Approximately 80% of women of reproductive age experience one or more physical and emotional symptoms during the premenstrual phase of the menstrual cycle (Hylan et al., 1999; Steiner et al., 2003). When these symptoms are confirmed over two or more cycles and significantly impact daily life, the condition is classified as premenstrual syndrome (PMS). PMS symptoms often worsen significantly during the late luteal phase, typically starting around 6 days before the onset of menstruation and peaking approximately 2 days before menstruation begins (Meaden et al., 2005; Pearlstein et al., 2005). These symptoms generally subside or disappear with the onset of menstruation. The most distinctive feature of PMS is the cyclical nature of its symptoms, closely associated with the menstrual cycle.

PMS symptoms are diverse, including physical symptoms such as abdominal pain, bloating, breast tenderness or swelling, and fatigue. Emotional symptoms include depressed mood or dysphoria, anxiety or tension, affective lability, and irritability. Among these, irritability is reported as the most severe complaint and tends to appear slightly earlier than other symptoms. Typically, women experience the same symptoms consistently from one cycle to the next (Bloch et al., 1997).

The diagnostic criteria for PMS are primarily based on assessing physical, psychological, and behavioral symptoms that recur in relation to the menstrual cycle. Globally, the criteria established by organizations such as the American College of Obstetricians and Gynecologists (ACOG) and Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) are commonly used. Based on these diagnostic criteria, the premenstrual symptoms screening tool (PSST) was developed to assess symptoms severity related to PMS and premenstrual dysphoric disorder (PMDD) as well as their impact on daily life (Steiner et al., 2003). The prevalence of PMS among all women is estimated at approximately 50%, with symptoms ranging from mild to severe (Direkvand-Moghadam et al., 2014). However, the diagnosis of PMS often proves challenging as it relies on the subjective evaluation of symptoms. Additionally, many women do not seek medical consultation regarding their symptoms, leading to delayed or undiagnosed cases (del Mar Fernández et al., 2019).

PMS is influenced by menstrual cycle-related sex hormones such as estrogen and progesterone. Receptors for these hormones are abundantly expressed in brain regions such as the amygdala and hypothalamus (Del Río et al., 2018), and they are known to interact closely with neurotransmitters, significantly affecting women’s emotions and mood (Del Río et al., 2018; Fruzzetti and Fidecicchi, 2020). While these sex hormones fluctuate cyclically, they decrease sharply during the late luteal phase, leading to a significant worsening of PMS symptoms. However, the exact mechanisms underlying the development of PMS remain unclear. This is because there is no difference in blood concentrations of sex hormones, or any structural abnormalities, between women with and without PMS (Backstrom et al., 1983; Rubinow and Schmidt, 1995; Rubinow and Schmidt, 2019). Consequently, it is challenging to identify clear pathological findings specific to PMS (Gollenberg et al., 2010). Instead, women with PMS may exhibit hypersensitivity to fluctuations in sex hormone levels within the normal range (Steiner et al., 2003; Yonkers et al., 2008).

Understanding how this heightened sensitivity to hormonal fluctuations affects emotional states in women with PMS is critical. We hypothesize that women with PMS may experience difficulties in the processing of internal state information, such as sensitivity to visceral conditions and sex hormone levels, which is referred to as interoception. Interoception is involved in updating the internal model of the body and the emergence of emotions (Khalsa et al., 2018). Interoceptive accuracy is associated with the intensity of emotional experiences (Garfinkel and Critchley, 2013); interoceptive awareness significantly influences the severity of menstrual and premenstrual symptoms (Borlimi et al., 2023). It is crucial to examine women with PMS for their interoceptive accuracy and awareness to better understand the severe discomfort they experience. We assessed subjective emotional states, interoceptive accuracy, and interoceptive awareness during the late luteal phase, a period characterized by a sharp decline in sex hormone levels.

Interoceptive accuracy was assessed using the Heartbeat Counting Task (HCT; Schandry, 1981); meanwhile, interoceptive awareness, reflecting interoceptive body consciousness, was evaluated through self-reports using the Multidimensional Assessment of Interoceptive Awareness (MAIA; Mehling et al., 2012). Subjective emotional states were measured using the Positive and Negative Affect Schedule (PANAS; Watson et al., 1988) and stress scores on the Visual Analogue Scale (VAS).

Interoceptive signals are transmitted to the brain through the autonomic nervous system (ANS) and contribute to the maintenance of homeostasis (Tsakiris and Critchley, 2016). The ANS, composed of the sympathetic and parasympathetic branches, plays a crucial role in unconsciously regulating basic bodily functions (Critchley and Harrison, 2013). However, PMS has been associated with ANS imbalances, and heart rate variability (HRV) has decreased in women with severe PMS (Matsumoto et al., 2007a, 2007b). When this regulatory mechanism is disrupted, the effects can extend throughout the body, leading to the development of complex and multifaceted symptoms, such as those seen in PMS (Matsumoto et al., 2007a, 2007b). One of the exacerbating factors of PMS is stress (Gollenberg et al., 2010). Acute stress during the luteal phase in women with PMS, in addition to chronic menstrual stress, significantly exacerbates autonomic dysfunction (Meng et al., 2022). Imbalance in autonomic nervous function disrupts the proper transmission of interoceptive signals, leading to inaccurate recognition of the body’s internal states and emotional instability (Critchley and Harrison, 2013).

Thus, clarifying how stress impairs the appropriate recognition of bodily information in women with PMS is important, as this contributes to the experience of more severe negative emotions. We hypothesize that stress during the late luteal phase disrupts the transmission of interoceptive signals and impairs the stress recovery process in women with PMS through ANS imbalance. By examining changes in subjective emotions, interoceptive accuracy, and ANS activity during, before, and after stress, including the recovery period, we investigated the influences of stress in women with PMS.

In this study, we compared subjective emotional states, interoceptive sensations, and ANS indicators between the PMS group and the without PMS group during the luteal phase. Additionally, we examined the effects of the TSST on stress responses and their recovery. The results were discussed to explore the characteristics of PMS in the luteal phase and the relationship between PMS and stress.

Based on the screening results of the study participants, 23.6% were identified as having PMS. The prevalence of PMS among all women is estimated to be around 50%, although this rate varies depending on biological factors such as social environment and age (Direkvand-Moghadam et al., 2014). In this study, the PSST was used for PMS screening. The PSST consists of two components: the severity of physical and emotional symptoms (measure of severity), and the impact of premenstrual symptoms on daily life (impact of premenstrual symptoms) (Steiner et al., 2003). The score derived from the severity of physical and emotional symptoms is referred to as the “PMS score,” with higher scores indicating more severe premenstrual symptoms. The distribution of PMS scores among the study participants ranged from a minimum of 12 to a maximum of 42 (Supplementary Figure S1).

Phase 1 corresponds to the baseline before stress. In this study, the measurements of psychological and physiological indices in women with PMS and those without PMS were compared. The results of Phase 1 showed no significant differences in positive affect states between the two groups. However, negative affect states and subjective stress levels were significantly higher in women with PMS. These findings are largely consistent with previous research (Liu et al., 2017), supporting the accurate assessment of the unpleasant symptoms experienced by women with PMS during the luteal phase and the reliable identification of PMS.

Measurements of interoception were conducted using the HCT to assess interoceptive accuracy and the MAIA to evaluate interoceptive awareness. The results showed that these two scores exhibited opposite trends. Specifically, the PMS group scored higher on the HCT but lower on the MAIA compared to the without PMS group. These findings suggest that women with PMS demonstrate higher interoceptive accuracy but lower interoceptive awareness.

Interoceptive accuracy is associated with the intensity of emotional experiences (Wiens et al., 2000). This relationship arises because improved detection of bodily information enables a clearer understanding of one’s emotions (Nord and Garfinkel, 2022). Moreover, higher interoceptive accuracy facilitates better emotion regulation in response to negative affects (Füstös et al., 2013; Pollatos et al., 2014; Kever et al., 2015).

By contrast, low interoceptive awareness has been reported to reduce bodily awareness and trust, making it harder to recognize one’s emotions. This, in turn, may hinder effective emotional processing and increase the likelihood of chronic stress (Leech et al., 2024). In this study, the PMS group showed opposite trends in the results of the HCT and MAIA. This aligns with the finding that beliefs about interoceptive aptitude do not necessarily match actual interoceptive accuracy. Furthermore, this mismatch suggests a maladaptive state in terms of mental health (Garfinkel et al., 2015; Garfinkel et al., 2016; Nord and Garfinkel, 2022).

No significant differences were found in ANS activity, responsible for transmitting interoceptive signals, between the two groups. The lack of attenuation in power spectral components during the luteal phase among PMS group in this study may be attributed to differences in experimental methods, particularly in the definition and determination of menstrual cycle phases and participant selection criteria (Matsumoto et al., 2007a, 2007b). In Matsumoto’s study, PMS was identified using a questionnaire that assessed physical and psychological complaints associated with the menstrual cycle, and comparisons were made with PMDD, a more severe form of PMS. Differences in screening methods, including the use of the PSST in this study and the more detailed classification of premenstrual symptoms by severity, are likely to have significantly influenced the results.

Finally, the absence of significant differences in ANS activity between the two groups highlights that women with PMS experience a mismatch between two distinct dimensions: interoceptive accuracy and interoceptive awareness. The mismatch between interoceptive predictions and actual interoceptive signals generates prediction errors and contributes to the emergence of emotions (Friston, 2010). Emotions and affective sensory states are believed to arise from interoceptive states of physiological arousal, as well as the precision and predictability of these signals (Barrett and Simmons, 2015; Inui, 2018).

This suggests that women with PMS may experience severe discomfort during the late luteal phase, such as depressed mood or dysphoria, anxiety or tension, affective lability, and irritability. This may be due to their ability to accurately perceive hormonal fluctuations while having reduced awareness of interoceptive sensations.

We conducted a simple regression analysis with PMS scores, which indicate the severity of PMS symptoms, as the dependent variable, and scores measuring interoceptive awareness as the independent variables. The independent variables included the HCT score, the total score of the MAIA, and its eight subscales. As a result, the predictor of PMS symptoms was identified as the HCT score. These results suggest that in predicting PMS symptoms, the objective perception of physiological body signals (as measured by the HCT) is important, while subjective interoceptive awareness (as assessed by MAIA) does not have a direct impact. Therefore, it is considered crucial to focus specifically on the “accuracy” aspect of interoceptive awareness to understand symptoms and mechanisms related to PMS.

The results of the simple regression analysis showed a small effect size (R² = 0.056), indicating a limited influence. One possible explanation for this outcome is that PMS symptoms exhibit significant individual differences and are influenced by complex and multifaceted characteristics, including emotional and psychological symptoms (Otsubo, 2018). Interoceptive sensory signals are transmitted to the brain via the ANS; however, this regulatory mechanism is disrupted by stress.

Since stress is one of the factors that exacerbate PMS, this study examined the effects of stress by comparing the mean values of various indices (subjective emotional scores, HCT scores, and ANS indices) between the PMS and the without PMS groups. The stress-free phase was set as the baseline for this comparison. The results for subjective emotional states showed that, as expected, a main effect of group was observed for subjective negative emotions and subjective stress scores. While the PMS group showed higher scores, no significant differences were observed between the groups in terms of subjective positive emotions.

Regarding the main effect of time, significant changes were observed in subjective positive effects, subjective negative effects, and subjective stress scores. These scores showed variation between the stress-free phase and immediately after stress, 30 min post-stress, and 60 min post-stress. However, no significant interaction effects were observed. These results suggest that stress induction was conducted appropriately.

The results of the HCT showed that the PMS group scored higher than the without PMS group. Additionally, a main effect of time was observed. Specifically, in the without PMS group, scores tended to increase from the baseline compared to those immediately after stress and 60 min post-stress induction. Conversely, no significant interaction effects were found. The HCT is reported to improve one’s ability to accurately perceive their own heartbeat when repeated (Schillings et al., 2022). The average scores of the HCT in this study were higher compared to those reported in previous studies (0.291–0.611; Desmedt et al., 2020). These results suggest that the findings of the HCT in this study may include both repetition and ceiling effects resulting from repeated implementations of the task.

The results of ANS activity showed no significant main effects of group for any indices. A main effect of time was observed significantly for heart rate, RR interval, HF, RMSSD, and pNN50.

Regarding heart rate, both the PMS group and the without PMS group showed decreases but the timings differed. In the without PMS group, a gradual decrease was observed from the pre-stress phase, whereas in the PMS group, a decrease was observed from immediately after stress to 60 min post-stress. Similarly, mean RR increased in both groups. However, in the without PMS group, a gradual increase was observed from the pre-stress phase, while in the PMS group, a marked increase was observed at 60 min post-stress compared to the pre-stress and immediately-after-stress phases. The HRV analysis revealed that HF increased from the pre-stress phase to 60 min post-stress, and RMSSD and pNN50 showed substantial increases in the PMS group from immediately after stress to 60 min post-stress.

However, for LF/HF, neither the main effect of time nor the interaction was significant. Therefore, it is necessary to carefully examine the effects of stress induction because HRV indices are considered indices of neural modulation and should be interpreted as reflecting changes in tonic neural activity relative to the mean (La Rovere et al., 2020). Thus, the LF/HF results may suggest that both groups exhibited similar ANS reactivity to stress.

To examine the degree of ANS response (Meng et al., 2022), Δ values of indices that showed a main effect of time in the frequency domain analysis were used. Although no group differences were observed significantly, both the main effect of time and the interaction were significant. ΔHF(P3 − P1) was greater than ΔHF(P2 − P1), and this effect was observed in the PMS group.

The observation that HF significantly increased exclusively in the PMS group during the recovery phase, despite the absence of significant differences between the groups at baseline, cannot be straightforwardly interpreted as a mere increase in parasympathetic activity or relaxation effects. This is because no differences in HF were significantly observed between the PMS and without PMS groups at baseline, indicating that the ANS activity of both groups was relatively stable in the absence of stress.

Under the application of a stress load, sympathetic activity typically becomes dominant, leading to a state of physiological arousal. However, this response is regulated by the baroreflex (baroreceptor reflex), which increases parasympathetic (vagal) activity in response to elevated blood pressure, resulting in a decrease in heart rate and suppression of sympathetic activity, even at the ventricular level (La Rovere et al., 2020). Following the relief of stress, the ANS typically initiates a process to restore homeostasis.

In this study, however, excessive parasympathetic activation was observed in the PMS group. This may be attributed to an overreaction of the baroreflex or vagal reflex, leading to an increase in HF. The prolonged increase in HF, exhibiting rebound effects (Mezzacappa et al., 2001), may lead to chronic fatigue and discomfort, making recovery difficult. Therefore, the phenomenon of increasing HF during the recovery phase is considered to reflect ANS imbalance experienced by women in the PMS group (Matsumoto et al., 2007a, 2007b). Additionally, these findings suggest that stress serves as an exacerbating factor for PMS symptoms.

These results suggest that both groups exhibited similar ANS reactivity to stress. However, a rebound effect was observed exclusively in the PMS group compared to the without PMS group, indicating that the PMS group may have impaired autonomic regulation timing, which could manifest as a rebound phenomenon during the recovery process. Furthermore, the results suggest that the PMS group would require more time for long-term adjustment during the recovery phase of stress response regulation, highlighting the potential for increased physiological burden and risk associated with stress in women with PMS.

In conclusion, women with PMS in the late luteal phase experience more negative emotional states and perceive greater stress compared to those without PMS. In the evaluation of interoception, high scores on the HCT and low scores on the MAIA suggest that women with PMS demonstrate high interoceptive accuracy, but their awareness of internal states is low. This mismatch may result in strong emotional experiences driven by accurate bodily information, while a reduced awareness and trust in internal states makes it difficult to integrate and interpret bodily sensations and signals as emotions. Thus, they may find it difficult to accurately recognize their own emotions. Additionally, the HF rebound effect observed exclusively in the PMS group after stress indicates impaired autonomic regulation timing, which may increase physiological burden and stress-related risks. Consequently, this could hinder their ability to process emotions appropriately and may contribute to a tendency toward chronic stress.

This study has several limitations. First, as the sample size is limited, caution must be exercised in generalizing the results. Second, data collection relied on self-reported measures, which may introduce bias. Third, the comparison of mean values for P1, P2, and P3 was conducted using only samples with significant stationarity, which limited the sample size. Considering these points, participant recruitment should be conducted more broadly to ensure generalizability. The prevalence of PMS varies across age groups, and its symptoms are diverse. Furthermore, while this study focused on the late luteal phase and was conducted in a laboratory setting, it is important to examine PMS conditions and the effects of stress during other phases, particularly the late follicular phase. Comparing these findings with the results of this study would provide valuable insights into understanding long-term changes.