Figure 3.

Integrating the proposed value-driven model with the Bayesian model of mirror neuron system (MNS) processing. Following processing in visual brain regions, information flows along established MNS pathways (Iacoboni, 2005) as well as along emotion processing/salience regions (anterior insula, anterior cingulate cortex [ACC], amygdala) and reward processing regions (substantia nigra reticulate [SNr], substantia nigra compacta [SNc], ventral tegmental area [VTA], ventral striatum/nucleus accumbens [NAc]). Here we integrate Bayesian models, which include prediction error signals (green arrows), and generative model processing (blue arrows, Kilner et al., 2007), with emotion and salience processing (pink arrows) and reward processing (red arrows). Indirect reward processing is depicted with dashed red lines. All of these processes modulate the MNS. The parietal MNS is thought to include the posterior parietal cortex (PPC) while the frontal MNS is thought to include the ventral premotor cortex (vPMC) and the inferior frontal gyrus (IFG). STS = superior temporal sulcus. For brevity, we don’t include every region of salient, emotion and reward systems, only primary nodes most likely to be directly related to the MNS. We also note that components of the reward system also process saliency, as discussed in “Toward an Explanatory Model” section.