Fig. 5.

Dynamics of adaptation and precision optimisation. By applying Dynamic Causal Modelling to EEG data acquired in a roving oddball paradigm, Garrido, Kilner, Kiebel et al. (2009) provided evidence for distinct neuronal mechanisms underlying different time-scales of repetition suppression. The model that best explained the observed data suggests that intrinsic connections (modelling gain or precision effects) show biphasic plasticity effects, while ascending connections from lower (A1: primary auditory cortex) to higher (STG: superior temporal gyrus) auditory regions are modulated by a monotonic function of stimulus repetition. This is consistent with the notion that the expected precision of prediction error decreases after the presentation of a novel stimulus and recovers with its repetitions, while prediction error signalling is greatest for the first presentation of an (unpredicted) deviant stimulus and decreases with repetition.