Figure 2.

Cortical and Subcortical Pathways of Salient Information

(A) Grand-mean data (n = 12) show enhanced responses to surprising compared to predictable auditory events. Responses to predictable sounds were similar across contextual manipulation of facial expressions. Surprised evoked fields increased with the emotional salience of facial expressions and were most deflected in the context of fearful faces.

(B) Scalp topography for surprise-evoked fields in a fearful context peaking at 185 ms showed a bilateral dipolar pattern over the temporal cortex.

(C) Source activity predicted by the dual-route (CS) (in red) and the cortical (C) (in blue) models at all network nodes shows enhanced early amygdala activity for model CS as compared to (C), whereas activity in auditory cortex remains similar.

(D) Graphical description of the models. Model CS includes both cortical and subcortical pathways that convey information from the auditory thalamus (MGB) indirectly, (through A1), or directly to the amygdala. Model (C) includes the cortical pathway only, precluding the subcortical pathway to amygdala.

(E) Bayesian model comparison reveals that the dual-route model explains the group data overall better than the cortical model alone, especially in early temporal windows. Solid black line corresponds to 50% probability, and the dotted black lines correspond to 90% and 10% probabilities. See also Figures S1 and S2.