3.2.2 P3 (350-450 ms)
The analyses on the P3 component detected a significant main effect of context, F (2, 38361) = 8.03, p < .001, and a significant interaction for flanker × context, F (2, 38363) = 6.05, p = .002. Follow-up tests showed that the forced switching context (M = 5.54 ± 17.53 μV) elicited a smaller P3 effect compared to both natural switching (M = 6.28 ± 14.47 μV; Estimate = -.76, SE = .19, t = -3.99, p < .001) and voluntary switching (M = 5.99 ± 14.51 μV; Estimate = -.44, SE = .19, t = -2.28, p = .034). Differences between the natural and voluntary switching contexts did not reach significance (Estimate = .33, SE = .19, t = 1.71, p = .088).
Planned pairwise comparisons within each context revealed a reversed flanker effect in the forced switching context (Estimate = 1.61, SE = .42, t = 3.81, p < .001), such that there was a stronger P3 effect on congruent trials (M = 6.37 ± 17.60 μV) compared to incongruent trials (M = 4.71 ± 17.42 μV). However, this effect was not significant in the natural and voluntary switching contexts (t s < 1.48, p s > .14). Moreover, in incongruent trials, the forced switching context (M = 4.71 ± 17.42 μV) elicited smaller P3 amplitude compared to both natural switching (M = 5.96 ± 14.61 μV; Estimate = -1.26, SE = .27, t = -4.66, p < .001) and voluntary switching (M = 5.79 ± 14.45 μV; Estimate = -1.07, SE = .27,t = -3.95, p < .001). Differences between the natural and voluntary switching contexts were not significant (Estimate = .19, SE = .27, t = .70, p = .483).
To recap, we observed a smaller N1 effect in the switch-incongruent condition of the natural switching context compared to the other two switching contexts. Differences between voluntary and forced switching contexts were not significant. Importantly, the P3 amplitude of incongruent trials in the forced switching context was smaller than both natural and voluntary switching contexts. In addition, an atypical flanker effect was detected on the P3 effect in the forced switching context, whereas such effect was not found in the other two switching contexts. These findings indicate that context-dependent language control modulates executive control.