Fig. 3: Attenuated gamma increase and alpha-beta suppression during encoding unsigned precision-weighted prediction errors about stimulus outcomes in bipolar disorder.

A Source reconstruction of MEG signals was carried out with linearly constrained minimum norm variance (LCMV) beamforming. The statistical analysis of convolution GLM results targeted brain regions implicated in decision-making under uncertainty and reward processing [43, 71,72,73,74,75], associated with impairments in the fronto-striatal reward circuitry in BD [5, 9, 76, 77]: caudal and rostral ACC, OFC (lateral and medial portions: lOFC, mOFC), SFG, caudal and rostral MFG, M1. Panel a illustrates these regions using anatomical labels from the neuroanatomical Desikan-Killiany atlas (DK), utilised to parcellate the cerebral cortex of each participant based on their individual T1-weighted MRI. B Left and centre panels display within-subject effects in time-frequency (TF) images representing oscillatory amplitude responses to unsigned precision-weighted PEs about stimulus outcomes. TF images cover the 4–100 Hz range, including theta (4–6 Hz), alpha (8–12 Hz), beta (14–30 Hz), and gamma (32–100 Hz) activity. The TF images were normalised by subtracting the mean and dividing by standard deviation (SD) of the activity in the [−300, −50] ms pre-outcome interval, and thus are presented in SD units. Significant within-subject effects are outlined in black for the HC (left) and BD (centre) groups (cluster-based permutation tests, negative cluster within 0.5–0.9 s post relative to pre-outcome baseline, P_FWER = 0.001, 0.024 in each group, respectively. Although no within-subject effects in BD were observed in the illustrated SFG label, there were effects across other ROIs). The right panel shows the between-group differences, significant in a cluster-based permutation test (positive cluster within 8–30 Hz, P_FWER = 0.0130; negative cluster within 60–100 Hz, P_FWER = 0.0090; N = 21 BD and 27 HC independent samples). The time point 0 s marks the onset of outcome presentation. C, D Panels Illustrate between-group effects in the alpha (C) and beta (D) ranges, attributed to more pronounced alpha and beta suppression in HC than in BD participants during encoding of unsigned pwPE on level 2. Effects are depicted in ROIs including the cACC, lOFC, SFG, M1. E Similar to panels C and D but in the gamma range, showing that unsigned pwPE were associated with increases in TF amplitude in gamma range for HC participants, yet with gamma attenuation in BD participants, and across a similar range of ROIs. Labels denote the rostral anterior cingulate cortex, rACC; caudal ACC, cACC; superior frontal gyrus, SFG; lateral and medial orbitofrontal cortex, lOFC and mOFC; primary motor cortex, M1; caudal and rostral middle frontal gyrus, cMFG, rMFC.