Note. The figure shows a significant interaction effect. Bar plots indicate group means and whiskers standard errors. CS+ – stimulus paired with unconditioned stimulus; CS- – stimulus not paired with unconditioned stimulus; UCS – unconditioned stimulus; SCR – skin conductance response. * = p<.05; ** = p.01 *** = p<.001.
Furthermore, an additional three-way interaction term with RMSSD indicated that this effect was moderated by vagally mediated heart rate variability. While the interaction term was only marginally significant, t(125.7) = 1.67, p = 0.098, adding the effect significantly improved the overall model fit, indicating a relevant explanation of the data through the three-way interaction. Figure 3 depicts the three-way interaction, illustrating the effect when all other effects are held constant. It revealed that greater fear discrimination during the luteal phase (resulting in a larger difference between UCS and CS-, as well as CS+ and CS-) is associated with higher vmHRV. Conversely, individuals with very low vmHRV in the luteal phase exhibit less distinction than those in the follicular phase. Post-hoc testing for the beta weights of the interaction showed that this effect is driven by the difference in slopes of the CS- and the UCS in the luteal phase group, tratio(1018) = 2.90, p < 0.05. While the standardized UCS slope in this phase is 0.40, the slope for the CS- is -0.18. This indicates that for each standard deviation (SD) lower vmHRV during the luteal phase, there is a corresponding decrease of 0.4 SDs in the SCR response to the UCS and an increase of 0.2 SDs in the SCR response to the CS-. During the follicular phase, the level of fear discrimination is not linked to resting vmHRV.
Figure 3 Skin conductance by condition, menstrual cycle phase and vagally mediated heart rate variability