Fig. 6: Brain hierarchies during recognition of auditory sequences revealed by Dynamic Causal Modelling (DCM).

a Graphical depiction of the six alternative models employed in DCM analysis. The connections between the six selected automated anatomical labelling (AAL) regions of interest (ROIs) (left Heschl’s gyrus [LHG], right Heschl’s gyrus [RHG], left hippocampus [LHP], right hippocampus [RHP], anterior cingulate gyrus [ACC], medial cingulate gyrus [MC]) are illustrated within a brain template, and through a graphical representation. Our hypothesised model of brain hierarchies during recognition of memorised and varied musical sequences is enclosed in a red box. Blue, red and black circles illustrate subsequently higher levels of brain hierarchy. b Posterior probability, protected exceedance probability and Bayesian omnibus risk (BOR) indicating the model with the highest evidence across the population (n = 83; inference computed using random-effects Bayesian model selection) for the memorised (M) sequence. We conducted four independent DCM analyses, one for each tone (excluding the first tone, which was common across all experimental conditions). Our hypothesised model (model one) had the highest evidence for tones two, three and four, while model three reported the highest evidence for tone five. c Posterior probability, protected exceedance probability and BOR indicating the model with the highest evidence across the population (n = 83; inference computed using random-effects Bayesian model selection) for the novel (N) conditions. Here, we focused on the first tone which introduced the variation in the sequence (i.e. tone 2 for novel T1 [NT1], tone 3 for novel T2 [NT2], tone 4 for novel T3 [NT3], tone 5 for novel T4 [NT4]). Our hypothesised model (model one) had the highest evidence for tones two, three and four, while model six reported the highest evidence for tone five. Source data are provided as a Source Data file.