Surprised expressions provide an important comparison expression for fear. For example, both expressions share features (e.g., eye-widening) consistent with the detection of an important eliciting event. Though both expressions leave the nature of their eliciting event ambiguous, fearful expressions provide additional information concerning the predicted negative valence of the event.
The animal literature documents that the medial prefrontal cortex (mPFC) provides an important regulatory input to the amygdala, communicating cortical representations of valence as they relate to predicted outcomes. For example, in rats, amygdala-mediated conditioned responses that normally decrease during extinction trials that follow aversive conditioning (tone now predicts no shock) persist in animals with lesions of the mPFC. Thus, mPFC inputs to the amygdala can signal a new more positive interpretation of a once negative predictive stimulus. Thus, if the amygdala encountered a stimulus that has sometimes predicted positive outcomes and sometimes predicted negative outcomes (like an extinguished tone or a surprised face), mPFC inputs would assist in assessment of this stimulus.
The Figure presented here summarizes the results of our Kim et al (2003) study (see Publications link). These three activations were identified in terms of their correlations with valence ratings of surprised facial expression blocks. Activity within the amygdala and dorsal mPFC demonstrated positive correlations with these ratings, that is, subjects who interpreted surprised faces as more negative showed higher signal changes here. Activity within the ventral mPFC showed a negative correlation with these ratings, that is, subjects who interpreted surprised faces as more positive showed higher signal changes here. Note also that activity at each of these three loci is significantly correlated with activity at each of the two other loci.
Thus, subjects who interpreted surprised faces negatively, tended to show higher amygdala and dorsal mPFC activity and lower ventral mPFC activity; while subjects who interpreted surprised faces positively, tended to show higher ventral mPFC activity and lower amygdala and dorsal mPFC activity.
It is important to note that these are final averaged relationships. Thus, though they offer evidence of functional connectivity, they do not necessitate direct connectivity. They tell us little about the nature (e.g., excitatory, inhibitory, etc.), direction (amygdala to mPFC, or the reverse, etc.), order (which areas respond first, how do they change over time, etc.), or weighting (are some connections 'heavier' and more prominent in determining this behavioral effect, etc.). Still, they are consistent with models of prefrontal-limbic regulation, they parallel the animal extinction literature, and they provide candidate structures that can be investigated in future studies.