Fig. 5: Coding of relevant and irrelevant information in MD regions under Control and Active TMS conditions.

Box plot summary depicting minimum and maximum values, lower and upper quartiles and median scores. Individual data points indicated by open circles. Lighter-coloured bars depict coding under control TMS, and darker-coloured bars depict coding under Active TMS trials. a–c show coding of relevant information (e.g., colour during the colour task) under Control and Active conditions, collapsed across feature (colour, form). d–f show coding of irrelevant information (e.g., colour during the form task) under Control and Active conditions, also collapsed across feature. All bars represent coding of identical stimulus information, variation in the strength of coding is driven by TMS intensity and whether the information was relevant for the participant’s current task. Due to outliers (>3 SD from the condition mean), we performed a log transformation on the unstimulated MD region data before statistical testing. The data displayed are in the untransformed form prior to log transformation. An ANOVA on the unstimulated MD regions (a, d) showed a significant TMS × relevancy interaction. TMS reduced coding of relevant features in unstimulated MD regions, but did not modulate coding of irrelevant information (BF₁₀ = 0.24). The ANOVA for right dlPFC (factors: TMS, Feature and Relevancy; c, f) showed no significant main effects or interactions. The significance markings for individual bars indicate whether coding was significantly greater than chance in each condition separately (permutation test). *P < 0.05. In this figure only, ** is equal to P < 0.008 (to correct for multiple comparisons in 6 unstimulated MD regions). N = 20 participants.