Avoidance Learning Across the Menstrual Cycle: A Conceptual Replication

Hormonal transitions across the menstrual cycle may modulate human reward processing and reinforcement learning, but previous results were contradictory. Studies assessed relatively small samples (n < 30) and exclusively used within-subject designs to compare women in hormonally distinct menstrual cycle phases. This increased the risk of sporadic findings and results may have been disproportionally affected by expectancy effects. Also, replication studies are widely missing, which currently precludes any reliable inferences. The present study was intended as a conceptual replication of a previous study [(1), Neuropsychologia 84; n = 15]. There, we had observed a reduction in avoidance learning capacity when women were in the high estradiol state of the late follicular phase as compared to the mid luteal phase with enhanced progesterone influence. These results conformed to the idea that estradiol and progesterone may antagonistically modulate dopaminergic transmission as a dopamine agonist and antagonist, respectively. Heightened progesterone in the luteal phase thereby supported the ability to learn from the negative outcomes of one's actions, while the follicular rise in estradiol interfered with this capacity. Here, we re-examined the above described within-subject difference between the follicular and the luteal phase in a between-subjects design. Seventy-five women were tested once with a probabilistic feedback learning task, while being either in the follicular (36 women) or luteal phase (39 women), and were compared for phase-related differences in behavior. Secondly, we combined the new data with data from three previous studies from our laboratory that used the same task and menstrual cycle phases. This meta-analysis included only data from the first test day, free of any biasing expectancy effects. Both analyses demonstrated the consistency of the decline in avoidance learning in the follicular relative to the luteal phase. We also showed that this decline reliably occurred in all of the included samples. Altogether, these results provide evidence for the consistency of a behavioral difference and its apparent association with a transient change in hormonal state that occurs in the natural menstrual cycle. Our findings may also open new avenues for the development of reliable between-subjects test protocols in menstrual cycle research.