Influence of habitual physical activity on cognitive performance of healthy young adults after passive heat exposure

Abstract

Objective

Regular physical activity is positively associated with enhanced cognitive performance, whereas excessive heat stress may negatively impact cognition. The role of habitual physical activity on cognitive function under heat stress is unclear. Thus, we investigated the influence of physical activity status on the cognitive performance of healthy individuals after acute passive heat exposure.

Methods

Our study involved non-randomised quasi-experimental controlled trials where 28 participants (active: n = 13, 4 females, less-active: n = 15, 5 females) underwent 45-min hot water immersion (HWI). Participants immersed their lower limbs in 42 °C water in a seated position for 15 min before they wore a disposable raincoat and continued immersion for 30 min. Estimated core temperature via non-invasive CORE sensor, heart rate, ratings of thermal sensation and thermal comfort were monitored. Pre- and post-immersion, the participants performed Stroop test and Iowa gambling task (IGT) to assess selective attention, executive function and decision-making.

Results

After HWI, both active and less-active participants attained hyperthermia (38.5 ± 0.4 °C and 38.3 ± 0.4 °C, respectively; p = 0.247). During HWI, the heart rate profiles, and subjective ratings were not different between groups regardless of physical activity status. Both groups exhibited faster reaction times and preserved accuracies in the Stroop test post-HWI, with no performance differences between groups. Similarly, for IGT, both active and less-active groups exhibited improved net scores post-HWI with no differences between groups.

Conclusion

When exposed to modest levels of heat strain (≤38.5 °C), both physically active and less-active groups did not show impairment in executive function and risk-taking behaviour.

1. Introduction

In the World Meteorological Organization's press release published January 2025,¹ 2024 was recorded as the warmest year with global temperatures exceeding 1.5 °C above pre-industrial baseline. This rising heat poses a major concern for countries, particularly in the tropical and subtropical regions where heat stress impairs human physical performance.² Likewise, some aspects of cognitive function can be impaired by heat stress, resulting in poorer work performance, or increased occupational errors and workplace accidents when heat exposure becomes unavoidable.³ A meta-analysis paper reported that productivity losses increase by 2.6 % for every 1 °C beyond 24 °C wet-bulb globe temperature.⁴ Workers who perform physically demanding tasks, wear personal protective and/or work in enclosed environments face heightened risks of experiencing severe heat strain. The poorer cognition caused by heat strain can engender significant productivity losses and elevate the likelihood of occupational hazards, especially for those who work in high-risk environments.⁵

Excessive heat stress can compromise various cognitive domains including selective attention, vigilance, cognitive flexibility and executive function. For instance, a study⁶ found that mild passive heating (+1.3 ± 0.2 °C in esophageal temperature) degraded executive function and inhibitory processing abilities during Go/No-go tasks in healthy young adults. In a separate study,⁷ it was found that environmental thermal stress similarly impacts cognitive functions including memory, attention and processing speeds of older adults above 60 years, especially those with lower levels of physical activity. In the occupational context, prolonged exposure to work-related heat stress impairs cognitive performance where workers’ response times were found to be increased with an elevated rate of errors during the Stroop test and Continuous Performance test.^3,8 Greater perceived fatigue and discomfort were also reported when working in hot environments.⁹ Heightened impulsivity and risk-taking behaviours are driven by warmer temperatures, leading to increased work errors.^10,11 Additionally, other studies have reported a decline in vigilance and cognitive flexibility in firefighters exposed to high temperatures,¹² and reduced sustained attention and executive function in soldiers in hot desert environment (42–43 °C).¹³ To develop effective mitigation strategies for heat-related cognitive deficits, it is pertinent to identify the determinants of cognitive functioning under heat stress.

Participation in regular physical activity is known to be beneficial for cognition where an inverse relationship exists between the level of physical activity and the likelihood of cognitive decline and dementia.^14,15 Habitual physical activity promotes higher fitness level,¹⁶ which has been associated with better academic performance,¹⁷ cognitive performance¹⁸ and cognitive control¹⁹ in adolescents. Young adults with lower cardiorespiratory fitness performed poorer during complex cognitive tasks compared to those with higher cardiorespiratory fitness.²⁰ The cognitive benefits related to aerobic fitness could be attributed to the neurophysiological adaptations to the stimulus from long-term exercise training. Beyond aerobic exercises, resistance training is likewise shown to promote structural brain changes and enhance cognitive function.²¹ Regular exercise stimulates neurogenesis,²² cerebral angiogenesis,²³ synaptogenesis in cerebral structures and improves cerebrovascular reserve,²⁴ which support better brain function. For instance, well-trained individuals demonstrate improved cerebral blood flow velocity²⁵ and regulation,²⁶ allowing them to maintain brain perfusion even with fluctuations in blood pressure under physical exertion or heat stress, a mechanism that is less efficient in untrained individuals. Moreover, regular bouts of physical activities, especially when performed outdoors, can elicit partial heat acclimatisation in physically active individuals²⁷ which significantly improve their heat tolerance and may contribute to better resistance against cognitive performance loss.²⁸ Therefore, habitual physical activity may confer a protective role in preserving cognitive function under heat stress.

Despite the known impact of heat stress on cognition, it is unclear if habitually active individuals exhibit increased resilience against heat-related cognitive deficits when compared with the less-active individuals. This is especially pertinent with the increasing recognition that regular engagement in physical activity is a likely protective factor against climate stressors.²⁹ Therefore, we aimed to evaluate and compare the cognitive performance between active and less-active individuals following acute heat exposure via hot water immersion. We hypothesised that habitually active individuals can better preserve their cognitive performance under heat stress compared with their less-active counterparts.

2. Methods

2.1. Participants

Ethical approval was obtained from the Singapore Institute of Technology Institutional Review Board (Ref no: RECAS-0118) and the study was conducted from August to December 2023. Healthy males and females aged between 21 and 35 years old were recruited. Eligible participants were either (1) active individuals that participated in regular physical activity >3 times per week with a total of duration of moderate/vigorous exercise ≥250 min per week, and endurance trained (involving mostly aerobic exercise), or (2) less-active participants that participated in physical activity ≤3 times per week with a total duration of moderate/vigorous exercise ≤150 min per week. The frequency and duration of moderate and vigorous physical activity performed per week, the type of exercise training, and the participants’ sports background were self-reported via an online survey. Participants with skin disease, skin allergies, skin injuries, burn scars, any contraindications for heat application (e.g., acute inflammation, venous thrombosis, or Raynaud disease), any thermoregulatory disorders such as impaired sweating, prior history of heat exhaustion or heat stroke, or any cognitive impairments or learning disabilities and existing cardiovascular disorders were excluded. Participants provided their written consent prior to the commencement of the study.

2.2. Experimental design

The study was conducted as non-randomised quasi-experimental controlled trials where active and less-active participants were exposed to an acute heating intervention (hot water immersion; HWI) with cognitive assessments conducted pre- and post-HWI (Fig. 1). The selected passive heating protocol allowed us to investigate the independent effect of heat on cognition, compared to an exertional protocol which could be confounded by the effects of exercise (e.g. increased arousal). One day prior to experiment, participants were instructed to abstain from exercise and refrain from ingesting alcoholic, caffeinated, and carbonated drinks to avoid confounding sympathetic activity and vasomotor effects that influence thermoregulation. They were instructed to consume at least 500 mL of water 2 h before the experiment to ensure euhydration. A safety briefing was conducted to remind on test termination criteria for HWI and potential adverse symptoms. Participants received instructions for the computerised cognitive assessments prior to the researcher providing demonstrations.

Fig. 1.

Schematic diagram of the experimental procedures involving a 45-min dual phase hot water immersion protocol via lower limb immersion in 42 °C water. Pre- and post-assessments include two cognitive tests including Stroop test and Iowa gambling task (IGT).

Upon arrival at the study site, participants’ height and body mass were assessed with a stadiometer (SECA 213, Hamburg, Germany) and a portable weighing scale (Tanita HD-382, Tokyo, Japan). Body mass index (BMI) and body surface area (BSA) were computed from their height and body mass values.³⁰ In addition, their resting physiological parameters were measured. Baseline heart rate (HR), blood pressure, and oxygen saturation level (SpO₂) were recorded via vitals monitor cart with a pulse oximeter (Spot Vital Signs, Welch Ally, New York, USA). Mean arterial pressure (MAP) was derived using the following equation: MAP = 1/3 systolic blood pressure + 2/3 diastolic blood pressure. Estimated core temperature (Est-T_core) was assessed using a CORE body temperature sensor (greenTEG AG, Rümlang, Switzerland) fitted using a chest strap positioned below the sternum level with the sensor placed on the left side of chest. A HR sensor (Polar H10 HR Sensor, Polar Electro, Kempele, Finland) was placed on the same chest strap for continuous monitoring. Additionally, their oral temperatures (T_oral) were measured with a digital thermometer (KT-DT04B, Medpro, Singapore; accuracy ± 0.1 °C) placed under the tongue. Subjective ratings such as the rating of thermal sensation (RTS; 8-point rating scale) and the Bedford thermal comfort scale (RTC; 7-point rating scale) were self-reported by participants during the experiment. Before commencing HWI, participants performed two cognitive assessments - Stroop test and Iowa Gambling Task (see details under cognitive assessments section). No familiarisation trial was given for cognitive assessments to minimise possible learning effects, but participants were briefed with detailed instructions on how to perform these tests.

2.3. Dual-phase hot water immersion

The HWI intervention involved a dual-phase protocol (Fig. 1) which was conducted in an air-conditioned room with thermostat maintained at 25 °C. In Phase I, participants fully immersed both lower limbs to hip level in 42 °C water for 15 min in a plastic upright immersion tub. The tub water temperature was maintained at 42.0 ± 0.4 °C. Est-T_core, T_oral, HR and SpO₂ were recorded every 5 min. BP, RTS, and RTC were recorded at 15-min intervals. Researchers verified participants’ well-being before proceeding into Phase II. In Phase II, participants donned a disposable insulative raincoat with the hood on while continuing to immerse for another 30 min. The intermission between Phase I and Phase II was less than 1 min for all participants. A graded heating protocol with lower limb immersion was adopted in this study to avoid undesirable symptoms in participants from rapid heating (e.g. headaches, dizziness and severe numbness in extremities), which were observed in earlier studies involving whole-body HWI. Also, the passive nature of HWI removes the confounding effect of exercise in exertional protocol. During the immersion process, participants were given fixed aliquots of ambient temperature water (1.5 g/kg/h fluid replenishment rate) to drink every 15 min to avoid dehydration. The immersion was immediately ceased once participants attained Est-T_core of 39.0 °C, or when the maximum immersion duration of 45 min was reached (all participants successfully completed 45 min of HWI without early termination in our study).

Immediately after the HWI, participants performed the post-intervention cognitive tests while remaining seated in the drained tub, with raincoat on but its hood off. This ensured that participants did not lose heat rapidly during the tests. After exiting the tub, the participants towel-dried themselves before their body mass was measured, and they were seated on a chair for 15 min before their vitals (Est-T_core, T_oral, HR, BP and SpO₂) were re-assessed.

2.4. Cognitive assessments (Stroop test and Iowa Gambling Task)

The primary outcomes are the cognitive performance results from Stroop test and Iowa Gambling Task (IGT), reflecting changes in selective attention, executive function and decision-making (relevant to risk-taking behaviour). Cognitive assessments were performed on a computer via a software platform (Inquisit 5 Lab version 5.0.14.0). The Stroop colour word test³¹ consists of neutral, congruent, and incongruent stimuli. Participants were required to press the respective key on the keyboard which corresponds with the colour of the words appearing on the screen. Participants were instructed to respond correctly as quickly as possible. For the neutral stimuli, only a coloured box is displayed. For congruent stimuli, the word and the colour match each other (e.g., the word “green” is displayed in green) whereas for incongruent stimuli, the word and colour do not match each other (e.g., the word “green” is displayed in red). Reaction time (RT) and accuracy (ACC) towards both stimuli were recorded and analysed. The Stroop test has been shown to be sensitive to the alteration of cognitive functions and can identify deficits in selective attention or executive function (inhibitory control) under heat stress conditions.^3,32 There is good test-retest reliability of Stroop test for the response latencies for same-day administration (r = 0.93).³³

As for the IGT,³⁴ participants were required to use the mouse to click on one of the four decks of cards displayed on the computer screen. Choosing one card deck could result in the delivery or removal of game money from the participant's “bank”. Each deck offered different odds of earning or losing game money, varying in immediate gain, expected long-term gain, and schedule of losses. The decks were reshuffled after each attempt and participants were given 100 selection attempts. The final net IGT score (number of advantageous decks chosen – number of disadvantageous decks chosen) and the progressive IGT score every 20 attempts (5 periods in total) were recorded and analysed. As IGT involves adaptive learning over time to learn about the profit or loss probability of each deck, the progressive IGT score allows for understanding of the temporal shift in participants' decision-making and risk-taking behaviour within each task. The net IGT score reflected the participants' decision-making abilities in uncertainty and their capacity in risk-benefit analysis. A positive net score (net advantage) suggests sound and logical decision-making, while a negative net score (net disadvantage) suggests difficulties in learning from feedback and choosing advantageous options. A negative net score serves as an indicator of risk-taking behaviour with the tendency for participants to favour immediate gains with high risks over long-term outcomes. Given that heat strain has been shown to influence risk-taking behaviour,¹¹ IGT provides a nuanced approach in our study to assess decision-making capacity and risk-taking behaviours in young adults.³⁵ Studies have reported moderate-to-good same-day test-retest reliability (r = 0.36 to 0.61) for IGT.³⁶

2.5. Statistical analysis

Estimation of the required sample size was performed using G∗Power application (G∗Power 3.1.9.7, gpower.hhu.de). Sample size was determined a priori to detect significant change based upon the findings on cognitive performance changes in a previous study involving healthy participants exposed to passive heat in a hot environmental chamber (50 °C, 50 % relative humidity) as compared to normothermic control in temperate (25 °C, 50 % relative humidity) condition.³⁷ Using the determined effect size (d = 1.57), power analysis indicated that 20 participants was sufficient to provide 0.9 power (β) for two-tailed T-test with α of 0.05 for significance.

Statistical Package for Social Sciences (SPSS) version 29.0.2.0 (SPSS Inc., Chicago, IL, USA) was used to compute all data. Shapiro-wilk test and Levene's test were conducted to assess data normality and equal variances. Chi-square test was used to compare the distribution of demographic variables in both groups. For between-group analysis, Student's T-test was used to compare differences in physical activity duration, body mass, BMI, BSA, end Est-T_core and end T_oral between groups. To control the familywise error rate, Bonferroni correction was applied post-hoc to correct for multiple comparisons. Two-way multivariate analysis of variance (MANOVA) was used to compare the RT and ACC values across congruent, incongruent and neutral stimuli in Stroop test, using Wilks' lambda (Λ) as the multivariate test statistic. Mixed model two-way ANOVA with repeated measures was used to compare the Est-T_core, T_oral, HR, MAP, subjective ratings, net and progressive IGT scores across timepoints between active and less-active groups. Where significant effects were established, pairwise differences were identified using Bonferroni correction for multiple comparisons. To replace one missing T_oral value (at 45 min) for analysis, single data imputation was performed using linear interpolation taking reference from adjacent mean values. Mauchly's test was used to assess data sphericity, and the Greenhouse–Geisser correction was applied as necessary. Effect sizes were computed using Hedges' g, rank-biserial correlation, r and partial eta-squared, η²_p to quantify the magnitude of differences accordingly. 95 % confidence intervals (CI) are presented for the mean differences in cognitive performance variables. Significance level was set at p < 0.05 for all statistical analyses. Values are depicted as mean ± SD.

3. Results

3.1. Participants

A total of 28 participants (active: n = 13, age 24 ± 2 years, 4 females; less-active: n = 15, age 25 ± 3 years, 5 females) completed the study and were included for analysis (see Table 1 for participants’ characteristics).

Table 1.

Participant characteristics.

Characteristics	Active (n = 13)	Less-active (n = 15)	p value
Sex (M:F)	9:4	10:5	0.885
Age (years)	24 ± 2	25 ± 3	0.537
Weight (kg)	63.6 ± 12.3	70.3 ± 21.7	0.334
Height (m)	1.68 ± 0.11	1.67 ± 0.07	0.738
BMI (kg/m2)	22.5 ± 2.5	25.2 ± 6.8	0.185
BSA (m2)	1.72 ± 0.21	1.77 ± 0.26	0.528
Resting heart rate (bpm)	69 ± 11	72 ± 14	0.325
Resting MAP (mmHg)	86 ± 7	88 ± 8	0.771
MVPA per week (min)	468 ± 254	142 ± 92	<0.001∗

BMI = body mass index, BSA = body surface area, F = female, M = male, MAP = mean arterial pressure, MVPA = moderate-to-vigorous physical activity, ∗ denotes significance with Bonferroni-adjusted threshold set at p < 0.006.

3.2. Est-T_core and T_oral temperature

During immersion, there was a main effect of time on the change of Est-T_core for both groups (F_1.4,36.8 = 137.7, p < 0.001, η²_p = 0.841) with no interaction effect (p = 0.518). However, there was no main effect of physical activity status (F_1,26 = 0.904, p = 0.351, η²_p = 0.034) on the change of Est-T_core across time (Fig. 2A). There were no differences in the Δ Est-T_core from pre-to post-HWI between active and less-active groups (1.3 ± 0.4 °C vs. 1.1 ± 0.4 °C respectively; p = 0.384, g = 0.334). After HWI, the end Est-T_core was not different between active (38.5 ± 0.4 °C) and less-active groups (38.3 ± 0.4 °C; p = 0.247, g = 0.480).

Fig. 2.

Profiling of physiological parameters during hot water immersion including the change from baseline (delta; Δ) in (A) estimated core temperature (Est-T_core), (B) oral temperature (T_oral) and (C) heart rate (HR), and (D) mean arterial pressure (MAP) of the participants. The open and shaded circles denote the data from active and less-active participants respectively. Data are presented as mean with standard deviation as error bars.

Similarly, for the T_oral profile during immersion, there was a main effect of time on the change of T_oral in both groups (F_2.9,75.9 = 203.9, p < 0.001, η²_p = 0.887) with no interaction effect (p = 0.078). Likewise, there was no main effect of physical activity status (F_1,26 = 2.906, p = 0.100, η²_p = 0.101) on the change of T_oral across time (Fig. 2B). There were no differences in the Δ T_oral from pre-to post-HWI between active and less-active groups (2.0 ± 0.5 °C vs. 1.6 ± 0.5 °C respectively; p = 0.076, g = 0.698). After HWI, the end T_oral was not different between active (38.5 ± 0.4 °C) and less-active groups (38.3 ± 0.5 °C; p = 0.243, g = 0.529).

3.3. Physiological and subjective parameters

For the body mass changes post-HWI, the active group experienced a greater reduction in body mass (0.8 ± 0.5 %) when compared with the less-active group (0.3 ± 0.3 %; p < 0.001, g = 1.249) despite similar fluid intake at 1.5 g/kg/h and identical immersion duration (all completed the 45-min HWI). The active group had a higher mean whole body sweat rate (24.9 ± 11.7 ml/min) when compared with the less-active group (13.7 ± 7.4 ml/min; p = 0.005, g = 1.166).

For the HR profile during immersion, there was a main effect of time on the change of HR in both groups (F_3.8,100.7 = 200.2, p < 0.001, η²_p = 0.885) with no interaction effect (p = 0.632). There was no main effect of physical activity status (F_1,26 = 0.284, p = 0.599, η²_p = 0.011) on HR change across time (Fig. 2C). For blood pressure changes, there was a main effect of time on the change of MAP in both groups (F_1.8,45.6 = 8.780, p < 0.001, η²_p = 0.252) with no interaction effect (p = 0.205). There was no main effect of physical activity status (F_1,26 = 1.266, p = 0.271, η²_p = 0.046) on MAP changes across time (Fig. 2D).

There was no difference in all subjective ratings between active and less-active groups across time (p ≥ 0.486), indicating no differences in thermal sensation (end RTS: 5.7 ± 1.0 vs. 5.6 ± 0.9) and thermal comfort (end RTC: 1.5 ± 1.0 vs. 1.8 ± 0.9) during passive heating, regardless of physical activity status.

3.4. Cognitive assessment – Stroop test

Pre-to-post HWI comparison of mean RT and ACC within each group are depicted in Table 2 and individual data plots are presented in Supplementary Fig. 1. Both groups showed faster RT to congruent, incongruent and neutral stimuli after HWI (F_3,50 = 10.574, p < 0.001, Wilks' Λ = 0.612). In contrast, there were no pre-to-post differences in Stroop test ACC in congruent, incongruent and neutral stimuli within each group (F_3,50 = 0.750, p = 0.528, Wilks' Λ = 0.957). Notably, both groups exhibited a high response accuracy (>93 %) for all three stimuli at pre- and post-HWI assessments.

Table 2.

Stroop test reaction time and response accuracy for congruent, incongruent and neutral stimuli during pre- and post-immersion assessments.

	Active (n = 13)		Less-active (n = 15)
Reaction Time (ms)
	Pre	Post	Pre	Post	Pooled Pre-Post Mean Difference	MANOVA p value
Congruent	994 ± 130	793 ± 82	897 ± 189	720 ± 97	−189 [-260, −117]	<0.001a
Incongruent	1194 ± 247	887 ± 151	1105 ± 227	852 ± 153	−281 [-388, −174]	<0.001a
Neutral	991 ± 208	802 ± 113	897 ± 182	718 ± 133	−184 [-272, −96]	<0.001a
Response Accuracy (%)
	Pre	Post	Pre	Post	Pooled Pre-Post Mean Difference	MANOVA p value
Congruent	99.2 ± 2.1	97.8 ± 3.6	98.8 ± 2.8	97.9 ± 2.9	−1.2 [-2.8, 0.5]	0.159
Incongruent	97.0 ± 3.7	94.5 ± 6.8	94.0 ± 5.6	93.3 ± 8.6	−1.6 [-5.2, 2.0]	0.380
Neutral	98.6 ± 2.2	98.1 ± 3.0	98.1 ± 2.6	97.9 ± 2.9	−0.4 [-1.9, 1.1]	0.596

^adenotes significance for pre to post changes across both groups (Bonferroni-corrected). Data presented as mean ± SD and mean differences with 95 % confidence intervals in square brackets.

For between-group comparisons, there was no difference in the magnitude of RT change (Δ RT) for congruent, incongruent and neutral stimuli after HWI between active and less-active participants (F_3,24 = 0.282, p = 0.838, Wilks' Λ = 0.966; Fig. 3A). Similarly, there was no difference observed in the magnitude of ACC change (Δ ACC) between active and less-active participants for congruent, incongruent and neutral stimuli after HWI (F_3,24 = 0.156, p = 0.925, Wilks' Λ = 0.981; Fig. 3B).

Fig. 3.

Stroop test performance changes (delta; Δ) for (A) reaction time and (B) response accuracy between the active and less-active groups. The open and shaded circles denote the individual data points for active and less-active participants respectively. Data are presented as median with interquartile range (boxes) with the minimum and maximum values (whiskers).

3.5. Cognitive assessment – Iowa Gambling Task

For both groups, there was a main effect of time (F_1,26 = 9.743, p = 0.004, η²_p = 0.273) on the net IGT mean scores with no interaction effect (p = 0.610) (Fig. 4A). The net IGT mean scores shifted positively after immersion for both the active and less-active groups (pooled mean difference: +18, 95 % CI [6, 30]). There was no main effect of physical activity status (F_1,26 = 0.450, p = 0.508, η²_p = 0.017) on the net IGT score change across time. Within the IGT test, there was no difference in the progressive IGT scores across the 20-attempt periods for pre-HWI test (F_4,104 = 1.813, p = 0.132, η²_p = 0.065; Fig. 4B). On the contrary, for post-immersion test, both groups showed improvements in decision-making (less risk-taking) across time with higher IGT scores across the periods (F_4,104 = 3.797, p = 0.006, η²_p = 0.127; Fig. 4C).

Fig. 4.

Iowa gambling task (IGT) net mean score for (A) overall attempts, and the progressive IGT mean score every 20 attempts for (B) pre-immersion attempts and (C) post-immersion attempts between the active and less-active groups. Individual data points are not presented here to facilitate clearer comparative analysis of group-level trends (refer to Supplementary Fig. 2 for the individual data plots). Data are presented as mean with standard deviation as error bars.

4. Discussion

We investigated the changes in cognitive performance of habitually active and less-active adults after 45-min passive heat exposure. We demonstrated that both groups experienced similar thermal and cardiovascular strain after HWI, even though the active participants lost more body mass and were more dehydrated. Our findings indicated that (1) the modest heat strain (≤38.5 °C) induced by HWI did not adversely affect cognitive performance; instead, it (2) improved reaction time during the Stroop test and reduced risk-taking behaviour during the IGT, and (3) these heat-induced cognitive changes did not differ according to participants’ self-reported physical activity levels.

Consistent with our previous finding,³² all the participants responded quicker in the Stroop test under a hyperthermic state, likely due to the improved contractility with elevated muscle temperatures and heightened arousal. However, there were no differences in their inhibitory control post-HWI when responding to incongruent stimuli with semantic interference during the Stroop test. This is in contrast with the findings from earlier studies^6,8 where heat stress exposure has resulted in the impairment of executive function. This could be related to the modest level of hyperthermia (mean increment of 1.1–1.3 °C) induced in our study via the 45-min lower limb HWI, which may not affect cognitive performance in young healthy individuals. The HWI protocol used in our study was intended to produce heat stress levels comparable to those typically found in real-world situations, where most people do not experience a significant rise in core temperature (greater than 1.5 °C). Beyond which, thermal discomfort is significant and this drives humans to adopt cool-seeking behaviours (e.g. look for shade or cool spaces)³⁸ or reduce activity level by self-pacing to prevent further rise in body temperatures.³⁹ In occupational settings, high body temperatures (e.g. ACGIH upper limit for acclimatised workers to discontinue work at 38.5 °C⁴⁰) are likewise actively avoided or mitigated to protect workers. Several field studies examining the impact of occupational heat stress have shown that workers’ body temperatures rarely exceed 38.0 °C,^9,11,41 yet impairments to cognition have been consistently demonstrated in similar settings.^3,11,42 Although higher order cognition –compared to simpler functions such as attention– are more susceptible to perturbation by hyperthermia,⁴³ we did not find deficits in executive function related to inhibitory control in both groups.

Corroborating the lack of impairments to Stroop test performance under modest heat strain, both groups did not demonstrate poorer IGT performance after acute heat exposure. On the contrary, both groups showed improved IGT net scores post-HWI. Based upon the inverted U-shaped relationship between cognitive performance and heat strain,⁴⁴ an initial increase in body temperature can benefit cognitive performance due to increased arousal and a Q10 effect driving an increased cerebral metabolic rate of oxygen which supports brain metabolic needs. However, beyond the putative threshold of 39 °C, the cognitive performance is likely progressively impaired.⁴⁴ In our study, our participants only attained moderate hyperthermia where their end Est-T_core remained ≤38.5 °C which could explain the lack of impairment in IGT performance. Moreover, their improved IGT scores could be reflective of an adaptive learning effect. In IGT, the participants had to make optimal decisions with no clear incentives or rewards, resulting in increased reliance on the experiential system to achieve suboptimal and satisficing (“look for the best” or “through gut feel”) decisions.⁴⁵ Therefore, the objective decision-making behaviour is expected to improve over time when there is more information collected regarding the task, with less heuristics and automatic responses (i.e. an adaptive learning effect). On the other hand, the lack of differences between groups could be due to interindividual differences in risk-taking tendencies and their individual abilities to adapt to the IGT test. While the two groups report different weekly MVPA durations, individual differences in training, heat acclimatisation and fitness status (e.g. type of physical activity, time spent outdoors and different exercise intensities) may affect their physiological heat tolerance and thereby modulate their resistance against cognitive deficits under heat stress. These factors could reduce the ability to detect small group differences. Additionally, their self-reported durations are subjected to reporting biases and future studies should profile the prior physical activity levels of participants via actigraphy.

Despite significant differences in their physical activity levels, there were unexpectedly similar physiological responses to acute heat stress exposure in both groups. While this study did not directly assess participants’ aerobic capacity, fitter individuals with regular exercise training (not higher VO₂max per se) are known to have superior vasodilatory responses, earlier onset of sweating and greater sweat rates and thus, will experience a smaller rise in T_core during uncompensable heat stress.^46,47 This was aligned to our results that the active group exhibited significantly higher whole body sweat rates and completed the immersion with greater body mass loss (0.5 % more). The similar Est-T_core increment from HWI may result from the active individuals having lower body fat percentages, despite similar BMI and BSA to the less-active group. Since the heat conductivity of fat is 50 % lower than that of muscle,⁴⁸ the heat transfer efficiency during HWI was expected to be higher for the active participants with less subcutaneous adipose tissues to act as an insulating layer to impede heat gain from exogenous sources.⁴⁹ Furthermore, the active group had a lower mean body mass (difference of 6.7 kg albeit not statistically significant) corresponding to a smaller heat storage capacity which likely contributed to their faster Est-T_core rise. This could have negated the thermoregulatory benefits brought about by exercise training adaptations.

Several limitations of this study need to be acknowledged. Firstly, the magnitude of rise in body temperature after HWI was modest which could have limited the heat-related cognitive deficits. Considering that cognitive impairments are more likely to manifest beyond 38.5°C⁴⁴ and heat stress is found to impair cerebral perfusion when T_core rises by more than approximately 1.2 °C,^50,51 future studies can explore a high T_core temperature such as 38.8 °C (>1.2 °C rise from 37.5 °C) as a thermal threshold to terminate the heat stress protocol and to assess cognitive performance thereafter. Secondly, we did not control for the menstrual cycle of the female participants which could have influenced their baseline body temperatures but the whole body heat loss capacity was unlikely to be compromised by the menstrual phase.^52,53 Thirdly, cognitive tests were only conducted before and after heat exposure, which may introduce an order effect. However, based on practice gains reported in literature, the change in cognitive performances from pre to post-heating in our study exceeds the typical learning gains and known coefficients of variation (∼8 %; 3–7 %) for Stroop test reaction time and IGT net score respectively.54, 55, 56, 57 With moderate to good test-retest reliability (r = 0.31–0.93) of the cognitive tests, the observed differences in performance in our study are likely valid.

It is also pertinent to consider that the body temperature of participants in this study was estimated with CORE sensors which are convenient and easy for non-invasive temperature monitoring. Given the acceptable reliability,⁵⁸ CORE sensors can be used to assess group mean responses but not for individualised accuracy. Furthermore, CORE sensors are shown to perform well in tracking rates of change,⁵⁹ thus the delta T_core differences observed in this study between active and less-active participants are likely valid. However, it is less accurate than direct measurements of deep body temperature at thermally sensitive tissue sites (e.g. rectal or gastrointestinal temperatures), and its validity has yet to be confirmed for passive heating protocol via HWI. Lastly, it is crucial to note that the current study focuses on young, healthy adults. The impact of exercise training is likely more pronounced in vulnerable populations, such as older adults⁷ and future research should investigate the effects of habitual physical activity and/or aerobic fitness on the resilience of older adults against heat-related cognitive deficits.

5. Conclusion

Our findings indicated that (1) the modest heat strain (≤38.5 °C) induced by HWI did not adversely affect cognitive performance; instead, it (2) improved reaction time during the Stroop test and reduced risk-taking behaviour during the IGT, and (3) these heat-induced cognitive changes did not differ according to participants’ self-reported physical activity levels.

Ethics statement

Ethical approval was obtained from the Singapore Institute of Technology Institutional Review Board (Ref no: RECAS-0118) and the conduct of this study adhered to the ethical principles outlined in the Declaration of Helsinki. Written informed consent was obtained from all participants.

Author contributions

FZ and RJG contributed to data collection, data analysis and interpretation, and initial drafting of the manuscript. ICCL, TAD, JKWL and SSC contributed to interpretation of study results, and critical revision of manuscript. PLC and XRT were involved in experimental conception and design, interpretation of study results, and critical revision of manuscript. All authors have read and approved the final version of the manuscript.

Declaration of generative AI and AI-assisted technologies in the writing process

The author(s) did not use generative AI technologies for preparation of this work.

Funding

Thomas Deshayes is financially supported by the Fonds de Recherche du Québec – Santé (FRQS).

Conflict of interest

The authors declared no conflicts of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

figs2.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.