Although the nature of the DCM connectivity values is very different from GLM based PPIs, it is striking that the basic result was consistent—viewing angry relative to neutral faces produced a greater change in connectivity between the ventral ACC and amygdala. However, the DCM results go further by demonstrating that this effect is restricted to connectivity from the ventral ACC to amygdala but not vice versa
, and that this was significantly correlated with reward–drive. It is of interest that this directionality accords with a recent and detailed anatomical study in monkeys (Ghashghaei et al., 2007
), which demonstrated that the ventral ACC sends proportionally more projections to the amygdala than it receives; consistent with the ventral ACC's role in the extinction of negative emotions. Similarly, additional comparative research shows that the ventral ACC gates the transfer of the information in the amygdala from the basolateral nuclei (containing face-sensitive cells) to the central nucleus (involved in the expression of emotional arousal) by acting on GABA-ergic interneurons (Maren and Quirk, 2004; Rosenkranz and Grace, 2002
). Previous research has shown that the pathway from ventral ACC to the amygdala is important in mediating the resolution of conflict in an emotional Stroop task (Etkin et al., 2006
). However, whereas Etkin et al. showed that the right amygdala was involved in emotional conflict detection, our own study identified the left amygdala in processing signals of aggression. Our current findings are consistent with previous studies in both healthy and psychiatric individuals (intermittent explosive disorder) showing that the activation of the left amygdala (when viewing facial signals of aggression) was significantly modulated by reward–drive (Beaver et al., 2008
) or by the severity of aggressive behaviour (Coccaro et al., 2007
DCM also allowed us to test whether facial information is projected in parallel into the ventral ACC and amygdala, or whether the processing is serial (i.e., facial information enters the amygdala and is then projected to the ventral ACC, or vice versa
). Previous animal and human literature provides evidence in favor of parallel inputs (Eimer and Holmes, 2007; Eimer et al., 2003; Etkin et al., 2006; Kawasaki et al., 2001; Leonard et al., 1985; Oya et al., 2002; Rolls, 2007
). In particular, both the amygdala and ACC have robust and direct connections with secondary visual areas (e.g. posterior occipital cortex, fusiform gyrus, and superior temporal gyrus) involved in the perception of facial signals of emotion (Aggleton et al., 1980; Catani et al., 2002, 2003; McDonald and Mascagni, 1996
). This sensory information is conveyed rapidly and influences the neuronal activity in the amygdala and prefrontal cortex at approximately the same time, as demonstrated by single cell recordings in monkeys showing similar short latencies for faces (~ 110–220 ms) in both regions (Leonard et al., 1985; Rolls, 2007
). Similarly, ERP experiments and in-depth implant electrode recording in humans have demonstrated rapid (~ 150 ms) frontocentral potentials in response to emotional faces and scenes (Eimer and Holmes, 2007; Eimer et al., 2003; Kawasaki et al., 2001; Oya et al., 2002
These results suggest that the rapid processing of emotional facial expressions within the anterior prefrontal cortex could occur independently and in parallel with processing in the amygdala. Consistent with this idea, Bayesian model selection (Penny et al., 2004
) showed very strong evidence in favor of the parallel model (faces ‘injected’ simultaneously in both ventral ACC and amygdala) compared with serial models (faces ‘injected’ independently in either the amygdala or ventral ACC alone).
On the basis of our findings and previous research, we propose that the processing of facial signals of emotion by the amygdala and ventral ACC involves at least two stages. An initial input to both regions provides a rapid and coarse analysis of affective content. Following possible transfer of information from the amygdala to the ventral ACC, a more complex evaluative process of the socio-affective meaning of a facial expression in relation to the appropriate context and the individual's temperament and personality is implemented by the ventral ACC which projects the results of these computations to the amygdala; this may include gating the transfer of information between basolateral amygdala and central nucleus, as demonstrated by comparative research (Maren and Quirk, 2004; Rosenkranz and Grace, 2002
). Of particular interest, it is this latter process that is captured by the anger-mediated effects of reward–drive from the ventral ACC to the amygdala.
fMRI does not have sufficient spatial resolution to detect specific amygdala nuclei. So we cannot be certain that ventral ACC afferents affect interactions between the basolateral and central nuclei of the amygdala, as opposed to other nuclei. However, as suggested by others (Aggleton, 1985; Emery et al., 2001; Stefanacci and Amaral, 2002
), the basolateral amygdala is in an ideal position to act as a neural integrator that attempts to match an environmental emotional stimulus with a particular social context conveyed by the ventral ACC. Individual differences in the gating signal from the ventral ACC to the amygdala could therefore lead to differences in how an aggressive signal is perceived and interpreted, ultimately influencing the variability of behavioural reactions.
We have shown that the interactive effects of reward–drive and processing angry expressions in the amygdala and ventral ACC is found for contrasts comparing anger with neutral (current study) or other negative expressions (sadness) (Beaver et al., 2008
). Moreover, there is little evidence that reward–drive personality affects the experience or perception of other emotions (i.e., happiness, fear, or sadness) (Carver, 2004; Putman et al., 2004
). In view of these findings, we focused on angry and neutral expressions only. However, we do not wish to deny that the amygdala and ventral ACC have a wider role in emotional behaviour, and in emotion regulation in particular (Davidson et al., 2000; Drevets, 1999
). The key role of the ventral ACC in controlling negative affects has been highlighted by paradigms testing fear extinction in both humans and animals (Delgado et al., 2006; LeDoux, 2003; Phelps et al., 2004
) and human studies of anxiety suppression (Petrovic et al., 2005
). Hence, our claim is that the reduced influence of viewing aggressive (compared with neutral) faces on the connectivity from the ventral ACC to amygdala (but not vice versa
) in high reward–drive individuals, fits with the idea that emotion regulation plays a key role in aggression (Davidson et al., 2000
). However, the influence of reward–drive on other negative emotions, such as disgust, whose recognition is also affected by aggressive disorders (Best et al., 2002
), should be considered in future research.
A further consideration is that the angry and neutral expressions were presented in short blocks intermixed with null events. Although, this provided the optimal design for both PPI and DCM analyses, it might affect the predictability of emotional faces. Hence, a question for future research is whether a fully randomised event-related design might produce similar findings.