Figure 8.

DCM models and model evidence. A, Invariable configuration of all network models. Taste and tasteless events (collapsed across expectancy) were used as driving inputs into the anterior insula (AI), and all models were specified to have full steady-state connectivity between the nodes in the network. B, Exceedance probabilities of all 31 possible models. The exceedance probability is the probability that a model is more likely than any of the other models given the observed fMRI data. Exceedance probabilities showed significant evidence in favor of model 19. C, Configuration of model 19. For model 19, modulation of AI by VS and IPS and modulation of VS by AI was specified. Average estimates (arbitrary units, ±SD) of modulation strength across subjects of each of the modulatory inputs (unexpected and expected events collapsed across stimulus) are depicted next to each of the intrinsic connections.