Severity of depressive symptoms moderates the sympathoinhibitory effect of local skin warming following exposure to a social stressor

of


Introduction
Research on the pathogenesis of depression has focused primarily on the central nervous system, and (non-psychological) treatments, such as pharmacologic approaches and electric or magnetic stimulation (e.g., transcranial magnetic stimulation [TMS]), focus overwhelmingly on the brain.However, the side-effects and limited efficacy of existing antidepressant treatments, combined with the fact that depression is now the leading cause of disease burden globally (Friedrich, 2017), have led to an interest in development of novel interventions.A "bottom-up" conceptualization of depression, which emphasizes the relevance of peripheral physiology to both the aetiology and treatment of depression, may provide solutions (Hale et al., 2013).Indeed, depression is associated with altered (and usually impaired) functioning across multiple sensory nervous system pathways, and there is evidence that depressive symptoms can be ameliorated or exacerbated by stimulation of these same pathways (for review, see Canbeyli, 2022).The thermosensory system may hold promise for understanding and treating depression, with some researchers proposing that exposure to warm stimuli directly influences cortical structures involved in affective processing (i.e., the affective warmth hypothesis; Raison et al., 2015).
Depression is associated with dysregulation of multiple aspects of thermosensory functioning, including thermoregulation and affective processing of thermal stimuli.For example, there is a well-established association between depression and elevated core body temperature (Avery et al., 1999), suggesting impaired thermoregulatory cooling processes.Consistent with this observation, an often-reported side-effect of antidepressant medications is increased sweating, implicating thermoregulatory pathways.Depressed individuals rate warm stimuli as more unpleasant (Strigo et al., 2008), and have a lower threshold for experiencing warmth as unpleasant (Ushinsky et al., 2013).The latter finding was insensitive to happy or sad mood induction, suggesting that an "affective bias" against warm stimuli may be a chronic feature of depression.The above research was on clinical populations, and, to the best of our knowledge, there is no research comparing the influence of depressive symptoms on thermosensory functioning in clinical vs non-clinical populations.Stimulation of thermosensory pathways by non-noxious warming influences affect beyond a simple estimation of the pleasantness or unpleasantness of a thermal stimulus.For example, local skin warming (e. g., via a warm cup), has been reported to reduce self-reported negative affect after a social stressor (IJzerman et al., 2012).There is some evidence that the affective response to non-noxious local skin warming is due to inhibition of the sympathetic branch of the autonomic nervous system.Preclinical research has demonstrated a connection between non-noxious skin warming and activation of brain regions that are implicated in the inhibition of sympathetic activity (Hale et al., 2011).Local skin warming has been shown to inhibit markers of sympathetic nervous system activity following stressor exposure (Kober et al., 2003).Research on local skin warming with no stressor exposure has been more equivocal, with warm footbaths resulting in both a decrease (Yamamoto et al., 2008) and increase (Uebaba and Xu, 2004) in sympathetic activity.
There is evidence that thermosensory pathways can be directly recruited to treat depression.Multiple studies have demonstrated that elevating core body temperature in depressed individuals into a fever range (i.e., whole-body hyperthermia) results in an antidepressant effect that lasts for at least several weeks, although the mechanism of action is unknown (Hanusch et al., 2013;Janssen et al., 2016;Naumann et al., 2020).The stimulus used to elevate core body temperature has ranged from a form of infrared sauna (e.g., Hanusch et al., 2013;Janssen et al., 2016) to hot baths (e.g., Naumann et al., 2020).
Currently, we are not aware of any research that has investigated whether the sympathetic nervous system is implicated in the link between depression and altered functioning of the thermosensory nervous system.However, there is evidence linking exposure to warm stimuli with changes in the sympathetic nervous system (Hale et al., 2011;Kober et al., 2003), evidence that depression is associated with differential responding to warm stimuli (Strigo et al., 2008), and that depression is associated with changes to sympathetic nervous system activity (Hu et al., 2016).
The above findings offer a paradoxical picture in which depressed individuals experience 1) impaired functioning of thermosensory pathways and an 'affective bias' against warm stimuli, and 2) particular benefits as a result of stimulation of thermosensory pathways, namely in the form of afferent thermosensory signaling.This raises the question as to whether depression is associated with either A) decreased responsiveness to the affective benefits of warm stimuli, or B) experiencing particular affective benefits as a result of exposure to warm stimuli (i.e., in the form of whole-body hyperthermia).Regarding the former proposition, it may be that depressed individuals have a lack of 'biosensors' for non-noxious stimuli in the periphery, leading to diminished sensory information for warm stimuli (Hale et al., 2013).Here, the antidepressant effect of whole-body hyperthermia may be due to its ability to 'reset' the thermosensory system, including affective responding.So, initially, a depressed individual is not able to gain the affective benefits of a warm cup of tea, likely resulting in diminished motivation to make the tea in the first place.However, following treatment with whole-body hyperthermia, their thermosensory functioning is 'reset', allowing them to experience the pleasant affect associated with the warm tea (and other pleasant stimuli), perhaps resulting in increased engagement.Alternatively, it may be that depressed individuals get particular benefit from warm stimuli, possibly due to their sympathoinhibitory effects on the autonomic nervous system.This would be consistent with research that shows an association between depression and sympathetic nervous system hyper-reactivity (Hu et al., 2016), suggesting that depressed individuals may need external input to help inhibit sympathetic nervous system activity.In this scenario, a depressed individual would receive particular affective benefits from a warm cup of tea (or other warm stimulus) as it would help inhibit their over-active sympathetic nervous system.Here, the effectiveness of whole-body hyperthermia may be due to a more profound resetting of the autonomic nervous system, resolving sympathetic nervous system hyper-reactivity.
In this experiment we tested the hypothesis that high levels of depressive symptoms are associated with altered functioning of the affective branch of the thermosensory system in a non-clinical sample.Our goal was not to test if warming led to a sustained antidepressant effect.Treatments that have used warming to generate an antidepressant effect elevate core body temperature into a fever range, something the local skin warming used in this study would not do.Our purpose was to activate thermosensory pathways with non-noxious local skin warming, thus allowing an investigation into whether the functioning of these pathways is influenced by severity of depressive symptoms.To test the hypothesis, we investigated the effect of local skin warming on markers of sympathetic nervous system activity and perceived negative affect following exposure to a social stressor.We assessed whether severity of self-reported depressive symptoms in a non-clinical population moderated the effect of local skin warming on subjective affect and physiological stress response.While we were using a non-clinical sample, considering the high prevalence of depression in the general population (ABS, 2022), we expected the sample to include individuals with levels of depressive symptoms consistent with a diagnosis of a mood disorder.

Participants
The sample was drawn from a healthy, non-clinical population and consisted of 90 participants (65 female, M age = 23.61years, SD age = 5.67, age range = 18 -45 years).Participant sex was identified by selfreport questionnaire, with the question "what is your sex", with the options "male", "female" or "intersex".Inclusion criteria consisted of being 18 years old or older.Exclusion criteria consisted of experiencing cold or flu symptoms or cardiac or thyroidal health issues (due to their possible impact on physiological measures), use of medication known to impact the thermosensory systems, such as beta-blockers and antihistamines, consumption of caffeinated, alcoholic or soft drinks, eating, smoking or engaging in physical exercise within 1 h of the experiment commencing (due to the impact on saliva samples), and lack of proficiency in English (in order to complete the self-report questionnaires accurately and to provide informed consent).Institutional ethics approval was granted from the La Trobe University Human Ethics Committee (HEC18122) and written consent was obtained by participants prior to commencement of the experiment.

Procedure
Participants were recruited via email, text and social media using scripts and via face-to-face recruitment at the university campus.Participants were asked to refrain from caffeine for 2 h prior to the experiment, and from smoking, exercise, food and drink (excluding water) 1 h prior to the experiment, as these factors may confound the measurement of EDA (Boucsein, 2012) and sAA (Rohleder and Nater, 2009).All participants were tested between 12:00 and 17:00 in the same M.P. Tyler et al. climate-controlled room (24.3 ± 0.5 • C, mean ± SD) with sessions lasting approximately 40 min.Two cinema vouchers valued at $15 each were given to participants to reimburse them for their time.
Phase 1 (T1): Baseline measures and pre-experiment questionnaires.Electrodes to measure EDA were attached immediately following a brief overview of the study, gaining informed consent, and screening participants for exclusion criteria, which in all took approximately 3-5 min.The baseline measurement of EDA started approximately 30 s after electrodes were fitted.The baseline measurement for EDA consisted of the average of the measure over the baseline period while participants were completing the self-report measures, which took approximately 5-8 min.The first saliva sample was collected immediately after the participants completed the self-report measures.
Phase 2 (T2): Exposure to stressor.Participants then completed an arithmetic task, which involved counting backwards from 2036 by 13 out loud for 5 min while being interrupted by the experimenter with negative evaluative feedback and being told they were video recorded (see supplementary material for detail).Similar procedures that include negative verbal feedback on arithmetic performance have been shown to reliably elicit a stress response (Glynn et al., 2002;Glynn et al., 2007).
Phase 3 (T3): Warming.Following completion of the arithmetic task, participants were prompted by a slide on a computer monitor to place their right arm on the heat blanket (UTK Far Infrared Jade Heating Pad).Electrodermal and fingertip temperature sensors were placed on the left hand.The blanket measured 38 cm × 48 cm and was covered by a thin, cotton cloth for hygiene.After 6 min resting their arm on the blanket, Fig. 1.A) Schematic illustration of the experimental timeline.Saliva samples for assessment of salivary alpha-amylase activity were taken at the end of T1 (baseline), immediately following T2 (stressor) and immediately following T3 (warming).Electrodermal activity was determined by the average level during the completion of self-report measures (baseline), average level during T2 (stressor) and average level during T3 (warming).Fingertip temperature was determined by a single time point immediately before the end of T1 (baseline), a single time point immediately before the end of T2 (stressor), a time point half-way through T3 (mid-point warming) and immediately before the end of T3 (end-point warming) B) Thermal photos of the heat blanket during the local skin warming or control conditions of the experiment.Photos taken with a FLIR TG165 thermal camera.
participants were prompted by another slide to raise their arm off the blanket.Participants then rated the pleasantness of the blanket on a Visual Analogue Scale (mentioned below) on the computer screen.Electrodermal activity and fingertip temperature were measured continuously throughout the 6-minute warming phase.
Participants were randomized into the local skin warming condition (in which the heat blanket was switched on; n = 45) or control condition (in which the heat blanket was not switched on; n = 45).To eliminate the risk of experimenter expectation influencing results (Klein et al., 2012), custom software randomized participants and automatically turned the blanket on if the participant was randomized to the local skin warming condition, ensuring the experimenter was blinded to condition.Randomization occurred approximately 45 min before the experiment started, which allowed enough time for the blanket to reach a stable, maximum temperature (see Fig. 1 for thermal photos of the heat blanket during the blanket phase of the experiment).As part of providing informed consent, participants were provided an overview of the experimental procedure by the experimenter, including that the blanket would either be warm or at room temperature.At no stage were participants informed that there were two different conditions (i.e., an experimental and control condition), nor that they had been allocated to either an experimental or control condition.However, participants were aware they were exposed to warming or not, as when they placed their arm on the blanket it would either be noticeably warm, or at room temperature.To mitigate the risk that a lack of warmth would be misinterpreted as a technological error, participants were told whether the blanket would be warm or at room temperature via automated slides.While great care was taken to avoid priming participants as to the nature of the warming, it is possible that they made their own assumptions about the experiment that led to placebo or nocebo experimental effects.We checked for indicators of placebo or nocebo experimental effects by reviewing any direct effects of local skin warming on measures of subjective affect.
In the local skin warming condition, the blanket was set to a constant warm temperature (mean ± SD, 44.8 ± 2 • C).In the control condition the blanket remained at room temperature (mean ± SD, 24.3 ± 0.9 • C).A sensor measured blanket temperature continuously at a rate of 10 Hz.
Phase 4 (T4): Post-experiment questionnaires.In this phase, participants completed the measure of negative affect, indicated whether they believed the stressor was genuine and were debriefed about the true nature of the experiment (i.e., the deception used during the stressor).Participants were provided with contact information for a low-cost counselling service should they want to access psychological support for any reason.

Self-report measures 2.3.1.1. Assessment of symptoms associated with depression.
The Center for Epidemiological Studies -Depression Scale Revised 14-item Version (CES-D-14) is a 14-item measure of symptoms associated with depression that have been experienced over the past week (Carleton et al., 2013).The CES-D-14 proposes to improve on the original CES-D measure by removing items that inflate scores in women, and by presenting a 3-factor structure (negative affect, anhedonia, and somatic symptoms) that are more appropriate for current diagnostic criteria for depression (Carleton et al., 2013;Jiang et al., 2019;Yang et al., 2015).Previous research has indicated a score of 16 on the CES-D-14 as the optimal cut-off to identify individuals at risk of clinical depression, with good sensitivity and specificity (Jiang et al., 2019).Cronbach's alpha for the present study was .90 for global score, showing high internal consistency.
2.3.1.2.Negative affect.Participants rated how they felt following the completion of the arithmetic task and warming phase ("how did you feel during the task") using a measure of negative affect developed by Feldman et al. (2004).The negative affect measure consists of 9-items in which negative emotions are ranked on 5-point Likert scales ranging from 0 'not at all' to 4 'extremely', with higher scores indicating greater negative affect.Cronbach's alpha of the scale for the present study was .93,showing high internal consistency.

Affective responses to local skin warming. Unipolar Visual
Analogue Scales (VAS) were used to measure subjective pleasantness of local skin warming.The scale ranged from '0′ ("not pleasant at all") to '100′ ("extremely pleasant").The scale was presented to participants on a computer monitor immediately following the warming phase.VAS measures have been validated in previous research on the thermosensory system (Leon et al., 2008).

Rumination.
Research has indicated that rumination, a core feature of depression, can prolong physiological stress responses following stressor exposure, and that distraction can interrupt rumination and thus inhibit stress responses (e.g., Gerin et al., 2006).Thus, any exploration of how depression moderates the effect of warm stimuli on physiological stress responses needs to address whether the effects are indeed a function of activation of thermosensory pathways, or distraction from rumination.To help determine if any possible inhibition in sympathetic activity was simply a function of the warm stimuli providing a distraction from rumination, participants completed The Rumination Questionnaire (Mellings and Alden, 2000), a 5-item measure of negative rumination with good reliability and validity (Samtani and Moulds, 2017).The measure assessed the extent to which participants ruminated during the warming phase to determine whether warming simply distracted participants from rumination.The measure was included specifically for the purpose of excluding distraction as a source of possible changes to sympathetic activity.The Rumination Questionnaire was adapted to make the questions relevant to the specific stressor used in this study (i.e., the arithmetic task).Ratings were made on 7-point Likert scales (e.g., ranging from 0 "not at all", to 6 "constantly"), with higher scores indicating greater rumination.Cronbach's alpha of the scale for the present study was .83,showing high internal consistency.

Sleep disturbance.
We included a measure of sleep disturbance in order to control for its influence on outcome variables.The Pittsburgh Sleep Quality Index (PSQI) is a 19-item measure of global sleep disturbance (Buysse et al., 1989).The items are grouped into seven 'component scores' (e.g., sleep latency, sleep duration etc.) and the global sleep disturbance score is calculated by summing the seven component scores, with higher scores indicating greater sleep disturbance.The PSQI has been found to have good reliability and validity (Mollayeva et al., 2016).

Physiological measures
2.3.2.1.Electrodermal activity.Electrodermal activity (EDA) is a noninvasive measure that uses activity of eccrine sweat glands on the palmar surface of the hand to assess sympathetic activity (Boucsein, 2012).We sampled EDA at a rate of 10 Hz using stainless steel sensors attached to the proximal phalanges of the index and ring finger of the left hand.These sensors were connected to a GSR amp (AD Instruments, Bella Vista, NSW, Australia), and recorded through a PowerLab 4-25 T (AD Instruments).The subsequent data were recorded directly into an Excel spreadsheet using hardware and software designed by technical staff at La Trobe University.Raw scores were transformed into Z-scores to standardize the data and adjust for inter-individual differences not resulting from differences in sympathetic activation (e.g., greater skin thickness leading to a lower range of EDA values; Boucsein, 2012).EDA data were then filtered with a lowpass 2nd order Butterworth filter to M.P. Tyler et al. remove movement artifacts.Data were visually inspected and adjusted to ensure timing was consistent with pre-filtered data.

Fingertip temperature.
Distal skin temperature (e.g., fingertip temperature), is influenced by vasoconstriction, which in turn is induced by sympathetic nervous system activity, making fingertip temperature a non-invasive measure of sympathetic activity (Kistler et al., 1998;Vinkers et al., 2013).Skin temperature sensors were placed on the distal phalanx (i.e., tip) of the middle finger of the left hand.Skin temperature sensors were designed and made by La Trobe technical staff.The sensors had an accuracy of 0.25 • C and collected data at a rate of 10 Hz continuously through the experiment.The same sensors were used to measure ambient air temperature in the study.Data were recorded directly in an Excel spreadsheet as outlined above.

Salivary Alpha
Amylase.Salivary alpha amylase (sAA) is a salivary enzyme involved in digestion that can be used as a valid and reliable non-invasive measure of sympathetic activity, with exposure to acute stressors resulting in higher concentrations (Rohleder and Nater, 2009).Concentrations are measured in U/ml, according to manufacturer's instructions.Salivary alpha amylase was analyzed using a Sali-metrics® 96 Well Kinetic Enzyme Assay Kit (No. 1-1902; State College, PA, USA) and the same procedures described in O' Donnell et al. (2015).Optical density measurements were performed at 450 nm with a Syn-ergyTM HT Multi-Detection Micro-Plate Reader (Bio-Tek Instruments Inc., Winooski, VT).Concentrations of the select compounds were calculated using KC4 v3.4 Software (Bio-Tek Instruments).The assay has a lower detection limit of 0.1 U/ml with intra-and inter-assay coefficients of variations < 8%.

Statistical Method
All data were analyzed using IBM SPSS Statistics version 27.0 (IBM Corp, 2020).Data were checked for outliers, missing data and violation of normality.Outliers identified via box plots and z scores exceeding 3.29 (Tabachnick et al., 2014, p. 107) were windsorized to the next highest score + 1, or the next lowest score − 1 in order to reduce their disproportionate influence on inferential analyses (Tabachnick et al., 2014, p. 111).Technological error resulted in missing data for 2 participants for both fingertip temperature and EDA (both in the local skin warming condition).Experimenter error resulted in missing data for 8 participants (3 in control and 5 in warming condition) and 1 participant with missing data on the negative affect scale.The age variable was found to have a positive skew and was therefore transformed into an ordinal variable then log transformed to resolve the violation of normality.Results did not differ between transformed and untransformed data, so we used the untransformed age data in the analyses.
Power analysis (G * Power; Faul et al., 2009) for a moderated regression with 5 predictors, with a statistical power of 80%, α = .05and a medium effect size of f 2 = .15,showed that a sample size of 68 would be sufficient to detect whether severity of depressive symptoms moderated the effect of local skin warming, indicating the study was not underpowered.
Pearson's correlation coefficients were calculated to determine associations between key variables.One-way ANOVAs were used to compare mean pleasantness ratings of the blanket temperature between the control and warming condition, and level of rumination during the warming phase between the conditions.Repeated-measures ANOVAs were used to assess if the stressor was effective in eliciting changes in sympathetic activity.Separate moderated regressions were conducted using the macro-program PROCESS 3.4.1 (Hayes, 2017) with bootstrapping (N = 5000) to determine the impact of local skin warming on sympathetic activity and negative affect, and the possible moderating influence of severity of depressive symptoms using the global score on the CES-D-14.For electrodermal activity, we used the average over the 6-minute duration of the warming phase.For fingertip temperature, we were able to analyze two specific time-points: 1) midway through the warming phase, and 2) at the end of the warming phase, thus allowing some insight into changes in moderation effects as a function of time.Electrodermal activity fluctuates more rapidly than skin temperature, so a mean value was considered a more valid and reliable measurement of sympathetic nervous system activity.Based on previous research identifying that sex, age and sleep disturbance can impact physiological stress responses to an acute stressor (Kelly et al., 2008;Rohleder and Nater, 2009) and negative affect (Hamilton et al., 2007;Kelly et al., 2008;Scott et al., 2013), these variables were included as covariates for all moderated regressions.

Descriptives
Based on the recommended cut-off score of 16 on the CES-D-14, 35.6% of the control condition and 33.3% of the warming condition reported levels of depressive symptoms that indicated they were at risk of clinical depression (Fig. 2).See Table 1 for correlations between the variables.

Manipulation checks
First, a series of manipulation checks were performed to determine: 1) whether local skin warming was non-noxious; 2) whether the stressor was perceived as genuine; and 3) whether the stressor elicited a physiological stress response.
Local skin warming was perceived as non-noxious.A one-way ANOVA demonstrated that participants in the warming condition rated the blanket as more pleasant than participants in the control condition F(1, 86) = 27.18,p < .0001,η p 2 = .24.Furthermore, the range for pleasantness ratings in the warming condition (31.84-99.74)demonstrated that all participants found the warming at least somewhat pleasant.Together, these findings indicate that all participants experienced the local skin warming as non-noxious.
Most participants perceived the stressor as genuine.Only 8 participants (17.8%) in the warming condition, and 5 participants (11.1%) in the control condition reported that they did not believe that the negative feedback during the stressor was genuine.Only 3 participants (6.7%) in the warming condition and 5 participants (11.1%) in the control condition reported that they did not believe that the video was recording them.Responses were broadly similar for participants who reported that they did not believe the stressor was genuine, so their data were included in the analysis.The stressor elicited a physiological stress response.Repeatedmeasures ANOVAs were used to determine changes in sympathetic nervous system activity from baseline to following exposure to the stressor.Overall, fingertip temperature and EDA both showed changes consistent with activation of sympathetic nervous system activity in response to the stressor (Fig. 3).Fingertip temperature decreased, F(1, 86) = 30.40,p < .001,η p 2 = .26,and average EDA increased, F(1, 86) = 981.07,p < .001,η p 2 = .92.Salivary alpha amylase (sAA) did not show a significant increase from baseline to immediately post-stressor F(1, 80) = 1.00, p = .33,η p 2 = .01.There was no difference in stress response between the conditions for fingertip temperature, F(1, 86) = .08,p = .77,η p 2 = .00,nor sAA F(1, 77) = .93,p = .33,η p 2 = .01,while a group difference did emerge for EDA, F(1, 86) = 5.78, p = .02,η p 2 = .06, showing that the control condition experienced a greater increase in electrodermal activity during the stressor phase.These results indicate that the stressor elicited a physiological stress response, specifically in the form of sympathetic nervous system activation.

Does severity of depressive symptoms moderate the impact of local skin warming on physiological stress response and self-reported affect?
Overall, severity of depressive symptoms moderated the effect of local skin warming on finger-tip temperature.Fig. 4 shows the correlation between CES-D-14 and fingertip temperature for the two conditions.Global depressive symptom scores moderated the impact of local skin warming on fingertip temperature both at the midpoint of the warming phase (b =.22, t (81) = 2.15, p = .03),and the endpoint of the warming phase (b =.30, t (81) = 2.06, p = .04).See Fig. 5 for graphical representation of the moderation effect.A direct effect of global depressive symptom scores on fingertip temperature also emerged at the midpoint (b = − .36,t (81) = − 2.08, p = .04)and endpoint of the warming phase (b = − .49,t (81) = − 2.34, p = .02),showing that higher levels of depressive symptoms predicted lower percentage change in fingertip temperature.There was no direct effect of local skin warming on fingertip temperature.
When a moderator is statistically significant, the Johnson-Newman technique allows the exact computation of cut-off values that separate sub-groups (Bauer and Curran, 2005).The Johnson-Newman significance regions indicated that only participants (both control and local skin warming groups) with scores higher than 19 on the CES-D-14 (i.e., the top 20th percentile) at midpoint of the warming phase and higher than 25 on the CES-D-14 (i.e., the top 8th percentile) by the end of the warming phase, showed significant (p < .05)positive associations between CES-D-14 and fingertip temp.Both these cut-offs are above the CES-D-14 clinical cut-off of 16.To determine if the proportion of those participants showing a significant positive association between CES-D-14 and fingertip temperature differed as a function of condition (control vs local skin warming), a Chi-Square analysis was conducted for both points (Table 2).The Chi-Square analysis showed that there was a higher proportion of respondents showing a significant positive association in the local skin warming condition at end point, demonstrating that local skin warming helped inhibit sympathetic nervous system activity for participants with high levels of depressive symptoms.
There was a direct effect of local skin warming on sAA, (b = -12.17,t (75) = -2.05,p < .05),showing that local skin warming resulted in a greater decrease in sAA from the end of the stressor phase to the end of    the warming phase.Severity of depressive symptomology did not moderate this effect (b = -1.26,t (75) = -1.56,p = .07).There was no moderating effect of global depressive symptom scores on the relationship between local skin warming and self-reported negative affect, nor electrodermal activity (see Table 3).Also, there were no direct effects of local skin warming nor level of depressive symptom on these measures.Importantly, there was no direct effect of local skin warming on negative affect, indicating a lack of placebo or nocebo experimental effects.There was no difference in the level of rumination between the local skin warming and control conditions, nor did depression moderate the levels of rumination between the conditions.

Discussion
Local skin warming inhibited markers of sympathetic activity, namely fingertip temperature and sAA, following exposure to a social stressor, with severity of depressive symptoms moderating the effect on fingertip temperature.Exposure to the social stressor decreased fingertip temperature and increased electrodermal activity, both indicating increased sympathetic nervous system activity.Results from fingertip temperature changes suggest that, whereas participants in the control condition with high levels of depressive symptoms experienced prolonged sympathetic activity, this prolonged stress response was interrupted for participants with high levels of depressive symptoms who were exposed to local skin warming.Results from sAA changes indicate that local skin warming inhibited sympathetic activity, but no moderation effect of severity of depressive symptoms emerged.There was no direct effect of local skin warming, nor moderation effect of severity of depressive symptoms, on electrodermal activity.Local skin warming did not reduce self-reported negative affect, nor did severity of depressive symptoms moderate this relationship.Together, these data suggest that the social stressor was effective in generating changes in sympathetic tone, that local skin warming inhibited some markers of physiological stress responses, but not electrodermal activity nor subjective affect, and that this effect was limited to participants with high levels of depressive symptoms for fingertip temperature.
Local skin warming increased fingertip temperature, but only for participants with high levels of depressive symptoms.The cut-off scores for differential responding to local skin warming were above the clinical cut-off of 16 on the CES-D-14.In the control condition, high levels of Fig. 5. Graphs showing that severity of depressive symptoms moderated the impact of local skin warming on change in fingertip temperature from immediately poststressor to (A) midway through the warming phase and (B) to the end of the warming phase.Note: increase in fingertip temperature represents decreased sympathetic nervous system activity.Abbreviations: CES-D-14, Center for Epidemiologic Studies Depression Scale Revised 14-item version.CES-D-14 scores were mean-centered to perform the moderated regression.

Table 2
A comparison of the proportion of participants with a significant positive association between CES-D-14 and change in fingertip temperature during the warming phase.Abbreviations: CES-D-14, Center for Epidemiological Studies -Depression Scale Revised 14-item Version.Mid-point measurement was the percentage change in fingertip temperature from the end of the stressor phase, to midway through the warming phase (i.e., 3 min after beginning of warming phase).End-point measurement was the percentage change in fingertip temperature from the end of the stressor phase to the end of the warming phase (i.e., 6 min after beginning of warming phase).Note.Sex dummy coded as '1′ = male, '2′ = female.Abbreviations: CES-D-14 = Center for Epidemiological Studies -Depression Scale Revised 14-item Version; PSQI, The Pittsburgh Sleep Quality Index; sAA, Salivary alpha amylase.CES-D-14 scores were mean-centered to perform the moderated regression.
depressive symptoms led to lower fingertip temperature, which can be a marker of prolonged activation of the sympathetic nervous system, following exposure to the social stressor.However, in the local skin warming condition, participants with high levels of depressive symptoms showed increased fingertip temperature relative to participants with high levels of depressive symptoms in the control condition, suggesting that warming may have inhibited sympathetic nervous system activity.Our findings on the fingertip temperature measure suggest that there was a tendency for high levels of depressive symptoms to be associated with a prolonged physiological stress response, and that local skin warming buffered against this association.While previous studies have demonstrated that local skin warming can inhibit sympathetic nervous system activity (e.g., Kober et al., 2003), this is the first study we are aware of that has shown a specific benefit to people with high levels of depressive symptoms.
While we hypothesize that the fingertip temperature findings are a results of local skin warming activating thermosensory pathways that connect to sympathoinhibitory regions of the brain, an alternative explanation for our results is that local skin warming simply distracted participants from thinking negatively about the stressor (i.e., from ruminating).Indeed, the observation that greater severity of depressive symptoms led to lower fingertip temperature in the control condition, suggesting prolonged sympathetic nervous system activity, is consistent with research showing that rumination, a core feature of depression, can lead to prolonged physiological stress responses following stressor exposure (e.g., Glynn et al., 2002, Gerin et al., 2006).For example, research by Gerin et al. (2006) demonstrated that participants with high trait rumination experienced prolonged stress activation following recall of an anger-inducing event, whereas participants with low-trait rumination showed spontaneous recovery from physiological stress responses.Gerin et al. (2006) found that participants with high trait rumination who were distracted following stressor exposure (e.g., by completing a puzzle) showed less rumination and reduced physiological stress responses.However, while distraction reduced the level of rumination in the study by Gerin et al. (2006), local skin warming did not reduce level of reported rumination in our study, suggesting an alternative mechanism of action than distraction.Furthermore, local skin warming is not a cognitively demanding stimulus, and so would be unlikely to directly interrupt cognitive processes.Rather, the results on changes to fingertip temperature raise the intriguing possibility that local skin warming interrupted the prolonged physiological stress response associated with rumination, even as the rumination itself continued.
Our findings contradict previous research showing an affective bias against warm stimuli amongst clinically depressed individuals (Strigo et al., 2008).Instead, we show that greater severity of depressive symptoms is associated with experiencing specific affective benefits from local skin warming.The discrepancy in findings may be due to differences in the duration (e.g., seconds vs minutes) and delivery method (e.g., metal thermode vs heat blanket) of warm stimuli between previous studies and our study.There may also be a qualitative difference between how the thermosensory system responds to warm stimuli for healthy individuals with high levels of depressive symptoms (i.e., our population), and clinically depressed individuals (i.e., the population used in research showing an affective bias against warm stimuli).Importantly, previous research showing an affective bias against warm stimuli in depressed individuals did not include stressor exposure.It may be that the affective benefits of local skin warming only emerge in the context of a stressor triggering a (prolonged) physiological stress response in individuals with high levels of depressive symptoms.There is non-experimental, naturalistic research showing that depressive symptoms are associated with an affective bias towards non-noxious warm stimuli, with studies showing that greater mood disturbance increases self-exposure to warm showers (Bargh and Shalev, 2012) and warm baths (Parker and Crawford, 2007).Our findings indicate that individuals with high levels depressive symptoms may be seeking non-noxious warm stimuli in an attempt to inhibit an otherwise overactive sympathetic nervous system; a form of self-medication or self-regulatory strategy after exposure to stressors.
While severity of depressive symptoms moderated the effect of local skin warming on one marker of sympathetic nervous system activity (i.e., fingertip temperature), there was only a direct effect of local skin warming on sAA and no effect on electrodermal activity.The differential findings across markers of sympathetic nervous system activity emphasize the need to interpret the findings with caution.The finding that local skin warming resulted in a greater decease in sAA (regardless of severity of depressive symptoms) is consistent with previous research showing sympathetic inhibition following exposure to local skin warming (Kober et al., 2003).Limiting the analysis of electrodermal activity to averaged data over the 6-minute duration of the warming phase, and sAA to the end-point of the warming phase, may have reduced our ability to detect group differences that only occurred at the early stages of the warming phase.We measured sympathetic nervous system activity with indirect markers (i.e., electrodermal activity, fingertip temperature and sAA), and differential findings may be due to other physiological processes influencing these measures.For instance, fingertip temperature is influenced by blood pressure (Tai et al., 2022), and local release of nitric oxide (i.e., which regulates vasodilation; Levine et al., 2012) and sAA can be influenced by parasympathetic nervous system activity (Bosch et al., 2011).
The observation that local skin warming elevated skin temperature in participants with high levels of depressive symptoms, but did not reduce self-reported negative affect, is consistent with a large body of literature that has shown a discrepancy between measures of psychological and physiological stress (for review, see Campbell and Ehlert, 2012).In the current study, self-reported negative affect was measured retrospectively (approximately 7-8 min after stressor exposure).Some research (e.g., Hellhammer and Schubert, 2012) has found that there is only consistency between physiological and subjective assessments of stress when subjective stress measures occur during stressor exposure.
Participant allocation to condition and heat blanket activation were automated in our study, allowing experimenters to be blind to condition and thus removing the influence of experimenter expectation on the results (Klein et al., 2012).As mentioned above, participants were not blind as to whether they were exposed to local skin warming or not.Every effort was made to ensure participants were not primed as to the nature of the local skin warming.However, it is possible that participants made their own assumptions about the experiment that led to placebo (e.g., "the warming is meant to be pleasant") or nocebo (e.g., "the warming is meant to be painful") effects.If there was a placebo or nocebo experimental effect, we would expect to see a direct effect of local skin warming on measures of subjective affect.Our findings showed no direct effect of local skin warming on affect, suggesting that the results are unlikely due to a placebo or nocebo effect.While results for fingertip temperature and electrodermal activity indicate the stressor increased sympathetic activity, there was no change in sAA.Baseline measurement of sAA occurred just before participants began the arithmetic task (i.e., the stressor) and immediately following assessment of oral temperature.Accordingly, there may have been a spike in sAA due to arousal associated with anticipation of the arithmetic task, or in response to the experimenter taking oral temperature.If there was a spike in sAA at baseline measurement, it would have masked any subsequent stress response occurring following stressor exposure and may explain why no increase in sAA was observed following stressor exposure.Neither electrodermal activity nor fingertip temperature would be prone to such a spike at baseline, as EDA was averaged over the entire baseline period and fingertip temperature changes too slowly to be influenced by such an acute spike in sympathetic activity (Kistler et al., 1998).Depression shows high comorbidity with other psychopathologies and somatic disorders (Steffen et al., 2020), which were not controlled for, and thus may have acted as confounds in the present study.Left-hand dominant participants may have found it more difficult to maneuver the mouse when completing the self-report measures, as all participants were required to use their right hand (i.e., due to electrodes being connected to the left hand).We did not assess whether participants were left or right hand dominant and so were unable to control for any possible influence of this on the results.
Previous research on the relationship between depression and the thermosensory system has focused on clinical populations, while our study used a non-clinical sample.While there is an understandable emphasis on interventions to treat clinical populations (i.e., the most acutely unwell, often hospitalized, individuals), we wanted to provide insights with our research that may be relevant for interventions targeted at individuals with high levels of depressive symptoms who are outside clinical settings (e.g., living in the community).Indeed, around one third of our sample reported depressive symptoms that have been found to be consistent with a diagnosis of depression.However, an important limitation of our findings is that participants were drawn from a non-clinical, high-functioning population, and the generalizability to clinical populations needs to be tested.

Conclusions
In summary, our study provides evidence that local skin warming may buffer the association between high levels of depressive symptoms and prolonged physiological stress response to a social stressor.However, evidence of the buffering effect of local skin warming was limited to one marker of sympathetic activity, fingertip temperature, while other markers showed evidence suggestive of either a direct effect on sympathetic activity, or no effect, indicating a need for caution in interpreting these findings, pending further research in this area.The findings do emphasize the importance of considering the ways in which peripheral physiology, and especially the thermosensory system, is affected by, and could be recruited to treat, depression.Future research with a larger sample (increased statistical power) will enable more detailed analyses of continuous data (e.g., electrodermal activity and fingertip temperature), such as comparing the slopes of the measures over time during the recovery phase.Depression as a construct includes multiple subsets of symptoms, and future research should further investigate the specific impact of these symptom clusters on the effect of local skin warming on stress responses.Controlling for the influence of comorbid mental and somatic disorders in future studies will further articulate the specific influence of depression on affective responses to warm stimuli.Future research should also investigate the effect of local skin warming on subjective stress during stressor exposure (as against retrospective report), and whether severity of depressive symptoms moderates this relationship.Future studies should investigate the effects of local skin warming on other physiological responses to stressor exposures, including increases in circulating peripheral blood mononuclear cells and proinflammatory cytokines, such as interleukin 6, which are thought to be secondary to stress-induced activation of the sympathetic nervous system (Böbel et al., 2018;Miller and Raison, 2016), and are exaggerated in individuals with a diagnosis of depression (Pace et al., 2006).Finally, there is some evidence that activation of thermosensory pathways with cold stimuli may also have an antidepressant effect (e.g., Van Tulleken, Tipton et al., 2018).Indeed, preclinical research has demonstrated that exposure to cold temperature activates the same subsets of neurons in the central nervous system that are hypothesized to mediate the antidepressant effect of exposure to warm temperature (Kelly et al., 2011).Research on the effect of depression on responses to cold stimuli would clarify if changes to thermosensory functioning are limited to warm thermal stimuli.
Our findings raise the intriguing possibility that local skin warming could interrupt the prolonged stress response that depressed individuals experience following exposure to a stressor, by activating sympathoinhibitory regions in the brain (Hale et al., 2011).While it remains to be tested whether clinically depressed populations experience the same benefit of local skin warming observed in our study, there are some clinical implications of our findings to consider.For instance, there is evidence that an over-active sympathetic nervous system predicts a lack of response to standard psychological and pharmacological depression treatments (Pallich et al., 2022).If subsequent research strengthens the finding that local skin warming inhibits sympathetic activity for individuals with high levels of depressive symptomology, it could be explored as a possible adjunct treatment to facilitate inhibition of the sympathetic nervous system, possibly increasing the efficacy of standard treatments for depression.Finally, there is an emerging literature suggesting that individuals with clinical depression experience a marked antidepressant effect when exposed to whole-body hyperthermia (Hanusch et al., 2013;Janssen et al., 2016), but the mechanism of action is still unclear (Hanusch and Janssen, 2019).Our findings suggest that a possible mechanism of action for this effect may be changes to autonomic nervous system activity, though it remains to be determined whether the current findings using healthy individuals reporting levels of depressive symptoms that indicate they are at risk of a diagnosis of depression translate to a clinical population.

Fig. 2 .
Fig. 2. Graph showing individual data points for the CES-D-14, together with the mean and standard deviation for each group.The dashed line indicates the clinical cut-off for depression of 16 using the CES-D-14.Abbreviations: CES-D-14, Center for Epidemiological Studies -Depression Scale Revised 14-item Version; SE, Standard Error.

Fig. 3 .
Fig. 3. Graphs showing changes in (A) fingertip temperature, (B) electrodermal activity (EDA), and (C) salivary alpha amylase (sAA) following stressor exposure.Change in fingertip temperature was assessed by comparing the difference between a single timepoint immediately before the end of the baseline phase with a single timepoint immediately before the end of the stressor phase.Change in electrodermal activity was assessed by comparing average level of electrodermal activity during the baseline phase to the average level during the stressor phase.Change in sAA was assessed by comparing levels of sAA in a saliva sample taken immediately following the baseline phase to a saliva sample taken immediately following the stressor phase.Error bars indicate 1 SE.*p < .05,* **p < 0.001.

Fig. 4 .
Fig. 4. Scatterplots showing association between severity of depressive symptoms and percentage change in fingertip temperature from immediately post-stressor to (A) midway through the warming phase and (B) to the end of the warming phase.Note: increase in fingertip temperature represents decreased sympathetic nervous system activity.Abbreviations: CES-D-14, Center for Epidemiologic Studies Depression Scale Revised 14-item version.

Table 1
Correlations comparing depressive symptoms, covariates (age, gender and sleep disturbance), change in physiological indicators of sympathetic activity (i.e., from immediately post-stressor to warming phase), pleasantness of warming and level of negative affect and rumination following stressor exposure in control (C) and warming (W) conditions.

Table 3
Moderated Regression Analyses Assessing the Relationship Between Global Depressive Symptom Level and Changes in Physiological and Self-Report Measures from the Stressor Phase to the Warming Phase.