Fig. 5: Causal impact of alPFC47 disruption is selective for metacognitive judgements when coordinating with the poor partner.
a TMS (cTBS) (N = 21) was directed to left alPFC47 ([x, y, z] = [− 38, 40, − 10] based on previous work12) or vertex (control) on separate days. b Preference for choosing the self task in the social metacognitive judgement stage when coordinating with a poor partner after alPFC47 stimulation (centre) and no-stimulation (left). The overlaid black line indicates when performances by the self and partner are equated with one another based on the first simple RDK task (i.e., baseline perceptual performance) (left). The differences between alPFC47 stimulation and no-stimulation (right) show that, after alPFC47 cTBS, participants chose the self-task more often even when the poor partner’s task was optimal. c Impact of alPFC47 stimulation on social metacognition as quantified by meta-d’. alPFC47 disruption impaired meta-d’ on Poor, but not Good, Partner trials (N = 21). The significant difference in meta-d’ between Good and Poor Partners without stimulation disappeared with stimulation. **p = 0.0095, ***p = 1.54 × 10−4 (two-sided). d Impact of alPFC47 cTBS on social metacognitive judgements as a function of constant bias (left), self-task probability (middle), and partner’s task probability (right) on trials with the Poor Partner (blue) and Good Partner (red) (N = 21). alPFC47 disruption changed the regression coefficient on Poor, but not Good, Partner trials. The significant difference in the coefficient between Good and Poor Partners without stimulation disappeared with stimulation. **p = 0.011, *p = 0.023. †p = 0.049, significant interaction in two-way ANOVA (stimulation, no-stimulation × Good, Poor Partner). e Causal impact of alPFC47 cTBS on participants’ assessment of the merits of tackling challenge trials as opposed to inevitable trials for the moderate coherence condition (see Fig. 2d). In the absence of alPFC47 cTBS, on Poor Partner trials, participants performed challenge trials better than inevitable trials, but this was not the case on Good Partner trials. This difference, however, was abolished by alPFC47 cTBS. We demonstrated that this was the case when analysing data from 19 out of 21 participants who chose the self-task on trials adjacent to the black line in panel (b). †p < 0.05. *p = 0.031 (t test against zero with Bonferroni correction).
