Exploring internal body heat balance to understand thermal sensation

ABSTRACT

A biological perspective is used to understand thermal sensation. The main premise is that thermal sensation serves an organism for the regulation of body temperature. A biological concept related to this premise is the physiological thermoneutral zone (TNZ). Within the TNZ the body can adjust body tissue insulation to maintain thermal balance and a stable core temperature. The approach presented here is based on the assumption that humans express neutral thermal sensation near the centre of their TNZ. To test this hypothesis, dTNZ_op is defined as the distance between measured operative temperature and the centre of the TNZ, and dTNZ_sk as the distance between measured mean skin temperature and the centre of the TNZ. The TNZ centre is calculated with a biophysical model using measured data from a climate chamber study with 16 female subjects. Regression between observed thermal sensation votes (TSV) and dTNZ_x revealed that the intercept corresponds with a slightly higher-than-neutral TSV and a strong linear relationship between TSV and dTNZ_op and dTNZ_sk. This approach shows great potential to improve the understanding of human thermal sensation in the context of physiology.

KEYWORDS:

• biology
• homeostasis
• indoor environmental quality
• physiology
• thermal comfort
• thermal perception
• thermal sensation
• thermoregulatory behaviour

View correction statement:

Correction

Introduction

The aim of this paper is to propose a biophysical model to understand thermal sensation from the perspective of internal body heat balance. The main premise is that thermal sensation (as a driver for behaviour) can serve an organism in homeostasis and allostasis of body temperature. Homeostasis concerns maintenance of a constant milieu intérieur of the body and allostasis is the process of achieving internal balance through physiological or behavioural change (Ramsay & Woods, 2014).

Fundamental works by John B. Pierce Foundation fellows have shown that thermal sensation can be understood from the perspective of heat balance between the body and its thermal environment (Gagge, Nishi, & Gonzalez, 1972; Gagge, Stolwijk, & Hardy, 1967). Their results indicate that averaged thermal sensation is linearly related to averaged mean skin temperature in cool conditions. However, in warm conditions this relation is less apparent, but dominated by skin wettedness. The explanation for this discrepancy is that in colder environments body heat exchange is dominated by sensible heat loss (i.e. conduction, convection and radiation), whereas in warmer environments heat exchange is dominated by latent heat loss (i.e. evaporation) (Gagge et al., 1972). The physical fundamentals that underlie sensible and latent heat loss are included in the standard effective temperature (SET) and John B. Pierce Foundation fellows have shown that thermal sensation scales linearly with SET both in cold and warm environments (Gagge et al., 1972).

The SET thus provides an approach to understand thermal sensation from the perspective of heat balance between the body and the environment. This paper investigates whether the same principles hold from the perspective of internal heat balance between the body core and skin tissues. The reason for this approach is that the body can only measure its own tissues, and cannot measure the surrounding operative temperature or relative humidity directly.

Body heat generated by tissues is transported to the environment by conduction through tissues and convection through blood flow (Pennes, 1948) (Figure 1). In a cold environment the body reduces blood flow to limbs and skin through vasoconstriction, which increases tissue insulation (Burton & Edholm, 1955; deGroot & Kenney, 2007). Vice versa, in a warm environment the body increases blood flow through vasodilation, which decreases body tissue insulation. The range of tissue insulation a lean healthy body can provide is between 0.032 m²K/W (i.e. maximum vasodilation) and 0.122 m²K/W (i.e. maximum vasoconstriction) (Burton & Edholm, 1955; Veicsteinas, Ferretti, & Rennie, 1982). It may thus be that tissue insulation required to support internal heat balance is related to thermal sensation.

Figure 1. Schematic representation of a biophysical heat balance model. From left to right, heat balance is satisfied when metabolic heat production equals heat loss. The temperature gradient between the core and skin temperature is determined by metabolic rate and tissue insulation. Likewise, the temperature gradient between skin and air temperature is determined by heat loss and clothing insulation. Note that metabolic heat production also occurs in tissues, not only in the body core. Source: Kingma and van Marken Lichtenbelt (2015).

In the biological sciences, thermal sensation is considered to drive thermoregulatory behaviour for energy conservation in mammals (Schlader, Simmons, Stannard, & Mundel, 2011). Mammals grow and reproduce from the food and water they can obtain. To maximize these, it is advantageous to spend as little energy and water as possible on the maintenance of body core temperature (Porter, 2001; Scholander, Hock, Walters, & Irving, 1950; Speakman & Krol, 2010). Mammals can do this by maximizing the time spent in the thermoneutral zone (TNZ), which is defined as the range of operative temperatures where the body can maintain body core temperature without any regulatory changes to metabolic heat production or sweating (IUPS, 2001) (Figure 2).

Figure 2. The thermoneutral zone (TNZ) is the range of operative temperatures associated with basal metabolic rate required to support life functions, and minimal water loss. The range of operative temperatures should be interpreted for a given constant relative humidity. Below the lower critical temperature (LCT), metabolic rate increases to maintain body core temperature. Above the upper critical temperature (UCT), water loss increases due to regulatory sweating and may coincide with heat-induced thermogenesis (e.g. Q10 effect: metabolic rate scales with tissue temperature according to Arrhenius law).

Within the TNZ the body adapts heat loss and body tissue insulation via the regulation of skin blood flow. This is also a thermoregulatory measure, but is not associated with increased metabolic rate or water loss. The cooler end of the TNZ corresponds to decreased skin blood flow and colder skin temperatures; vice versa for the warmer end of the TNZ (see the grey area in Figure 3(a)). For nude men the air temperature of the TNZ has been reported to range between 26 and 32°C, with corresponding mean skin temperature between 33.5 and 34.5°C (Erikson, Krog, Andersen, & Scholander, 1956; Hardy & Dubois, 1937). For men clothed in winter attire the lower end of the TNZ is reported to decrease to about 14°C (Erikson et al., 1956). Outside the TNZ, the body core temperature can be maintained, e.g. by shivering on the cold side or sweating on the warm side. However, this increases energy expenditure and for sweating the water loss as described below.

Figure 3. The thermoneutral zone (TNZ) is depicted with the grey area and represents the combinations of skin temperature and operative temperature for which heat loss equals basal metabolic rate (corrected for respiratory heat loss) and core temperature is in a normal range (e.g. 36.5–37.5°C): (a) dependence of TNZ position on clothing, metabolism, tissue insulation (i.e. body fat, muscle and skin blood flow), air speed and relative humidity (adapted from Kingma & van Marken Lichtenbelt, 2015); (b) TNZ of two individuals; the operative temperature is in the centre of the TNZ for person A, which corresponds to a neutral thermal sensation; the operative temperature is at a lower operative temperature than the TNZ centre for person B, which corresponds to a cold thermal sensation for person B; (c) current state of an individual relative to the TNZ centre expressed as operative temperature distance to the TNZ centroid (dTNZ_op; see the link to (d)); and (d) the hypothesis of this paper: thermal sensation versus dTNZ: neutral thermal sensation occurs when the distance to the TNZ centroid is zero; a positive distance relates to a warm sensation (see the link to (c)).

Staying within the TNZ reduces the chemical energy required to maintain body temperature because in cold environments energy expenditure is increased by (non-)shivering thermogenesis, and in hot environments energy expenditure increases because the body needs to work harder to transport excess heat away from core tissues, and because warmer tissues have a higher metabolic rate (i.e. Q₁₀ effect: metabolic rate scales with tissue temperature) (Burton & Edholm, 1955). Relative to resting, the metabolic rate has been reported to increase by approximately 10% due to increased temperature (Cannon & Keatinge, 1960; Gagge et al., 1967), but may be compensated by lower physical activity and relaxation (Flouris & Schlader, 2015). Furthermore, in a hot environment water loss increases due to sweating. Thus, the centre of the TNZ may be considered as the safest state for an animal from an energetic and hydration perspective: it provides most internal flexibility to cope with future thermal challenges at the minimal cost of nutrients. Therefore, the approach presented here is based on the assumption that humans, as mammals, express a neutral thermal sensation near the centre of their physiological TNZ.

It should be noted that the exact positioning of the TNZ depends on air velocity, clothing, metabolism and tissue insulation (e.g. body fat, muscle tissue and skin blood flow) (deGroot, Havenith, & Kenney, 2006; Kingma & van Marken Lichtenbelt, 2015; Rennie, 1988). As shown in Figure 3(b), individual differences in the aforementioned parameters can therefore explain why a specific operative temperature is perceived to be as neutral for one individual (person A in Figure 3(b)) and cold or warm for another (person B in Figure 3(b)). In addition, note that earlier attempts have been made to include body tissue insulation for simple steady-state models of thermal comfort (Humphreys, 1970).

Temperature-sensitive neurons populate the body skin and core tissues (Benzinger, 1969). The body may use the information provided by these neurons as a heuristic to assess its thermal state. It is hypothesized that this is done by evaluating the distance between the actual mean skin temperature and the skin temperature corresponding to the centre of the TNZ or the TNZ centroid (see dTNZ_sk in Figure 3(c)). Parallel to the body, temperature-sensitive devices can be used to measure the skin temperature of individuals and derive the distance to their TNZ. As can be seen in Figure 3(a), skin and operative temperatures are closely related within the steady-state TNZ. Therefore, it is also possible to derive the operative temperature distance between that corresponding to the TNZ centroid of a person and the actual operative temperature of the room (dTNZ_op) (Figure 3(c)).

In this paper, the hypotheses are tested that:

• neutral thermal sensation corresponds to zero distance from the TNZ centroid, expressed as (a) dTNZ_sk and (b) dTNZ_op

• thermal sensation and the distance from the TNZ centroid are linearly correlated, expressed as (a) dTNZ_sk and (b) dTNZ_op.

A graphical representation of the model and the hypothesis is shown in Figures 3(c, d) respectively.

Methods

The analysis in this study is performed by reusing data from a study on thermal sensation in young adult females (Jacquot, Schellen, Kingma, van Baak, & van Marken Lichtenbelt, 2014). The participant characteristics and study procedure are shortly summarized; for details see the aforementioned papers.

Participants and protocol

Participants consisted of 16 healthy young adult females (age: 23 ± 4 years, height: 1.69 ± 0.06 m, weight: 66 ± 8 kg) who were lightly clothed (underwear, cotton/polyester sweatpants, sport socks and cotton t-shirt) while performing light office work (mainly studying).

All measurements took place between May and August 2012 in a climate-controlled respiration chamber without connection to the outdoor environment. A respiration chamber is an indirect calorimeter, i.e. it measures the heat production of the person in the room by the production of CO₂ and consumption of O₂. All participants were exposed to two temperature drifts in random order: the cooling and warming protocols. Each protocol starts with a 45-min baseline at 24°C, followed by 120 min at 4 K/h cooling or warming. All participants had a minimum of one day and a maximum of one month between the two protocols.

Thermal sensation was assessed every 15 minutes on a continuous seven-section ASHRAE thermal sensation interval scale, which ranged from –3 cold to 3 hot.

Operative temperature and relative humidity were measured at 0.7 and 1.1 m height using wireless sensors (iButton, DS1923, Maxim Integrated Products, San Jose, California, USA). Air velocity was measured by a portable anemometer (Testo 415, Testo BV, Almere, the Netherlands). Skin temperature was measured with iButtons (DS1922L) at the 14 positions as described by ISO 9886 standard (ISO, 2004).

Energy expenditure (metabolic rate) was measured by indirect calorimetry (Maastricht Instruments, Maastricht, the Netherlands), and recordings of baseline CO₂ production and O₂ uptake were converted into their heat equivalent using the Weir equation (Weir, 1949). Body surface area was calculated based on subjects’ height measured using a stadiometer (Seca, Hamburg, Germany) and circumferences of the different body components and limbs (head, neck, chest, abdomen, bottom and from the right sight of the body: upper arm, lower arm, hand, upper leg, lower leg and foot) using a circumference measuring tape (Seca).

Biophysical model

The theoretical centre of the TNZ is calculated with a biophysical model (Kingma et al., 2014a) (see Figure 3(c) for an example of the calculated TNZ and its centre). Note that the applied model considers the entire range of tissue insulation that is biologically possible (including modest variation in core temperature and metabolic rate). This is different than traditional thermophysiological models that predict a single value of tissue insulation for a given thermal state (e.g. the thermoregulatory control modules of Fiala, Lomas, and Stohrer (2001), Gagge (1973), Stolwijk (1971), Tanabe, Kobayashi, Nakano, Ozeki, and Konishi (2002), and Wissler (1971). This means that with the model applied in this paper all possible heat balance solutions are covered. The latter allows for calculating the centre of all possible heat-balance solutions, while the body only adjusts tissue insulation.

To calculate body core temperature, the model assumes a steady-state heat balance within the body and between the body and its environment (Figure 2). Model variables are given in Table 1.

Table 1. Model variables and values used in the biophysical model for this paper.

Description	Symbol	Value or range	Unit
Metabolic rate	M	46–50^a	W/m^2
Respiratory heat loss fraction	arsp	0.08	Fraction
Tissue insulation	Ibody	0.032 (minimum)–0.112 (maximum)	m^2W/K
Body surface area	A	1.88^a	m^2
Body core temperature	Tc	36.5–37.5	°C
Clothing insulation	Icl	0.1054	m^2W/K = (0.68 clo)
Air velocity	Vair	0.09^a	m/s
Relative humidity	Rh	0.5^a	Fraction
Lewis relation	λ	2.2	°C/mmHg
Skin wettedness	w	0.06	Fraction

Note: ^aBased on measurements during the experimental study.

The mathematical procedure is as follows:

1. Define the variables as given in Table 1.

2. Correct the metabolic rate for respiratory heat loss, such that only heat transport from the body core to skin tissues is considered:$M_{\min }=(1-a_{\mathrm{rsp}})\times M_{\min }$$M_{\max }=(1-a_{\mathrm{rsp}})\times M_{\max }$

3. Calculate the minimal and maximal skin temperatures (Ts_min, Ts_max) that support the internal heat balance:$\text{T}{\text{s}}_{\min }=\text{T}{\text{c}}_{\min }-M_{\max }\times I_{\mathrm{body},\max }$$T{s}_{\max }=T{c}_{\max }-M_{\min }\times I_{\mathrm{body},\min }$

4. Define a 200 × 200 matrix for mean skin temperature (T_sk) and operative temperature (T_op):$T_{\mathrm{sk}}\text{between}{28}^{\circ }\text{C}\text{and}{38}^{\circ }\text{C}$$T_{\mathrm{op}}\text{between}{14}^{\circ }\text{C}\text{and}{32}^{\circ }\text{C}$

5. Calculate T_c for each combination of T_sk and T_op in the above-defined matrix according to Kingma et al. (2014b):$T_c=\left({\displaystyle \frac{I_{\mathrm{body}}}{(1-a_{\mathrm{rsp}})}}\right)\times \left(\right(\left({\displaystyle T_{\mathrm{sk}}-T_{\mathrm{op}}(I_{\mathrm{cl}}+I_{\mathrm{op}}))+Q_{\mathrm{e}}}\right)+T_{\mathrm{sk}}$where Q_e is evaporative heat loss; and I_op is operative insulation (the inverse of the operative heat transfer coefficient); for more details, see the above-mentioned paper. The result is a 200 × 200 matrix with values of T_c.

6. Calculate the heat loss (Q_out) for each combination of T_sk and T_op in the matrix according to Kingma et al. resulting in a 200 × 200 matrix with values of Q_out:$Q_{\mathrm{o}\mathrm{u}\mathrm{t}}=A\times \left(\right(\left({\displaystyle T_{\mathrm{s}\mathrm{k}}-T_{\mathrm{o}\mathrm{p}}(I_{\mathrm{c}\mathrm{l}}+I_{\mathrm{o}\mathrm{p}}))+Q_{\mathrm{e}}}\right)$

7. The TNZ is obtained by filtering out those points of the resulting two matrices with values of T_c and Q_out for which T_c is within core temperature bounds (Table 1), and Q_out is within metabolic rate bounds (corrected for respiratory heat loss, as described in step 2).

8. The operative temperature centre of the thermoneutral zone (TNZcenter_op) is defined as the horizontal average of the TNZ. Accordingly, the skin temperature centre of the thermoneutral zone (TNZcenter_sk) is defined as the vertical average of the TNZ.

9. dTNZ_sk is defined as the difference between the observed mean skin temperature at the time of the vote and the calculated TNZcenter_sk and is thus calculated by:$\text{dTN}{\text{Z}}_{\mathrm{sk}}=T_{\mathrm{sk}}-\text{TNZcente}{\text{r}}_{\mathrm{sk}}$

10. dTNZ_op is defined as the difference between the observed operative temperature at the time of the vote and the calculated TNZcenter_op and is thus calculated by:$\text{dTN}{\text{Z}}_{\mathrm{op}}=T_{\mathrm{op}}-\text{TNZcente}{\text{r}}_{\mathrm{op}}$

Statistics

For each measurement sample (1) the independent variables (dTNZ_op and dTNZ_sk) are calculated following the procedure described above. Values are represented as individual points and as the mean per voting time point ±95% confidence interval of the mean.

The hypotheses are tested by linear regression analysis using the following models:

• Y_sensation = a₀ + a₁ X_dTNZsk

• Y_sensation = a₀ + a₁ X_dTNZop

Recapitulating the introduction, this paper tests the hypotheses that:

• Neutral thermal sensation corresponds to zero distance from the TNZ centroid.

• Thermal sensation is linearly correlated to the distance from the TNZ centroid.

The first hypothesis is falsified when the model constant (a₀) significantly differs from zero. The second hypothesis is falsified when the slope (a₁) does not significantly differ from zero. The statistics are considered significant when alpha < 0.05. The explained variance is expressed by the R² value.

The regression analyses are performed on population means per voting time point. To test whether the hypotheses are also valid on an individual level, the same regression analyses were repeated for each subject on individual data.

All data analysis are conducted using Matlab 2014a for mac, Mathworks, USA.

Results

Mean thermal sensation versus mean distance from thermoneutral centroid is presented in Figure 4.

Figure 4. (a) Mean thermal sensation per voting time point versus the mean distance from the thermoneutral centre (dTNZ_op) per voting time point. Error bars denote the 95% confidence interval of the mean for thermal sensation (vertical) and dTNZ_op (horizontal); (b) mean thermal sensation per voting time point versus the mean distance from the thermoneutral centre (dTNZ_sk) per voting time point. Error bars denote the 95% confidence interval of the mean for thermal sensation (vertical) and dTNZ_sk (horizontal).

Linear regression shows a significant linear relation between observed thermal sensation votes (TSV) and calculated mean dTNZ_op (Figure 4(a), p < 0.001, r² = 0.98) as well as mean dTNZ_sk (Figure 4(b), p < 0.001, r² = 0.95). Furthermore, for dTNZ_op the intercept is significantly greater than zero (a₀ = 0.18 ± 0.07) and the slope significantly differs from zero (a₁ = 0.28 ± 0.02). For dTNZ_sk the intercept is also significantly greater than zero (a₀ = 0.14 ± 0.11) as well as the slope (a₁ = 1.09 ± 0.12).

Consequently, the calculated centroid of the TNZ corresponds to a slightly warmer-than-neutral sensation. Furthermore, thermal sensation is associated with the distance to the calculated thermoneutral centroid.

Individual thermal sensations versus individual distance from thermoneutral centroid are presented in Figure 5. All individuals show a significant linear relation between thermal sensation and distance from the thermoneutral centroid (for dTNZ_op (Figure 5(a)), p < 0.01, 0.72 < r² < 0.87), and for dTNZ_sk (Figure 5(b)), p < 0.01, 0.68 < r² < 0.96)). The averages of the coefficients obtained through the linear regression analyses of each individual (for dTNZ_op (Figure 5(a)), a₀ = 0.18 ± 0.21 and a₁ = 0.28 ± 0.06), and for dTNZ_sk (Figure 5(b)), a₀ = 0.17 ± 0.28 and a₁ = 1.08 ± 0.24)) are in line with those presented for the regression to mean values presented above.

Figure 5. (a) Individual thermal sensation per voting time point versus the individual distance from the thermoneutral centre (dTNZ_op) per voting time point; (b) individual thermal sensation per voting time point versus the individual distance from the thermoneutral centre (dTNZ_sk) per voting time point.

Discussion

This study shows that the physiological thermoneutral centre is associated with a slightly warmer-than-neutral TSV and that thermal sensation is strongly related to the distance to the thermoneutral centre, expressed as either operative temperature or mean skin temperature (further referred to as dTNZx).

The R² values presented above signify an extreme good linear relationship between thermal sensation and dTNZx. The relation is particularly strong at a group level (as shown by high r² values in Figure 4), and also holds for individuals (high r² values in Figure 5). However, between individuals there is considerable variation in the slope between thermal sensation and dTNZx. Limitations to interpretation are discussed below. The results shown here must be regarded as preliminary and with great care for the following reasons.

Firstly, the high linearity between dTNZx and thermal sensation may be caused by the dynamic protocol where operative temperature gradually decreases or increases. Basically, the continuous thermal input for the body could be associated with a high signal-to-noise ratio of thermal status versus other inputs, and thus makes it easier for the body to assess its own thermal state. It is also plausible that a different thermal transient would induce a different slope of the relation between dTNZx and thermal sensation.

Secondly, this analysis is based on only 16 young adult females observed under controlled conditions in a climate chamber. This means that the number of subjects is low, the expected variance due to age and gender differences is limited, and there are hardly any context related factors increasing the variance (de Dear & Brager, 1998; Schweiker & Wagner, 2015). With respect to gender and age, earlier studies have reported that young adult women feel colder than young adult men in cool conditions, however there is an overlap in air temperatures that both genders rate as neutral (Parsons, 2002). Likewise, the elderly have also been reported to rate an environment in general colder than their younger counterparts (Schellen, van Marken Lichtenbelt, Loomans, Toftum, & de Wit, 2010; Taylor, Allsopp, & Parkes, 1995).

Thirdly, the averaging per voting time point reduced variance for the analysis. However, when non-averaged data are used per time point, the explained variance drops to about R² = 0.85, which is still high compared with other measures.

Fourthly, the model, as all models, is a simplification of reality and, for instance, does not take into account that skin temperature varies over different limbs, that is, only mean skin temperature is considered in the calculation of heat balance. In reality the temperature difference between proximal (e.g. torso) and distal (e.g. hands) skin sites is largest in a cold environment and smallest in a warm environment; and these differences may have an impact on the actual heat exchange between the body and the environment (Kingma, Frijns, Saris, van Steenhoven, & van Marken Lichtenbelt, 2011; Liu, Lian, Deng, & Liu, 2011). Despite these limitations, the approach presented here looks promising for understanding human thermal sensation. Therefore, it should be explored through a broader sample with a higher variety in participant characteristics. Furthermore, the influence of dynamical versus steady-state protocols on the model prediction accuracy should be tested. Thus, in summary, future analyses need to address issues related to context-related aspects.

Biophysical model as an extension to the PMV model

The biophysical model presented here is a steady-state heat balance model. Therefore, and given its intended context of use for the control of indoor thermal environments, it can be regarded as a physiological extension of the PMV (predicted mean vote) model. Two crucial differences are (1) the biophysical model requires the actual metabolic rate for accurate representation of the TNZ position, in contrast to the PMV model which performs with a standard reference table value for metabolic rate; and (2) the biophysical model incorporates tissue insulation, which is a function of body composition (i.e. fat and muscle tissue) and the ability of the body to adjust skin blood flow.

Implications of changes in tissue insulation and metabolic rate on overall heat balance are given below. For a given metabolic rate tissue insulation determines the temperature difference between body core and skin temperature (see step 3 of the mathematical procedure). Given that body core temperature does not deviate much from 37°C during normal conditions, tissue insulation thus determines the skin temperature and indirectly influences the body’s ability for heat loss to the environment. That is, for equal metabolic rate and body surface area, a lean person will have a higher skin temperature than an obese person. Indeed, experiments have shown that mean skin temperature is significantly lower in obese versus lean subjects both in neutral (24°C) and mild cold (16°C) environments (Wijers, Saris, & van Marken Lichtenbelt, 2010).

Likewise, if metabolic rate increases while tissue insulation remains constant, the TNZ shifts to lower operative temperatures to accommodate for increased heat loss requirement. The TNZ also becomes wider and shifts the centre of the TNZ to lower temperatures (see again step 3 of the mathematical procedure, and Figure 6).

Figure 6. Shift in thermoneutral zone (TNZ) as a consequence of increased physical activity (dark grey) versus the rest (light grey). A higher metabolic rate causes the TNZ to shift to lower operative temperatures to accommodate for increased heat loss requirement; the TNZ also widens (see step 3 in the mathematical procedure), and consequently the centre of the TNZ to shift to lower temperatures.

According to the findings of this paper, this would mean that also neutral thermal sensation is shifted to a lower mean skin temperature (and lower operative temperature). In the schematic example of Figure 6, if a person starts in a resting condition and then increases physical activity while the operative temperature remains the same, the TNZ shifts to cooler operative temperatures, and that person places him/herself outside the TNZ on the right side. This means that extra heat loss is required to maintain thermal balance (e.g. through sweating). Indeed, in the literature the relation between increased physical activity, lower skin temperature and neutral thermal sensation is explained as an effect of evaporative cooling due to increased sweating (Gagge, Stolwijk, & Saltin, 1969). The biophysical analysis in this paper does not contradict the finding in the literature, but complements the rationale by explaining that lower skin temperature for neutral thermal sensation at higher physical activity is also expected from the perspective of internal body heat balance.

In case no measurements of metabolic rate are available, it can be estimated based on individual/subpopulation characteristics (i.e. height, weight, age and gender) using, for instance, Harris and Benedict (1918) for basal metabolic rate (note basal metabolic rate is associated with 0.8 MET). Furthermore, body surface area can be estimated from height and weight with Mosteller (1987). For tissue insulation the range presented in this paper is based on lean young adults during extreme thermal exposures (Burton & Edholm, 1955; Hayward & Keatinge, 1981; Veicsteinas et al., 1982). Literature values are corrected for respiratory heat loss where required.

Comfort, the neutral zone, further hypotheses and potential applications

Current models of thermal sensation and comfort assume that thermal sensation and thermal comfort are dependent on body core and skin tissues (Fiala, 1998; Zhang, Arens, Huizenga, & Han, 2010). The popular consensus is that actual temperatures are compared with one or more set points or reference levels (Schweiker & Shukuya, 2009). The difference between the actual and reference value is an error signal, which scales with thermal sensation and thermal comfort and conceptually can be interpreted as the drive for a thermoregulatory action (either autonomous or behaviourally). Interestingly, the explicit existence of an absolute set point is not required from a neurophysiological perspective. Instead, the dynamic balance between warm- and cold-sensitive neural afferents may already contain sufficient information to distinguish warm from cold and vice versa (Filingeri, 2016; Kingma, Schellen, Frijns, & van Marken Lichtenbelt, 2012; Kingma et al., 2014a).

The concept of TNZ and dTNZ assumes a range of conditions, which do not require further thermoregulatory adjustments except adjustments of the skin blood flow. Based on the results presented above, the closer such conditions are to the centroid of TNZ, the more likely they will be regarded as thermally neutral.

However, the position of the TNZ centroid may change with changes in environmental conditions (e.g. air speed and relative humidity), clothing or physiological conditions (e.g. metabolic rate or skin blood flow). Following current models of thermal sensation and thermal comfort this means that the reference temperature is a function of these parameters as well. This hypothesis is in line with the discussion on the variability of the reference level based on findings from biology, psychology and neuroscience (Schweiker & Shukuya, 2009), and supports the hypothesis that the body is able to learn which constellations of exposures and thermal state (combining conscious and unconscious information) correspond to optimal homeostasis (Keramati & Gutkin, 2014). If such thermal experiences are not repeated, the body may forget whether or not a thermal state is favourable. A hypothesis to be tested could be that the lack of such thermal experiences due to uniform thermal indoor environments may be one reason for the differences in the range of conditions perceived as comfortable between air-conditioned and naturally ventilated buildings (de Dear & Brager, 1998).

The biophysical model leaves room for individual variation not only in thermal sensation but also in thermal preference due to long-term adaptation (e.g. geographical adaptation) and short-term adaptation or acclimatization (e.g. seasonal or shorter adaptive processes) (Hori, 1995). This means that through adaptation and acclimatization the body learns a new optimal position relative to its thermoneutral centroid. With respect to the above findings, it could be hypothesized that people perceive conditions slightly above the centre of the TNZ as neutral as this is the typical condition during summer season (see also Pallubinsky et al., 2017, on warmth acclimatization in this special issue of Building Research & Information). Quantifying the drivers and their effects related to people’s individual parameters is a challenge still to be faced. Nevertheless, the authors believe that the basis for analysing and understanding such research needs to be a biophysical approach.

Beyond the interpretation with respect to thermal sensation, preference and comfort, the approach presented here can be linked to the concept of alliesthesia (de Dear, 2011). Leaving conditions outside the TNZ towards its centre should be related to a perception of relaxation: the actions performed (thermoregulatory or non-thermoregulatory) lead to the goal to get back to the safest state for an animal from an energetic and hydration perspective, the TNZ. The result of such a change might lead to pleasure as it ‘occurs whenever sensation indicates the presence of a stimulus which helps to correct an internal trouble’ (Cabanac, 1971, p. 1104).

In addition, the biophysical model could be extended by the framework for an adaptive heat-balance model, which includes effects of behavioural (non-thermoregulatory) and psychological adaptation and acclimatization processes (Schweiker & Wagner, 2015). At the same time, physiological adaptation could be included in the biophysical model directly and not solely through an alteration of metabolic rate as an input variable to a heat-balance model.

With the rise of wearables assessing more and more personal parameters from those who wear them – including physiological values such as skin temperature – and the increased connectivity between devices through the internet of things, the biophysical model could be meaningful for future applications in the built environment. Based on the information given by sensors connected to building automation systems together with those derived by the wearables, either the dTNZ_op or the dTNZ_sk could be estimated for an individual. Such an estimate could then be used within control algorithms for the indoor climate control. Such an approach would make the biophysical model of interest not only for researchers in the field of thermal comfort but also for developers within the fields of building automation, wearables and their connection. However, as pointed out above, this would require a model based on a broader dataset to prove its generalizability. In addition, issues related to the protection of individual health-related data would need to be solved before such an application.

In conclusion, this paper shows that the thermoneutral centre as calculated by a biophysical analysis is associated with a slightly warmer-than-neutral TSV and that thermal sensation is strongly related to the distance to the thermoneutral centre. The biophysical approach presented does look promising with respect to understanding and predicting thermal sensation, and also to be the basis for future research related to thermal comfort, alliesthesia, and non-thermoregulatory and psychological behavioural adaptive processes.

Acknowledgements

The authors express their gratitude to Paul Schoffelen, Loek Wouters and Marc Souren, Metabolic Research Unit Maastricht (MRUM), for their assistance and technical support. Finally, the authors would like to express their appreciation to all volunteers for their participation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article was originally published with errors, which have now been corrected in the online version. Please see Correction (http://doi.org/10.1080/09613218.2017.1299996)

Additional information

Funding

This work was supported by the AgentschapNl: [grant number INTEWON: EOSLT10033], TKI Energo, TKI Solar Energy: [grant number TRECO: TEGB13023], and H2020-EE-2016-RIA-IA: [grant number Mobistyle 723032].