3.3 Results from the prediction models
Adhering to the same EEG pre-processing standards, the brain activity from the preceding prcture-naming trials were put into a family of regularized RR models as independent variable to reveal its predictability on the brain activity during the current flanker trials. The predictive results regarding the impact of language control on executive control across three switching contexts are displayed in Figure 3A.
Results showed that robust brain-brain relationships were observed in the forced switching context for the modeling of congruent trials and incongruent trials. More precisely, language processing notably anticipated brain activity in subsequent congruent trials (-10 – 800ms:r = .57, p < .001). Additionally, for the modeling of incongruent trials, a negative relationship was significantly detected in the two time windows (272 - 640ms: r = -.43, p = .007; 648 - 772ms: r = -.44, p = .007). Furthermore, planned paired-sample t -tests showed that the prediction value of language processing in the forced switching context for congruent trials in flanker task was significantly stronger than natural or voluntary switching contexts (t s > 30.96,p s < .001).
Figure 3 . Prediction results of ridge regression models between the three language switching contexts and flanker trials (A) and specification curve for congruent and incongruent trials separately (B).