Background:
Emotional face perception activates a distributed neural system including not only visual areas but limbic and prefrontal regions. Previous studies suggested that there was a core network for face perception that includes fusiform face area. The functional relationship between the core network and other brain regions, however, remains unclear.

Methods:
Twenty three (female 12, male 11) healthy subjects (Japanese, right-handed) participated in the functional magnetic resonance imaging (fMRI) experiment with a 3T MRI scanner (Philips Achieva). The mean age was 20.9 (18.3 - 23.9). The task was emotional face perception, in which the participants pressed button 1 for male faces and button 2 for female faces. The face expressions included neutral, contempt/sad, disgust/fear, and happy. The imaging was composed of two sessions, each of which contained 140 volumes (TR = 3 s). Data were preprocessed and analyzed using SPM8 software.

Results:
The activation maps showed the activity in the fusiform face area and superior temporal sulcus, which were suggested to be the core system for face perception (Fairhall and Ishai 2006). Motor areas, such as the premotor cortex and supplementary motor area, were also activated.
Dynamic Causal Modeling included these areas as well as the hippocampus, amygdala, middle frontal gyrus, and inferior frontal gyrus. The effective connectivities between these regions, however, exhibited a large individual variability. There was also an inter-session variability.

Conclusion:
A large-scale, heterogeneous network participated in emotional face perception. However, individual as well as inter-session variabilities prevented deterministic description of the network. Functional structure of this network is to be studied.