Fitness level, but not sex, affects exercise-induced pain modulation

Janne Ina Nold; Tahmine Fadai; Zora Pia Gerbers; Christian Büchel

doi:10.7554/eLife.109061

. 2026 Oct 30;15:RP109061. doi: 10.7554/eLife.109061

Fitness level, but not sex, affects exercise-induced pain modulation

Janne Ina Nold ^1,^✉, Tahmine Fadai ², Zora Pia Gerbers ¹, Christian Büchel ¹

Editors: José Biurrun Manresa³, Timothy E Behrens⁴

PMCID: PMC13722413 PMID: 42961450

Abstract

In a previous study by our group (Nold et al., 2025), we investigated exercise-induced pain modulation after high-intensity compared to low-intensity exercise in a heterogeneous sample of diverse fitness levels. Exploratory analyses suggested an interaction of sex, fitness level, and drug treatment, indicating that males showed increasing hypoalgesia after high- compared to low-intensity exercise with increasing fitness levels, which was diminished when naloxone was administered. In contrast, these effects were not evident in females. These exploratory findings warranted further investigation to determine if and to what extent exercise-induced pain modulation depends on fitness level and/or sex. In this current study, we investigated an all-female sample (N=21) of high fitness levels using a similar paradigm as in the previous study, comparing heat and pressure pain ratings after high-intensity and low-intensity exercise. Our data show an interaction of exercise intensity and stimulus intensity in heat pain, with greater pain relief following high-intensity exercise, especially at the highest stimulus intensity. Despite results for pressure pain not reaching significance, a similar trend was evident. These results suggest that females at a high fitness level also show exercise-induced pain modulation for high-intensity compared to low-intensity exercise. Furthermore, we pooled the data of the current and previous study, providing evidence for fitness level but not sex to moderate exercise-induced pain modulation in heat pain. These findings suggest that exercise-induced pain modulation depends on fitness level but not on sex.

Research organism: Human

Introduction

In a previous study (Nold et al., 2025), we investigated the effect of high-intensity (HI) and low-intensity (LI) aerobic exercise on pain perception in a heterogeneous sample of both sexes and diverse fitness levels (N=39, 21 females), where exploratory analyses revealed a three-way interaction of sex, fitness level, and drug treatment. Males showed greater hypoalgesia after HI compared to LI exercise with increasing fitness levels. This effect was reduced when the µ-opioid antagonist naloxone was administered. These effects were not evident in females. Taken together, these results point to the nuanced interplay of individual factors, such as fitness level and sex underlying exercise-induced hypoalgesia.

Repeated aerobic exercise has been shown to be associated with the activation of endogenous neuromodulatory systems implicated in pain modulation (for review see Gajic et al., 2026), such as the endogenous opioid system (Koltyn et al., 2014; Saanijoki et al., 2022) and endocannabinoid system (Matei et al., 2023). Correspondingly, previous research in humans (Geva and Defrin, 2013; Schmitt et al., 2020; Sluka et al., 2018) suggests that fitness level plays a crucial role in exercise-induced pain modulation, where subjects with higher fitness levels showed increased hypoalgesia following exercise compared to subjects with lower fitness levels (Schmitt et al., 2020). Correspondingly, higher self-reported fitness levels were shown to be associated with reduced pain ratings (Geva and Defrin, 2013), reduced temporal summation (Naugle and Riley, 2014), and greater opioid release in the orbitofrontal cortex and insula (Saanijoki et al., 2022). Crucially, most of these studies Saanijoki et al., 2022; Schmitt et al., 2020 have been conducted in healthy all-male samples, limiting the translation to females, where the association of physical fitness and pain perception after exercise remains less clear (de Bruijn et al., 2011).

Despite the majority of chronic pain patients being female (Fillingim et al., 2009; Mogil, 2012; Rice et al., 2019), evidence on potential sex differences in exercise-induced pain modulation is surprisingly limited (Koltyn et al., 2014). Of those studies that have considered potential sex differences in exercise-induced pain modulation, some (Brellenthin et al., 2017; Koltyn et al., 2014; Niwa et al., 2022; Smith, 2004) could show that there are no sex differences, whereas others (Koltyn, 2000; Sternberg et al., 2001; Vaegter et al., 2014) suggest that hypoalgesia following exercise is more pronounced in females, potentially due to sex differences in the descending antinociceptive circuits (Fullerton et al., 2018; Mogil, 2012), but overall results remain inconclusive (Rice et al., 2019).

This current behavioural study aims to build on our previous findings (Nold et al., 2025) by investigating exercise-induced pain modulation in an independent all-female sample (N=21) with a high fitness level using the same paradigm as in the previous study, without a pharmacological intervention. Importantly, fitness levels (quantified by the weight-corrected functional threshold power [FTP]) in the previous study (Nold et al., 2025) differed significantly (p=0.003) between females (M=1.58, SD=0.44) and males (M=2.03, SD=0.40), thus, warranting further investigation into whether exercise-induced hypoalgesia after HI compared to LI aerobic exercise can also be observed in females with a higher fitness. Disentangling if and to what extent exercise-induced hypoalgesia depends on fitness level and sex is a crucial step to determine exercise-based interventions for (chronic) pain populations.

Results

The final sample included N=21 healthy female participants with higher fitness (weight-corrected FTP: M=2.25 W/kg, SD=0.41; Training volume: M=13.10, SD=12.28). The distribution of menstrual cycle phases (based on self-report and divided into three phases: follicular, ovulatory, and luteal as well as hormonal contraceptives) was equal across (χ2(3)=1.66, p=0.65) as well as within days (Day 1: χ2(3)=2.26, p=0.52; Day 2: χ2(3)=0.40, p=0.94; Figure 1—figure supplement 1A). Furthermore, the expectation of how acute exercise affects different pain domains was assessed (Lindheimer et al., 2020) and yielded no significant effects overall (joint pain: t(20) = –1.78, p=0.09, muscle pain: t(20) = 0.15, p=0.88). However, participants indicated that they expect acute aerobic exercise to reduce whole-body pain (t(20) = –2.83, p=0.01; Figure 1—figure supplement 1B). Finally, mood ratings were assessed using the Profile of Mood States (POMS) questionnaire (Curran et al., 1995) before the experiment commenced (pre) and after it was completed (post). There were no significant differences in pre- and post-levels of fatigue (t(20) = –0.50, p=0.62), discontent (t(20) = 1.84, p=0.08), and drive (t(20) = 0.58, p=0.57). However, participants rated significantly reduced dejection post compared to pre experimental testing (t(20) = 3.61, p=0.002; Figure 1—figure supplement 1C).

Successful pain calibration and exercise intervention

We calibrated pain intensity levels corresponding to 30, 50, and 70 on a Visual Analogue Scale (VAS), where VAS 0 marks the pain threshold (Figure 1—figure supplement 2A and B). On the experimental day, we introduced an online rating scale (real-time ratings while the stimulus is ongoing) where the pain threshold now corresponded to VAS 50 (Figure 1—figure supplement 2C and D) to more accurately track the temporal dynamics of pain perception through continuous ratings and incorporate non-painful sensations. Participants perceived the stimulus intensities as significantly different for pressure (β=1.13, SE=0.07, t(565.20)=17.04, p<2×10^–16; Supplementary file 1, table S7) and heat pain (β=1.76, SE=0.08, t(530.58)=21.46, p<2×10^–16; Supplementary file 1, table S9). The anticipated intensities of VAS 50 and VAS 70 were rated above the pain threshold on the online scale, indicating that they were perceived as painful (Figure 2—figure supplement 1), despite the ratings being lower than the anticipated intensities, potentially due to habituation effects between days (Ellerbrock et al., 2015; May et al., 2012). We achieved a significant difference between HI and LI exercise conditions in absolute power (t(20)=19.34, p=2.043e-14, d=0.97; Figure 1A), relative power (%FTP; t(20)=42.35, p<2.2e-16, d=7.97; Figure 1B), heart rate (HR; t(13)=20.98, p=2.073e-11, d=3.45; Figure 1C), and rating of perceived exertion (RPE; t(20)=17.17, p=1.942e-13, d=4.61; Figure 1D). This confirms that the exercise intensity protocols (55% FTP at LI exercise and 100% at HI exercise) elicited a different power output, relative power, physiological demand (HR), and perceived demand (RPE) and, overall, successful implementation of the exercise intervention.

Figure 1—figure supplement 1. — (A) Absolute power (Watts), (B) Relative power (%FTP), (C) heart rate in beats per minute (bpm), and (D) rating of perceived exertion (RPE; BORG scale) during LI (green) and HI (purple) cycling were all significantly different. p-values were calculated using paired t-tests (two-tailed, (absolute/relative) power: N=21, heart rate: N=14, BORG rating: N=21). n.s.=not significant, *p<0.05, **p<0.01, ***p<0.001. Error bars represent standard error of the mean (SEM).

Greater pain relief after high-intensity exercise at higher stimulus intensities

In the current study, participants provided online ratings throughout the stimulus duration. We calculated the maximum pain ratings (in the following referred to as pain ratings) across the whole stimulus duration (0–17 s), which closely correspond to the pain ratings provided in the previous study (Nold et al., 2025), where participants were asked to rate the most painful sensation after the stimulus concluded. To statistically test the effect of exercise intensity (HI and LI) on pain ratings, we calculated separate Linear Mixed Effect models (LMER) with the dependent variables overall pain ratings, pressure pain ratings, and heat pain ratings, respectively. All models included exercise intensity or the interaction of exercise intensity and stimulus intensity, as well as trial and block as fixed effects, and subject as a random effect.

There was no main effect of exercise intensity on overall pain ratings (β=–3.29, SE=2.15, t(1133.64)=–1.53, p=0.14) as well as no main effect of exercise intensity on pressure pain ratings (β=–3.05, SE=2.15, t(570.11)=–1.42, p=0.16; BF₁₀=0.46; Figure 2A, top) or heat pain ratings (β=–4.13, SE=2.93, t(538.69)=–1.41, p=0.16; Figure 2A, bottom). Bayesian model comparison showed that the model, including exercise intensity was not better supported than the null model for the model across modalities (BF₁₀=0.62), as well as in pressure pain (BF₁₀=0.64) or heat pain (BF₁₀=0.78) only. This indicates only anecdotal evidence in favour of the null model, suggesting that the data were largely insensitive to distinguish between models with and without exercise intensity.

Figure 2. — (A) Pressure (top; p=0.16) and heat pain ratings (bottom; p=0.16) for LI (green) and HI (purple) exercise. (B) A significant interaction between exercise intensity and stimulus intensity was observed in heat (p=0.001) but not pressure pain (p=0.33). Post-hoc t-tests revealed that this interaction in heat pain was driven by the highest stimulus intensity (Visual Analogue Scale, VAS 70; p=0.0002). (C) Differences between high-intensity (HI) and low-intensity (LI) exercise (LI–HI) for pressure (top) and heat (bottom) pain ratings visualised at each stimulus intensity. Dots depict subject-specific pain ratings averaged across trials. p-values were calculated using post-hoc t-tests for the respective Linear Mixed Effect models (LMER) models. Error bars depict the SEM (pressure: N=21, heat: N=20).

Figure 2—figure supplement 1. — (A) Pressure (top; p=0.16) and heat pain ratings (bottom; p=0.16) for LI (green) and HI (purple) exercise. (B) A significant interaction between exercise intensity and stimulus intensity was observed in heat (p=0.001) but not pressure pain (p=0.33). Post-hoc t-tests revealed that this interaction in heat pain was driven by the highest stimulus intensity (Visual Analogue Scale, VAS 70; p=0.0002). (C) Differences between high-intensity (HI) and low-intensity (LI) exercise (LI–HI) for pressure (top) and heat (bottom) pain ratings visualised at each stimulus intensity. Dots depict subject-specific pain ratings averaged across trials. p-values were calculated using post-hoc t-tests for the respective Linear Mixed Effect models (LMER) models. Error bars depict the SEM (pressure: N=21, heat: N=20).

However, an interaction of exercise intensity and stimulus intensity on overall pain ratings was evident (β=–0.22, SE=0.10, t(1119.88)=–2.22, p=0.03). This interaction was driven by the interaction of exercise intensity and stimulus intensity during heat pain (β=–0.36, SE=0.11, t(529.95)=–3.32, p=0.001; Figure 2B and C). Post–hoc t-tests revealed that the effect was especially evident at the highest stimulus intensity (p=0.0002; Figure 2B and C). For pressure pain, this interaction of exercise intensity and stimulus intensity was not significant (β=–0.09, SE=0.09, t(564.99)=–0.97, p=0.33; Figure 2B and C) but a similar trend was evident when considering the post-hoc t-tests (Figure 2B and C), where pain relief following HI compared to LI was greater at the highest stimulus intensity (p=0.05). Bayesian model comparisons indicated evidence against an interaction between exercise intensity and stimulus intensity across modalities. For the combined dataset, the interaction model was less supported than the main-effects model (BF_₁₀=0.22), indicating that the data were approximately 4.5 times more likely under the additive model. This pattern was consistent for both heat (BF_₁₀=0.33) and pressure pain (BF_₁₀=0.19), suggesting moderate evidence in favour of additive effects over interactive modulation. The full model outputs and post-hoc t-tests can be found in Supplementary file 1, table S1–S10. For a comprehensive overview, the online ratings at each stimulus intensity (Figure 2—figure supplement 1) and averaged across stimulus intensities (Figure 2—figure supplement 2) following HI and LI exercise for heat and pressure pain separately are also visualised.

Sample differences in exercise-induced hypoalgesia

We compared the weight-corrected FTP values of the current sample with those of the previous sample (Nold et al., 2025) (N=39, 21 females) for males and females separately using two-sample t-tests. The females employed in the current study (Figure 3A; dark blue) had a significantly higher FTP compared to the females of the previous study (light blue) (t(40)=–5.00, p=1.23×10^–5; Figure 3A). There was no significant difference in the FTP between the females in this study (dark blue) and males in the previous study (red) (t(36)=1.65, p=0.11; Figure 3A). The self-reported training volume was significantly higher in the females of the current study (dark blue) compared to the females (light blue) (t(22)=3.45, p=0.003, Figure 3B) and males (red) (t(24)=2.82, p=0.009; Figure 3B) of the previous study. Furthermore, analyses comparing the females of the previous study and the females of the current study concerning height, weight, and BMI revealed no significant differences but a significant age difference (t(32.98)=3.27, p=0.003) (Figure 3—figure supplement 1). Overall, this suggests that this current study included females of overall higher fitness levels compared to the previous study. In the following, females from the current study will be referred to as females with higher fitness levels due to the significantly higher FTP values than females from the previous study, as well as significantly higher training volume compared to the males and females from the previous study. These labels should be interpreted relative to the samples included in the current and previous studies rather than as absolute classifications.

Figure 3. — (A) Weight-corrected functional threshold power (FTP Watts divided by kilograms) differed significantly between the females in the current study and the females in the previous study (p=1.23×10^–5) but not between the females in the current study and males in the previous study (p=0.11). (B) Training volume (hours per week) differed significantly between the females in the current study and the females (p=0.003) and males (p=0.009) of the previous study. n.s.=not significant, *p<0.05, **p<0.01, ***p<0.001.

Figure 3—figure supplement 1. — (A) Weight-corrected functional threshold power (FTP Watts divided by kilograms) differed significantly between the females in the current study and the females in the previous study (p=1.23×10^–5) but not between the females in the current study and males in the previous study (p=0.11). (B) Training volume (hours per week) differed significantly between the females in the current study and the females (p=0.003) and males (p=0.009) of the previous study. n.s.=not significant, *p<0.05, **p<0.01, ***p<0.001.

Consistent with the previous study (Nold et al., 2025), the effects in the current study were more prominent in heat pain. Therefore, all subsequent analyses for heat pain are reported in the main manuscript and for pressure pain in the supplements (Figure 4—figure supplement 1, Figure 5—figure supplement 1). To allow for better comparability between the previous and current study, we rescaled the maximum pain ratings of the current study to a common scale of 0–100 to correspond with the previous study.

To determine whether the effect of exercise intensity differs between the participant groups contributing to the pooled dataset, we examined the interaction of exercise intensity and subgroup (males from the previous study, females from the previous study and females from the current study). Frequentist LMERs were fitted across all stimulus intensities (VAS 30, 50, 70) and separately for VAS 70 only, as previous analyses suggested the exercise effect to be most pronounced at the highest stimulus intensity. To complement these analyses, we performed Bayesian model comparisons to quantify the relative evidence for models, including exercise intensities and subgroup effects compared with simpler alternative models.

For the model across all stimulus intensities, there was no significant interaction of exercise intensity and subgroup (β=–1.63, SE=1.68, t(1897)=–0.97, p=0.33; Figure 4A), with overall low absolute mean difference pain ratings between LI and HI exercise (Figure 4B; males (previous): –0.54; females (previous): –1.74; females (current): 2.26). However, as the previous analyses revealed the effect to be most pronounced at the highest stimulus intensity, the model for VAS 70 heat pain revealed a significant interaction of exercise intensity and subgroup (β=–5.96, SE=1.53, t(562)=–3.89, p<0.001; Figure 4C), where the females of the current study showed a reduced pain perception after high compared to low intensity exercise (absolute mean difference: 6.70; Figure 4D) compared to the males (absolute mean difference: –4.71; Figure 4D) and females of the previous study (absolute mean difference: –1.11; Figure 4D). The full LMER model outputs and post-hoc t-tests can be found in the supplements (Supplementary file 1, tables S11–S18).

Figure 4. — (A) Heat pain ratings averaged across all stimulus intensities (Visual Analogue Scale VAS 30, 50, and 70) in each subgroup for low-intensity (LI) (green) and high-intensity (HI) (purple) exercise. There was no significant interaction of exercise intensity and subgroup on heat pain ratings (p=0.33). (B) Differences between HI and LI exercise (LI–HI) of pain ratings across all stimulus intensities are visualised for each subgroup. (C) Heat pain ratings at VAS 70 in each subgroup for LI (green) and HI (purple) exercise. There was a significant interaction of exercise intensity and subgroup on heat pain ratings (p<0.001). (D) Differences between HI and LI exercise (LI–HI) of pain ratings at VAS 70 are visualised for each subgroup. Dots depict subject-specific pain ratings averaged across trials and blocks. p-values were calculated using post-hoc t-tests for the respective Linear Mixed Effect (LMER) models. Error bars depict the SEM (N=59).

Figure 4—figure supplement 1. — (A) Heat pain ratings averaged across all stimulus intensities (Visual Analogue Scale VAS 30, 50, and 70) in each subgroup for low-intensity (LI) (green) and high-intensity (HI) (purple) exercise. There was no significant interaction of exercise intensity and subgroup on heat pain ratings (p=0.33). (B) Differences between HI and LI exercise (LI–HI) of pain ratings across all stimulus intensities are visualised for each subgroup. (C) Heat pain ratings at VAS 70 in each subgroup for LI (green) and HI (purple) exercise. There was a significant interaction of exercise intensity and subgroup on heat pain ratings (p<0.001). (D) Differences between HI and LI exercise (LI–HI) of pain ratings at VAS 70 are visualised for each subgroup. Dots depict subject-specific pain ratings averaged across trials and blocks. p-values were calculated using post-hoc t-tests for the respective Linear Mixed Effect (LMER) models. Error bars depict the SEM (N=59).

Bayes model comparison for the models across all stimulus intensities indicated evidence against a model, including exercise intensity alone relative to the null model (BF₁₀=0.24). However, adding subgroup information substantially improved the model fit relative to the exercise intensity only model (BF₁₀=1239) as well as relative to the null model (BF₁₀=321), indicating strong evidence for an interaction between exercise intensity and subgroup. Bayes model comparison for the models, including only heat pain ratings at VAS 70 mirrored these results, with evidence against the model, including exercise intensity alone relative to the null model (BF₁₀=0.22) but a substantially improved model fit when including the subgroup relative to the exercise intensity alone model (BF₁₀=4497) and the null model (BF₁₀=1038).

Fitness level, but not sex, affects exercise-induced pain modulation in the pooled sample

To investigate the effects of fitness level (indicated by weight-corrected FTP) and sex on exercise-induced pain modulation, we pooled participants across studies and first conducted frequentist linear regression analyses corresponding to those used in the previous study. These analyses used participant-level difference scores (pain following low-intensity exercise minus pain following high-intensity exercise) as the outcome variable. We then performed Bayesian model comparison and Bayesian regression analyses on the pooled heat pain ratings to estimate the effects of exercise intensity, fitness, sex, and their interactions. Unlike the frequentist analyses, which focused on difference pain scores, the Bayesian models estimated posterior distributions of the regression parameters directly, yielding posterior mean estimates, 95% credible intervals (CrIs), and probabilities of effect direction without reliance on null-hypothesis significance testing (Kruschke, 2021).

The linear model, including fitness level, indicated a significant effect of fitness level on difference pain scores (β=4.99, SE=2.05, t(56)=2.43, p=0.02; Figure 5A) corresponding to a correlation of r=0.28, p=0.03. The model, including fitness level and sex, also indicated a significant interaction (β=11.12, SE=4.69, t(56)=2.37, p=0.02; Figure 5B) with a correlation of r=0.57, p=0.01 for males from the previous study (Figure 5B, red line): and a non-significant correlation for females pooled across the previous study and current study (Figure 5B, pink line) of r=0.17, p=0.28. The model, including fitness level and subgroup also revealed a significant interaction (β=–7.99, SE=2.76, t(56)=–2.89, p=0.005; Figure 5C), with weak and non-significant correlations for females from the previous (r=0.08, p=0.72; Figure 5C, light blue line) and current study (r=–0.15, p=0.54; Figure 5C, dark blue line). The full LMER model outputs can be found in the supplements (Supplementary file 1, tables S19–S21). Corresponding analyses using training volume as a measure of fitness level have been included in the supplements (Figure 5—figure supplement 2).

Figure 5. — (A) Subject-specific differences in heat pain ratings (dots) between low-intensity (LI) and high-intensity (HI) exercise conditions (LI–HI exercise pain ratings) and corresponding regression line pooled across all stimulus intensities. Fitness level (measured as functional threshold power; FTP) showed a significant main effect on difference pain ratings (p=0.02) and a positive relation to heat pain ratings (r=0.28, p=0.03). (B) A significant interaction of fitness level×sex on difference pain ratings (p=0.02) with a significant relation in males (red line; r=0.57, p=0.01) but not in females (pink line; pooled across current and previous study; r=0.17, p=0.28). (C) Significant interaction between fitness level×subgroup (p=0.005) with non-significant correlations for females from the previous (r=0.08, p=0.72; light blue line) and current study (r=–0.15, p=0.54; dark blue line). (D) Posterior density distribution, (E) posterior mean estimates and 95% Credible Intervals (CrI) and (F) posterior probability for the effect of exercise intensity (β=1.50, 95% CrI [–5.88, 8.76], *Pr(β<0)*=0.34), fitness level moderation (β=–2.41, 95% CrI [–5.79, 1.00], *Pr(β<0)*=0.92) and sex moderation (β=0.13, 95% CrI [–4.91, 5.06], *Pr(β<0)*=0.48) on heat pain ratings. The dashed vertical line indicates the null effect (β=0), corresponding to no association between the predictor and pain ratings.

Figure 5—figure supplement 1. — (A) Subject-specific differences in heat pain ratings (dots) between low-intensity (LI) and high-intensity (HI) exercise conditions (LI–HI exercise pain ratings) and corresponding regression line pooled across all stimulus intensities. Fitness level (measured as functional threshold power; FTP) showed a significant main effect on difference pain ratings (p=0.02) and a positive relation to heat pain ratings (r=0.28, p=0.03). (B) A significant interaction of fitness level×sex on difference pain ratings (p=0.02) with a significant relation in males (red line; r=0.57, p=0.01) but not in females (pink line; pooled across current and previous study; r=0.17, p=0.28). (C) Significant interaction between fitness level×subgroup (p=0.005) with non-significant correlations for females from the previous (r=0.08, p=0.72; light blue line) and current study (r=–0.15, p=0.54; dark blue line). (D) Posterior density distribution, (E) posterior mean estimates and 95% Credible Intervals (CrI) and (F) posterior probability for the effect of exercise intensity (β=1.50, 95% CrI [–5.88, 8.76], *Pr(β<0)*=0.34), fitness level moderation (β=–2.41, 95% CrI [–5.79, 1.00], *Pr(β<0)*=0.92) and sex moderation (β=0.13, 95% CrI [–4.91, 5.06], *Pr(β<0)*=0.48) on heat pain ratings. The dashed vertical line indicates the null effect (β=0), corresponding to no association between the predictor and pain ratings.

For the Bayesian hierarchical models, model comparison indicated evidence against a model containing exercise intensity only relative to the null model (BF₁₀=0.24). Adding fitness (BF₁₀=1.61) or sex (BF₁₀=1.48) provided only anecdotal improvements over the model, including exercise intensity only. The inclusion of study (‘previous,’ ‘current’) substantially improved the model fit (BF₁₀=2454). Finally, the full model was strongly favoured over both the exercise intensity-only model (BF₁₀=644) and the null model (BF₁₀=158).

For the exercise intensity effect, posterior mean estimates indicated substantial uncertainty regarding the influence on heat pain ratings (β=1.50, 95% CrI [–5.88, 8.76]; Figure 5D and E) and a posterior probability of a negative effect of Pr(β<0)=0.34 (Figure 5F). The posterior distribution for the interaction of exercise intensity ×fitness level was predominantly negative (β=–2.41, 95% CrI [–5.79, 1.00]; Figure 5D and E) with a posterior probability of 92% (Pr(β<0)=0.92) of a negative effect (Figure 5F). Although the credible interval included zero, the posterior distribution suggested a probable negative moderation effect of fitness on the relationship between exercise intensity and fitness level on heat pain ratings. The tendency toward a negative exercise intensity ×fitness level interaction is compatible with the frequentist analyses showing that higher fitness was associated with greater exercise-induced analgesia (Figure 5A), although the two analyses address different model parameters and should not be interpreted as directly equivalent. Finally, the posterior estimates for the exercise intensity ×sex interaction provided no meaningful evidence for an effect (β=0.13, 95% CrI [–4.91, 5.06]; Figure 5D and E), with a posterior probability of Pr(β<0)=0.48 (Figure 5F), indicating equal support for positive and negative effects.

Discussion

In this study, we investigated exercise-induced pain modulation between high-intensity (HI) and low-intensity (LI) exercise in a sample of females (N=21) of relatively higher fitness levels. In a previous study by our group (Nold et al., 2025), exploratory analyses suggested a three-way interaction of fitness levels, sex, and drug treatment, where males showed hypoalgesia after HI compared to LI exercise with increasing fitness levels, which was diminished when naloxone was administered. These effects were not evident in females. In this current study, we aimed to explore whether exercise-induced pain modulation is also evident in females of higher fitness levels and whether this is moderated by fitness level or sex. We found that, for heat pain, females in the current study showed pain relief after HI compared to LI exercise, especially for highly painful stimuli. A similar trend was observed for pressure pain. To place these findings in the context of our earlier observations, we pooled data from the current and previous studies and examined the influence of fitness and sex on exercise-induced pain modulation. Across the combined sample, both frequentist and Bayesian analyses indicated that fitness was associated with exercise-induced hypoalgesia, whereas there was little evidence for sex-dependent effects. Together, these findings provide support for a role of fitness level in explaining variability in exercise-induced pain modulation, while offering little evidence that exercise-induced pain modulation differs systematically between males and females.

Effect of exercise and stimulus intensity on pain in the current sample

Corresponding to our previous study (Nold et al., 2025), there was no overall effect of exercise intensity on heat or pressure pain in the current study. However, we identified an interaction of stimulus intensity and exercise intensity that was driven by heat pain, with greater pain relief after HI compared to LI exercise at high stimulus intensity (VAS 70). A similar trend was evident in pressure pain, but it did not reach significance. These findings align with the results of our previous study (Nold et al., 2025), where exercise-induced pain modulation after HI compared to LI exercise was more prominent in heat than pressure pain.

Pooled analyses of the current and previous study sample (Nold et al., 2025)

We confirmed that the females of the current sample showed significantly higher fitness level (as measured in weight-corrected FTP) than the females of the previous study (Nold et al., 2025). Furthermore, the absolute training volume was significantly higher compared to the males and females of the previous study, suggesting overall higher fitness levels of the females of the current study. Interestingly, although the training volume of the females of the current study was significantly higher compared to the males of the previous study, the FTP did not significantly differ, which suggests the FTP is a more sensitive measure of fitness level compared to self-reported training volume. Previous research has shown that the FTP and FTP20 tests are reliable and convenient methods to estimate approximate measures of VO_2max (Denham et al., 2020) and that the FTP test is a useful test for performance prediction in moderately trained cyclists (Sørensen et al., 2019). Furthermore, as pain is a dynamic experience, we utilised real-time pain ratings in the current study as opposed to static post-stimulus pain ratings (Nold et al., 2025) to better capture the temporal dynamics of pain. Previous research (Koyama et al., 2004) has shown that, in heat pain, real-time ratings account for R²=0.89 of the variability of post-stimulus rating, where much of the variability for the post-stimulus intensity pain rating is accounted for by the peak response during the online ratings (adj. R²=0.3–0.4), suggesting good comparability of both pain measures. Since results were more pronounced in heat pain in the previous and current study, our analyses in the main manuscript focused on heat pain, but results for pressure pain are reported in the supplements.

Evidence for fitness level, but not sex, effects on exercise-induced hypoalgesia

To identify whether exercise intensity affects pain ratings differently in the samples, we analysed the interaction of exercise intensity and subgroup (males from the previous study, females from the previous study, and females from the current study). We found that females from the current study showed a significant difference between HI and LI exercise heat pain ratings at the highest pain intensity, but this was not evident in the males or females from the previous sample. To further evaluate the moderating roles of fitness level and sex in exercise-induced pain modulation, we examined these factors across the pooled sample. Both frequentist and Bayesian analyses converged in suggesting that fitness level is associated with variability in exercise-induced pain modulation, whereas evidence for sex-dependent effects was limited. Specifically, individuals with higher fitness levels tended to show greater hypoalgesia following HI compared with LI exercise, while sex did not meaningfully contribute to explaining differences in exercise-induced pain modulation. These findings extend our previous exploratory observations and suggest that fitness level may be a more important determinant of exercise-induced pain modulation than sex. Given the remaining uncertainty surrounding the magnitude of the fitness-related effects, some uncertainty regarding the magnitude of this effect remains.

A multitude of findings on exercise-induced hypoalgesia have been derived from earlier studies with male-dominated samples with higher fitness levels (Geisler et al., 2019; Haier et al., 1981; Janal et al., 1984; Scheef et al., 2012), and results on sex dependent effects in exercise-induced hypoalgesia are still equivocal (Rice et al., 2019). Generally, (repeated) aerobic exercise has been shown to be associated with the activation of endogenous neuromodulatory systems (Gajic et al., 2026), such as endogenous opioid (Koltyn et al., 2014; Saanijoki et al., 2022), endocannabinoid (Matei et al., 2023), serotonergic, and noradrenergic pathways (Anderson and Shivakumar, 2013; Beserra et al., 2018). Concerning the serotonergic and noradrenergic pathways, repeated aerobic exercise induces regulatory adaptations in glucocorticoid receptors in the hippocampus, prefrontal cortex, and thalamus that potentially contribute to more efficient hypothalamic-pituitary-adrenal (HPA) axis regulation and reduced cortisol reactivity (Beserra et al., 2018). Likewise, positron emission tomography (PET) studies suggest that a higher aerobic fitness level is associated with reduced MOR binding in cortical (ACC, insula) and subcortical (ventral striatum) areas that are part of the descending pain modulatory system (DPMS) as well, suggesting that trained individuals show greater endogenous opioid release after aerobic exercise (Saanijoki et al., 2022). Overall, repeated aerobic exercise, resulting in a higher fitness level, might contribute to shaping (i.e. by increasing receptor density) or more strongly engaging already existing endogenous neuromodulatory systems and, thus, altering pain perception (Gajic et al., 2026; Song et al., 2022).

Concerning the mechanisms of sex potentially influencing exercise-induced pain modulation, the literature remains scarce and rather inconclusive. Since exercise has been linked to the release of endogenous opioids (Haier et al., 1981; Janal et al., 1984) and, more recently, endocannabinoids (Crombie et al., 2018; Siebers et al., 2023; Siebers et al., 2021), sex differences in these neuromodulatory systems might also contribute to the different expressions of exercise-induced pain modulation between the sexes. In terms of endogenous opioid signalling, previous studies indicate that men show larger mu-opioid receptor (MOR) activation in pain-related regions than women during sustained pain (Zubieta et al., 2002). Furthermore, sex differences in opioid analgesia remain controversial (Fullerton et al., 2018), pointing towards sex differences in the descending antinociceptive circuit. Specifically in the PAG, male rats showed higher MOR expression and binding compared with female rats (Loyd et al., 2008). Interestingly, increasing evidence in rodents also points towards sex differences in the expression and function of the endocannabinoid system in pain (Blanton et al., 2021), such as lower CB1 receptor density in the prefrontal cortex and amygdala (Castelli et al., 2014) or lower 2-AG concentration in the PAG (Levine et al., 2021) in female than male rodents. These regions are critical parts of the DPMS (Bingel and Tracey, 2008; Borras et al., 2004; Eippert et al., 2009; Petrovic et al., 2002) and are also involved in mediating exercise-induced pain modulation (Geisler et al., 2019; Nold et al., 2025; Scheef et al., 2012).

Of those studies that have included both sexes in their samples (Naugle et al., 2014; Niwa et al., 2022), a dose-response relationship of aerobic exercise intensity and the analgesic response could be shown in the overall sample, where high-intensity (70% heart rate reserve [HRR]) aerobic exercise produced larger hypoalgesic effects compared to moderate-intensity (50% HRR) exercise (Naugle et al., 2014; Niwa et al., 2022) and low-intensity (30% HRR) exercise (Niwa et al., 2022). Importantly, despite including both sexes, these studies did not include sex as a factor in their analyses, potentially confounding the observed effects (Tesarz et al., 2012). Of those studies that have explicitly accounted for sex as a moderating variable in exercise-induced hypoalgesia, some studies could show a hypoalgesic response from exercise in both sexes (Koltyn et al., 2014; Koltyn et al., 2001), whereas others (Gajsar et al., 2017; Lemley et al., 2016) found a stronger hypoalgesic response in females. When administering an aerobic exercise protocol, one study found the hypoalgesic response following exercise to be more pronounced in females compared to males (Vaegter et al., 2014).

Despite some studies showing comparable or greater hypoalgesic effects in females, these studies rarely account for fitness levels or athletic status (Tesarz et al., 2012). Generally, it has been shown that athletes exhibit increased pain tolerance and decreased pain intensity compared to non-athletic controls (Geva and Defrin, 2013; Tesarz et al., 2012). Among the few studies that have explicitly considered both fitness levels and sex in the hypoalgesic response to exercise, one study (Sternberg et al., 2001) found that treadmill running induced hypoalgesia in the cold pressor test in females, regardless of athletic status, but not in males. Notably, both male and female athletes showed a significant decrease in forearm withdrawal latency from a radiant heat device after treadmill exercise (Sternberg et al., 2001). Further corroborating this, female athletes experienced more pronounced hypoalgesia following aerobic exercise, especially during cold pressor and thermal pain tests (Smith, 2004). For isometric exercise, one study (Black et al., 2017) found that females demonstrated analgesic responses regardless of their physical activity levels, which is in contrast to our previous findings (Nold et al., 2025), where females did not show an analgesic response after aerobic exercise. Thus, the extent of sex differences might also vary depending on the type of exercise, the pain stimulus administered, and the measurement method employed (Tesarz et al., 2012), emphasising the complexity of sex-specific pain modulation mechanisms. Overall, female athletes tend to report pain similarly to their male counterparts (Tesarz et al., 2012), suggesting that athletic training may mitigate typical sex differences in the hypoalgesic response after exercise and potentially reduce disparities observed in the general population. Our results advance this by showing that fitness level, but not sex, moderates the hypoalgesic response to heat pain, even after short bouts of aerobic exercise.

Limitations

On the experimental day, heat stimuli with a target intensity of VAS 30 were rated below the pain threshold. The anticipated intensities of VAS 50 and VAS 70 were rated above the pain threshold, suggesting that they were perceived as painful. However, the lower pain ratings on the experimental day as opposed to the calibration day are likely due to habituation effects between days (Ellerbrock et al., 2015; May et al., 2012). Furthermore, despite assessing exercise-related expectations and identifying participants to expect decreased whole-body pain following exercise, we did not experimentally manipulate those expectations, precluding causal conclusions regarding their relative contribution to the observed effects. Although expectations can modulate pain perception (Colloca et al., 2018; Jones et al., 2017), positive exercise-related expectations alone do not appear to augment exercise-induced hypoalgesia (Vaegter et al., 2020), suggesting that exercise-related expectations do not account for our findings. Moreover, as we compared HI and LI exercise but no sedentary rest condition, we cannot draw conclusions on the overall effect of exercise as opposed to no exercise on pain. Thus, it could be possible that even LI exercise suffices to evoke a hypoalgesic response (Niwa et al., 2022) or that individuals with higher fitness levels can show a distinction in the analgesic response based on the exercise intensity. However, previous research has suggested that HI exercise produces greater hypoalgesia compared to LI exercise (Jones et al., 2019; Naugle et al., 2014; Niwa et al., 2022; Smith, 2004; Vaegter et al., 2014). Therefore, we chose the LI exercise as the control condition instead of rest to contribute to the understanding of the dose-response relationship between exercise intensities and hypoalgesic effects. Future research should investigate whether LI exercise suffices to induce an analgesic response.

Conclusion

In this study, we found that participants with relatively higher fitness levels (as measured by weight-corrected FTP) experience greater pain relief following high-intensity compared to low-intensity exercise, especially at high pain intensities. Together with our previous findings, these results suggest that fitness level may play a more central role than sex in explaining variability in exercise-induced pain modulation.

Methods

Participants

Overall, N=22 female participants were recruited for this study from local cycling clubs and by (online) advertisements. Participants were required to be aged between 18 and 50 years and have a body mass index (BMI) ranging from 18 to 30, and to regularly participate in structured cycling training sessions per week. One participant was excluded due to a technical failure on the experimental day, resulting in incomplete data. For one participant, heat pain calibration failed, resulting in N=20 participants for heat pain data and N=21 participants for pressure pain data (for participant characteristics, see Table 1). It is crucial to note that the N=21 females included in the current study did not participate in the previous study (Nold et al., 2025).

Table 1. Participant characteristics from the previous (Nold et al., 2025) and current study.

	Previous study (Nold et al., 2025)			Current study
	Overall (M±SD)	Females (M±SD)	Males (M±SD)	Females (M±SD)
N	39	21	18	21
Age (years)	26.03 (4.83)	25.33 (5.10)	26.83 (4.35)	32.29 (7.64)
Weight (kg)	70.95 (12.14)	63.33 (7.53)	79.83 (10.38)	63.6 (8.09)
Height (cm)	177.10 (9.08)	170.52 (6.35)	184.78 (4.58)	170 (5.95)
BMI	22.51 (2.64)	21.80 (2.59)	23.33 (2.53)	21 (1.93)
FTP (Watts/kg)	1.79 (0.48)	1.58 (0.44)	2.03 (0.40)	2.25 (0.41)
Training Volume (h/w)	4.44 (3.24)	3.88 (2.63)	5.08 (3.81)	13.10 (12.28)

Open in a new tab

M=Mean; SD=Standard deviation; kg=kilogram; cm=centimetre; FTP=Functional threshold power. h/w=hours per week.

Experimental paradigm

The experimental paradigm was similar to that of the original study (Nold et al., 2025), using the same study setup and equipment (CPAR and TSA 2 and Wahoo KICKR bike). More precisely, participants completed four cycling blocks in total, two HI and two LI exercise blocks, presented in a pseudo-randomised order. Furthermore, the same calibration procedures were used to calibrate heat and pressure pain as well as the functional threshold power using the FTP₂₀ test (Allen and Coggan, 2012; Allen and Coggan, 2006) as implemented and validated in previous studies (Borszcz et al., 2018; McGrath et al., 2019; Nold et al., 2025). However, participants in the current study did not receive a pharmacological intervention and did not undergo fMRI scans; therefore, they received the painful stimuli outside the MR scanner whilst lying in a supine position. To monitor the painful stimuli more closely, participants provided online (continuous) ratings whilst the pain stimulus was ongoing on a VAS scale ranging from ‘no sensation’ (0) to ‘almost unbearably painful’ (150), with VAS 50 marking ‘minimally painful’ (pain threshold). Furthermore, we administered the same questionnaires as described in the previous study. For a detailed description of the study procedures and experimental paradigm, please refer to Nold et al., 2025. Overall, the current study consisted of one calibration day (Day 1) and one experimental day (Day 2).

Behavioural data acquisition

During cycling, heart rate (HR), maintained power (in watts), and cadence (in revolutions per minute [RPM]) were acquired continuously. In N=7 participants, HR data were not recorded due to a technical failure, but were monitored throughout the cycling. The rating of perceived exertion (RPE) on the BORG scale (Borg, 1998), ranging from ‘no exertion’ (6) to ‘maximal exertion’ (20), was provided after each cycling block. Furthermore, the elapsed time between each cycling and pain within each block was 5 min to correspond with the mean time required in the previous study. Following the cycling, participants received a total of 18 pain stimuli, with 9 heat and 9 pressure stimuli applied in an alternating fashion and with their respective intensities (30, 50, 70 VAS) applied in a randomised order. Whilst the pain was ongoing, participants were asked to rate their currently perceived pain/non-painful sensation using the left and right buttons of a button box (Logitech). When their perception did not change, the cursor should remain at the same position. The online rating time exceeded the stimulus length by 2 s to capture changes in perception when the stimulus ramped down. The online ratings were sampled at 70–75 samples per second and interpolated at 0.9 s.

Statistical analyses

The behavioural statistical analyses were performed in MATLAB and RStudio (Version 2021.09.1). We used the lmer and emmeans functions from the lme4 package (Version 1.1–35.1) to conduct linear mixed-effects (LMER) (Bates et al., 2014; Hox and Moerbeek, 2010) and post-hoc t-tests in R, respectively. Furthermore, the rstatix package (Kassambara, 2019) was used to conduct one- and two-sample t-tests. For the Bayesian approach, all analyses were performed in R using the package brms and rstan that implements Bayesian multilevel regression models via Hamiltonian Monte Carlo sampling in Stan (Stan Development Team, 2024).

To investigate whether the distribution of menstrual cycle phases as well as hormonal contraceptives differed within and between days, χ2-tests were calculated. Furthermore, we evaluated whether there was a significant effect of expectation about acute aerobic exercise on different pain dimensions and calculated one-sample t-tests (two-tailed). Furthermore, we conducted paired-samples t-tests comparing pre- and post-mood ratings on the dimensions of dejection, fatigue, discontent, and drive. To compare the FTP, self-reported fitness levels, training volume, age, weight, height, and BMI between the previous sample (males and females) and the current sample, we calculated two-sample t-tests. To evaluate the exercise parameters within the current study, watts and HR were averaged across time and blocks for each participant and for both exercise intensities. The RPE was also averaged across blocks for participants and exercise intensity. We calculated one-sample t-tests comparing the respective measures between HI and LI exercise.

Linear models

For visualisation purposes, we displayed the online ratings (averaged across subjects, blocks, trials, and stimulus intensities) and the SEM following HI and LI exercise for heat and pressure pain separately (Figure 2—figure supplement 1). For a more comprehensive understanding, the averaged online ratings at each stimulus intensity were visualised in Figure 2—figure supplement 2. To statistically evaluate the online pain ratings and improve consistency with the post-stimulus pain ratings reported in the previous study (Nold et al., 2025), we calculated the maximum rating by averaging the online rating across the whole stimulus duration (0–17 s) and extracting the peak rating. The peak pain ratings across the whole stimulus duration would most closely correspond to the post-stimulus pain ratings provided in the previous study, where participants were asked to rate the most painful sensation throughout the stimulus duration. We calculated separate LMER models with the dependent variables overall pain ratings, heat pain ratings, and pressure pain ratings, respectively. All LMER models included trial and block as fixed effects and subject as a random effect. In a first LMER model, we included exercise intensity as a fixed effect; in a second LMER model, the stimulus intensity and exercise intensity, as well as their interaction, served as fixed effects. Finally, we conducted post-hoc t-tests and applied a Tukey adjustment to correct for multiple comparisons.

To gain a better understanding of the overall effect of exercise intensity, fitness level, and sex across the current and previous study (Nold et al., 2025), we pooled the datasets. From the previous study, only data from the control (saline) condition were included in the current analyses. To allow for comparability of the previous and current study, as well as the subgroups (males of the previous study, females of the previous study, and females of the current study), we rescaled the maximum pain ratings of the current study to a common scale of 0–100 to correspond with the previous study. To investigate the effect of exercise intensity and subgroup on pain ratings across all stimulus intensities, as well as at VAS 70 only, we conducted an LMER model. The interaction of exercise intensity and subgroup was used to assess whether the magnitude of exercise-induced pain modulation differed between the subgroups. To investigate the effect of fitness level, sex, and subgroup on exercise-induced pain modulation, we calculated participant-level difference scores (pain ratings following low-intensity exercise minus pain following high-intensity exercise), such that larger values reflected greater hypoalgesia following high-intensity exercise. Fitness level, fitness level×sex, and fitness level×subgroup were, respectively, included in three different linear models as predictors. The treatment order was included as a fixed effect in each model. Furthermore, we calculated the correlation coefficient using Pearson’s product-moment correlation for each of the respective models. These analyses correspond to the analyses conducted in Nold et al., 2025.

Bayesian analyses

We conducted Bayesian analyses to complement and extend frequentist analyses by quantifying evidence for competing hierarchical models and estimating posterior distributions of regression coefficients (β) in the pooled samples model. All analyses modelled pain ratings as the dependent variable using Gaussian linear mixed-effects models. To account for repeated measurements within participants, all models included subject–specific intercepts. Furthermore, trial and block were included as fixed effects in all models. Models were organised in a hierarchical comparison framework consisting of a null model (including only nuisance covariates), main effects models (exercise intensity/VAS intensity) and interaction models (interaction terms). Overall, weakly informative priors were specified for all models (Table 2). Prior specifications were identical across models, with the only difference in intercepts depending on the rating scale used (current study: 0–150; pooled dataset: 0–100) (for details, see Table 2). All models were estimated using four Markov chain Monte Carlo (MCMC) chains with 4000 iterations per chain (2000 warm-up). For more complex interaction models, adapt_delta was increased (up to 0.995), and maximum tree depth was increased (up to 15) to ensure stable sampling. Convergence was assessed using the potential scale reduction statistic (Ȓ ≈ 1.00), trace plots, and effective sample sizes.

Table 2. Prior specification for Bayesian mixed-effects models.

Parameter type	Parameter	Prior distribution	Rationale
Fixed effects (main models)	Regression coefficients (β)	Normal (0, 8)	Weakly informative, allowing broad plausible effects
Fixed effects (interaction models)	Regression coefficients (β)	Normal (0, 5)	Stronger shrinkage to reduce overfitting in higher-order terms
Intercept (current study)/ Intercept (pooled dataset)	Intercept	Normal (75,30)/ Normal (50, 20)	Midpoint matches the observed scale of outcomes in each dataset
Random effects SD	Subject-level SD	Student-t (3, 0, 10)	Weakly informative, allows heterogeneity across participants
Residual SD	σ	Student-t (3, 0, 10)	Weakly informative, constrains extreme variance

Open in a new tab

β=regression coefficient; SD=standard deviation; σ=residual standard deviation. Normal (μ, σ) indicates a normal prior with mean μ and standard deviation σ; Student-t (ν, μ, σ) indicates a Student-t prior with degrees of freedom (ν), location (μ), and scale (σ). Subject-level SD represents between-participant variability captured by random effects. Priors were weakly informative unless otherwise noted.

Bayesian model comparison

Bayesian model comparison was conducted using the Bayes Factor (BF₁₀) estimated via bridge sampling for all frequentist linear model analyses. The Bayes Factor quantifies the relative evidence provided by the data for one model over another by comparing their marginal likelihoods. Specifically, BF₁₀ expresses the evidence for the more complex model relative to a simpler model. Generally, BF₁₀>1 indicates support for the more complex model over the simpler model and vice versa for BF₁₀<1. According to the conventional evidence thresholds (Lovric, 2025), a 1<BF₁₀<3 indicates anecdotal/weak evidence, 3<BF₁₀<10 indicates moderate evidence, and BF₁₀>10 indicates strong evidence in favour of the complex model. BF₁₀ close to 1 indicates insensitive or inconclusive evidence for the complex over the simpler model. We computed the Bayes factor for the main effects model vs the null model, the interaction model vs. main effect model, and the full model vs. null model.

Posterior inference

For the final analysis investigating the effect of exercise intensity, fitness level, and sex in the pooled dataset, posterior inference was conducted. The model included pain ratings as the dependent variable with exercise intensity, fitness level, and sex, as well as subgroup as fixed effects, along with their interactions with exercise intensity. Again, trial and block were included as fixed effects along with subject-specific intercepts. Importantly, this model did not rely on the difference in pain ratings as in the frequentist analyses but on raw pain ratings. Therefore, this should be treated not as confirmatory but as complementary analyses. All posterior summaries were based on four MCMC chains with satisfactory convergence (Ȓ ≈ 1.00) and sufficient effective sample sizes across parameters. Unlike frequentist statistics, Bayesian inference does not rely on point estimates and p-values but rather a full posterior distribution of each model parameter, allowing for direct probability statements about effect size and direction of effects. Posterior summaries focused on the fixed-effect regression coefficients (β), specifically: (i) the main effect of exercise intensity, (ii) its moderation by fitness level, (iii) sex, and (iv) subgroup (males from the previous study, females from the previous study, females from the current study). Posterior samples were extracted and reported from the fitted model and used to compute and visualise in three complementary ways. First, the posterior point estimates were computed as the posterior mean of each fixed-effect regression coefficient (β) representing the expected regression coefficient under the posterior distribution. Second, uncertainty was summarised using 95% credible intervals, defined as 2.5 and 97.5% quantiles of the posterior distribution. These intervals indicate the range containing 95% of the posterior probability for the regression coefficient. Third, the posterior probabilities of direction were computed for the fixed effect regression coefficients (β) by calculating the proportion of posterior samples below (Pr(β<0)) for the main effect of exercise intensity, and its interaction terms with fitness level, sex, and subgroup. The posterior probabilities provide a direct probability statement about the direction of each effect.

Code availability

This study was programmed using MATLAB 2021b and Psychophysics Toolbox (Version 3.0.19). For data visualisation, we used RStudio (Version 2026.05.0). The custom behavioural analysis pipelines are available on the public repository https://github.com/jannenold/hotspin_analyses_behavioural, copy archived at Nold, 2026 after publication.

Acknowledgements

We thank the participants for taking part in the study. Furthermore, we thank Alexandra Tinnermann for her helpful comments. C.B. and J.I.N. are supported by ERC-AdG-883892-PainPersist. C.B. is supported by DFG SFB 289 Project A02 (Project-ID 422744262–TRR 289).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Janne Ina Nold, Email: j.nold@uke.de.

José Biurrun Manresa, National Scientific and Technical Research Council (CONICET), National University of Entre Ríos (UNER), Argentina.

Timothy E Behrens, University of Oxford, United Kingdom.

Funding Information

This paper was supported by the following grants:

European Research Council 10.3030/883892 to Janne Ina Nold, Christian Büchel.
Deutsche Forschungsgemeinschaft SFB 289 Project A02 422744262-TRR 289 to Christian Büchel.

Additional information

Competing interests

No competing interests declared.

Senior editor, eLife.

Author contributions

Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing.

Data curation, Validation, Investigation, Writing – review and editing.

Data curation, Investigation, Writing – review and editing.

Conceptualization, Resources, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing – original draft, Writing – review and editing.

Ethics

All participants gave informed written consent. The study was approved by the Ethics Board of the Hamburg Medical Association (PV7456/2020-10144-BO-ff). We support inclusive, diverse, and equitable conduct of research.

Additional files

MDAR checklist

elife-109061-mdarchecklist1.docx^{(89KB, docx)}

Supplementary file 1. Supplementary tables S1–S30.

elife-109061-supp1.docx^{(48.2KB, docx)}

Data availability

The dataset for the raw behavioural data generated in the current study is available on the public repository https://github.com/jannenold/hotspin_analyses_behavioural, copy archived at Nold, 2026. All necessary data to evaluate the results of the study are included in the manuscript and supplementary materials.

References

Allen H, Coggan A. Racing and Training with a Power Meter. VeloPress; 2006. [Google Scholar]
Allen H, Coggan A. Training and Racing with a Power Meter. VeloPress; 2012. [Google Scholar]
Anderson EH, Shivakumar G. Effects of exercise and physical activity on anxiety. Frontiers in Psychiatry. 2013;4:27. doi: 10.3389/fpsyt.2013.00027. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. Journal of Statistical Software. 2014;67:1–48. doi: 10.18637/jss.v067.i01. [DOI] [Google Scholar]
Beserra AHN, Kameda P, Deslandes AC, Schuch FB, Laks J, de MH. Can physical exercise modulate cortisol level in subjects with depression? A systematic review and meta-analysis. Trends in Psychiatry and Psychotherapy. 2018;40:360–368. doi: 10.1590/2237-6089-2017-0155. [DOI] [PubMed] [Google Scholar]
Bingel U, Tracey I. Imaging CNS modulation of pain in humans. Physiology. 2008;23:371–380. doi: 10.1152/physiol.00024.2008. [DOI] [PubMed] [Google Scholar]
Black CD, Huber JK, Ellingson LD, Ade CJ, Taylor EL, Griffeth EM, Janzen NR, Sutterfield SL. Exercise-induced hypoalgesia is not influenced by physical activity type and amount. Medicine & Science in Sports & Exercise. 2017;49:975–982. doi: 10.1249/MSS.0000000000001186. [DOI] [PubMed] [Google Scholar]
Blanton HL, Barnes RC, McHann MC, Bilbrey JA, Wilkerson JL, Guindon J. Sex differences and the endocannabinoid system in pain. Pharmacology, Biochemistry, and Behavior. 2021;202:173107. doi: 10.1016/j.pbb.2021.173107. [DOI] [PMC free article] [PubMed] [Google Scholar]
Borg G. Borg’s Perceived Exertion and Pain Scales. Human Kinetics; 1998. [Google Scholar]
Borras MC, Becerra L, Ploghaus A, Gostic JM, DaSilva A, Gonzalez RG, Borsook D. fMRI measurement of CNS responses to naloxone infusion and subsequent mild noxious thermal stimuli in healthy volunteers. Journal of Neurophysiology. 2004;91:2723–2733. doi: 10.1152/jn.00249.2003. [DOI] [PubMed] [Google Scholar]
Borszcz FK, Tramontin AF, Bossi AH, Carminatti LJ, Costa VP. Functional threshold power in cyclists: validity of the concept and physiological responses. International Journal of Sports Medicine. 2018;39:737–742. doi: 10.1055/s-0044-101546. [DOI] [PubMed] [Google Scholar]
Brellenthin AG, Crombie KM, Cook DB, Sehgal N, Koltyn KF. Psychosocial influences on exercise-induced hypoalgesia. Pain Medicine. 2017;18:538–550. doi: 10.1093/pm/pnw275. [DOI] [PMC free article] [PubMed] [Google Scholar]
Castelli MP, Fadda P, Casu A, Spano MS, Casti A, Fratta W, Fattore L. Male and female rats differ in brain cannabinoid CB1 receptor density and function and in behavioural traits predisposing to drug addiction: effect of ovarian hormones. Current Pharmaceutical Design. 2014;20:2100–2113. doi: 10.2174/13816128113199990430. [DOI] [PubMed] [Google Scholar]
Colloca L, Corsi N, Fiorio M. The interplay of exercise, placebo and nocebo effects on experimental pain. Scientific Reports. 2018;8:14758. doi: 10.1038/s41598-018-32974-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
Crombie KM, Brellenthin AG, Hillard CJ, Koltyn KF. Endocannabinoid and opioid system interactions in exercise-induced hypoalgesia. Pain Medicine. 2018;19:118–123. doi: 10.1093/pm/pnx058. [DOI] [PMC free article] [PubMed] [Google Scholar]
Curran SL, Andrykowski MA, Studts JL. Short form of the profile of mood states (POMS-SF): psychometric information. Psychological Assessment. 1995;7:80–83. doi: 10.1037/1040-3590.7.1.80. [DOI] [Google Scholar]
de Bruijn ST, van Wijck AJM, Geenen R, Snijders TJ, van der Meulen WJTM, Jacobs JWG, Veldhuijzen DS. Relevance of physical fitness levels and exercise-related beliefs for self-reported and experimental pain in fibromyalgia: an explorative study. Journal of Clinical Rheumatology. 2011;17:295–301. doi: 10.1097/RHU.0b013e31822c5196. [DOI] [PubMed] [Google Scholar]
Denham J, Scott-Hamilton J, Hagstrom AD, Gray AJ. Cycling power outputs predict functional threshold power and maximum oxygen uptake. Journal of Strength and Conditioning Research. 2020;34:3489–3497. doi: 10.1519/JSC.0000000000002253. [DOI] [PubMed] [Google Scholar]
Eippert F, Bingel U, Schoell ED, Yacubian J, Klinger R, Lorenz J, Büchel C. Activation of the opioidergic descending pain control system underlies placebo analgesia. Neuron. 2009;63:533–543. doi: 10.1016/j.neuron.2009.07.014. [DOI] [PubMed] [Google Scholar]
Ellerbrock I, Wiehler A, Arndt M, May A. Nocebo context modulates long-term habituation to heat pain and influences functional connectivity of the operculum. PAIN. 2015;156:2222–2233. doi: 10.1097/j.pain.0000000000000297. [DOI] [PubMed] [Google Scholar]
Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley JL. Sex, gender, and pain: a review of recent clinical and experimental findings. The Journal of Pain. 2009;10:447–485. doi: 10.1016/j.jpain.2008.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Fullerton EF, Doyle HH, Murphy AZ. Impact of sex on pain and opioid analgesia: a review. Current Opinion in Behavioral Sciences. 2018;23:183–190. doi: 10.1016/j.cobeha.2018.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gajic P, Kovac I, Lubinsky G, Knezevic NN. Exercise-induced hypoalgesia: cellular and molecular mechanisms linking pain modulation and stress regulation-a narrative review. Cells. 2026;15:858. doi: 10.3390/cells15100858. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gajsar H, Titze C, Hasenbring MI, Vaegter HB. Isometric back exercise has different effect on pressure pain thresholds in healthy men and women. Pain Medicine. 2017;18:917–923. doi: 10.1093/pm/pnw176. [DOI] [PubMed] [Google Scholar]
Geisler M, Eichelkraut L, Miltner WHR, Weiss T. An fMRI study on runner’s high and exercise-induced hypoalgesia after a 2-h-run in trained non-elite male athletes. Sport Sciences for Health. 2019;16:159–167. doi: 10.1007/s11332-019-00592-8. [DOI] [Google Scholar]
Geva N, Defrin R. Enhanced pain modulation among triathletes: a possible explanation for their exceptional capabilities. PAIN. 2013;154:2317–2323. doi: 10.1016/j.pain.2013.06.031. [DOI] [PubMed] [Google Scholar]
Haier RJ, Quaid K, Mills JSC. Naloxone alters pain perception after jogging. Psychiatry Research. 1981;5:231–232. doi: 10.1016/0165-1781(81)90052-4. [DOI] [PubMed] [Google Scholar]
Hox JJ, Moerbeek M. Multilevel Analysis: Techniques and Applications. Routledge; 2010. [DOI] [Google Scholar]
Janal MN, Colt EWD, Clark CW, Glusman M. Pain sensitivity, mood and plasma endocrine levels in man following long-distance running: effects of naloxone. Pain. 1984;19:13–25. doi: 10.1016/0304-3959(84)90061-7. [DOI] [PubMed] [Google Scholar]
Jones MD, Valenzuela T, Booth J, Taylor JL, Barry BK. Explicit education about exercise-induced hypoalgesia influences pain responses to acute exercise in healthy adults: a randomized controlled trial. The Journal of Pain. 2017;18:1409–1416. doi: 10.1016/j.jpain.2017.07.006. [DOI] [PubMed] [Google Scholar]
Jones MD, Nuzzo JL, Taylor JL, Barry BK. Aerobic exercise reduces pressure more than heat pain sensitivity in healthy adults. Pain Medicine. 2019;20:1534–1546. doi: 10.1093/pm/pny289. [DOI] [PubMed] [Google Scholar]
Kassambara A. Rstatix: pipe-friendly framework for basic statistical tests. Version 1.1.0Rstatix. 2019 doi: 10.32614/CRAN.package.rstatix. [DOI]
Koltyn KF. Analgesia following exercise: a review. Sports Medicine. 2000;29:85–98. doi: 10.2165/00007256-200029020-00002. [DOI] [PubMed] [Google Scholar]
Koltyn KF, Trine MR, Stegner AJ, Tobar DA. Effect of isometric exercise on pain perception and blood pressure in men and women. Medicine and Science in Sports and Exercise. 2001;33:282–290. doi: 10.1097/00005768-200102000-00018. [DOI] [PubMed] [Google Scholar]
Koltyn KF, Brellenthin AG, Cook DB, Sehgal N, Hillard C. Mechanisms of exercise-induced hypoalgesia. The Journal of Pain. 2014;15:1294–1304. doi: 10.1016/j.jpain.2014.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Koyama Y, Koyama T, Kroncke AP, Coghill RC. Effects of stimulus duration on heat induced pain: the relationship between real-time and post-stimulus pain ratings. PAIN. 2004;107:256–266. doi: 10.1016/j.pain.2003.11.007. [DOI] [PubMed] [Google Scholar]
Kruschke JK. Bayesian analysis reporting guidelines. Nature Human Behaviour. 2021;5:1282–1291. doi: 10.1038/s41562-021-01177-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lemley KJ, Senefeld J, Hunter SK, Hoeger Bement M. Only women report increase in pain threshold following fatiguing contractions of the upper extremity. European Journal of Applied Physiology. 2016;116:1379–1385. doi: 10.1007/s00421-016-3389-8. [DOI] [PubMed] [Google Scholar]
Levine A, Liktor-Busa E, Lipinski AA, Couture S, Balasubramanian S, Aicher SA, Langlais PR, Vanderah TW, Largent-Milnes TM. Sex differences in the expression of the endocannabinoid system within V1M cortex and PAG of Sprague Dawley rats. Biology of Sex Differences. 2021;12:60. doi: 10.1186/s13293-021-00402-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lindheimer JB, Szabo A, Raglin JS, Beedie C, Carmichael KE, O’Connor PJ. Reconceptualizing the measurement of expectations to better understand placebo and nocebo effects in psychological responses to exercise. European Journal of Sport Science. 2020;20:338–346. doi: 10.1080/17461391.2019.1674926. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lovric M. International Encyclopedia of Statistical Science. Springer; 2025. [DOI] [Google Scholar]
Loyd DR, Wang X, Murphy AZ. Sex differences in micro-opioid receptor expression in the rat midbrain periaqueductal gray are essential for eliciting sex differences in morphine analgesia. The Journal of Neuroscience. 2008;28:14007–14017. doi: 10.1523/JNEUROSCI.4123-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Matei D, Trofin D, Iordan DA, Onu I, Condurache I, Ionite C, Buculei I. The endocannabinoid system and physical exercise. International Journal of Molecular Sciences. 2023;24:1989. doi: 10.3390/ijms24031989. [DOI] [PMC free article] [PubMed] [Google Scholar]
May A, Rodriguez-Raecke R, Schulte A, Ihle K, Breimhorst M, Birklein F, Jürgens TP. Within-session sensitization and between-session habituation: a robust physiological response to repetitive painful heat stimulation. European Journal of Pain. 2012;16:401–409. doi: 10.1002/j.1532-2149.2011.00023.x. [DOI] [PubMed] [Google Scholar]
McGrath E, Mahony N, Fleming N, Donne B. Is the FTP test a reliable, reproducible and functional assessment tool in highly-trained athletes? International Journal of Exercise Science. 2019;12:1334–1345. doi: 10.70252/RQOO7391. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mogil JS. Sex differences in pain and pain inhibition: multiple explanations of a controversial phenomenon. Nature Reviews. Neuroscience. 2012;13:859–866. doi: 10.1038/nrn3360. [DOI] [PubMed] [Google Scholar]
Naugle KM, Naugle KE, Fillingim RB, Samuels B, Riley JLI. Intensity thresholds for aerobic exercise-induced hypoalgesia. Medicine & Science in Sports & Exercise. 2014;46:817–825. doi: 10.1249/MSS.0000000000000143. [DOI] [PMC free article] [PubMed] [Google Scholar]
Naugle KM, Riley JL. Self-reported physical activity predicts pain inhibitory and facilitatory function. Medicine and Science in Sports and Exercise. 2014;46:622–629. doi: 10.1249/MSS.0b013e3182a69cf1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Niwa Y, Shimo K, Ohga S, Tokiwa Y, Hattori T, Matsubara T. Effects of exercise-induced hypoalgesia at different aerobic exercise intensities in healthy young adults. Journal of Pain Research. 2022;15:3615–3624. doi: 10.2147/JPR.S384306. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nold JI, Fadai T, Büchel C. Acute aerobic exercise intensity does not modulate pain potentially due to differences in fitness levels and sex effects: results from a pharmacological fMRI study. eLife. 2025;14:RP102392. doi: 10.7554/eLife.102392. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nold JI. Hotspin_analyses_behavioural. swh:1:rev:e087127ab84c076d9e9623d7e591b3356951da09Software Heritage. 2026 https://archive.softwareheritage.org/swh:1:dir:d851ab73fdf2024d15f28e6e6afc6fc225822d4c;origin=https://github.com/jannenold/hotspin_analyses_behavioural;visit=swh:1:snp:dfd24da3e76835f2ddacc8f8e009db67896fff61;anchor=swh:1:rev:e087127ab84c076d9e9623d7e591b3356951da09
Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia-- imaging a shared neuronal network. Science. 2002;295:1737–1740. doi: 10.1126/science.1067176. [DOI] [PubMed] [Google Scholar]
Rice D, Nijs J, Kosek E, Wideman T, Hasenbring MI, Koltyn K, Graven-Nielsen T, Polli A. Exercise-induced hypoalgesia in pain-free and chronic pain populations: state of the art and future directions. The Journal of Pain. 2019;20:1249–1266. doi: 10.1016/j.jpain.2019.03.005. [DOI] [PubMed] [Google Scholar]
Saanijoki T, Kantonen T, Pekkarinen L, Kalliokoski K, Hirvonen J, Malén T, Tuominen L, Tuulari JJ, Arponen E, Nuutila P, Nummenmaa L. Aerobic fitness is associated with cerebral μ-opioid receptor activation in healthy humans. Medicine & Science in Sports & Exercise. 2022;54:1076–1084. doi: 10.1249/MSS.0000000000002895. [DOI] [PubMed] [Google Scholar]
Scheef L, Jankowski J, Daamen M, Weyer G, Klingenberg M, Renner J, Mueckter S, Schürmann B, Musshoff F, Wagner M, Schild HH, Zimmer A, Boecker H. An fMRI study on the acute effects of exercise on pain processing in trained athletes. Pain. 2012;153:1702–1714. doi: 10.1016/j.pain.2012.05.008. [DOI] [PubMed] [Google Scholar]
Schmitt A, Wallat D, Stangier C, Martin JA, Schlesinger-Irsch U, Boecker H. Effects of fitness level and exercise intensity on pain and mood responses. European Journal of Pain. 2020;24:568–579. doi: 10.1002/ejp.1508. [DOI] [PubMed] [Google Scholar]
Siebers M, Biedermann SV, Bindila L, Lutz B, Fuss J. Exercise-induced euphoria and anxiolysis do not depend on endogenous opioids in humans. Psychoneuroendocrinology. 2021;126:105173. doi: 10.1016/j.psyneuen.2021.105173. [DOI] [PubMed] [Google Scholar]
Siebers M, Biedermann SV, Fuss J. Do endocannabinoids cause the runner’s high? evidence and open questions. The Neuroscientist. 2023;29:352–369. doi: 10.1177/10738584211069981. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sluka KA, Frey-Law L, Hoeger Bement M. Exercise-induced pain and analgesia? Underlying mechanisms and clinical translation. Pain. 2018;159:S91–S97. doi: 10.1097/j.pain.0000000000001235. [DOI] [PMC free article] [PubMed] [Google Scholar]
Smith LD. Haverford College; 2004. The effects of competition and exercise on pain perception. [Google Scholar]
Song JS, Yamada Y, Kataoka R, Wong V, Spitz RW, Bell ZW, Loenneke JP. Training-induced hypoalgesia and its potential underlying mechanisms. Neuroscience & Biobehavioral Reviews. 2022;141:104858. doi: 10.1016/j.neubiorev.2022.104858. [DOI] [PubMed] [Google Scholar]
Sørensen A, Aune TK, Rangul V, Dalen T. The validity of functional threshold power and maximal oxygen uptake for cycling performance in moderately trained cyclists. Sports. 2019;7:217. doi: 10.3390/sports7100217. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stan Development Team RStan: the R interface to Stan. Version 2.32.7Stan. 2024 https://mc-stan.org/
Sternberg WF, Bokat C, Kass L, Alboyadjian A, Gracely RH. Sex-dependent components of the analgesia produced by athletic competition. The Journal of Pain. 2001;2:65–74. doi: 10.1054/jpai.2001.18236. [DOI] [PubMed] [Google Scholar]
Tesarz J, Schuster AK, Hartmann M, Gerhardt A, Eich W. Pain perception in athletes compared to normally active controls: a systematic review with meta-analysis. Pain. 2012;153:1253–1262. doi: 10.1016/j.pain.2012.03.005. [DOI] [PubMed] [Google Scholar]
Vaegter HB, Handberg G, Graven-Nielsen T. Similarities between exercise-induced hypoalgesia and conditioned pain modulation in humans. Pain. 2014;155:158–167. doi: 10.1016/j.pain.2013.09.023. [DOI] [PubMed] [Google Scholar]
Vaegter HB, Thinggaard P, Madsen CH, Hasenbring M, Thorlund JB. Power of words: influence of preexercise information on hypoalgesia after exercise-randomized controlled trial. Medicine and Science in Sports and Exercise. 2020;52:2373–2379. doi: 10.1249/MSS.0000000000002396. [DOI] [PubMed] [Google Scholar]
Zubieta JK, Smith YR, Bueller JA, Xu Y, Kilbourn MR, Jewett DM, Meyer CR, Koeppe RA, Stohler CS. mu-opioid receptor-mediated antinociceptive responses differ in men and women. The Journal of Neuroscience. 2002;22:5100–5107. doi: 10.1523/JNEUROSCI.22-12-05100.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.109061.3.sa0

eLife Assessment

José Biurrun Manresa ¹

This study proposes that fitness level, but not sex, influences exercise-induced hypoalgesia in women. However, the evidence to support this claim is incomplete: the conclusions rely on a small-effect interaction that emerges only under specific conditions, the findings are inconsistent across pain modalities and stimulus intensities, and the absence of a baseline condition limits the interpretability of results as reflecting true hypoalgesia. Additionally, the methods by which fitness level was categorized across cohorts can be questioned. The results of this study could be useful to improve future experimental designs exploring these factors in more depth.

eLife. doi: 10.7554/eLife.109061.3.sa1

Reviewer #2 (Public review):

Anonymous

I appreciate the substantial amount of work the authors have invested in revising the manuscript. The response is extensive, and the addition of pooled frequentist and Bayesian analyses represents a considerable and technically sophisticated effort to address the concerns raised during review. The authors have also improved the use and presentation of the continuous pain-rating data and have responded constructively to several methodological points.

Nevertheless, with respect to my original review, I do not think the revisions have yet aligned the manuscript's central claims with the specificity and uncertainty of the evidence. The additional analyses provide some evidence consistent with a contribution of fitness to heat-pain modulation, but they cannot fully resolve the limited internal replication, selected sample or differences in measurement procedures between studies. Nor do they establish that sex has no effect on pain modulation. In my view, the evidence supports a modest, exploratory conclusion that fitness may contribute to variability in heat-pain modulation under specific conditions, rather than the more categorical conclusion that exercise-induced pain modulation depends on fitness but not sex. The manuscript would therefore be strengthened by reflecting this distinction more consistently in the title, abstract, Discussion and Conclusion. My remaining concerns, organised in relation to the original review, are detailed below.

(1) The core interpretation may not be fully supported by the data

(1.1) Lack of baseline condition

I am satisfied with the authors' substantive response to Comment 1.1, however I note that the authors are still attributing their findings to "hypoalgesia" in the Discussion.

(1.2) Lack of internal replication across conditions

The pooled analyses do not resolve the original lack of internal replication. The substantive effect remains largely restricted to heat pain at the highest stimulus intensity and is not clearly reproduced across stimulus intensities or for pressure pain. The authors' response describes the evidence more cautiously, stating that fitness "contributes to variability in exercise-induced pain modulation". However, the title and principal conclusions do not reflect the exploratory, modality-specific and statistically uncertain nature of the evidence.

(1.3) Deviations from the original study and selective use of data

I appreciate the addition of analyses using the continuous pain ratings, which addresses my concern that the richness of the data from the current study was not being fully used. However, the decision to pool the current and previous datasets introduces a related and potentially more consequential comparability problem. The previous study used retrospective post-stimulus ratings, whereas the current study used continuous ratings from which peak values were extracted. Rescaling these measures to a common numerical range does not establish their measurement equivalence. This is particularly important because study membership is closely confounded with the participant characteristics central to the manuscript: the higher-fitness female group was recruited and assessed in the current study, whereas the comparison groups were assessed using the earlier procedure. Consequently, apparent effects of fitness may partly reflect differences in outcome acquisition between studies. Indeed, including study membership improves model fit substantially more than including fitness.

(1.4) Over-reliance on null results regarding sex differences

I do not think the new Bayesian analyses resolve my original concern regarding the interpretation of the null sex effect. The authors state that the analyses provide no evidence for a moderating role of sex, but they do not report a Bayes factor clearly supporting the null hypothesis for the exercise intensity × sex interaction. In fact, Figure 5 clearly shows that the association between fitness and pain modulation is significant in males but not females. Panels D-F indicate uncertainty regarding a differently specified exercise intensity × sex interaction, rather than evidence for its absence. The figure therefore suggests that the role of sex remains unresolved and may qualify the association attributed to fitness. The authors should therefore distinguish absence of evidence from evidence of absence and temper the conclusion that exercise-induced pain modulation depends on fitness "but not sex".

(2) Limited sample and lack of diversity

The manuscript does not acknowledge the limited generalisability of this small, highly selected sample of physically active women recruited primarily from cycling clubs. Although the rebuttal states that the findings should be interpreted within the context of the sample studied, this qualification is absent from the Limitations section. It should be added, and the conclusions restricted accordingly.

(3) Methodological choices limit interpretability of the data

(3.1) The use of a non-standard VAS scale (0-150 with a fixed pain threshold at 50) is unconventional and may influence how participants report pain, while limiting comparability with related literature.

The authors' response explains the rationale for the modified scale but does not fully address its implications for measurement comparability. In particular, setting all ratings below 50 to zero is not merely rescaling: it collapses distinct responses into a single value and may alter the distribution of the outcome. The cited study addresses the relationship between real-time and retrospective ratings, but does not validate this scale or transformation.

eLife. 2026 Oct 30;15:RP109061. doi: 10.7554/eLife.109061.3.sa2

Author response

Janne Ina Nold ¹, Tahmine Fadai ², Zora Pia Gerbers ³, Christian Büchel ⁴

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The current study is a follow-up to a previously published study by the same research group (Nold et al. 2025). In the previous study, the authors had included a set of exploratory analyses which assessed the effects of fitness level (denominated by a relative FTP), sex, and drug treatment (Naxolone versus placebo). In this previous study, the authors state that "exploratory analysis showed a significant main effect of fitness level on differences in pain ratings in the [saline] condition... suggesting increased hypoalgesia with increasing fitness levels, pooled across all stimulus intensities".

In the current study, the authors have recruited an additional 22 female participants (21 included in analysis) from local cycling clubs to assess if fitness level does indeed impact exercise-induced hypoalgesia responses to experimental thermal and pressure pain models.

Strengths:

The current study has the potential to present a convincing argument about the effect of fitness level and potentially other factors (e.g., sex) on exercise-induced hypoalgesia responses. Combining data across two of their primary studies would be highly fruitful to the research community interested in this area. Specifically, it has the potential to inform sports medicine practitioners and how they administer exercise protocols to help those experiencing pain with a further consideration for the fitness level (and maybe sex) of their patients.

Weaknesses:

However, the current study makes several bold claims about the role of fitness level and sex on exercise-induced hypoalgesia, which I do not believe that this study on its own - or in conjunction with the previously published study by the same authors - can make at present. Namely, the current study does not appear to conduct any specific analyses between the cohorts from either study (current and present). The results mention a difference in the group mean values in "fitness level" between cohorts, but the analysis itself on pain responses/exercise-induced hypoalgesia is limited only to the cohort from the current study. If the authors wanted to provide a convincing argument that fitness level has an effect on exercise-induced hypoalgesia, then the analysis of this study would have to include an analysis between the groups considered to be of "high" and "low" fitness level. I do not think the current study does this. Instead, it makes an assumption from the previous study (Nold et al. 2025) which only states that "exploratory analysis showed a significant main effect of fitness level on differences in pain ratings in the [saline] condition... suggesting increased hypoalgesia with increasing fitness levels, pooled across all stimulus intensities". The analysis of this study would have to include fitness level "high fitness" versus "low fitness" of participants across both studies in its statistical model to properly discern if fitness level has an impact on exercise-induced hypoalgesia.

A similar comment can be made with respect to sex differences, as these have not been assessed in the analysis of this study either.

Another area of weakness in this study is how "fitness level" has been demarcated across participants. One issue is how authors have assumed that the current cohort is 'fit', whereas the previous cohort was 'less fit', meaning that the authors could be coming to false conclusions about fitness level. In detail, figures within the current study show a large overlap between the 'fit' and 'less fit' cohorts, where some participants have a higher relative functional threshold power (FTP) in the 'less fit' cohort than the 'fit' cohort and vice versa. Therefore, I believe the authors should better demarcate between those that are in the 'more fit' and 'less fit' groups according to a validated and well-established criterion from the kinesiology and sport science literature. That being said, I think this may be problematic in some ways as FTP is considered a relatively poor measure to denote fitness levels, a limitation highlighted in the previous study's review.

Altogether, whilst I commend the researchers on their body of work across the two studies, the current methods and analysis provide an incomplete assessment of their primary research question, and therefore, I would urge the authors to reconsider some of their methods/analysis and the framing of their results to better reflect the main research question they have attempted to answer. Likewise, I would recommend that readers ensure they consider the current results with caution until the authors have addressed some areas of concern which currently limit their main conclusions.

We sincerely thank the reviewer for their thorough and insightful comments and for the time invested in reviewing our work. In response, we have made several important revisions, particularly to address the major concerns regarding fitness level and sex differences.

Specifically, we have:

(1) Conducted additional pooled analyses to directly evaluate fitness and sex effects.

We pooled the data from the previous and current studies and conducted linear mixed-effects models investigating the effect of exercise intensity between the subgroups, as well as specifically testing the effect of fitness level and sex across the pooled data. This allowed us to directly test the effects of fitness level and sex on a common scale rather than infer these effects from separate cohorts. To quantify evidence both for and against the presence of exercise-induced pain modulation and its moderation by fitness, sex, and study sample, we estimated Bayes factors comparing nested models of increasing complexity. In addition, we examined the posterior distributions of the regression coefficients (β) and subgroup- specific (males from previous study, females from previous study, females from current study) exercise effects to estimate the magnitude and uncertainty of these effects. The Bayesian analyses provided some evidence that fitness level moderated exercise-induced pain modulation, but sex did not. Importantly, these findings were consistent with the overall pattern of results obtained in the frequentist analyses. All corresponding statistics in the main manuscript and Supplementary Materials have been updated accordingly. To improve transparency, the pooled analyses are now presented in the dedicated Results subsections and accompanied by new figures (Figures 4 and 5). Furthermore, we have adapted the Title, Discussion and Conclusion to more closely reflect the results.

(2) Clarified the assessment of fitness level.

We expanded our consideration of fitness level and training status by discussing training volume as an additional indicator of training status. We also acknowledge that fitness is a continuous construct and that some overlap in FTP values exists between cohorts. To address this concern, we conducted pooled analyses using FTP as a continuous predictor rather than relying solely on cohort membership. For consistency and comparability with the previous study, FTP remained our primary indicator of fitness level, while analyses using training volume are now reported in the Supplementary Materials.

(3) Expanded the mechanistic background on the effect of fitness level and sex on EIH.

We expanded the mechanistic background in both the Introduction and Discussion to provide a clearer theoretical rationale for how fitness level and sex may influence exercise-induced hypoalgesia (EIH), drawing on relevant experimental and clinical literature.

Reviewer #2 (Public review):

This study addresses an important question regarding exercise-induced modulation of pain in women, but the conclusions appear to be based on relatively limited and selective evidence. The authors report an interaction between exercise intensity and stimulus intensity, which they interpret as evidence for exercise-induced hypoalgesia and conclude that fitness, but not sex, modulates this effect. However, this main result relies on a relatively small interaction that emerges only under specific conditions, with inconsistent findings across pain modalities and stimulus intensities, and an analysis approach that does not fully exploit the continuous pain ratings collected. The lack of a baseline condition further limits the interpretability of the findings as reflecting hypoalgesia, and overall, the data provide a rather constrained basis for drawing broader conclusions.

Strengths:

(1) The focus on women is important and timely, particularly given the ambiguity in prior findings and the historical bias toward male-dominated samples.

(2) The attempt to revisit previous findings in a new cohort is valuable in principle.

Weaknesses:

(1) The core interpretation may not be fully supported by the data

The central claim-that the results demonstrate exercise-induced hypoalgesia and its dependence on fitness but not sex-does not appear to be fully supported by the evidence presented.

(1.1) Lack of baseline condition

The absence of a no-exercise baseline substantially limits interpretation. The study compares high- and low-intensity exercise, but without a baseline, it is not possible to determine whether either condition produces hypoalgesia or hyperalgesia relative to calibration. The observed HI-LI difference, therefore, reflects only a relative contrast between exercise intensities, not an absolute reduction in pain. As a result, attributing the findings to "hypoalgesia" may be difficult to justify fully.

(1.2) Lack of internal replication across conditions

The reported effect is highly specific and does not clearly generalise across the experimental design. It emerges significantly only for heat pain at the highest stimulus intensity, with no clear effects for other intensities and for pressure pain. Moreover, the main statistical result is a relatively small interaction effect with a modest p value, which translates into a difference of approximately 6-8 VAS units on a 150 scale. This combination-a small effect size, limited statistical strength, and restriction to a single condition-substantially weakens the evidence for a robust or generalisable effect.

(1.3) Deviations from the original study and selective use of data

Although framed as a follow-up to previous work, the current study introduces substantial methodological changes, particularly in the acquisition and scaling of pain ratings (continuous vs post-hoc ratings, modified VAS with sub-threshold range). Despite collecting rich continuous data, the analysis focuses on peak responses to approximate the previous study. While this may aid comparability, it results in a strong emphasis on a single data point (highest intensity), rather than leveraging the full dataset. This limits both interpretability and comparability.

(1.4) Over-reliance on null results regarding sex differences

The conclusion that fitness, but not sex, modulates exercise-induced pain may not be directly supported by the data presented. The current study includes only highly fit women, and comparisons with men or less-fit women rely on non-significant differences in a previous cohort. The absence of a significant difference does not provide evidence for equivalence, and no formal statistical support for a null effect is provided. As such, conclusions about the absence of sex differences would unfortunately benefit from more cautious interpretation.

(2) Limited sample and lack of diversity

The dataset is narrow in scope, comprising a small sample (N = 21) of healthy, highly fit women. Key demographic characteristics (e.g. age range, BMI distribution) are not fully presented, explored or discussed. This limits generalisability and makes it difficult to draw broader conclusions about exercise-induced pain modulation in women, as the main focus of the study.

(3) Methodological choices limit the interpretability of the data

Several methodological decisions would benefit from stronger justification:

(3.1) The use of a non-standard VAS scale (0-150 with a fixed pain threshold at 50) is unconventional and may influence how participants report pain, while limiting comparability with related literature.

(3.2) Participants explicitly reported expecting exercise to reduce pain, introducing a potential confound that is not presently addressed.

(3.3) A more comprehensive use of the full time series of pain ratings would provide a stronger and more transparent basis for interpretation of the present findings.

We sincerely thank the reviewer for their time and effort invested in reviewing our manuscript and appreciate the comments. In response to the points raised by the reviewer, we have now updated our work. Specifically, we have:

(1) Pooled samples and strengthened inferential framework:

We pooled the current and previous datasets to provide a stronger basis for the primary interpretation of the findings. On this combined dataset, we conducted both frequentist linear mixed-effects models and complementary Bayesian analyses, allowing us to directly test effects of fitness level, sex, and exercise intensity within a unified framework and to quantify both effect sizes and evidence for/against key hypotheses. The Title, Results, Methods and Discussion have been adapted to more closely reflect the results.

(2) Addressed sample limitations through pooled analysis:

To mitigate limitations related to sample size and composition in the individual studies, we extended all primary inferential analyses to the pooled dataset. This approach increases statistical power, improves the stability of estimates, and enables a more reliable examination of between-group effects across fitness level and sex, thereby strengthening the generalisability of the findings.

(3) Expanded analyses of continuous pain ratings:

To address methodological concerns regarding the use of a summary pain metric, we supplemented the primary peak-based analyses with additional linear mixed-effects models on the full time-series data. Specifically, we modelled time × exercise intensity interactions for both heat and pressure pain, thereby capturing the dynamic structure of pain responses and ensuring that conclusions are supported by the richness of the continuous rating data. These analyses have been added to the Supplementary Materials.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Synopsis

The current study is a follow-up to a previously published study by the same research group (Nold et al. 2025). In the previous study, the authors had included a set of exploratory analyses which assessed the effects of fitness level (denominated by a relative FTP), sex, and drug treatment (Naxolone versus placebo).

In the current study, the authors have recruited an additional 22 female participants (21 included in analysis) from local cycling clubs to assess if fitness level impacts exercise-induced hypoalgesia responses to experimental thermal and pressure pain models.

All the analysis of this study seems to be specific to the "fit" cohort from this study. No analysis or figures seem to show differences between cohorts from the current and previous studies. For this study, the authors found that maximum pain ratings after application of a thermal stimulus and pressure pain stimulus were lower after a high-intensity exercise compared to a lower-intensity exercise. There was also an exercise intensity × stimulus intensity interaction when thermal stimulations were applied, suggesting that the exercise-induced hypoalgesia response varies across exercise intensities as the stimulus intensity increases/decreases. The same cannot be said for pressure pain according the statistical analysis outputs of this study, although inspection of Figure 3B (top row) does illustrate that at higher stimulus intensities, there is a difference between exercise intensities for maximal pain ratings, whereas there is no difference at lower stimulus intensities (30 and 50 VAS). However, observing the changes in maximum pain ratings (Figure 3C) does show similar changes across all stimulus intensities for pressure pain, unlike thermal pain. Therefore, the authors claim that there is an effect of fitness level on the exercise-induced hypoalgesia response.

Overall, the authors have completed a neat supplementary study to their previous study. However, there are several areas that I believe should be addressed within the current manuscript.

First, the authors regularly state "...the hypoalgesic response to exercise depends on fitness level but not on sex". As mentioned earlier, the results section of the previous study states that "exploratory analysis showed a significant main effect of fitness level on differences in pain ratings in the [saline] condition... suggesting increased hypoalgesia with increasing fitness levels, pooled across all stimulus intensities". In this study, the authors do not seem to have completed a direct comparison between the two study cohorts. The results for the current study only seem to discuss the effects of exercise intensity within this study cohort. If this study is to be treated as a stand-alone study, then the data that is presented should make it clear that the effect of fitness level is being assessed. Currently, the results and the figures only illustrate the effects of exercise intensity and stimulus intensity, not fitness level (or sex), which are mentioned in the title, abstract, and most discussions.

Second, on the topic of sex-related effects, as authors repeatedly state ""...the hypoalgesic response to exercise depends on fitness level but not on sex", I think it is worth noting that this statement would require an a priori assessment of sex differences on exercise-induced hypoalgesia responses. As the previous study (exploratory) and this study (no sex difference analysis) do not include this a priori assessment, I do not think they can conclude that sex is not a factor in the exercise-induced hypoalgesia response, though they can surmise that fitness level may have an influencing role if it was specifically tested in this study. If the authors wish to judge the role of sex, there ought to analysis on this specifically.

Third, whilst the authors have tested the effects of "fitness level", I urge the authors to consider how they have demarcated between fitness levels in this study and any other (validated) criterion that could be used in place of their current method to distinguish between "fit" and "less fit" participants. To explain the cohort of the current study is judged to be "fitter" than the previous study based on the group mean relative FTP. Yet, observing Figure 1, there are several participants with lower relative FTP in the "fit" cohort than the "less fit" cohort, and vice versa. I would recommend the authors to consult with some kinesiologists/sport scientists about this issue. As the central topic of this study is the effect of fitness level, this warrants careful consideration, as the current method to split the groups is problematic, especially as there is clear overlap between the two groups in the metric that the authors have used (relative FTP).

Fourth, I think it would be highly beneficial for the authors to add more discussion around the theoretical and mechanistic accounts about why fitness level (and sex) would impact exerciseinduced hypoalgesia responses. The current manuscript provides a nice account of the current literature in terms of the direction of each study's results, yet they do not explore why fitness level or sex has such effects. This would be highly valuable to a reader to set the tone of the study rationale in the introduction and provide more depth to the authors' discussion of their results.

Below I have highlighted all my comments that have been separated into MAJOR and MINOR comment sections.

I hope my comments are of use to the authors, and I would like to thank them and the journal for the opportunity to review their paper on a truly interesting and insightful topic.

We thank the reviewer for their insightful comments and appreciate the time invested in reviewing our work. In the following, we have addressed all comments raised by the reviewer and marked all changes in the manuscript.

Major Comments

(1) In my opinion, it would enhance the manuscript if the authors provided a theoretical or mechanistic account of why fitness levels or sex would impact the exercise-related hypoalgesia effects. Though some reasoning about the mechanisms of fitness level and sex is mentioned in the discussion of the previous study (Nold et al. 2025), and even if the overall research findings are mixed at present, I do not think it would be overkill to provide the rationale for why the authors do/do not predict a certain response in the current study after their analyses. On this note, the authors may argue that the manuscript content from L52-61 is an account of the previous literature. However, whilst this literature sets out the directions of previous research findings, as I read this, I do not think there is a theoretical framework to underpin why fitness level or sex would impact exercise-induced hypoalgesia (EIH). References already included in the manuscript should have some discussions around this (e.g., Sluka et al. 2018; Koltyn 2000; Koltyn et al. 2014).

We thank the reviewer for their comment. We agree that the mechanistic rationale for why fitness level and sex may influence EIH could be described in more detail. Accordingly, we have expanded the Introduction/Discussion to provide a more comprehensive overview of the potential mechanisms underlying these effects, drawing on previous EIH literature (e.g., Sluka et al., 2018; Koltyn, 2000; Koltyn et al., 2014) as well as evidence relating to changes in endogenous modulatory systems due to repeated exercise (Anderson & Shivakumar, 2013; Gajic et al., 2026; Matei et al., 2023; Saanijoki et al., 2022; Song et al., 2022) as well as sex differences in these neuromodulatory systems potentially contributing to sex differences in EIH (Blanton et al., 2021; Castelli et al., 2014; Fullerton et al., 2018; Levine et al., 2021; Loyd et al., 2008; Zubieta et al., 2002).

(2) The study sets out an aim, but I do not see any (a priori) hypotheses in the intro or methods. If no hypotheses were registered, this can be reported in the discussion and factored into the discussion about why certain results would/would not be expected.

The reviewer is correct that no formal, pre-registered hypotheses were specified for this study. As this study was designed as a follow-up to our previous study, which reported no exercise-induced pain modulation in relatively less fit females, the present investigation was primarily exploratory in nature and aimed to examine whether exercise-induced pain modulations would be observed in a sample of females with significantly higher fitness levels. We have now clarified this in the discussion and acknowledge the exploratory nature of the study when interpreting the findings.

(3) Is there a sample size justification, sensitivity analysis, or post-hoc calculation for the additional 21 participants to detect effects for fitness level and sex?

We thank the reviewer for this question. No formal a priori sample size calculation was performed for the recruitment of the additional 21 participants. The current study was designed as a follow-up to our previous work, in which female participants were, on average, slightly less fit than male participants. To better disentangle the effects of fitness and sex, we therefore aimed to recruit a comparable number of female participants with higher fitness levels. Accordingly, we targeted a sample size similar to that of the female subgroup in the previous study.

(4) I have a wider question about where there has truly been a test between fitness levels.

Definition and measurement of "fitness level".

Is there an appropriate rationale that relative FTP is an appropriate measure to distinguish between fitness levels? Understandably, this is done for consistency between the two studies, and there may be mention of the use of the FTP measure in the original study; however, it would be worth justifying the use of the relative FTP measure to categorise fitness level with previous literature to back this up.

We thank the reviewer for this very valid and important point regarding both the validity of FTP as a proxy for fitness and the distinction between fitness levels. Previous research has shown that the FTP and FTP20 tests are reliable and convenient methods to estimate approximate measures of VO2max (Denham et al., 2020) and that the FTP test is a useful test for performance prediction in moderately trained cyclists (Sørensen et al., 2019). Furthermore, we argue that FTP represents a continuous performance measure in our studies instead of a categorical variable distinguishing between fit and unfit individuals, and we have strengthened the manuscript by providing a clearer justification for its use as a field-based proxy of endurance fitness, supported by previous literature linking FTP to more direct physiological markers such as VO_2max (Denham et al., 2020). Since we have used FTP in the previous sample as a measure of fitness level, we maintained this for consistency in the current study.

We also acknowledge that, despite statistically significant group differences, there is meaningful overlap between the males of the previous study and the females of the current study and that “fitness level” should be interpreted as a continuum rather than a strict categorical separation. Accordingly, the labels “fit” and “unfit” should be interpreted relative to the samples included in the current and previous studies rather than as absolute classifications. To reflect this more accurately, we have revised the manuscript to refer to participants as having higher or lower fitness levels, which should be interpreted within the context of the pooled sample rather than as discrete fitness categories.

Classification of fitness level.

In L110-111 - The statistical analysis does show a difference between groups, but Figure 1A does show that there is a crossover with around half the sample from each study cohort. That is, the current study had ~16 participants who had a lower relative FTP than the highest relative FTP from the previous study. When removing the highest participant (outlier) in the previous study, there are still 7 participants (a third of the sample) with a lower relative FTP than participants considered of a 'lower' fitness level than the previous study.

Essentially, on the group level, there is a statistical difference in relative FTP, but there is notable overlap between the two groups.

We thank the reviewer for this observation. Our aim was not to create entirely non-overlapping fitness categories, but rather to recruit a female sample with significantly higher average fitness levels than the females included in the previous study. As shown in the revised manuscript, females from the current study had, on average, significantly higher relative FTP values than females from the previous study, whereas no significant difference was observed between females from the current study and males from the previous study. We also report training volume as an additional indicator of training status; however, because training volume was self-reported, we consider FTP to provide a more objective and sensitive measure of fitness. The partial overlap in FTP values between samples is not unexpected and, in fact, facilitates comparisons across studies by ensuring that the samples are not entirely distinct populations.

Could another distinction be made based on:

- A set value (e.g., 2.0 W/kg-1);

- Another value from previous literature, and the analysis be re-run with this classification of participants? The approach to use a set value is consistent with kinesiology literature that uses set values to demarcate fitness level for other more common, measures like VO2max (see ACSM guidelines and de Pauw et al. 2013);

- Or by dissociating between training volume and slightly tweaking the wording to "training status/volume" (although I am cognisant that this would introduce an incongruence in terminology between the authors' studies, but other studies that have been referenced [e.g., Geva and Defrin, do specifically test between training volume]).

We thank the reviewer for their suggestions. We would like to refrain from making another distinction based on a set value and dividing the group into “fit” and “unfit” individuals. Instead, we have pooled the samples and treated FTP as a continuous variable, which preserves statistical power and avoids the loss of information associated with dichotomising a continuous measure.

To address the reviewers' suggestion of including training volume as a measure of fitness level, we have repeated our new analyses, namely the LMER model using training volume as a predictor for difference pain ratings (LI – HI Exercise pain ratings), in the pooled sample. These analyses yielded no significant effects for training volume, the interaction of training volume x sex, or training volume x subgroup in either heat or pressure pain. Furthermore, the training volume is potentially influenced by outliers in the current study sample. For completeness, we have added these analyses and according visualisation to the Supplements (Fig. S6).

We believe that the FTP provides a more sensitive and objective measure of cardiovascular fitness level in our study, as it is not susceptible to recall bias, differences in training intensity and differences in training efficacy. We also note that other studies (Geva & Defrin, 2013) did not rely solely on self-reported training volume as a measure of fitness level, but recruited participants based on training history and athletic status (e.g. competitive triathletes vs. active controls).

For these reasons, we consider the FTP the most appropriate primary indicator of fitness level in the current (and previous) study, while still reporting training volume in the main manuscript to consider.

The above suggestions would also allow to select matched pairs from the male cohort when comparing effects between sexes. Altogether, though, I believe this comment to be the most important in my review of this paper, as the primary question depends strongly on how fitness level is defined and categorised by the researchers, and this requires careful consideration and validation from previous literature if such decisive conclusions are to be drawn from the results. Therefore, a clearer description of the reason why certain effects are expected (theory/mechanisms) and how participants have been categorised will be highly valuable for readers. It may be useful for the authors to reach out to kinesiologists/sport scientists within their peer groups to help with this comment.

To further address the concern regarding between-cohort comparability, we have now pooled the samples from the previous and current study and re-analysed the data fitness level as a continuous variable rather than relying on cohort-based comparisons. Specifically, we conducted regression analyses examining fitness level, sex, and their interaction, as well as subgroup analyses including males from the previous study, females from the previous study, and females from the current study. These analyses were complemented by Bayesian model comparison and posterior inference. The results are now reported in the Results section. We also acknowledge that fitness is best viewed as a continuum rather than a discrete characteristic and have revised the manuscript accordingly, replacing references to “fit” and “unfit” participants with descriptions of relatively higher and lower fitness levels within the pooled sample.

Finally, as mentioned in the previous comment (major comment #1), we have expanded the Introduction and Discussion to more clearly outline the mechanistic and theoretical rationale for expecting effects of fitness level and sex on exercise-induced pain modulation, drawing on relevant literature (see major comment #1). We believe these additions substantially strengthen both the conceptual justification and the analytical robustness of the manuscript.

(5) The abstract states that "...further investigation to determine if and to what extent exercise induced hypoalgesia depends on fitness level and/or sex." However, I cannot see any analysis relating to sex in the results of this study that takes the female sample of this study and the male sample of the previous study and then carries out another LNER model? I also see no report of results about sex in any of the figures/tables. If sex is part of the abstract and main research question, then I would argue it ought to be in the results and feature in figures within the main manuscript, not supplementary material.

We agree with the reviewer that, given the emphasis on fitness and sex in the Abstract and research aims, analyses directly examining sex effects should be presented in the main manuscript. In response, we pooled the samples from the previous and current studies and repeated the linear mixed-effects regression analyses used in the previous study to examine the effects of fitness and sex on exercise-induced pain modulation. We further complemented these analyses with Bayesian model comparison and parameter estimation to assess the moderating effects of fitness level and sex in the pooled dataset. The results of these pooled analyses are now reported in the corresponding Results section and are presented in the main manuscript as Figure 5. The Abstract, Results, Discussion, and Conclusions have been revised accordingly.

(6) The current abstract and main manuscript make it seem that this study completed an additional analysis on the effect of sex, but this does not appear to be the case. Statements like "together with our previous findings, this suggests that exercise-induced hypoalgesia depends on fitness level but not on sex" are quite decisive, although it does not seem that any analysis was completed between sexes in an a priori fashion. The only data I can see that assesses between sex effects is an exploratory analysis from the previous study, which was part of a larger exploration of sex - fitness level - and drug interactions.

Therefore could the authors either:

- Run this analysis between sexes;

- Or make it explicit that this study did not conduct any specific a priori analysis on sex differences in EIH at high and low intensities.

- And/or temper how decisive their conclusions on the (null) effects for sex, unless there are statistical analyses to support this.

We thank the reviewer for this suggestion. We agree that the manuscript did not include a dedicated analysis of sex effects in the combined dataset. In response, we pooled the samples from the previous and current studies and conducted additional frequentist (LMER) and Bayesian (Bayes Factor and posterior distributions) analyses examining the effects of fitness and sex on exercise-induced pain modulation. The frequentist analyses indicated an association between fitness and EIH, but not sex, while the Bayesian analyses similarly provided little evidence for sex-dependent effects. These new analyses have been added to the manuscript, and the Abstract, Results, Discussion, and Conclusions have been revised accordingly.

Section-specific Comments:

(1) Abstract

L38-40: See major comment #6 above.

As described above, we have now conducted additional analyses on the moderating effect of fitness level and sex using frequentist and Bayesian analyses and adapted the Abstract accordingly.

(2) Introduction

L68 - What is meant by "robust" in this sentence? Robust to describe the direction and magnitude of the differences in effects between sexes seems a little misrepresentative.

We thank the reviewer for their comment and agree that the wording is vague. To emphasise clarity, we have rephrased this to “[…] whereas others (Koltyn, 2000; Sternberg et al., 2001; Vaegter et al., 2014) suggest that hypoalgesia following exercise is more pronounced in females, but overall results remain inconclusive (Rice et al., 2019).”

(3) Results

L124 - What is the online rating scale? It is the word "online" in particular that is confusing here, as it is described in the methods, but just seems to be a standard handheld device with a cursor.

We thank the reviewer for their comment. In this case, “online” refers to the continuous ratings whilst the stimulus is ongoing, using a handheld rating device and using button presses. We clarified this further at the beginning of the Results and Methods sections to include “real-time (continuous) pain ratings” to avoid confusion.

L136 - I think there is a missing word here before the set of parentheses. RPE?

We thank the reviewer for detecting this and have added “RPE” before the parentheses.

L136-138 - I would amend to read something like... "This confirms the exercise intensity protocols elicited a different output, relative power, physiological demand (HR), and perceived demand (RPE)", just to be clear on what the differences have been physiologically and psychologically for the participants when exercising at the two intensities.

We agree with the reviewer and have rephrased the sentence according to their suggestion for clarity.

(4) Discussion

L190 - "results suggest that fit females show a hypoalgesic response to HI compared to LI exercise". I think there is a word missing in this sentence to make it clear what the results are.

We have revised the sentence in accordance with the new (pooled) analysis results.

L250-253 - The authors have given a fantastic account of the previous literature and the trends that each study has observed. However, in line with major comment #1, could the authors include some discussion to speculate why there are expected differences between training level/fitness level and sex (plus interactions)? Some studies that are referenced are theoretical papers, and even if the current literature is unsure about the likely mechanisms behind fitness and sex differences in EIH responses, stating this would also be useful to a reader but the authors could signpost them to specific references that would broaden their understanding after reading this study.

We have now provided a more detailed account of the potential mechanistic background on why fitness level and/or sex could influence exercise-induced modulation in the Discussion (see major comment #1).

(5) Limitations

L258-259 - If I have understood this sentence correctly and the content in L301-304, the heat stimulus (temperature of the thermal probe) that was applied to induce a 30 VAS pain intensity rating at rest actually induced no pain at all, as 30 VAS is considered below the pain threshold of "minimally painful". In retrospect, this seems like an odd choice of measurement.

It means that 30 VAS is simply a measure of warm (non-painful but detecting heat) and perturbation to the skin (pressure but not hurting) in the heat and pressure pain conditions, respectively. Therefore, to include the 30 VAS values in the analysis to dissociate whether HIGH or LIGHT intensities, fitness level, or sex have an impact on EIH seems misplaced, as there is no actual pain for which a hypoalgesia response can be tested at this intensity.

Making this clearer in the results or discussion would be worthwhile for the readers, in my opinion.

We thank the reviewer for highlighting this. In both the previous and current studies, participants were calibrated using the same procedure and rating scale (0 – 100) to identify painful stimulus intensities corresponding to VAS 30, 50, and 70. On the main experimental day, participants in the current study rated on a modified scale ranging from 0 to 150, with 50 marking the pain threshold, such that ratings below 50 reflected a non-painful rating. Importantly, the calibrated VAS 30, 50, and 70 stimuli refer to the target pain intensities established during the calibration procedure and not to the numerical values displayed on the experimental rating scale. Participants were familiarised with the scale and anchor points prior to testing and viewed only the visual analogue scale rather than numerical values during the experiment. We have adapted the wording, referring to VAS 30, 50, and 70 as intensities and not numerical ratings.

(6) Conclusion

L276-277 - Similar mention as before, can the authors please make it clear if there is a sex effect in their analyses or how this has properly been tested in a prior study through an a priori assessment to be certain that "fitness level rather than sex plays a central role in modulating EIH".

We have now tested the effect of sex in frequentist and Bayesian analyses in the pooled samples and revised the conclusion in accordance with the results.

(7) Methods

L284 - What denotes an individual being "physically active"?

We thank the reviewer for this question. “Physically active” was defined based on training volume, operationalised as regular participation in structured cycling training sessions (i.e., consistent engagement in planned cycling training units per week). This criterion was used to ensure that participants had an established and ongoing training routine relevant to the study context. We have clarified this in the referenced section.

Table 1 - Female participants in the current study are noticeably older than the other cohorts from the previous study. While all are considered "young" (<40 y.o.), if one were to take a matched pair of females from each cohort (previous vs current) in terms of relative FTP, then the years of experience and exposure to training are likely a strong influence on EIH responses, not "fitness level" as it is described in this study. Just a speculative thought.

The reviewer raises an interesting point. However, since we have not acquired any information on the years of training and volume, we are unable to answer this question in the current framework. Future research should investigate the effect of training years (which potentially strongly correlates with fitness level) on exercise-induced pain modulation.

L313-315 - For future reference, there are some nice papers on how to administer the RPE scales with clear instructions to be certain that perceived effort is being measured appropriately (Halperin & Emanuel 2020; Halperin & Visgotsky 2025; Malleron et al. 2023). I always have a habit of signposting to these studies with regard to RPE because it can often be misrepresented in many different areas of research where RPE scales are used.

We thank the reviewer for this suggestion and for highlighting these relevant papers. In the present study, participants were extensively familiarised with the rating of perceived exertion scale, and the instructions closely followed the original definition and recommendations provided by Borg (1998). The scale was presented in accordance with its intended anchor points to promote accurate reporting of perceived exertion. We have now added a citation (Borg, 1998) in the Methods section to clarify this procedure.

(8) Code Availability

Thank you for including open source code in your submission. I am very appreciative of researchers who champion open research principles.

We appreciate the reviewer’s comment and support of open research practices. We have updated the code containing the updated analyses and datasets.

Reviewer #2 (Recommendations for the authors):

(1) Core interpretation may not be fully supported by the data

We thank the reviewer for this comment. To investigate the role of fitness level and sex in exercise-induced pain modulation, we have now pooled the samples from the previous and current study and conducted frequentist (linear mixed effects models) analyses. Specifically, we tested for the effect of exercise intensity between the subgroups, as well as (consistent with those conducted in the previous study Nold et al., 2026) testing the effect of fitness level and sex across the pooled data. This allowed us to directly test the effects of fitness level and sex on a common scale rather than infer these effects from separate cohorts. Furthermore, we conducted Bayesian analyses to quantify evidence both for and against the presence of exercise-induced pain modulation and its moderation by fitness, sex, and study sample, utilising the Bayes factors comparing nested models of increasing complexity. In addition, we examined the posterior distributions of the regression coefficients (β) and subgroup-specific exercise effects to estimate the magnitude and uncertainty of these effects. The Bayesian analyses provided some evidence that fitness level moderated exercise-induced pain modulation, but sex did not. Importantly, these findings were consistent with the overall pattern of results obtained in the frequentist analyses. All corresponding statistics in the main manuscript and Supplementary Materials have been updated accordingly. To improve clarity, the pooled analyses are now presented in dedicated Result subsections and accompanied by new figures (Fig. 4 and Fig. 5).

(1.1) Lack of baseline condition

We thank the reviewer for this important comment. We agree that, in the absence of a no-exercise control condition, our design does not allow us to determine whether either exercise condition produced hypoalgesia or hyperalgesia relative to no exercise. Rather, the observed differences reflect the relative effects of high- versus low-intensity exercise on pain ratings.

Our study was not designed to assess the overall effect of exercise compared with rest, but rather to investigate whether exercise intensity influences exercise-induced pain modulation. Specifically, we hypothesised that high-intensity exercise would produce greater pain modulation than low-intensity exercise. This approach was motivated by previous work demonstrating intensity-dependent differences in post-exercise pain responses (Jones et al., 2019; Vaegter et al., 2014).

In response to the reviewer's comment, we have revised the manuscript to clarify this distinction and have replaced several instances of "exercise-induced hypoalgesia" with the more neutral term "exercise-induced pain modulation" where appropriate. We also explicitly acknowledge in the Limitations that the lack of a no-exercise condition precludes conclusions regarding absolute hypoalgesic/hyperalgesic effects of either exercise condition.

(1.2) Lack of internal replication across conditions

We thank the reviewer for this comment. We agree that the observed effect was most pronounced for heat pain at the highest stimulus intensity and was not consistently observed across all stimulus intensities or pain modalities. Importantly, however, this pattern mirrors the findings of our previous study, in which exercise intensity effects were likewise evident for heat pain but not pressure pain. We therefore interpret the current findings as a replication of the modality-specific pattern observed previously rather than an isolated positive result.

To further address concerns regarding robustness and sample size, we have now pooled the data from the previous and current studies and conducted additional frequentist and Bayesian analyses examining the moderating roles of fitness level and sex on exercise-induced pain modulation. These analyses yielded converging evidence that fitness level contributes to variability in exercise-induced pain modulation, thereby strengthening confidence in the overall findings.

With respect to pressure pain, previous meta-analyses have reported comparable or even larger exercise effects for pressure pain than heat pain (Naugle et al., 2012). However, most studies included in these analyses employed handheld pressure algometry, whereas our study used cuff pressure stimulation. Consequently, the pressure pain measures used here may not be directly comparable to those reported in the broader literature.

(1.3) Deviations from the original study and selective use of data

We agree with the reviewer that we have adapted the rating scale to reflect a real-time pain rating. To provide a comprehensive account of participants’ pain experience, we have included analyses corresponding to the visualisations on how online ratings evolve over the course of the pain administered in Supplemental Figure S7 – S8. Specifically, we conducted LMER models with exercise intensity × time on heat and pressure pain ratings, separately (Supplemental Fig. S7). The results yielded a significant interaction of exercise intensity × time on heat (β = -0.13, SE = 0. 06, t(7577) = 2.12, P = 0.03) but not pressure pain ratings (β = -0.07, SE = 0.10, t(7956) = -0.71, P = 0.48). Furthermore, we conducted LMER models with exercise intensity × time × stimulus intensity on heat and pressure pain ratings (Supplemental Fig. S8), suggesting no significant interaction on pressure pain ratings (β = -0.0002, SE = 0.004, t(2353) = -0.06, P = 0.95) but a significant interaction on heat pain ratings (β = -0.01, SE = 0.003, t(2201) = -3.24, P = 0.001).

As the reviewer correctly notes, to facilitate comparison with the previous study, we conducted our primary analyses using peak pain ratings. We acknowledge that this approach does not fully capture the richness of the continuous data but allows for comparability.

(1.4) Over-reliance on null results regarding sex differences

We thank the reviewer for this important point regarding the interpretation of null effects, particularly in relation to sex differences. We agree that the absence of a statistically significant difference does not, in itself, provide evidence for equivalence between groups.

To address this issue, we have pooled the samples and supplemented our frequentist analyses with Bayesian analyses. Bayes factors were computed to quantify the relative evidence for the null versus alternative hypotheses, providing a continuous measure of support that complements frequentist inference, particularly in cases of non-significant effects where evidence for the absence of an effect is of interest. Furthermore, posterior distributions of the regression coefficient of interest (in this case, the moderating role of sex on exercise-induced pain modulation) provided no evidence for a moderating role of sex on exercise-induced pain modulation. This approach allows for a more nuanced interpretation of the data, especially concerning null effects. We have adapted the Methods, Results, and Discussion accordingly.

(2) Limited sample and lack of diversity

We thank the reviewer for this comment. The study was intentionally designed to include only female participants with significantly higher fitness levels compared to the females from the previous study, as our previous study included females who were slightly less fit than males, making it difficult to separate the effects of sex and fitness. We, therefore, aimed to examine exercise-induced pain modulation in a more homogeneous sample. However, much of the EIH literature has been conducted in highly trained male participants (Janal et al., 1984; Saanijoki et al., 2022; Scheef et al., 2012). As such, studies in women of higher fitness are important to improve our understanding of potential sex differences in exercise-induced pain modulation. While the present findings should be interpreted within the context of the sample studied, when analysed and considered alongside our previous work, they contribute to a broader understanding of exercise-induced pain modulation across sexes and fitness levels.

(3) Methodological choices limit the interpretability of the data

(3.1) The use of a non-standard VAS scale (0-150 with a fixed pain threshold at 50) is unconventional and may influence how participants report pain, while limiting comparability with related literature.

We thank the reviewer for their comment. Since we introduced real-time pain ratings in the current study, while the pain is ongoing, we aimed to introduce a scale that allowed us to pinpoint the period when a painful stimulus reaches painfulness (exceeds 50 on the scale). To allow for comparability between the studies, we have calculated the peak pain rating provided during the real-time pain ratings, as this corresponds to the instruction participants received in the previous study to rate the most painful point of the previous stimulation. Previous research (Koyama et al., 2004) has shown that, in heat pain, real-time ratings account for R² = 0.89 of the variability of post-stimulus rating, where much of the variability for the post-stimulus intensity pain rating is accounted for by the peak response during the online ratings (adj. R² = 0.3 – 0.4), suggesting good comparability of both pain measures. Furthermore, as we have now pooled the samples and conducted analyses, we rescaled the maximum pain ratings so that ratings below 50 (below the pain threshold) were set to 0, corresponding to VAS 0 in the previous study, resulting in pain ratings ranging between 0 and 100.

(3.2) Participants explicitly reported expecting exercise to reduce pain, introducing a potential confound that is not presently addressed.

We thank the reviewer for this comment. Indeed, participants reported expecting lower whole-body pain in response to acute aerobic exercise, but no changes in joint or muscle pain. In the context of exercise, previous research (Colloca et al., 2018) showed that exercise-unrelated expectations and exercise itself induce comparable hypoalgesic effects, though they might work through distinct mechanisms. Studies directly manipulating exercise-related expectations showed that education about EIH, compared with general education about exercise and pain, could induce hypoalgesia following aerobic exercise (Jones et al., 2017). Interestingly, another study (Vaegter et al., 2020) investigated the interplay of exercise-related expectations (through verbal instructions) and resistance exercise on pain perception and showed that only explicit negative instructions induced hyperalgesia, whereas positive or neutral exercise-related instructions did not change the magnitude of EIH.

Since participants in the current study reported highly specific positive expectations only for whole-body pain, but not for joint or muscle pain, expectations are unlikely to fully explain our findings. Consistent with this, in our previous study using the same prospective expectation questionnaire, we found no significant expectation effects for any pain domain (Nold et al., 2025), although we acknowledge that the samples differed, especially with respect to fitness levels.

Nevertheless, the current study was not designed to disentangle the effects of exercise and expectations, as expectation ratings were measured prospectively but not manipulated. Therefore, we cannot determine to what extent positive expectations of exercise have contributed to the observed effects, and we have now acknowledged this as a limitation in the Discussion.

(3.3) A more comprehensive use of the full time series of pain ratings would provide a stronger and more transparent basis for interpretation of the present findings.

We thank the reviewer for this suggestion. As noted in our response to Comment 1.3, we have included analyses and visualisations of the full time series of online pain ratings in the Supplementary Materials (Fig. S7–S8), providing a more comprehensive understanding of pain dynamics over time.

Figures are very low in quality - artefacts, small text, insufficient labelling.

We thank the reviewer for highlighting this and, indeed, rendering the PDF reduced the quality of the images. For the reviewed version, we have been specifically attentive to image quality by increasing font sizes and adding labels.

There was not enough detail in the methods section for the number of cycling blocks, and how low and high intensity were distributed throughout the session.

We thank the reviewer for their comment. Participants completed four cycling blocks in total, two HI and two LI exercise blocks, presented in a pseudo-randomised order in the session. To improve clarity, we have added the relevant details to the Methods section.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

MDAR checklist

elife-109061-mdarchecklist1.docx^{(89KB, docx)}

Supplementary file 1. Supplementary tables S1–S30.

elife-109061-supp1.docx^{(48.2KB, docx)}

Data Availability Statement

[bib1] Allen H, Coggan A. Racing and Training with a Power Meter. VeloPress; 2006. [Google Scholar]

[bib2] Allen H, Coggan A. Training and Racing with a Power Meter. VeloPress; 2012. [Google Scholar]

[bib3] Anderson EH, Shivakumar G. Effects of exercise and physical activity on anxiety. Frontiers in Psychiatry. 2013;4:27. doi: 10.3389/fpsyt.2013.00027. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. Journal of Statistical Software. 2014;67:1–48. doi: 10.18637/jss.v067.i01. [DOI] [Google Scholar]

[bib5] Beserra AHN, Kameda P, Deslandes AC, Schuch FB, Laks J, de MH. Can physical exercise modulate cortisol level in subjects with depression? A systematic review and meta-analysis. Trends in Psychiatry and Psychotherapy. 2018;40:360–368. doi: 10.1590/2237-6089-2017-0155. [DOI] [PubMed] [Google Scholar]

[bib6] Bingel U, Tracey I. Imaging CNS modulation of pain in humans. Physiology. 2008;23:371–380. doi: 10.1152/physiol.00024.2008. [DOI] [PubMed] [Google Scholar]

[bib7] Black CD, Huber JK, Ellingson LD, Ade CJ, Taylor EL, Griffeth EM, Janzen NR, Sutterfield SL. Exercise-induced hypoalgesia is not influenced by physical activity type and amount. Medicine & Science in Sports & Exercise. 2017;49:975–982. doi: 10.1249/MSS.0000000000001186. [DOI] [PubMed] [Google Scholar]

[bib8] Blanton HL, Barnes RC, McHann MC, Bilbrey JA, Wilkerson JL, Guindon J. Sex differences and the endocannabinoid system in pain. Pharmacology, Biochemistry, and Behavior. 2021;202:173107. doi: 10.1016/j.pbb.2021.173107. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Borg G. Borg’s Perceived Exertion and Pain Scales. Human Kinetics; 1998. [Google Scholar]

[bib10] Borras MC, Becerra L, Ploghaus A, Gostic JM, DaSilva A, Gonzalez RG, Borsook D. fMRI measurement of CNS responses to naloxone infusion and subsequent mild noxious thermal stimuli in healthy volunteers. Journal of Neurophysiology. 2004;91:2723–2733. doi: 10.1152/jn.00249.2003. [DOI] [PubMed] [Google Scholar]

[bib11] Borszcz FK, Tramontin AF, Bossi AH, Carminatti LJ, Costa VP. Functional threshold power in cyclists: validity of the concept and physiological responses. International Journal of Sports Medicine. 2018;39:737–742. doi: 10.1055/s-0044-101546. [DOI] [PubMed] [Google Scholar]

[bib12] Brellenthin AG, Crombie KM, Cook DB, Sehgal N, Koltyn KF. Psychosocial influences on exercise-induced hypoalgesia. Pain Medicine. 2017;18:538–550. doi: 10.1093/pm/pnw275. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Castelli MP, Fadda P, Casu A, Spano MS, Casti A, Fratta W, Fattore L. Male and female rats differ in brain cannabinoid CB1 receptor density and function and in behavioural traits predisposing to drug addiction: effect of ovarian hormones. Current Pharmaceutical Design. 2014;20:2100–2113. doi: 10.2174/13816128113199990430. [DOI] [PubMed] [Google Scholar]

[bib14] Colloca L, Corsi N, Fiorio M. The interplay of exercise, placebo and nocebo effects on experimental pain. Scientific Reports. 2018;8:14758. doi: 10.1038/s41598-018-32974-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Crombie KM, Brellenthin AG, Hillard CJ, Koltyn KF. Endocannabinoid and opioid system interactions in exercise-induced hypoalgesia. Pain Medicine. 2018;19:118–123. doi: 10.1093/pm/pnx058. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Curran SL, Andrykowski MA, Studts JL. Short form of the profile of mood states (POMS-SF): psychometric information. Psychological Assessment. 1995;7:80–83. doi: 10.1037/1040-3590.7.1.80. [DOI] [Google Scholar]

[bib17] de Bruijn ST, van Wijck AJM, Geenen R, Snijders TJ, van der Meulen WJTM, Jacobs JWG, Veldhuijzen DS. Relevance of physical fitness levels and exercise-related beliefs for self-reported and experimental pain in fibromyalgia: an explorative study. Journal of Clinical Rheumatology. 2011;17:295–301. doi: 10.1097/RHU.0b013e31822c5196. [DOI] [PubMed] [Google Scholar]

[bib18] Denham J, Scott-Hamilton J, Hagstrom AD, Gray AJ. Cycling power outputs predict functional threshold power and maximum oxygen uptake. Journal of Strength and Conditioning Research. 2020;34:3489–3497. doi: 10.1519/JSC.0000000000002253. [DOI] [PubMed] [Google Scholar]

[bib19] Eippert F, Bingel U, Schoell ED, Yacubian J, Klinger R, Lorenz J, Büchel C. Activation of the opioidergic descending pain control system underlies placebo analgesia. Neuron. 2009;63:533–543. doi: 10.1016/j.neuron.2009.07.014. [DOI] [PubMed] [Google Scholar]

[bib20] Ellerbrock I, Wiehler A, Arndt M, May A. Nocebo context modulates long-term habituation to heat pain and influences functional connectivity of the operculum. PAIN. 2015;156:2222–2233. doi: 10.1097/j.pain.0000000000000297. [DOI] [PubMed] [Google Scholar]

[bib21] Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley JL. Sex, gender, and pain: a review of recent clinical and experimental findings. The Journal of Pain. 2009;10:447–485. doi: 10.1016/j.jpain.2008.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] Fullerton EF, Doyle HH, Murphy AZ. Impact of sex on pain and opioid analgesia: a review. Current Opinion in Behavioral Sciences. 2018;23:183–190. doi: 10.1016/j.cobeha.2018.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Gajic P, Kovac I, Lubinsky G, Knezevic NN. Exercise-induced hypoalgesia: cellular and molecular mechanisms linking pain modulation and stress regulation-a narrative review. Cells. 2026;15:858. doi: 10.3390/cells15100858. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib24] Gajsar H, Titze C, Hasenbring MI, Vaegter HB. Isometric back exercise has different effect on pressure pain thresholds in healthy men and women. Pain Medicine. 2017;18:917–923. doi: 10.1093/pm/pnw176. [DOI] [PubMed] [Google Scholar]

[bib25] Geisler M, Eichelkraut L, Miltner WHR, Weiss T. An fMRI study on runner’s high and exercise-induced hypoalgesia after a 2-h-run in trained non-elite male athletes. Sport Sciences for Health. 2019;16:159–167. doi: 10.1007/s11332-019-00592-8. [DOI] [Google Scholar]

[bib26] Geva N, Defrin R. Enhanced pain modulation among triathletes: a possible explanation for their exceptional capabilities. PAIN. 2013;154:2317–2323. doi: 10.1016/j.pain.2013.06.031. [DOI] [PubMed] [Google Scholar]

[bib27] Haier RJ, Quaid K, Mills JSC. Naloxone alters pain perception after jogging. Psychiatry Research. 1981;5:231–232. doi: 10.1016/0165-1781(81)90052-4. [DOI] [PubMed] [Google Scholar]

[bib28] Hox JJ, Moerbeek M. Multilevel Analysis: Techniques and Applications. Routledge; 2010. [DOI] [Google Scholar]

[bib29] Janal MN, Colt EWD, Clark CW, Glusman M. Pain sensitivity, mood and plasma endocrine levels in man following long-distance running: effects of naloxone. Pain. 1984;19:13–25. doi: 10.1016/0304-3959(84)90061-7. [DOI] [PubMed] [Google Scholar]

[bib30] Jones MD, Valenzuela T, Booth J, Taylor JL, Barry BK. Explicit education about exercise-induced hypoalgesia influences pain responses to acute exercise in healthy adults: a randomized controlled trial. The Journal of Pain. 2017;18:1409–1416. doi: 10.1016/j.jpain.2017.07.006. [DOI] [PubMed] [Google Scholar]

[bib31] Jones MD, Nuzzo JL, Taylor JL, Barry BK. Aerobic exercise reduces pressure more than heat pain sensitivity in healthy adults. Pain Medicine. 2019;20:1534–1546. doi: 10.1093/pm/pny289. [DOI] [PubMed] [Google Scholar]

[bib32] Kassambara A. Rstatix: pipe-friendly framework for basic statistical tests. Version 1.1.0Rstatix. 2019 doi: 10.32614/CRAN.package.rstatix. [DOI]

[bib33] Koltyn KF. Analgesia following exercise: a review. Sports Medicine. 2000;29:85–98. doi: 10.2165/00007256-200029020-00002. [DOI] [PubMed] [Google Scholar]

[bib34] Koltyn KF, Trine MR, Stegner AJ, Tobar DA. Effect of isometric exercise on pain perception and blood pressure in men and women. Medicine and Science in Sports and Exercise. 2001;33:282–290. doi: 10.1097/00005768-200102000-00018. [DOI] [PubMed] [Google Scholar]

[bib35] Koltyn KF, Brellenthin AG, Cook DB, Sehgal N, Hillard C. Mechanisms of exercise-induced hypoalgesia. The Journal of Pain. 2014;15:1294–1304. doi: 10.1016/j.jpain.2014.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Koyama Y, Koyama T, Kroncke AP, Coghill RC. Effects of stimulus duration on heat induced pain: the relationship between real-time and post-stimulus pain ratings. PAIN. 2004;107:256–266. doi: 10.1016/j.pain.2003.11.007. [DOI] [PubMed] [Google Scholar]

[bib37] Kruschke JK. Bayesian analysis reporting guidelines. Nature Human Behaviour. 2021;5:1282–1291. doi: 10.1038/s41562-021-01177-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] Lemley KJ, Senefeld J, Hunter SK, Hoeger Bement M. Only women report increase in pain threshold following fatiguing contractions of the upper extremity. European Journal of Applied Physiology. 2016;116:1379–1385. doi: 10.1007/s00421-016-3389-8. [DOI] [PubMed] [Google Scholar]

[bib39] Levine A, Liktor-Busa E, Lipinski AA, Couture S, Balasubramanian S, Aicher SA, Langlais PR, Vanderah TW, Largent-Milnes TM. Sex differences in the expression of the endocannabinoid system within V1M cortex and PAG of Sprague Dawley rats. Biology of Sex Differences. 2021;12:60. doi: 10.1186/s13293-021-00402-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] Lindheimer JB, Szabo A, Raglin JS, Beedie C, Carmichael KE, O’Connor PJ. Reconceptualizing the measurement of expectations to better understand placebo and nocebo effects in psychological responses to exercise. European Journal of Sport Science. 2020;20:338–346. doi: 10.1080/17461391.2019.1674926. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] Lovric M. International Encyclopedia of Statistical Science. Springer; 2025. [DOI] [Google Scholar]

[bib42] Loyd DR, Wang X, Murphy AZ. Sex differences in micro-opioid receptor expression in the rat midbrain periaqueductal gray are essential for eliciting sex differences in morphine analgesia. The Journal of Neuroscience. 2008;28:14007–14017. doi: 10.1523/JNEUROSCI.4123-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] Matei D, Trofin D, Iordan DA, Onu I, Condurache I, Ionite C, Buculei I. The endocannabinoid system and physical exercise. International Journal of Molecular Sciences. 2023;24:1989. doi: 10.3390/ijms24031989. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib44] May A, Rodriguez-Raecke R, Schulte A, Ihle K, Breimhorst M, Birklein F, Jürgens TP. Within-session sensitization and between-session habituation: a robust physiological response to repetitive painful heat stimulation. European Journal of Pain. 2012;16:401–409. doi: 10.1002/j.1532-2149.2011.00023.x. [DOI] [PubMed] [Google Scholar]

[bib45] McGrath E, Mahony N, Fleming N, Donne B. Is the FTP test a reliable, reproducible and functional assessment tool in highly-trained athletes? International Journal of Exercise Science. 2019;12:1334–1345. doi: 10.70252/RQOO7391. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib46] Mogil JS. Sex differences in pain and pain inhibition: multiple explanations of a controversial phenomenon. Nature Reviews. Neuroscience. 2012;13:859–866. doi: 10.1038/nrn3360. [DOI] [PubMed] [Google Scholar]

[bib47] Naugle KM, Naugle KE, Fillingim RB, Samuels B, Riley JLI. Intensity thresholds for aerobic exercise-induced hypoalgesia. Medicine & Science in Sports & Exercise. 2014;46:817–825. doi: 10.1249/MSS.0000000000000143. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib48] Naugle KM, Riley JL. Self-reported physical activity predicts pain inhibitory and facilitatory function. Medicine and Science in Sports and Exercise. 2014;46:622–629. doi: 10.1249/MSS.0b013e3182a69cf1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib49] Niwa Y, Shimo K, Ohga S, Tokiwa Y, Hattori T, Matsubara T. Effects of exercise-induced hypoalgesia at different aerobic exercise intensities in healthy young adults. Journal of Pain Research. 2022;15:3615–3624. doi: 10.2147/JPR.S384306. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] Nold JI, Fadai T, Büchel C. Acute aerobic exercise intensity does not modulate pain potentially due to differences in fitness levels and sex effects: results from a pharmacological fMRI study. eLife. 2025;14:RP102392. doi: 10.7554/eLife.102392. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib51] Nold JI. Hotspin_analyses_behavioural. swh:1:rev:e087127ab84c076d9e9623d7e591b3356951da09Software Heritage. 2026 https://archive.softwareheritage.org/swh:1:dir:d851ab73fdf2024d15f28e6e6afc6fc225822d4c;origin=https://github.com/jannenold/hotspin_analyses_behavioural;visit=swh:1:snp:dfd24da3e76835f2ddacc8f8e009db67896fff61;anchor=swh:1:rev:e087127ab84c076d9e9623d7e591b3356951da09

[bib52] Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia-- imaging a shared neuronal network. Science. 2002;295:1737–1740. doi: 10.1126/science.1067176. [DOI] [PubMed] [Google Scholar]

[bib53] Rice D, Nijs J, Kosek E, Wideman T, Hasenbring MI, Koltyn K, Graven-Nielsen T, Polli A. Exercise-induced hypoalgesia in pain-free and chronic pain populations: state of the art and future directions. The Journal of Pain. 2019;20:1249–1266. doi: 10.1016/j.jpain.2019.03.005. [DOI] [PubMed] [Google Scholar]

[bib54] Saanijoki T, Kantonen T, Pekkarinen L, Kalliokoski K, Hirvonen J, Malén T, Tuominen L, Tuulari JJ, Arponen E, Nuutila P, Nummenmaa L. Aerobic fitness is associated with cerebral μ-opioid receptor activation in healthy humans. Medicine & Science in Sports & Exercise. 2022;54:1076–1084. doi: 10.1249/MSS.0000000000002895. [DOI] [PubMed] [Google Scholar]

[bib55] Scheef L, Jankowski J, Daamen M, Weyer G, Klingenberg M, Renner J, Mueckter S, Schürmann B, Musshoff F, Wagner M, Schild HH, Zimmer A, Boecker H. An fMRI study on the acute effects of exercise on pain processing in trained athletes. Pain. 2012;153:1702–1714. doi: 10.1016/j.pain.2012.05.008. [DOI] [PubMed] [Google Scholar]

[bib56] Schmitt A, Wallat D, Stangier C, Martin JA, Schlesinger-Irsch U, Boecker H. Effects of fitness level and exercise intensity on pain and mood responses. European Journal of Pain. 2020;24:568–579. doi: 10.1002/ejp.1508. [DOI] [PubMed] [Google Scholar]

[bib57] Siebers M, Biedermann SV, Bindila L, Lutz B, Fuss J. Exercise-induced euphoria and anxiolysis do not depend on endogenous opioids in humans. Psychoneuroendocrinology. 2021;126:105173. doi: 10.1016/j.psyneuen.2021.105173. [DOI] [PubMed] [Google Scholar]

[bib58] Siebers M, Biedermann SV, Fuss J. Do endocannabinoids cause the runner’s high? evidence and open questions. The Neuroscientist. 2023;29:352–369. doi: 10.1177/10738584211069981. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib59] Sluka KA, Frey-Law L, Hoeger Bement M. Exercise-induced pain and analgesia? Underlying mechanisms and clinical translation. Pain. 2018;159:S91–S97. doi: 10.1097/j.pain.0000000000001235. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib60] Smith LD. Haverford College; 2004. The effects of competition and exercise on pain perception. [Google Scholar]

[bib61] Song JS, Yamada Y, Kataoka R, Wong V, Spitz RW, Bell ZW, Loenneke JP. Training-induced hypoalgesia and its potential underlying mechanisms. Neuroscience & Biobehavioral Reviews. 2022;141:104858. doi: 10.1016/j.neubiorev.2022.104858. [DOI] [PubMed] [Google Scholar]

[bib62] Sørensen A, Aune TK, Rangul V, Dalen T. The validity of functional threshold power and maximal oxygen uptake for cycling performance in moderately trained cyclists. Sports. 2019;7:217. doi: 10.3390/sports7100217. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib63] Stan Development Team RStan: the R interface to Stan. Version 2.32.7Stan. 2024 https://mc-stan.org/

[bib64] Sternberg WF, Bokat C, Kass L, Alboyadjian A, Gracely RH. Sex-dependent components of the analgesia produced by athletic competition. The Journal of Pain. 2001;2:65–74. doi: 10.1054/jpai.2001.18236. [DOI] [PubMed] [Google Scholar]

[bib65] Tesarz J, Schuster AK, Hartmann M, Gerhardt A, Eich W. Pain perception in athletes compared to normally active controls: a systematic review with meta-analysis. Pain. 2012;153:1253–1262. doi: 10.1016/j.pain.2012.03.005. [DOI] [PubMed] [Google Scholar]

[bib66] Vaegter HB, Handberg G, Graven-Nielsen T. Similarities between exercise-induced hypoalgesia and conditioned pain modulation in humans. Pain. 2014;155:158–167. doi: 10.1016/j.pain.2013.09.023. [DOI] [PubMed] [Google Scholar]

[bib67] Vaegter HB, Thinggaard P, Madsen CH, Hasenbring M, Thorlund JB. Power of words: influence of preexercise information on hypoalgesia after exercise-randomized controlled trial. Medicine and Science in Sports and Exercise. 2020;52:2373–2379. doi: 10.1249/MSS.0000000000002396. [DOI] [PubMed] [Google Scholar]

[bib68] Zubieta JK, Smith YR, Bueller JA, Xu Y, Kilbourn MR, Jewett DM, Meyer CR, Koeppe RA, Stohler CS. mu-opioid receptor-mediated antinociceptive responses differ in men and women. The Journal of Neuroscience. 2002;22:5100–5107. doi: 10.1523/JNEUROSCI.22-12-05100.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Fitness level, but not sex, affects exercise-induced pain modulation

Janne Ina Nold

Tahmine Fadai

Zora Pia Gerbers

Christian Büchel

Roles

Abstract

Introduction

Results

Successful pain calibration and exercise intervention

Figure 1. Implementation of high-intensity (HI) and low-intensity (LI) exercise.

Figure 1—figure supplement 1. Evaluation of menstrual cycle phase, expectation, and mood.

Figure 1—figure supplement 2. Calibrated stimulus intensities (30, 50, 70 visual analogue scale [VAS]) and online pain ratings for pressure and heat pain.

Greater pain relief after high-intensity exercise at higher stimulus intensities

Figure 2. The effect of exercise intensity and stimulus intensity on pain ratings.

Figure 2—figure supplement 1. Behavioural online ratings for pressure (top) and heat (bottom) pain throughout the stimulus duration after high- (purple) and low- (green) intensity exercise at all stimulus intensities (Visual Analogue Scale VAS 30, 50, 70).

Figure 2—figure supplement 2. Behavioural online ratings for pressure and heat pain throughout the stimulus duration after high- (purple) and low- (green) intensity exercise (averaged across all stimulus intensities and participants).

Sample differences in exercise-induced hypoalgesia

Figure 3. Comparing participant characteristics of the females of the current study (N=21; dark blue) with the females (N=21; light blue) and the males (N=18; red) of the previous study (Nold et al., 2025).

Figure 3—figure supplement 1. Comparing participant characteristics of the females of the current study (N=21; dark blue) with the females (N=21; light blue) and the males (N=18; red) of the previous study (Nold et al., 2025).

Figure 4. The effect of exercise intensity and subgroup (males from the previous study, females from the previous study and females from the current study) on heat pain ratings.

Figure 4—figure supplement 1. The effect of exercise intensity and subgroup (males from the previous study, females from the previous study and females from the current study) on pressure pain ratings.

Fitness level, but not sex, affects exercise-induced pain modulation in the pooled sample

Figure 5. Frequentist and Bayesian analyses of exercise intensity, fitness level and sex on heat pain across pooled study samples.

Figure 5—figure supplement 1. Effect of fitness level, sex, and subgroup on subject-specific differences in pressure pain ratings (dots) between low-intensity (LI) and high-intensity (HI) exercise conditions (LI–HI exercise pain ratings) pooled across samples and all stimulus intensities.

Discussion

Effect of exercise and stimulus intensity on pain in the current sample

Pooled analyses of the current and previous study sample (Nold et al., 2025)

Evidence for fitness level, but not sex, effects on exercise-induced hypoalgesia

Limitations

Conclusion

Methods

Participants

Table 1. Participant characteristics from the previous (Nold et al., 2025) and current study.

Experimental paradigm

Behavioural data acquisition

Statistical analyses

Linear models

Bayesian analyses

Table 2. Prior specification for Bayesian mixed-effects models.

Bayesian model comparison

Posterior inference

Code availability

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

eLife Assessment

José Biurrun Manresa

Roles

Reviewer #2 (Public review):

Anonymous

Roles

Author response

Janne Ina Nold

Tahmine Fadai

Zora Pia Gerbers

Christian Büchel

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases