Impact of maternal emotional state during pregnancy on fetal heart rate variability

Abstract

Background

The fetal autonomic nervous system (ANS) is believed to be negatively affected by maternal adverse emotional states. In this study, we evaluated how depression, anxiety and stress during pregnancy are related to fetal heart rate variability (HRV) as recorded with magnetocardiography (MCG). We also considered metabolic factors such as maternal adiposity and circulating levels of cortisol during gestation. Furthermore, we followed up these fetuses after birth, recording HRV and saliva levels of cortisol in these infants to establish any effects postpartum.

Methods

We calculated HRV in spontaneous MCG recordings from 32 healthy fetuses between 32 and 38 weeks of gestational age. Maternal emotional state was assessed using standardized questionnaires about anxiety, depression and stress. An overall indicator of maternal well-being was calculated by z-scoring each individual questionnaire and summation. We used a median split to divide the group into high and low z-scores (HZS and LZS), respectively. Standard HRV measures were determined in the time and frequency domain. T-test analyses were performed between LZS and HZS, with the HRV and the metabolic measures as the dependent variables.

Results

We found an impaired HRV in the HZS group both during pregnancy and after birth. No differences were observed between LZS and HZS for metabolic factors. Depression and anxiety symptoms seem to affect HRV differently. No relationship was found between maternal and infant cortisol levels.

Conclusions

On the basis of our results on different HRV parameters, we propose that maternal emotional state might affect the development of the fetal nervous system in utero.

Highlights

• •Maternal emotional states might impact fetal autonomic nervous system development.
• •Depression and anxiety seem to impact fetal heart rate variability differently.
• •Impaired heart rate variability seem to persist after birth.
• •These differences are not related to maternal metabolic factors such as adiposity.

1. Introduction

Pregnancy is characterized by many changes with regard to hormonal levels, professional and social environment, and financial situation. These factors could lead to intense emotional states, and there is currently growing evidence to suggest that increased maternal stress during pregnancy can lead to adverse effects on the physiological, metabolic and neuronal development of the fetus during gestation, with possible long-lasting effects [1]. Generally speaking, adverse maternal emotional states such as depression, anxiety, social stress, discrimination, and general prenatal distress can influence fetal and neonatal motor, cognitive, and social development [[2], [3], [4], [5], [6], [7]]. The severity of psychological symptoms is associated with the severity of many anthropometric outcomes such as preterm birth or low birth weight [8].

The evaluation of fetal well-being consists in monitoring the fetal autonomic nervous system (ANS). For example, higher fetal heart rate variability (fHRV) and lower fetal heart rate (fHR) over gestational age (GA) have been taken as indicators of fetal health and of fetal development [9]. A reduced fHRV is associated with fetal distress and impaired development of the ANS on both the sympathetic and parasympathetic branches (SNS and PNS respectively. For a review, see Ref. [10]). The fetal ANS quickly responds and adapts to environmental changes, and negative effects on fHRV have been reported following maternal stress, lifestyle, pathologies and drug exposure (see Ref. [11] for a review). Maternal chronic emotional state also influences fetal ANS, and both elevated [12] and decreased baseline fHR [13] have been linked to maternal depression. It is also recognized that maternal psychological stress is related to decreased fHRV in the second and third trimester of gestation (e.g. Refs. [13,14]), which is consistent with a delayed ANS maturation [10].

Psychological health and chronic stress during pregnancy are also associated with many hormonal, inflammatory and metabolic changes. Prenatal levels of depression or anxiety, for example, are linked to a higher concentration of cortisol [15] which, in turn, is associated with fetal motor activity [16], infants’ stress reactivity [15], and nervous system development [17]. In general, exposure to glucocorticoids hormones impacts fetal synaptogenesis, neurotransmitters’ functions on receptor expression, maturation of organs, birth weight [18], and the future development of cognitive functions [19] and emotion regulation [17]. In addition, maternal stress affects the maternal inflammatory response, and there is a positive linear relationship between stress levels and Interleukin-6 (IL6) concentrations (e.g. Ref. [20]). Cytokines such as IL6 are normally involved in the inflammatory response, and mediate healthy pregnancies. However, they can act as teratogens, affecting or disrupting placental developmental pathways and thereby leading to birth defects, pregnancy complications or premature birth [[21], [22], [23], [24]]. Finally, maternal psychological well-being, chronic inflammation and birth complications are also related to other maternal metabolic factors such as pre-pregnancy Body Mass Index (BMI) and Gestational Weight Gain (GWG) [25,26]. For example, excessive pre-pregnancy BMI and GWG are linked to preterm birth, caesarean delivery, increased birthweight and fetal growth, macrosomia, and others [[27], [28], [29]]. In addition, maternal weight has an independent effect on fetal autonomic functions [30,31]. Even though some works established a positive association between pre-pregnancy BMI or GWG with depression or anxiety levels [[32], [33], [34]], other works reported the inverse pattern or no association [35,36].

The aim of our current work is to evaluate how the overall maternal emotional and psychological state during pregnancy in a non-clinical population is related to fHRV, maternal circulating levels of cortisol and IL6, pre-pregnancy BMI and GWG. We also follow up the fetuses in the first two months of life to highlight possible effects in later life. We recorded HRV and saliva levels of cortisol in these infants. Our study was preregistered on OSF [37]. Notably, pre-pregnancy BMI and GWG were not listed as variables of interest in the preregistration protocol. However, we opted to include them in the current report. We expected that fetuses from mothers with worse emotional well-being would have higher mean fHR and lower fHRV. We expected the same results at the follow-up. We also expected a positive correlation between maternal circulating level of cortisol during pregnancy and infant saliva cortisol. In addition, we explored how pre-pregnancy BMI, GWG, maternal cortisol and IL6 levels are related to maternal emotional state. Finally, we explored how depressive and anxiety symptoms per se might affect our parameters of interest.

2. Methods

2.1. Participants

For our study, we enrolled 40 healthy, German speaking, adult mothers with singleton, uncomplicated pregnancies between 32 and 38 weeks of GA. We excluded those mothers who consumed alcohol or drugs during pregnancy, who had an acute diagnosed mental disorder, high-risk pregnancy, confirmed hemoglobin levels below 10 g/dl, urinary infections or inflammation, or whose fetus had congenital disorders. For the study follow-up, infants and mothers were invited for a second visit during the first two months of the offspring’s life. We included those newborns who were healthy and with an uncomplicated birth. We excluded any fetuses with congenital disorders. See Fig. 1 for a consort chart of the study design.

Fig. 1.

Study design. Inclusion and exclusion criteria, and variables collected during pregnancy (1st visit) and at the follow-up (2nd visit) are shown.

2.2. Procedure

The current study was conducted at the fMEG Center at the University of Tuebingen, Germany. The study was approved by the Ethical Committee of the Medical Faculty at the University of Tuebingen (No. 383/2019BO1). All the participants gave their informed consent in accordance with the Declaration of Helsinki. The procedure consists of two laboratory visits, one during pregnancy and one following delivery in the first two months of life. For the first visit, after arrival at the fMEG Center, a short interview was conducted during which demographic data were collected. A short ultrasound measurement to determine the fetal head position was carried out by our in-house midwife. The women were also requested to fill in the Profile of Mood States (POMS) and the Visual Analog Scale (VAS) questionnaires to evaluate current levels of mood and stress. The women were then placed on the biomagnetometer device to record fetal biomagnetic activity. After the recording, the ultrasound measurement was repeated to again assess fetal head position. The women were then asked to fill further questionnaires (in German). Finally, a basal fasting blood sample was collected.

For the second visit, only the biomagnetic recording was performed. In addition, the day before the second measurement, the mothers collected infant saliva samples to determine infant cortisol at two time-points (morning (8–9 am) and evening (8–9 pm)) and brought these to the appointment at the fMEG Center. The blood samples were analyzed in Tuebingen, Germany, while the saliva samples were analyzed in Dresden, Germany.

2.3. Fetal and infant magnetocardiography

The fetal and infant magnetocardiographic data were recorded using a fetal biomagnetometer device. This is a non-invasive device for the recording of brain and heart activity of fetuses during the last trimester of gestation, and of infants during the first months of life [38]. The datasets in the current study were recorded using a SARA (SQUID Array for Reproductive Assessment, VSM MedTech Ltd., Port Coquitlam, BC, Canada) system which is available at the fMEG Center at the University of Tuebingen, Germany. This device consists of 156 magnetic sensors and 29 additional reference sensors. All sensors are placed in a concave array which is designed to match the maternal abdomen. To restrict the influence of external magnetic fields, the device is placed in a magnetically shielded room (Vakuumschmelze, Hanau, Germany). For the fetal measurements, an ultrasound localization (Ultrasound Logiq 500MD, GE, UK) of the fetal head position is performed by the midwife, who is present throughout the entire examination. During the biomagnetic measurement, a sensor is placed at the position of the fetal head on the abdominal surface. After briefing and ultrasound, the woman is positioned on the device and we check that she is comfortable. Measurements consist of 10 min of silent spontaneous recording, during which time no stimulation is given. The sampling frequency is 610,3516 Hz. The ultrasound localization is repeated after the measurement to detect possible changes in fetal head position. For the infant measurements, the infant lies comfortably in a cradle which is positioned on the biomagnetometer. The procedure does not otherwise differ from that for the fetal measurements.

2.4. Questionnaires and groups definition

The overall maternal emotional state was assessed by standardized questionnaires related to anxiety, depression, and stress. Nast et al. [39] described well-established questionnaires for investigating stress, and our selection is based on this work. A list of questionnaires and the concept measured by each one is provided in Table 1. The questionnaires were both computer based or paper pencil. For the computer-based questionnaires we used the software Unipark (www.unipark.de).

Table 1.

List of questionnaires and the psychological construct evaluated by each.

Category	Questionnaire
Acute stress	Profile of Mood states (POMS)
	Visual Analog Scale (VAS)
	Patient Health Questionnaire (PHQ-D)
Anxiety	State-Trait Anxiety-Inventory (STAI)
Depressive symptoms	Edinburgh Postnatal Depression Scale (EPDS)
Pregnancy-related stress	Prenatal Distress Questionnaire (PDQ)
Life events	Prenatal Life Event Scale (PLES)
Daily hassles	Perceived Stress Scale (PSS)
Chronic stress	Trierer Inventar zum Chronischen Stress (TICS)

2.4.1. LZS/HZS

On the basis of the different questionnaires, we calculated an overall indicator of maternal well-being to create two groups. Following the evaluation and scoring of each questionnaire separately, the values were z-scored. Later, the z-scores of each questionnaire were summed within subject. Note that POMS has four different scales, while PHQ-D has two. The following formula was used to calculate the overall stress score for each participant:

Stress_score = (0.25* POMS_depression/anxiety + 0.25 * POMS_fatigue + 0.25 * POMS_vigor+ 0.25 * POMS_hostility + VAS + 0.5 * PHQ-D + 0.5 * schweregrad_stress + STAI + EPDS + PDQ + PLES + PSS + TICS)/9.

Finally, on basis of the resulting overall z-score, participants were divided into a high or a low z-score (HZS and LZS, respectively) group by median split.

2.4.2. Anxiety and depressive symptoms

To evaluate the influence of anxiety and depression symptoms, we also investigated the STAI and EPDS questionnaires separately. The values of the questionnaires were z-scored. On the basis of their z-scores, the participants were allocated to a high or a low z-score group by median split.

2.5. Maternal fasting blood

During the first laboratory visit, 20 ml of maternal basal fasting blood is collected subsequent to the fetal recordings. The following parameters are determined: electrolytes, creatinine, urea, uric acid, glutamic oxaloacetic transaminase or aspartate aminotransferase, Glutamat-Pyruvat-Transaminase (GPT), Alanin-Aminotransferase (ALAT), gamma glutamyl transpeptidase, albumin, CRP, HDL/LDL, iron, ferritin, TSH, glucose, HbA1c, insulin, C-peptide, total cholesterol, triglycerides, full blood count (FBC), plasmatic coagulation factors, free fatty acids, GAD–II–antibodies, human chorionic gonadotropin (hCG), alpha-amylase, testosterone, cortisol, and Interleukin 6 (IL6). For the current paper, only cortisol and IL6 are analyzed further.

2.6. Infant saliva samples

The infant saliva samples are collected by the mothers the day before the second measurement at two different time points; between 8 and 9 am and between 8 and 9 pm [40,41]. Cortisol is collected by Eye Spears (VisispearTM Cellulose-Keiltupfer, 7 cm, Beaver-Visitec International, Ltd). Each spear consists of a plastic applicator shaft joined to a triangular absorbent sponge. The sponge is placed under the infant’s tongue for approximately 10 s, by which time the sponge is completely soaked through. The sponges are then placed in a tube and frozen until analysis. The mothers bring the two saliva samples to the fMEG Center when they come for the second measurement.

2.7. Preprocessing

The data preprocessing of MCG data and extraction of HRV parameters is carried out using Matlab R2016b (The MathWorks, Natick, MA, USA). R-peak detection of maternal and fetal heart is performed using FLORA [42]. Maternal heart interference was removed by FAUNA [43]. HRV parameters in the time and frequency domain were calculated in accordance with the example from Mat Husin and colleagues [31], and a summary and brief description of the associated autonomic function is listed in Table 2.

Table 2.

Description of assessed HRV parameters.

	Parameter	Description	Associated autonomic function
Time domain	HR	Mean heart rate	SNS accelerates the HR, while PNS slows the HR
	SDNN	Standard deviation of R-R intervals	Reflects the overall HRV. SNS and PNS both contribute to SDNN, and SDNN is correlated to LF
	RMSSD	Root mean square of successive differences between normal heartbeats	Reflects short term HRV as it describes the variance in HR. Estimates the vagally mediated changes in HRV
	NN10	number of pairs of successive NN intervals that differ by > 10 ms	Mainly related to PNS, and highly correlated to SDNN
Frequency domain	LF	Absolute power of the low (0.08–0.2 Hz) frequency bands	SNS and PNS both contribute to LF, reflecting blood pressure and HR accelerations
	HF	Absolute power of the high (0.4–1.7 Hz) frequency bands	Reflects PNS activity and is called the respiratory band since it corresponds to HR variations due to respiratory cycle
	LFHF	Ratio of LF to HF power	Estimates the sympathovagal balance. Possibly reflects the ratio between SNS and PNS activity, but this remains controversial

2.8. Statistics

All statistical analyses were performed using Matlab R2022a. Before proceeding with any of the following tests, outliers were removed using the function “rmoutliers” in Matlab. Our first aim was to test whether fetal HRV (HR, SDNN, RMSSD, NN10, LF, HF, LFHF) or maternal metabolic factors (maternal pre-pregnancy BMI, GWG, cortisol and IL6) differ between the HZS and LZS groups. To this end, having monitored for normal distribution, we performed a two-sample t-test using the function “ttest2” in Matlab. Significance level was set at p = 0.05. For all contrasts, the Holm-Bonferroni correction for multiple comparisons was applied. Second, to test whether HRV in infants differed from HZS and LZS groups, we also applied two-sample t-test, set the significance level at p = 0.05 and applied Holm-Bonferroni correction. Third, to calculate the relation between maternal circulating level of cortisol during pregnancy and infant saliva cortisol levels both in the morning and the evening, a linear model was defined as:

y = β0 + β1X + ε

where y is maternal cortisol, β0 is the intercept, β1 the regression coefficient, X the infant cortisol, and ε the error in the estimate. Model value of p < 0.05 was considered statistically significant. Fourth, to investigate whether fetal HRV or maternal metabolic factors were influenced by anxiety or depression alone, we z-scored the EPDS (depression) and STAI (anxiety) questionnaires individually, performed a median split, and compared our parameters between the high and low z-score groups. This entailed performing a further two-sample t-test, setting the significance level at p = 0.05 and applying Holm-Bonferroni correction.

Unlike in the preregistration, during which we specified one-sided unpaired t-tests and MANOVA analyses to test our hypotheses, we opted to use two-sample t-tests to harmonize the statistical tests.

3. Results

Of the 40 participants recruited, 32 participants were eligible for further analyses after data quality check. Reasons for exclusion from the study analysis were missing/incomplete questionnaires, lack of blood samples, or noisy fMCG recordings. Of the 32 fetuses in our study, 17 infants returned to the follow-up appointment and their data was of good quality (Fig. 2). Of the 17 infants, 12 had viable saliva cortisol data both in the morning and in the evening. In addition, we excluded the evaluation of IL6 levels from all further analysis since most of the women (N = 27/32) had values below the detection level and all reported in the metabolic panel as <2.7 (ng/l). For details on these fetuses and infants, see Table 3. In Supplementary Table 1, descriptive values of the single questionnaires are also presented.

Fig. 2.

Summary of participants during the study.

Table 3.

Description of our sample.

Variable	Range	Mean ± standard deviation
Fetal GA (weeks)	32 – 38	33.6 ± 1.9
Maternal age (years)	23 – 42	31.7 ± 5.1
Weight at birth (g)	2510 – 4450	3416.2 ± 384.7
Pre-pregnancy BMI (Kg/m^2)	17.5 – 32.7	22.9 ± 3.4
GWG (Kg)	4.5 – 26.4	11.2 ± 4.2
Maternal blood cortisol (nmol/L)	435.0 – 1006.0	653.5 ± 130.4
Infants age (days)	18 – 63	36 ± 12
Infant morning cortisol (nmol/L)	0.8 – 23.34	6.3 ± 6.4
Infant evening cortisol (nmol/L)	0.5 – 11.8	3.7 ± 3.2

3.1. Comparison of fetal HRV and metabolic factors between LZS and HZS

Following Holm-Bonferroni correction for multiple comparison, LF was found to be significantly lower in the HZS than in the LZS group (t(26) = 3.86, p = 0.0007, Fig. 3A). In Supplementary Table 2, the means and standard deviations of our parameters for the LZS and HZS groups are provided.

Fig. 3.

A) LF components of the HRV are lower in fetuses from HZS mothers. B) LFHF is higher in infants whose mothers belonged to the HZS group during pregnancy.

3.2. Comparison of HRV in infants between LZS and HZS

Following Holm-Bonferroni correction for multiple comparison, LFHF was significantly lower in the LZS than in the HZS group (t(12) = 3.62, p = 0.004, Fig. 3B). The means and standard deviations of our parameters for the LZS and HZS groups are shown in Supplementary Table 3.

3.3. Relation between maternal blood cortisol levels and saliva infant cortisol

Of the 32 participants involved in the current analysis, only 12 infants had recorded values of saliva cortisol both in the morning and in the evening. This model was therefore not significant for the correlation between maternal cortisol and infant morning cortisol (R² = 0.07, p = 0.42). Maternal cortisol and infant evening cortisol were, furthermore, not related to each other (R² = 0.03, p = 0.64).

3.4. Relationship between fetal and infant HRV and anxiety/depressive symptoms

Following Holm-Bonferroni correction, fetal LF was lower in the high EPDS than in the low EPDS group (t(26) = 3.41, p = 0.002). There were no significant differences in infant HRV parameters based on EPDS score. Fetal NN10 (t(28) = 3.00, p = 0.006) and HF (t(26) = 2.92, p = 0.007) were lower in the high STAI group than in the low STAI group. Finally, infants from the high STAI group had lower HF than the low STAI group (t(11) = 3.96, p = 0.002). No other differences were detected. Mean, standard deviation and p-value relative to the group differences are available in the Supplementary Table 4, Supplementary Table 5, Supplementary Table 6, and Supplementary Table 7.

4. Discussion

4.1. Overall maternal emotional state affects the fetal and infant ANS

Taking the overall maternal emotional state into consideration, a comparison of fetal HRV between low and high z-score groups revealed that the absolute power of the low (LF, 0.08–0.2 Hz) frequency bands was lower in HZS than in LZS (Supplementary Table 2). In addition, NN10 and HF tended to be lower in the HZS group, despite not reaching the significance level after correction for multiple comparisons. Low frequency components of the HRV reflect the activity of both the SNS and PNS and, in adults, a reduction in this parameter is associated with stress, negative emotional processing and cardiovascular complications (e.g. Ref. [44]). Moreover, we showed that NN10 and HF were reduced in those fetuses whose mothers suffered from higher anxiety symptoms, as measured by the STAI questionnaire (Supplementary Table 6).

The effects on HRV observed during pregnancy appear to persist after birth. In fact, at the follow-up, infants from the HZS group still displayed ANS alterations, with a higher LFHF than those who experienced lower levels of stress during pregnancy. Again, this effect is concordant with findings in adults, where stress elicits a reduction in parasympathetic activation, reflected by an increase in the LFHF [45,46]. However, one should be careful while associating lower HRV in adults to fetuses and infants. In fact, compared to adults, HRV in fetuses, neonates and infants is sympathetic dominant. Specifically, the sympathetic influence on the vascular system decreases during early development, while the parasympathetic modulation of heart activity increases [47].

In addition – and contrary to our hypothesis – we found no differences in the mean fetal heart rate on the basis of maternal emotional state. Literature reporting the effects of maternal stress, anxiety or depression on the fetal HR generally focuses on acute or severe emotional conditions, or laboratory-induced stressors [3]. In our sample, however, maternal distress is limited, and acute mental disorders precluded participation in the study. In addition, the lack of effects on fetal HR might depend on the short recording time (10 min).

4.2. No effect of emotional state on metabolic parameters

Emotional and psychological distress are closely related to a large number of inflammatory and metabolic alterations. In this work, no effects of the overall maternal emotional state during pregnancy were observed on maternal blood cortisol levels, pre-pregnancy BMI, or weight gain during pregnancy (Supplementary Table 2). We also collected IL6 concentrations, albeit 27 out of 32 participants had very low values, preventing further analyses. Again, we consider the lack of differences in our results as dependent on our cohort, which had consisted of women without clinically relevant or acute psychological disorders. Furthermore, no relationship between maternal cortisol levels during pregnancy and infant saliva cortisol measured both in the morning and in the evening could be established. However, only few infants had valid values and this, together with individual cortisol level fluctuations, prevents us from drawing any conclusion on the basis of these results.

4.3. Depression and anxiety symptoms have different effects on fetal ANS

One of our goals was to determine whether depression or anxiety scores affect the HRV parameters in different ways. On the basis of the EPDS questionnaire, which screens for depression symptoms, we established that fetuses from mothers with higher scores have lower LF (Supplementary Table 4). This reduction in LF power was also observed when the overall emotional state was taken into consideration, and might further support the idea that maternal negative emotional processing has an effect on the fetal PNS. However, since no differences persist in infants (Supplementary Table 5), the effect of depressive symptoms might be only transitory.

Anxiety symptoms, recorded with the STAI, appear to have different outcomes. In fact, fetuses from mothers who had reported a higher level of stress have lower NN10 and lower HF (Supplementary Table 6). This trend was already visible when considering the overall emotional state. The effect of maternal anxiety on the offspring seems to persist at the follow-up, with infants from the higher anxiety group having reduced HF (Supplementary Table 7).

As previously mentioned, a reduced PNS activation in adults is anticipated in response to psychological stress [45,46], and differential effects of maternal depression and anxiety symptoms on the fetus have already been reported [3,6,13,48].

5. Limitations and outlook

In this work, we detected effects of maternal emotional state during pregnancy on fetal and infant ANS, as quantified by means of HRV. However, if we focus our recordings on fetuses in the last trimester of gestation, we cannot conclude whether our results depend on exposition to maternal stress in the last trimester or on continuous exposure during the whole pregnancy. Similarly, stress exposure during sensitive windows for the fetal development might have different outcomes on fetal ANS development [11]. In addition, due to empty or missing samples, the collection of saliva cortisol in infants produced invalid data in approximately one third of our population. The correlation between maternal and infant cortisol was not significant, but the low number of participants and the high inter-personal variability in cortisol levels during the day prevent a clear interpretation of this finding. Finally, a longer follow-up would enable us to determine whether the effects on the offspring are persistent over time, and whether they relate to the offspring’s cognitive, emotional or motor activity levels.

6. Conclusions

Our results endorse the theory that, in a population of pregnant women without any acute psychological disorder, maternal emotional state in the prenatal period might affect fetal and infant ANS development. Overall, the altered heart rate variability in the offspring from mothers with unfavorable emotional state might highlight a negative effect of maternal stress on fetal ANS development, and depressive and anxiety symptoms seem to affect HRV parameters in different ways. However, at least in our sample, these differences do not relate to maternal metabolic factors such as circulating levels of cortisol and inflammatory markers, adiposity before pregnancy, and weight gain during pregnancy. Furthermore, no relation between maternal and infant cortisol levels was found, which might be due to the small sample size and fluctuation of the subjective cortisol levels. For decades, impaired mental health during pregnancy has been linked to adverse outcomes on both the mother and offspring’s well-being. With the current work, we propose that exposition to stress, even in the absence of an acute mental disorder, might affect the development of the fetal nervous system in utero. These findings emphasize that a careful and diverse evaluation of the maternal emotional state during the perinatal period should be part and parcel of clinical routine.

Author contributions

L.S. conducted the analysis, drafted and revised the manuscript, and prepared the figures. I.B. conceptualized the study, performed the preregistration and the recordings. K.S drafted and revised the manuscript. J.H. performed the recordings. N.K.S. conceptualized and supervised the work. H.P. supervised the work. All authors discussed the results and implications, reviewed and edited the manuscript and approved its final version.

Data availability statement

The data used for the current publication, together with the related codes used for analysis, are available from the corresponding author upon reasonable request.

Declaration of competing interest

The authors declare no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article.