Data Analysis

Behavioral performance was evaluated by measuring by accuracy (ACC) and reaction time (RT). Incorrect responses or RTs below 100 ms were excluded from both behavioral and ERP analyses. For ERP analysis, both matched and non-matched trials were included in the averaging of P2 and P3 components (43, 51). In our study, Fz and Pz were the sites that showed the largest amplitudes of P2 and P3, respectively. The mean amplitude of P2 was calculated 160–190 ms after the onset of stimuli. The mean amplitude of P3 was calculated 300–450 ms after the onset of stimuli.

In order to check whether stress induction was successful, one-way repeated measures analysis of variance (ANOVA) was conducted for salivary cortisol, with time as a within-subject measure. Greenhouse-Geisser corrected p-values were reported when sphericity assumptions were violated. Post-hoc comparisons were conducted using Bonferroni adjustments. Partial η² was provided as a measure of the effect size.

The magnitude of HPA response to acute stress was indexed by the percentage change in salivary cortisol (18, 52), which was calculated as follows: percentage cortisol increase = (CORT_peak – CORT_base)/CORT_base. The peak point was defined according to the group maximum point among the three post-TSST measures. The “percentage cortisol increase” was phrased as “cortisol reactivity” in the following to make the expression more concise.

Bivariate correlations were conducted among all potential predictor variables (behavioral and ERP index), outcome variable (cortisol reactivity), and controlled variables (age and neuroticism). For those behavioral and ERP index that were significantly correlated with cortisol reactivity, they were further included in a hierarchical regression in step 2 to examine how working memory predicts HPA response, with age and neuroticism being entered in Block 1 as control variables.