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; ITI – intertrial interval. ns = not significant; *** = p<.001.
 

Skin conductance response

A Box-Cox analysis indicated the necessity of logarithmic transformation of the SCR data due to left skewness. Therefore, the analyses were conducted using the log-transformed data (log(1+SCR)). In 4 participants, null reactions in UCS trials amounted to more than 50%, which led to their exclusion from the analysis (Lonsdorf et al., 2019).
The results of the model predicting SCR are presented in Table 2. The inclusion of a three-way interaction term (Cycle Phase x Condition x RMSSD) significantly improved the model fit, as evidenced by a Likelihood Ratio Test, χ2 (2, N = 58) = 8.42, p > .05. Adding age or RMSSD as main effects did not improve the model. There was no main effect of the menstrual cycle phase in the final model, t(75.55) = 0.82, p = .42, indicating no overall differential skin conductance levels between the phases.
 
Table 2 Results of mixed model predicting skin conductance
Predictors Estimates CI p
(Intercept) 1.06 1.02 – 1.10 <0.001***
condition [CS+] 0.06 0.03 – 0.09 <0.001***
condition [UCS] 0.24 0.20 – 0.28 <0.001***
phase [lut] 0.14 -0.20 – 0.47 0.413
condition [CS+] × phase [lut] -0.16 -0.42 – 0.10 0.219
condition [UCS] × phase [lut] -0.42 -0.74 – -0.11 0.009**
Condition [CS-] × phase [lut] × log(RMSSD) -0.04 -0.13 – 0.05 0.398
condition [CS+] × phase [lut] × log(RMSSD) 0.00 -0.09 – 0.09 0.962
condition [UCS] × phase [lut] × log(RMSSD) 0.09 -0.02 – 0.19 0.096+
Observations 1080    
Marginal R2 / Conditional R2 0.178 / 0.365  
 Note. The final model had the structure: log(skin conductance response) ~ condition * phase + log(RMSSD):condition:phase + (1|participant). CS+ – stimulus paired with unconditioned stimulus; CS- – stimulus not paired with unconditioned stimulus; UCS – unconditioned stimulus; RMSSD – root mean square of successive differences; lut – luteal phase.
 
However, a two-way interaction of Cycle Phase x Condition (UCS vs. CS-) yielded a significant effect (see Figure 2 for visualization including significance levels from post-hoc contrast testing), t(1020) = -2.625, p < .01. The effect mirrors the interaction observed in the startle response, albeit less distinctly pronounced. During the luteal phase, a heightened SCR is evident specifically in response to the UCS (unconditioned stimulus) compared to the follicular phase. Although the visual inspection confirms that this difference drives the interaction effect, post-hoc testing did not find a significant difference between the UCS values in the follicular phase group and the luteal phase group, t(125.7) = -1.084, p = 0.28.
 
Figure 2 Skin conductance by condition and menstrual cycle phase