviews: Murphy et al., 2003, Phan et al., 2002, Sergerie et al., 2008, Wager et al., 2003). More recently, neuroimaging studies started to explicitly examine the contribution of valence and arousal to neural activation supporting human emotion. This work has concentrated mainly on a limited number of brain regions, namely the amygdala and the prefrontal cortex.
With respect to the amygdala, the traditional view poses that this region is part of a brain system specialized in the detection of threatening stimuli (LeDoux, 1996), supported by studies indicating that fearful visual stimuli (regardless whether faces or pictures were used) induced increased activation in the amygdala (Baas et al., 2004, Phan et al., 2002). Subsequent work provided evidence that the role of the amygdala is perhaps less specific, as this region also responds to other negative emotions such as sadness and disgust (Stark et al., 2007, Wang et al., 2005, Goldin et al., 2005) and even to positively valenced stimuli (Beauregard et al., 2001, Ferretti et al., 2005, Fitzgerald et al., 2005, Garavan et al., 2001, Hamann and Mao, 2002, Kensinger and Schacter, 2006). This has led to the suggestion that the amygdala is not specifically sensitive to the emotional valence of a visual stimulus per se but instead to its intensity, or level of arousal (Phan et al., 2002, Phan et al., 2003, Wager et al., 2003). Indeed, studies controlling for possible confounding effects of valence have generally demonstrated amygdala activation related to the level of arousal of pictorial stimuli (Garavan et al., 2001, Kensinger and Schacter, 2006, Lane et al., 1999).
In contrast to the amygdala, which tends to be associated with quick and primary emotional responses, the prefrontal cortex (PFC) is generally associated with higher-order regulation (or evaluative aspects) of emotion (Ochsner & Gross, 2005). Evidence from neuroimaging studies suggests that different parts of the PFC play distinct roles in emotional evaluation. The medial PFC would be dominant for appetitive or rewarding stimuli (Elliott et al., 2003, O’Doherty et al., 2001) whereas the lateral orbital PFC is associated with processing of aversive stimuli (Kensinger and Schacter, 2006, O’Doherty et al., 2006, Small et al., 2001, but see Northoff et al., 2000, who found evidence for the opposite pattern). A second dominant hypothesis is that positive emotions tend to be lateralized towards the left hemisphere while negative emotions are associated with the right hemisphere (Davidson & Irwin, 1999), a notion that is supported by recent neuroimaging data (Murphy et al., 2003, Wager et al., 2003, Dolcos et al., 2004, Grimm et al., 2006, Nitschke et al., 2004), although there are negative findings as well (Lane et al., 1997, Teasdale et al., 1999). Concerning arousal, evidence points to involvement of the medial prefrontal cortex (Lane et al., 1999), in particular the dorsomedial regions (Anders et al., 2004, Dolcos et al., 2004, Grimm et al., 2006, Kensinger and Schacter, 2006). Finally, for regions outside amygdala and prefrontal cortex, there is evidence that the insula is particularly sensitive to negatively valenced stimuli (Anders et al., 2004; Lewis, Critchley, Rotshtein, & Dolan, 2007) and that the response of the extrastriate occipital cortex is modulated by both valence and arousal (Bradley et al., 2003, Lane et al., 1999, Mourao-Miranda et al., 2003).
In sum, the general picture that emerges suggests that (i) the amygdala and perhaps occipital cortex primarily support processing of arousal (as they seem to respond to both valences), (ii) the prefrontal cortex is most responsive to valence differences, and (iii) the insula is specifically involved in the processing of negative valence. So far, however, the majority of these studies did not consider the possibility that valence and arousal may interactively determine neural responding to emotional pictures. That this issue is not trivial is demonstrated by a recent fMRI study on olfactory processing showing that the effect of odor intensity on amygdala activation depended on whether the smell was pleasant, unpleasant or neutral (Winston, Gottfried, Kilner, & Dolan, 2005). Based on these data, the authors suggest that the amygdala is not specifically sensitive to either arousal or valence as such, but instead to a combination of these parameters that may reflect the overall emotional value of a stimulus (Winston et al., 2005). Using emotional words, similar findings were reported by Lewis et al. (2007), who demonstrated that the amygdala responded to increasing arousal but only for negative words. With respect to the pictorial domain, most neuroimaging studies used designs that did not allow the investigation of interactions between valence and arousal. Only recently studies started to control explicitly for possible confounding effects of arousal on valence (Dolcos et al., 2004, Kensinger and Schacter, 2006) and vice versa (Garavan et al., 2001, Mather et al., 2006). We know of two studies that explicitly considered valence-by-arousal interactions and these reported negative results (although it should be noted that these studies were performed with lower-resolution PET (Lane et al., 1999) or focused on the amygdala only (Garavan et al., 2001)).
Taken together, the number of studies examining the brain systems coding for valence and arousal in picture processing is increasing rapidly. However, few of these have attempted to investigate to what extent activity in these regions (and elsewhere in the brain) is influenced by arousal-by-valence interactions. The presence of such interactions may imply that emotion processing in the brain has a greater complexity than suggested previously. Therefore, the aims of this fMRI study were (i) to tease apart activity associated with processing valence and arousal at the level of whole-brain function, and (ii) to determine in which regions arousal-by-valence interactions take place. To this end, we used event-related functional MRI to measure whole brain responses to affective pictures in a sample of healthy female volunteers. We included only females as there is increasing evidence that men and women show distinct neural activations to emotional stimuli (e.g. Hagemann et al., 1999, Caseras et al., 2007, Hofer et al., 2006). A 2 (Valence) × 2 (Arousal) design allowed us to determine not only main effects of arousal/valence, but interaction effects as well. Based on the literature, we expected to find valence-specific responding in prefrontal cortex (with lateral regions responding to negative valence and medial regions activated by positive valence) and predominantly arousal-specific activations in amygdala and occipital cortex. Furthermore, we hypothesized that the most likely regions to find possible valence-by-arousal interactions are the amygdala, insula and occipital regions.
Section snippets
Subjects
Twenty-three right-handed female subjects (age: 20–25 years) were recruited by means of campus advertisements. Before scanning, participants completed the Dutch version of the Symptom Checklist (SCL-90) to screen for the presence of psychopathology. None of the subjects used psychotropic medication or had a history of psychiatric or neurological illness. All subjects had normal or corrected-to-normal vision. To ensure the effectiveness of the erotic pictures, we included only subjects who
Behavioural data
Mean accuracy and reaction times (RTs) on the indoor–outdoor classification task were analysed with a 2 (Valence) × 2 (Arousal) ANOVA. For both dependent variables, we found no significant interaction between Valence and Arousal (Accuracy: F1,22 = 3.2, ns; RT: F1,22 = 1.9, ns). There was, however, a significant main effect of Valence (Accuracy: F1,22 = 18.2, P < 0.001; RT: F1,22 = 19.42, P < 0.001), which was due to the fact that subjects displayed more classification errors and slower response times to
Discussion
The aim of the current study was to provide insight in brain areas supporting human neural responses to emotional scenes, and in particular to the dimensions that are presumed to underlie emotion: valence and arousal. So far, most neuroimaging studies have focused on the role of amygdala and prefrontal cortex, and to a lesser degree on occipital regions, in processing arousal and valence of affective pictures. There is increasing evidence, though, that brain regions may respond to particular
References (76)
- et al.
Lateralization of amygdala activation: A systematic review of functional neuroimaging studies
Brain Research Reviews
(2004) - et al.
Modulation of emotion by cognition and cognition by emotion
Neuroimage
(2007) - et al.
Neural correlates of social and nonsocial emotions: An fMRI study
Neuroimage
(2006) - et al.
Sex differences in neural responses to disgusting visual stimuli: Implications for disgust-related psychiatric disorders
Biological Psychiatry
(2007) - et al.
The functional neuroanatomy of emotion and effective style
Trends in Cognitive Science
(1999) - et al.
Dissociable effects of arousal and valence on prefrontal activity indexing emotional evaluation and subsequent memory: An event-related fMRI study
Neuroimage
(2004) - et al.
Dynamics of male sexual arousal: Distinct components of brain activation revealed by fMRI
Neuroimage
(2005) - et al.
The neural bases of amusement and sadness: A comparison of block contrast and subject-specific emotion intensity regression approaches
Neuroimage
(2005) - et al.
Segregated neural representation of distinct emotion dimensions in the prefrontal cortex – An fMRI study
Neuroimage
(2006) - et al.
The assessment of affective reactivity using films: Validity, reliability and sex differences
Personality and Individual Differences
(1999)