Introduction

In recent years, brown adipose tissue (BAT) has garnered increasing attention as a potential therapeutic target for metabolic syndrome, obesity, and malignancies1,2,3. The ability of cold exposure to activate BAT presents new opportunities for developing healthcare and living environments tailored to specific patient populations. In addition to BAT 4, human adipose tissue also includes white adipose tissue (WAT), which serves very different physiological functions. WAT is primarily located in the chest, subcutaneous limbs, and around visceral organs, with its main function being the storage of excess energy as triglycerides5. BAT is mainly found in the head and chest areas and promotes non-shivering thermogenesis in response to core body temperature reduction6. BAT was previously believed to exist primarily in newborns and young children, disappearing in adulthood7. Recent studies, however, have found that BAT is present in adults and declines with age8,9,10.

The physiological activation of BAT has been extensively studied, including through cooling11, drugs12, diet13, and mental stress14. Cold exposure, known to trigger BAT activation through sympathetic stimulation, is one of the most studied methods15. Extreme cold exposure primarily induces heat production through shivering, with BAT activation contributing much less to overall energy expenditure16. At milder levels of cooling, BAT becomes the primary source of heat production. Most studies recommend an ambient temperature range of 17–25 °C during cold exposure, as lower ambient temperatures can induce shivering14,17,18,19,20. Previous studies have reported measurable changes in subclavicular (SCV) skin temperature within 5 min of cold stimulation, with further changes occurring after prolonged exposure, typically lasting around 2 h11,21,22. Although the environment affects the results, studies have been successfully conducted outside the laboratory, including in homes21 and schools23.

The benefits of BAT are increasingly recognized. Cold exposure activates mitochondrial uncoupling protein 1 (UCP1) in BAT, reducing plasma lipid and glucose levels by increasing heat expenditure through glucose and fatty acid thermogenesis24. Studies have shown that just 2 h of cold exposure can activate BAT and alter signaling lipid levels, thereby improving cardiometabolic health25. BAT activation in colder environments significantly improves insulin sensitivity, which is crucial for regulating glucose levels in type 2 diabetes patients26,27,28. BAT also plays a positive role in preventing obesity and endocrine disruption29,30. A recent study found that cold acclimation-induced BAT activation significantly inhibited the growth of various solid tumors, including fibrosarcoma, breast cancer, melanoma, and pancreatic cancer1. In contrast, tumor growth resumed under cold exposure following BAT removal, due to the absence of UCP1. This introduces a novel concept in cancer therapy, offering potential clinical benefits for cancer patients.

Numerous methods have been used experimentally to assess BAT activity. Currently, human BAT is assessed using 18F-FDG-PET/CT31,32,33, single-photon emission computed tomography (SPECT)34, magnetic resonance imaging (MRI)35, near-infrared spectroscopy (NIRS)36, and infrared thermography (IRT)37. In particular, 18F-FDG-PET/CT has long been considered the gold standard for imaging and measuring BAT in the human body. However, its application is mainly limited by exposure to relatively high levels of ionizing radiation (~ 8 mSv) and research on healthy individuals is ethically limited22. Notably, IRT is a novel, non-invasive, and cost-effective method for detecting mammalian BAT, with increasing validation in the literature5,14,17,19,21,37. Studies on both children and adults have validated the equivalence of IRT with 18F-FDG PET/CT in imaging human BAT38,39. A limitation of IRT is that it detects only skin temperatures, potentially underestimating deeper BAT thermogenic activity. However, the SCV area, the largest and most active BAT storage region, is located superficially beneath the subcutaneous adipose tissue in the lateral neck, making it well-suited for IRT40,41,42. Overall, IRT is an effective method for detecting BAT in adults.

Current research on human BAT has primarily focused on its health benefits and activation mechanisms. However, existing studies have largely emphasized the outcome of whether BAT is activated, with limited attention to the dynamic activation process and how it is jointly modulated by environmental temperature and exposure duration, especially under mildly cold conditions relevant to everyday settings. In addition, subjective thermal perception has yet to be adequately integrated into current research frameworks.

As cold exposure gains recognition as a potential non-pharmacological health intervention, its effectiveness and real-world applicability depend not only on physiological outcomes but also on behavioral and comfort-related factors. Therefore, it is essential to investigate how environmental parameters and individual thermal perception interact to influence BAT activity, from both objective physiological and subjective experiential perspectives.

This study systematically investigates how ambient temperature, exposure duration, and subjective thermal sensation interact to shape the dynamic activation of human BAT under controlled cold-exposure conditions, thereby providing empirical evidence for designing indoor environments that foster metabolic health. Specifically, it addresses two core scientific questions:

  • How do cold exposure intensity and duration jointly influence the dynamic activation of human BAT?

  • Can subjective thermal sensation serve as a regulatory variable to improve comfort and acceptance while maintaining effective BAT activation?

Materials and methods

Ethics procedures

The study protocol received approval from the Ethics Committee of Lanzhou University Second Hospital (approval No.: 2024A-1013), and all procedures adhered to the Declaration of Helsinki43. Participants provided written informed consent before the procedure commenced.

Study population

Measurements were conducted at a university hospital in Lanzhou, China. Human BAT activity varies with individual factors such as age, body mass, and other individual characteristics. BAT stores decline with age, resulting in lower BAT tissue volume in the elderly compared to younger individuals21,44. BAT activity is lower in obese individuals following cold exposure, indicating an inverse relationship between BAT activity and BMI45. Additionally, the thickness of subcutaneous adipose tissue influences skin temperature, reducing IRT’s sensitivity in detecting BAT46. No significant difference in BAT activity exists between men and women in low BMI or younger age groups47. Currently, most studies assessing BAT activity using IRT methods focus on regions of interest (ROIs) in the clavicular area23,48. Nirengi et al.49 proposed categorizing subjects into high BAT (HBAT) and low BAT (LBAT) groups based on whether the skin temperature difference of the subclavicular (SCV) and external (EXT) ROIs exceeded 1.0 °C.

Based on these findings, the inclusion criteria for volunteers in this study included: no hyperlipidemia, no hypothyroidism or hyperthyroidism, no smoking, no family history of type 2 diabetes or other chronic diseases, no medications affecting thermoregulation, age under 25 years, and being an adult male classified as HBAT. Exclusion criteria included: obesity and any prior surgeries in the clavicle, chest, or abdominal regions47.

Participants’ height and body mass were measured with the SECA255 scale, which combines the SECA217 height scale and the SECA878 weight scale. Body mass index (BMI) was calculated using the formula: body mass divided by height squared (kg/m2). Ultimately, 12 young men aged 19 to 25 years were enrolled in the study. The characteristics of the participants are summarized in Table 1.

Table 1 Baseline characteristics of subjects.
Full size table

Temperature and experimental site

In this study, we use the term “ambient temperature” to refer specifically to the controlled and measured air temperature inside the experimental room (Room B). In contrast, “environmental temperature” refers more broadly to the perceived thermal condition that encompasses air temperature, radiation, and other surrounding thermal effects on the participant. This distinction ensures clarity in describing both objective environmental settings and subjective thermal perceptions.

The experimental room, shown in Fig. 1, is equipped with a constant ambient temperature and humidity air-conditioning system that has a temperature accuracy of ± 0.5 °C and a range of 14–30 °C. The room is divided into two sections: a waiting area (Room A) and a climate area (Room B).

Fig. 1.
figure 1

Layout of the experimental site.

Full size image

The average outdoor temperature in Lanzhou during the study period ranged from 18.4 °C to 22.2 °C and the humidity ranged from 45 to 55%. Four indoor environmental parameters were measured: air temperature (Ta), air velocity (Va), relative humidity (RH) and black globe temperature (Tg). The range and accuracy of the instruments (Table 2) adhere to the ASHRAE-5550. The instrument was set next to the chair at a height of 1.1 m. Ta, RH and Tg values were recorded every minute while Va values were recorded every 30 s in groups of 10 min and mean values were taken. The average indoor Va was measured as 0.1 ± 0.04 m/s, RH as 50 ± 3.16%, and the Tg is on average 0.3 °C lower than Ta. Top can be determined using Eqs. (1) and (2):

$$T_{op} = {\text{A}} \times T_{a} + \left( {1 - {\text{A}}} \right) \times T_{r}$$
(1)
$$T_{r} = \left[ {\left( {t_{g} + 273} \right)^{4} + \frac{{1.1 \times 10^{8} \times v_{a}^{0.6} }}{{\varepsilon_{g} \times D^{{_{0.4} }} }}\left( {t_{g} - t_{a} } \right)} \right]^{\frac{1}{4}} - 273$$
(2)

where A is the weighting factor of air temperature (Ta) and mean radiant temperature (Tr), εg is the black sphere emissivity, and D (75 mm) is the diameter of the black sphere thermometer. In our study, the airflow rates at the measurement points were all less than 0.2 m/s, and A were assigned a value of 0.5.

Table 2 Instrumentation measurement range and accuracy.
Full size table

Subjects remained seated in both the waiting and test areas with a metabolic rate of 1 met50. All participants wore the same clothing provided by the researchers to ensure consistency. Wooden chairs with a thermal resistance of 0.01 clo were used, and clothing insulation was standardized at 0.6 clo50. Room ambient temperatures were calculated for PMV values of 0, − 1, − 2, and − 3 using the UCB (University of California, Berkeley) model51. After inputting the values of Va, RH, metabolic rate (Met), and total thermal resistance into the UCB model, the Top was adjusted to yield PMV values. Subsequently, the ambient temperatures was calculated using Eqs. (1) and (2). Table 3 presents the PMV values corresponding to various indoor ambient temperatures. The corresponding experimental ambient temperatures were 25.5 °C (Neutral), 22.5 °C (Slightly Cool), 19.5 °C (Cool), and 17 °C (Cold).

Table 3 Values of PMV at different ambient temperatures.
Full size table

To ensure consistent conditions when subjects move from Room A to Room B, the ambient temperature in Room A was set to the neutral temperature (25.5 °C) calculated by the PMV model52, while the ambient temperature in Room B was adjusted according to the experimental requirements. In the experiment, only the ambient temperature in Room B was varied, while parameters related to the thermal, acoustic, and light environments were kept constant.

Design

Participants were assessed on the morning of the study day. Normal sleep was ensured the night before, while alcohol, stimulants, body lotions, and medications affecting peripheral circulation were avoided for 24 h. Moderate or strenuous physical activity was restricted for 48 h before assessment. They did not wear accessories such as watches, bracelets, or rings. Participants were instructed to drink 1 L of water (45 ± 1 °C) within 30 min prior to entering Room A53,54. Emotional stress and environmental adaptation may influence BAT activation55,56,57. To minimize emotional stress, participants were given a 30-min quiet rest period in Room A prior to testing, allowing them to acclimate before entering Room B for the experiment. Each participant received procedural instructions from the same researcher to reduce anxiety and other negative emotions.

Subjects entered Room B after sitting Room A for 30 min, they were exposed to cold in all three ambient temperature conditions in Room B for 120 min. To minimize sequence-related effects, the order of the three ambient temperature exposures was randomized and counterbalanced across participants using a Latin square design. Each session was conducted on a separate day with at least 48 h between sessions to prevent thermal adaptation. This strategy follows the approach of Zhang et al.58, which emphasizes randomized scheduling and day-to-day consistency to ensure experimental reliability. Based on observations and verbal reports, subjects showed no symptoms or signs of shivering in the cold.

IRT images acquisition and BAT activity

Subjects remained seated with a matte surface on the back wall and wore same clothing so that the supraclavicular region was visible, and the thermal imaging camera was placed horizontally 1.0 m away from them (Fig. 1). The ROI thermal images for analysis included: (a) BAT active regions: skin temperature (Tsk) in the right (TSCVR) and left supraclavicular (TSCVL) regions, which have been consistently identified in prior studies as key anatomical depots of BAT in adults, particularly under cold exposure6,22; (b) control region: Tsk in the sternum (TSTR), which lacks significant BAT activity as confirmed by PET-CT and serves as a stable reference site47. To maximize infrared radiation detection, the thermal imaging camera was aligned perpendicularly to the ROI, and the ROI was kept as large as possible within the field of view47,59. The graphical output clearly displayed the identified hotspots.

After subjects entered room B, the first thermal image was taken 10 min later, followed by images every 10 min until the end of the 120th minute (Fig. 2). This interval was chosen based on previous studies demonstrating that supraclavicular skin temperature changes associated with BAT activation occur within the first 10–60 min of cold exposure, with clear detection using 10-min imaging intervals22,59. Moreover, BAT-related thermal signals tend to plateau between 60 and 120 min37, making this full duration critical for capturing both the activation and stabilization phases of BAT thermogenesis. This sampling frequency provides adequate temporal resolution for detecting BAT dynamics while balancing participant comfort and measurement feasibility.

Fig. 2
figure 2

Experimental protocol.

Full size image

A FLIR E8 camera (FLIR System, Inc., Wilsonville, OR) was used, mounted on a tripod, with a thermal resolution of 320 × 240 pixels, a 3-inch LCD screen, and an emissivity of 0.98 (corresponding to human skin emissivity). Before each image acquisition, the distance between the subject and the camera was confirmed to be 1 m to ensure data consistency. Thermal imaging followed the TISEM protocol53, using FLIR Tools software (version 6.4, FLIR Systems Inc., Wilsonville, OR).

When the human body is exposed to cold, it activates BAT to produce heat, and the skin temperature difference between the BAT-active area and a reference skin area can be used to measure the extent of BAT activation. BAT activity was calculated using the formula: BAT activity (°C) = TSCV (the mean value of the TSCVR and TSCVL) − TSTR38,59. To calculate clavicle and sternum skin temperatures based on Refs.60,61, a 2 cm radius circle was placed above the clavicle (on both the left and right sides), and in the area between the midpoints of the third and fourth rib notches, as the ROIs (Fig. 3). Software determined the hottest 10% of points within each ROI, and the median (equivalent to the 95th percentile) was calculated for these points14,21,23.

Fig. 3
figure 3

Schematic diagram of the regions of interest used for infrared thermographic analysis. Regions of interest: 1—TSCVR, 2—TSCVL, 3—TSTR.

Full size image

Questionnaire

Subjects entered room B and completed the Thermal Sensation Vote (TSV), Thermal Comfort Vote (TCV)62, and Thermal Acceptance Vote (TAV)51 questionnaires at 30-min intervals. The questionnaires are presented in Table 4.

Table 4 Thermal comfort and sensation votes.
Full size table

Statistical analysis

Statistical analyses were performed using SPSS® version 29.0.1.0 (IBM, Armonk, NY, USA). Descriptive statistics, including mean and standard deviation (SD), were used to summarize the participants’ characteristics and study variables. A repeated-measures design was adopted to account for inter-individual variability, as each subject underwent all three ambient temperature conditions (Cold, Cool, Slightly Cool). The Shapiro–Wilk test was used to verify normality of the data.

For normally distributed data, two-way repeated-measures ANOVA was conducted with two within-subject factors: (1) Ambient temperature (3 levels): Cold (17 °C), Cool (19.5 °C), Slightly Cool (22.5 °C); (2) Exposure time (4 levels): 30, 60, 90, and 120 min. Where significant main effects or interactions were observed, Bonferroni-corrected post hoc comparisons were performed to identify pairwise differences63. If the data failed to meet normality assumptions, non-parametric tests such as Friedman ANOVA and Wilcoxon signed-rank tests were employed64. Statistical significance was set at p < 0.05. All graphical outputs and confidence intervals were visually examined to support statistical interpretations.

In addition to significance testing, Cohen’s d effect sizes were calculated based on the raw ΔT values to assess the magnitude of differences between temperature conditions. The pooled standard deviation was used in the calculation, and effect sizes were interpreted according to established thresholds: small (d = 0.2), medium (d = 0.5), and large (d ≥ 0.8)65,66.

To evaluate the cumulative thermogenic response over the full exposure period, the area under the curve (AUC) of ΔT was calculated for each condition using the trapezoidal method across 12 time points (10–120 min). A one-way repeated-measures ANOVA was conducted to assess AUC differences among the three ambient temperatures, followed by Bonferroni-corrected post hoc tests for pairwise comparisons. The AUC provides an integrated measure of BAT activation over time and complements the pointwise statistical analyses.

Results

Changes in skin temperature over time in the ROI regions

Table 5 shows the average Tsk obtained on the SCV and STR ROIs for the Cold, Cool and Slightly Cool conditions. Considering the SCV ROI, skin temperature differences under varying cold exposure conditions primarily occurred between the ambient temperatures of Cold and Slightly Cool. There was a difference between the SCV-Slightly Cool group and the SCV-Cold and SCV-Cool groups at 10, 20, 40, 50 (p < 0.001), 30 (p ≤ 0.002) and 60 min (p ≤ 0.007). There was a difference for the SCV-Cold and SCV-Cool groups at 40 (p = 0.031), 50 (p < 0.001) and 60 min (p < 0.001). Furthermore, there was a significant difference for the SCV-Cold group and the SCV-Cool and SCV-Slightly Cool groups at 60–120 min (p < 0.001).

Table 5 Supraclavicular (SCV) and sternum (STR) skin temperature (°C) for cold, cool, and slightly cool condition.
Full size table

Considering the STR ROI, there was a difference between the STR-Slightly Cool group and the STR-Cold and STR-Cool groups at 10, 20, 30, 40, 50 (p < 0.001), 60 (p ≤ 0.002),70 (p ≤ 0.008) and 80 min (p ≤ 0.005). Significant differences were found for the STR-Cold and STR-Cool groups at 40 (p = 0.004), 50, 60, 70 and 80 min (p < 0.001). Finally, there was a difference between the STR-Cold group and the STR-Cool and STR-Slightly Cool groups at 90–120 min (p < 0.001).

Changes in BAT activity over time

Figure 4 shows the differences skin temperature (ΔT) between SCV and STR across the Cold, Cool, and Slightly Cool conditions. In the Cold condition, significant differences were observed between the 10 min and other measurement points (p < 0.001), except for 20 min (p = 0.266). The 20 min time point also differed from others (p < 0.001), except at 30 min (p = 0.015). At 30 min, differences were noted compared to the 80, 100, and 120 min measurements (p = 0.015).

Fig. 4
figure 4

BAT activity over time under different cold exposure conditions (Symbols a–f indicate within-group differences over time; symbols #, $, and & indicate between-group differences at the same time point. cp < 0.001 of this measurementmoment vs. the others for the Cold condition. dp ≤ 0.03 of this measurementmoment vs. the others for the Cool condition. fp < 0.001 of this measurementmoment vs. the others except at 20 min (p ≤ 0.04) for the Slightly Cool condition. ep < 0.001 of this measurementmoment vs. the others except at 10 min for the Slightly Cool condition. ap ≤ 0.019 of this measurementmoment vs. the others for the Slightly Cool condition. #p ≤ 0.03 cold vs. Cool and Slightly Cool condition for the same measurementmomen. $p ≤ 0.006 Cool vs. Cold and Slightly Cool condition for the same measurementmomen. &p ≤ 0.031 Slightly Cool vs. Cold and Cool condition for the same measurementmomen).

Full size image

In the Cool condition, significant differences were found between the 10 min and other time points, including 30 (p = 0.002), 40 (p = 0.001), 50 (p = 0.03), 60 (p = 0.018), 70 (p = 0.001), and 80–120 min (p < 0.001). The 20 min point differed significantly from 80–120 min (p ≤ 0.011).

For the Slightly Cool condition, BAT activation varied significantly over time. Significant differences were observed between the 10-min mark and most subsequent time points (p < 0.001), except for 20 min (p = 0.04). Additionally, the 20-min measurement differed from most other time points (p < 0.001), and 30 min showed significant differences from 90–120 min (p < 0.001). The 90-min point differed from earlier measurements (10–70 min; p ≤ 0.028), while 110 min showed significant differences from all time points except 90, 100, and 120 min (p > 0.05). Finally, 120 min differed from most earlier points (p < 0.001), but not from 90–110 min (p > 0.05), suggesting stabilization of BAT activity after approximately 90 min.

In the between-group comparison under different ambient temperatures, significant differences were observed between the Slightly Cool condition and both the Cold and Cool conditions at several time points: 10, 20, 30, 40, 70, 80, and 100 min (p ≤ 0.031). The Cold condition showed greater BAT activation than the Cool and Slightly Cool groups at 50, 60, 70, and 90 min (p ≤ 0.03). Notably, all three groups differed significantly at 70 min (p ≤ 0.006). However, by 100 and 120 min, no significant differences in BAT activity were observed among the three groups (p > 0.05).

These results indicate that BAT activation under colder conditions is more rapid and pronounced during the first 40 min, particularly when comparing the Cold and Slightly Cool environments. Between 50 and 100 min, the Cold condition maintained higher BAT activity, supporting the temperature-dependence of activation. However, beyond 110 min, BAT activity tended to plateau, and the effect of ambient temperature diminished, with no significant differences observed between groups. This suggests a convergence in thermogenic response over time regardless of initial temperature differences.

Table 6 show the full effect size results. According to the effect size analysis (Cohen’s d), the ΔT differences among the three ambient temperature conditions demonstrated moderate to large effect sizes across multiple time points, particularly between the Cold and Slightly Cool conditions. Across all 12 time points (10–120 min), nearly all comparisons reached the threshold for a large effect (d ≥ 0.8), with several exceeding d = 1.5. The largest observed effect size was at 10 min (d = 2.01). Even at later time points (e.g., 120 min), when statistical differences weakened, effect sizes remained large (d = 0.89), suggesting a persistent influence of ambient temperature on BAT activation. Additionally, consistent moderate effect sizes (d ≈ 0.5–0.9) were found between the Cold and Cool conditions, as well as between the Cool and Slightly Cool conditions, indicating clear distinctions in BAT activation levels across cold intensities.

Table 6 ΔT effect sizes (Cohen’s d) between temperature conditions across time points.
Full size table

Table 7 presents the AUC of ΔT across the 120-min exposure period under different ambient temperature conditions. Descriptive statistics showed that AUC was highest under the Cold condition (17 °C: 112.6 ± 8.9), followed by Cool (19.5 °C: 97.7 ± 6.7) and Slightly Cool (22.5 °C: 83.6 ± 8.2). A one-way repeated-measures ANOVA revealed a significant effect of ambient temperature on AUC (F(2, 22) = 46.17, p < 0.001), with Bonferroni-corrected post hoc tests confirming significantly higher AUC under Cold vs. Cool (p = 0.005) and Slightly Cool (p < 0.001), as well as Cool vs. Slightly Cool (p < 0.002). These results suggest that lower temperatures not only accelerate early BAT activation but also sustain greater cumulative thermogenesis over time.

Table 7 Mean ± standard deviation of ΔT AUC values (°C·min) across three ambient temperature conditions.
Full size table

BAT activity and thermal sensation indicators under different exposure times and temperatures

Variations in each indicator over the course of exposure time

Figure 5 depicts the changes in BAT activity, TSV, TCV, and TAV with exposure time under the three ambient temperature conditions. BAT activity increases proportionally with longer exposure time, while subjective thermal comfort decreases as exposure time extends. At 22.5 °C, BAT activity increased from 0.7 at 30 min of exposure to 0.78 at 60 min (an 11.4% increase), 0.89 at 90 min (a 14.1% increase), and 0.97 at 120 min (a 9% increase). All thermal sensation indicators remained within the comfort range.

Fig. 5
figure 5

Trends in TSV, TCV, TAV, and BAT activity at different exposure times (the green shaded area in the figure, as indicated by the survey in Table 4, represents a slightly uncomfortable but acceptable range, which is defined as the comfort range in this study. This figure reflects time-dependent thermal physiological and perceptual changes, across uniform ambient temperature conditions).

Full size image

At an ambient temperature of 19.5 °C, BAT activity was 0.8 at 30 min, 0.84 at 60 min (a 5% increase), 0.95 at 90 min (a 13.1% increase), and 1.03 at 120 min (an 8.4% increase). After 30 min of exposure, individuals experienced a Slightly Cool TSV (− 1), a comfortable TCV (− 0.3), and an acceptable TAV (− 0.5). After 90 min of exposure, individuals’ TAV (− 1.1) slightly exceeded the unacceptable threshold.

At 17 °C, BAT activity reached 0.93 after 30 min of exposure, increased to 1.01 after 60 min (an 8.6% increase), reached 1.04 at 90 min (a 3% increase), and further increased to 1.11 at 120 min (a 6.7% increase). After 60 min of exposure, TAV (-0.9) remained within acceptable limits, but both TSV (− 1.8) and TCV (− 1.2) exceeded the comfort range. All subjective indicators exceeded the acceptable range at later time points.

The results indicate that at a constant cold exposure ambient temperature, BAT activity rises while thermal sensory indices decline as exposure time increases. In a cold environment, metabolic heat is transferred from the body core to the skin primarily through conduction and convection via blood flow, while heat is then lost from the skin surface to the environment through radiation, convection, and evaporation. At this stage, the body activates BAT to promote non-shivering thermogenesis and counteract core temperature loss67. However, the primary thermogenic mechanism-muscle shivering-was not present in this experiment, and the heat produced by BAT was unable to maintain body thermal balance. Meanwhile, according to the human body thermal balance equation68, the S-value (heat storage rate) is negative, and cumulative heat loss increases over time. Consequently, subjects feel progressively colder, reducing thermal comfort.

Under cold exposure, the time factor has a greater impact on BAT activity in environments perceived as Slightly Cool or Cool (e.g., 22 °C and 19 °C). In colder environments (e.g., 17 °C), body heat loss and BAT activation occur more rapidly. Thus, in Slightly Cool or Cool conditions, BAT activation is slower but increases more uniformly over time. In contrast, at lower ambient temperatures, BAT activates rapidly during early exposure, followed by a slower increase in activity as exposure time continues.

Impact of temperature conditions on BAT activity and thermal sensation

Figure 6 depicts the changes in BAT activity, TSV, TCV, and TAV in relation to ambient temperature (PMV) at a constant exposure time. After 30 min of exposure to the three ambient temperature conditions, the maximum difference in BAT activity was 0.23 (Cold vs. Slightly Cool) and the minimum difference was 0.1 (Cool vs. Slightly Cool). Additionally, both TCV and TAV remained within the comfort zone, while only the TSV in the Cold condition fell below the comfort range (-1.5).

Fig. 6
figure 6

Changes in BAT activity, TSV, TCV, and TAV with ambient temperature (PMV). Over the same time period (this figure isolates the influence of ambient temperature while holding exposure time constant).

Full size image

After 60 min of exposure, the maximum difference in BAT activity was 0.23 (Cold vs. Slightly Cool) and the minimum was 0.06 (Cold vs. Cool). At this point, only the TSV in the Cool condition and the TCV in the Cold condition exceeded the comfort zone, while the TAV remained within the acceptable range.

After 90 min of exposure, the effect of ambient temperature on BAT activity diminished, showing a maximum difference of 0.11 (Cold vs. Slightly Cool) and a minimum difference of 0.05 (Cold vs. Cool) across the three ambient temperature conditions. Similarly, at 120 min, the maximum difference in BAT activity was 0.14 (Cold vs. Slightly Cool) and the minimum was 0.06 (Cool vs. Slightly Cool). During both exposure periods, thermal sensory indices exceeded the comfort zone in all Cold and Cool conditions, except for TCV in the Cold condition.

Ambient temperature plays a crucial role in activating BAT in humans, but its effect on BAT activity varies with exposure time. Lower ambient temperatures lead to stronger BAT activity during shorter exposure times, but the impact of ambient temperature on BAT activity diminishes after 90 min of exposure. Analysis of the subjective thermal sensation voting revealed that individuals had high acceptance of the ambient temperature and reported no discomfort even at lower ambient temperatures during shorter exposure times. However, as exposure time increased, acceptance and comfort decreased due to body heat loss, with this process accelerating at lower ambient temperatures. After 90 min of exposure, subjects in the Cool and Cold conditions reported discomfort from hypothermia and exhibited progressively lower tolerance for the ambient temperature.

As summarised in Table 8, participants in the Slightly Cool condition consistently reported higher thermal comfort and acceptability than those in the Cool and Cold conditions. All three indices decreased progressively as ambient temperature fell, reflecting greater discomfort and lower acceptability in colder environments.

Table 8 Statistical comparison of TSV, TCV, and TAV at four time points under three ambient temperature conditions.
Full size table

Significant differences were observed in TSV, TCV, and TAV across the three thermal conditions at several time points (p < 0.05). Pairwise comparisons were further examined with Mann–Whitney U tests (Table 9). Most contrasts between Cold and Slightly Cool were highly significant (p < 0.001), particularly for TCV and TAV. Differences between Cool and Cold were less uniform, whereas comparisons between Slightly Cool and Cool still showed significant early-stage differences in TSV and TCV. Collectively, these results underscore the advantage of a Slightly Cool condition in sustaining acceptable thermal perception during prolonged cold exposure.

Table 9. Group comparisons of TSV, TCV, and TAV among three conditions at four time points (Mann–Whitney U test).
Full size table

Change of TAV, TCV and BAT activity with TSV

Figure 7 illustrates the variation of BAT activity, TAV, and TCV with TSV. The thermal sensations corresponding to TSV ranges of [− 3, − 2], [− 2, − 1], and [− 1, 0] are classified as ‘Cold’, ‘Cool’, and ‘Slightly Cool’, respectively. After 30 min of exposure, a significant difference in BAT activity was observed between the ‘Cold’ and ‘Slightly Cool’ conditions. At this stage, individuals reported higher satisfaction and acceptance of the exposure environment.

Fig. 7
figure 7

Changes in BAT activity, TCV, and TAV with TSV over the same time period (this figure illustrates how thermal responses vary across subjective thermal sensation categories, regardless of actual ambient temperature).

Full size image

At 60 min of exposure, the values of BAT activity, TAV, and TCV for each of the three cold sensory conditions remained relatively stable, except for a significant decrease in individual comfort in the ‘cold’ condition. BAT activity significantly increased at 90 min of exposure. In the ‘cold’ condition, TAV and TCV exceeded the comfort range, leading to a gradual decline in individual acceptance of the ambient temperature.

BAT activity progressively increased with exposure time, reaching a peak at 120 min, underscoring the cumulative and time-dependent nature of cold-induced thermogenesis. In environments categorized as “Slightly Cool”, BAT activation displayed a delayed onset but a sustained upward trend, ultimately attaining levels comparable to those observed under “Cold” conditions. Importantly, no significant difference was found in final BAT activation between the “Cool” and “Slightly Cool” conditions.

From the perspective of thermal perception, both TAV and TCV remained within the comfort zone throughout the exposure period in the “Slightly Cool” condition. In contrast, these values exceeded comfort thresholds within the first 30 min under both the “cool” and “cold” conditions, indicating reduced thermal tolerance despite higher BAT activation rates.

Collectively, these findings suggest that while lower ambient temperatures and colder thermal sensations may elicit faster BAT activation, extended exposure in a “Slightly Cool” environment can produce comparable cumulative thermogenic effects with significantly better comfort and acceptability. This supports the design of cold exposure protocols that balance physiological effectiveness with user tolerability, which is particularly relevant for long-duration or clinical applications.

Discussion

This study investigated the effects of ambient temperature and exposure duration on human BAT activation and examined the relationship between thermal sensation and thermogenic responses. Our findings demonstrate that both environmental temperature and exposure time significantly influence BAT activation, and that subjective thermal perception closely tracks thermogenic changes.

The physiological mechanisms underlying cold-induced BAT activation are well established: upon cold exposure, thermoreceptors relay signals to the central nervous system, activating sympathetic pathways that stimulate non-shivering thermogenesis via norepinephrine and UCP111,69. Previous studies have shown that BAT activity peaks around 30 min into cold exposure61,70, consistent with our observation of rapid ΔT increases during the initial phase. The sustained SCV–STR temperature gradient observed in our study reflects BAT-mediated thermogenesis and aligns with prior infrared imaging research17,21,32,38. The STR region, which was equally exposed and insulated, served as a reliable control by minimizing confounding effects from vasodilation or clothing insulation.

Subjective thermal responses (TSV, TCV, TAV) varied significantly across thermal conditions, particularly within the first 60–90 min. While PMV provides an objective measure of environmental thermal stress, subjective indicators better capture real-time individual thermoregulatory responses71,72. Studies have shown that PMV often underestimates thermal perception, particularly in warm climates73,74,75,76, further underscoring the value of subjective measures in thermal research.

We observed that thermal perception and BAT activation followed similar temporal patterns, with colder environments inducing more pronounced early responses. Pairwise comparisons (Table 8) revealed significant differences—especially between Cold and Slightly Cool conditions (p < 0.001), most notably in TCV and TAV scores. Although the Slightly Cool condition was associated with slower initial BAT activation, thermogenesis increased steadily, and ΔT levels at 120 min were statistically comparable to those under colder conditions.

These findings indicate that BAT activation is both temperature-dependent and time-sensitive. Large effect sizes (Cohen’s d ≥ 0.8) between Cold and Slightly Cool conditions at multiple time points highlight the physiological significance of these differences. The observed threshold-like activation pattern suggests that lower temperatures rapidly trigger BAT, while Slightly Cool environments elicit a delayed but sustained thermogenic response. This was confirmed by AUC analysis, which showed that cumulative BAT activation under Slightly Cool conditions approached levels observed in the Cold condition by the end of 120 min.

Importantly, the Slightly Cool condition was associated with significantly higher thermal comfort and acceptability throughout the exposure period. Although colder environments accelerated early BAT activation, they were linked to lower TCV and TAV scores, indicating decreased tolerability over time. These results suggest that mild cold exposure can balance effective metabolic stimulation with subjective comfort, which is critical for long-term adherence and real-world application. Previous studies have also reported that repeated or daily mild cold exposure improves BAT activity and metabolic adaptation16,77,78,79.

Our findings suggest that subjective thermal sensation may serve as a practical indicator of thermogenic state, and that prolonged exposure to a Slightly Cool environment (approximately 22.5 °C) for 120 min can effectively activate BAT while maintaining comfort. From a translational perspective, this strategy could be applied in clinical and residential settings (such as wintertime thermal regulation in hospital wards or nursing homes) to support metabolic health in populations with diabetes or obesity. Moreover, this approach aligns with energy conservation goals by reducing the need for intensive indoor heating.

Limitations and future research

Despite offering novel insights into designing healthy thermal environments, our study has some limitations.

Firstly, our sample size was limited, and we imposed strict restrictions when selecting subjects. In particular, obese participants were excluded because the thickness of subcutaneous adipose tissue in the SCV may affect the quality of thermal imaging46. Therefore, it is necessary to expand the range of subjects to include gender, age, and physical condition.

Secondly, the exposure time in our study was short. Cold acclimatization occurs in the human body after prolonged cold exposure, at which time individuals have different psychological and physiological responses to cold exposure stimuli80,81. Moreover, there are changes in human BAT activity after prolonged cold exposure. Therefore, it is worthwhile to investigate how the effect of ambient temperature on BAT activity differs in individuals after acclimatization to a cold environment.

Finally, a limitation of our study was the absence of biochemical measurements, such as UCP-1, a key indicator of BAT activity. Future studies could incorporate some biochemical measurements related to BAT activity.

Conclusion

To investigate the effects of cold environments and exposure times on BAT activity, TSV, TCV, and TAV in humans, we set up three cold exposure environments and four exposure times based on objective thermal indicators. The study yielded the following key findings.

  • BAT activity increased rapidly during the initial stage of cold exposure and continued to rise gradually over time.

  • Ambient temperature significantly influenced BAT activation within the first 30 min of exposure, with colder environments inducing a faster response. However, as exposure time increased, the differences between temperature conditions diminished and became statistically insignificant after approximately 110 min.

  • Subjective thermal sensation showed a strong temporal association with BAT activation. Variability in TSV, TCV, and TAV across conditions supports the role of psychological perception in modulating thermogenic responses.

  • A Slightly Cool environment may offer a practical balance between effective BAT activation and thermal comfort. Although activation occurred more slowly, the cumulative ΔT increase under this condition approached levels observed in colder environments while maintaining higher acceptability.

  • Given that thermal sensation varies based on age, gender, and individual physiology, using subjective perception as a reference for thermal environment design may be more appropriate than relying solely on ambient temperature.

  • These conclusions are based on a sample of young adult males with high BAT sensitivity and may not directly apply to other populations. Future studies involving more diverse groups and physiological markers are warranted.

This study involved a limited sample of healthy young males with presumed high BAT responsiveness, and the findings should therefore be interpreted with caution. Given that BAT activity and thermal perception vary with age, sex, and metabolic status, further research involving more diverse populations is warranted to validate and extend these findings.