Exploring the Use of Thermal Infrared Imaging in Human

Stress Research
Veronika Engert1*., Arcangelo Merla2,3., Joshua A. Grant1, Daniela Cardone2,3, Anita Tusche1,
Tania Singer1
1 Max Planck Institute for Human Cognitive and Brain Sciences, Department of Social Neuroscience, Leipzig, Germany, 2 Institute of Advanced Biomedical Technologies
(ITAB), G. d’Annunzio Foundation, Chieti, Italy, 3 Department of Neuroscience and Imaging, G. d’Annunzio University, Chieti-Pescara, Italy

Abstract
High resolution thermal infrared imaging is a pioneering method giving indices of sympathetic activity via the contact-free
recording of facial tissues (thermal imprints). Compared to established stress markers, the great advantage of this method is
its non-invasiveness. The goal of our study was to pilot the use of thermal infrared imaging in the classical setting of human
stress research. Thermal imprints were compared to established stress markers (heart rate, heart rate variability, finger
temperature, alpha-amylase and cortisol) in 15 participants undergoing anticipation, stress and recovery phases of two
laboratory stress tests, the Cold Pressor Test and the Trier Social Stress Test. The majority of the thermal imprints proved to
be change-sensitive in both tests. While correlations between the thermal imprints and established stress markers were
mostly non-significant, the thermal imprints (but not the established stress makers) did correlate with stress-induced mood
changes. Multivariate pattern analysis revealed that in contrast to the established stress markers the thermal imprints could
not disambiguate anticipation, stress and recovery phases of both tests. Overall, these results suggest that thermal infrared
imaging is a valuable method for the estimation of sympathetic activity in the stress laboratory setting. The use of this non-
invasive method may be particularly beneficial for covert recordings, in the study of special populations showing difficulties
in complying with the standard instruments of data collection and in the domain of psychophysiological covariance
research. Meanwhile, the established stress markers seem to be superior when it comes to the characterization of complex
physiological states during the different phases of the stress cycle.

Citation: Engert V, Merla A, Grant JA, Cardone D, Tusche A, et al. (2014) Exploring the Use of Thermal Infrared Imaging in Human Stress Research. PLoS ONE 9(3):
e90782. doi:10.1371/journal.pone.0090782
Editor: Momiao Xiong, University of Texas School of Public Health, United States of America
Received August 9, 2013; Accepted February 5, 2014; Published March 27, 2014
Copyright: ß 2014 Engert et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by Max Planck Gesellschaft. The funders had no role in study design, data collection and analysis, decision to publish, or
preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: [email protected]
. These authors contributed equally to this work.

Introduction contact-free recording of facial tissues with an easy-to-hide thermal

camera helps avoid sources of unsystematic data variation (e.g.,
Everyone has experienced the ‘‘rush of blood to the head’’ when participants’ knowledge of being tested or the stressful installation
confronted with a stressful situation. This form of stress reaction, of recording equipment on the participants’ body). This opens up
i.e., skin blood flow, can be measured with the pioneering method exciting research opportunities in the study of special populations
of high resolution thermal infrared (IR-) imaging. IR-imaging (i.e., showing difficulties in complying with the standard instru-
estimates variations in autonomic activity reflected by a complex ments of data collection). Despite a growing interest in the method,
interplay of heat exchange processes involving skin tissue, inner IR-imaging has yet to find access to the field of stress research. We
tissue, local vasculature and metabolic activity [1–5]. In detail, here tested the use of IR-imaging in the stress laboratory setting. In
during threat or arousal the sympathetic nervous system causes detail, we examined the sensitivity of IR-imaging data (hereafter
sweat secretions that lubricate the skin, achieving elasticity [6,7] referred to as facial thermal imprints) to two widely used
and sustaining temperature homeostasis in prolonged periods of laboratory tests, the Cold Pressor Test (CPT) and the Trier Social
vigorous activity [8–11]. Furthermore, vasoconstriction of the Stress Test (TSST). The CPT [16] is a pain induction technique;
skin’s blood vessels protects the body from possible hemorrhage the TSST [17] a psychosocial challenge. Both are considered stress
and excessive blood loss during injury [12–15]. These physiolog- tests and reliably induce sympathetic and HPA-axis activity (for
ical occurrences cause skin temperature to fluctuate. Thus, by CPT-related evidence see [18–20]; for TSST-related evidence see
observing the thermal infrared signal one can infer autonomic [21–23]. Rather than use a single stressor and focus on test-retest
arousal and further attempt to differentiate between the two reliability we chose to cover a wider spectrum of stressors
competing subdivision of the autonomic nervous system. (psychosocial versus physical). This decision was driven by the
Compared to established stress markers like heart rate or the fact that it is difficult to achieve robust stress responses when
hypothalamic-pituitary-adrenal (HPA) axis end-product cortisol, repeatedly administering a single stress test.
the great advantage of IR-imaging is its non-invasiveness. The

PLOS ONE | www.plosone.org 1 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Fifteen healthy males underwent CPT and TSST in pseudo- past year), habitual smokers (.5 cigarettes/week) and individuals
randomized order. Women were excluded to avoid the confound- reporting chronic illness (e.g., cardiovascular or thyroid conditions)
ing effects of hormonal status on cortisol levels [24]. Facial thermal psychological disorders (e.g. depression or anxiety) or taking
imprints were collected across anticipation, stress and recovery medication that influences HPA-axis regulation were excluded
phases of both stress tests. To allow for a comparison of thermal from the study. 15 participants (mean age 26.18 years; SD 3.36),
imprints with established stress markers, we further assessed the all of which had college degrees and BMIs ranging between 21 and
participants’ heart rate (regulated by sympathetic and parasym- 25, were included in the study. The study was approved by the
pathetic influences), heart rate variability (HRV; permitting to Research Ethics Board of Leipzig University (ethics number: 360-
disentangle sympathetic and parasympathetic contributions to 10-13122010) and performed in agreement with the Declaration
heart rate) and finger temperature (indicative of adrenergic of Helsinki. All participants signed informed consent and could
sympathetic activity). The salivary enzyme alpha-amylase as an withdraw from the study at any time.
indirect indicator of adrenergic sympathetic activity and the
hormone cortisol as an indicator of HPA-axis activity were Cold Pressor Test
additionally sampled from saliva (see Table 1 for a summary of the The CPT [16] is a widely used paradigm that has been applied
dependent variables). Conventional statistical tests and advanced to the study of pain since the 1930s [25]. Participants place their
multivariate pattern analyses (MVPAs) were applied to analyze the non-dominant hand or forearm into a tank of cold water (0–10uC)
resulting set of dependent variables over time. The MVPAs aimed and are instructed to remove it when the sensation becomes
at exploring the relative sensitivity of thermal imprints and intolerable. Typically, a ceiling of several minutes for hand/
established stress markers for characterizing the different phases of forearm submersion is imposed. For the current study, the main
the stress cycle. unit of the CPT device consisted of a ‘‘Mobicool’’ cooler (model
In summary, we could show that the majority of the facial B40) with a volume of 40 L which allowed for automatic
thermal imprints were change-sensitive in CPT and TSST. temperature maintenance (at 3uC (+/21uC). A water pump
Moreover, and in contrast to the established stress markers, the (Esotec GmbH, model ‘‘Fontana’’ with a flow rate maximum of
thermal imprints correlated with stress-induced mood changes. 500 L/h) was mounted on the main unit and submerged to ensure
The established stress markers conversely had an improved ability a constant circulation of water around the immersed forearm.
to disambiguate anticipation, stress and recovery phases of both
Finally, a custom made submergible, electronic button was placed
stress tests.
inside the cooler to record the duration of time the participant
stayed in the water (while a ceiling time of 3 min was imposed).
Materials and Methods
Participants Trier Social Stress Test
Male participants between 21 and 33 years of age were The TSST [17] is the most frequently administered psycholog-
recruited by posting ads on an electronic billboard of Leipzig city. ical paradigm to stimulate an endocrine stress response in the
Women were excluded to avoid the confounding effects of laboratory setting. Compared to other social evaluative laboratory
hormonal status on cortisol levels [24]. Given a potential influence stressors, it provokes the most robust HPA-axis response [26]. It
on our dependent variables, information about recreational drug also increases hemodynamic responses [23,27] and the activity of
use, medical and psychological history were assessed in a salivary alpha-amylase in a pattern resembling that of norepi-
telephone interview. Regular recreational drug users (cannabis nephrine [28]. The TSST consists of three phases. A preparatory
within the last two months, other recreational drugs within the anticipation phase (5 min) is followed by a stress phase in which

Table 1. Summary of the dependent variables.

Variable Description

Facial thermal imprints

FHT Forehead temperature
CRT Corrugator temperature
POBT Periorbital temperature
NTT Nose tip temperature
POT Perioral temperature
CHT Chin temperature
Established stress markers
HR Heart rate
LF Low frequency heart rate variability
HF High frequency heart rate variability
LH/HF Ratio of low to high frequency heart rate variability
FT Finger temperature
AA Alpha-amylase
COR Cortisol

doi:10.1371/journal.pone.0090782.t001

PLOS ONE | www.plosone.org 2 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

participants are asked to give a 5-min free speech for a simulated Biopac recording unit. All data were sampled at 500 Hz and
job interview followed by 5 min of difficult mental arithmetic. The analyzed with the software Acknowledge (version 3.9).
entire performance is audio- and videotaped and held in front of Heart rate (beats per min), HRV components (s2/Hz) and finger
an evaluative panel that does not provide any form of feedback. temperature change (slope) were extracted for each experimental
The TSST is usually preformed in a standing position. To provide phase (5 min for baseline, anticipation and recovery). For the
a better comparability with the CPT, we here asked participants to TSST stress phase, interview and mental arithmetic were averaged
sit down for the actual testing. to one 5-min stressor. For the CPT stress phase, a 5-min window
including a maximum of 3 min of actual cold pain and a
Assessment and analysis of thermal IR images minimum of 2 min of recovery was used. Hence, the cardiovas-
IR-imaging was performed using a digital thermal camera cular and thermal responses to the cold pain were likely
(FLIR SC3000, FlirSystems, Sweden) with a Focal Plane Array of underestimated.
3206240 QWIP detectors, 0.02 sec time resolution, 0.02 K
temperature sensitivity and the capability to collect thermal Assessment and analysis of salivary alpha-amylase
radiation in the 8–9 mm band. In order to null noise effects The salivary enzyme alpha-amylase which is mainly involved in
related to the sensor drift/shift dynamics and optical artifacts, the the digestion of starch in the oral cavity has received increasing
camera response was blackbody-calibrated. The sampling rate was attention as an indicator of sympathetic activity within the past
set at 5 frames/sec. Variations in cutaneous temperature of facial years [37]. Pharmacological studies have shown that its release is
regions of interest were analyzed using customized Matlab dependent on b-adrenergic transmission [38]. Using Salivette
programs (http://www.mathworks.com). Our primary regions of collection devices (Sarstedt, Nümbrecht, Germany), we sampled
interest, 1) the nose, 2) the corrugator (supraorbital) and 3) the alpha-amylase at 10 min before and at 10 and 20 min after the
forehead, were selected based on previous studies in primates and onset of CPT and TSST to fully capture peak activity and return
humans, and have been shown to respond to emotional or to baseline. Participants were instructed to place the saliva
distressing stimuli; responses in periorbital, perioral and chin collection swabs in their mouths and to refrain from chewing for
regions were also analyzed [1,2,5,29–31]. exactly 2 min [39]. Samples were stored at 230uC until analysis.
Initially, the thermal imprints were visually inspected to ensure Alpha-amylase activity (calculated and expressed in U/ml) was
adequate quality for all recordings. We then corrected the determined using an enzyme kinetic method [40,41].
thermograms for body movements. In cases of marked head
rotation, we skipped to the next available undistorted frame.
Assessment and analysis of salivary cortisol
Displacements between images were corrected frame by frame
Cortisol is the most important endocrine stress marker. It was
using anatomical landmarks based on the participants’ nose
sampled using the same collection devices as for alpha-amylase.
profiles [32]. Visual inspection of the temperature time-course in
Samples were taken at 10 min before stressor onset, at 10 min
the regions of interest evidenced short-lasting artifacts (potentially
after stressor onset and in 10-min intervals thereafter (until 70 min
associated with transitions from nasal to oral breathing or
after stressor onset) to fully capture hormone peak and return to
vocalizations) which did not affect the long-term temperature
baseline. Cortisol concentration (calculated and expressed in
evolution. Corrupted segments were filtered out following a
nmol/l) was determined using a time-resolved fluorescence
procedure by Iriarte et al. [33]. Data were then processed with a
immunoassay [42], with intra- and interassay variabilities of less
band-stop filter to remove breathing-dependent oscillations. To
than 10% and 12%.
provide evidence for a proper acclimatization of participants, we
verified at the intra-individual level that the skin temperature did
not vary significantly during the 5-min baseline phase. Mood assessment
Stress-induced mood changes were assessed using two Visual
Assessment and analysis of heart rate, heart rate Analogue Rating Scales (VARS) asking how anxious and angry
participants currently felt on a scale ranging from 1 (not at all) to
variability and finger temperature
10 (extremely). These VARS were administered at baseline (2
Heart rate and HRV were derived from a continuous
10 min) and at the end of the anticipation phase/immediately
electrocardiogram (ECG) recorded with Biopac equipment
before the onset of acute stress (0 min).
(Biopac Systems, Inc.) using a three-electrode array. The peak of
the R-wave was detected automatically to obtain a continuous
RR-interval tachogram, which was corrected for artifacts manu- Study design
ally when necessary. Heart rate was defined as the reciprocal of the Participants underwent CPT and TSST on two consecutive
RR-interval in units of beats per minute. Although heart rate days in pseudo-randomized order. Testing took place at identical
combines sympathetic and parasympathetic contributions, adren- time slots on both days, either between 10:00 and 12:00 am or
ergic sympathetic influences are dominant during stress. Following between 4:00 and 6:00 pm. Participants tested in the morning
guidelines put forth by the Task Force of the European Society for group were asked to get up at least 2 hours prior to coming to the
Cardiology [34], frequency domain analysis techniques were laboratory. Figure 1 gives a detailed overview of the study design.
employed to investigate variations in the heart rate time series
based on beat-to-beat intervals. The two main frequency bands of Statistical analysis
this signal are referred to as the low (0.04 to 0.15 Hz) and high Scoring of dependent variables. For the temperature
frequency (0.15 to 0.4 Hz) components. High frequency compo- measures, raw scores were defined as the mean slope per phase
nents are believed to be influenced by parasympathetic tone while (baseline, anticipation, stress and recovery). For heart rate and
low frequency components are a combination of sympathetic and HRV components, raw scores were defined as the arithmetic
parasympathetic activities [35]. The ratio of low to high frequency mean of the data within each experimental phase. To account for
components is used as a measure of sympathovagal balance individual differences in baseline activity, reactivity and rebound
[34,35]. Finger temperature, an indicator of sympathetic activity scores were calculated as the change from baseline in each
[36], was sampled from the third finger of the right hand with a measure and phase. Alpha-amylase and cortisol data were initially

PLOS ONE | www.plosone.org 3 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Figure 1. Timeline of the experiment. For the assessment of alpha-amylase (AA) and cortisol (COR), saliva was sampled at discrete measurement
timepoints between 210 min and 20 min/70 min relative to stressor onset. Visual Analogue Rating Scales (VARS) assessing stress-induced mood
changes were administered at baseline (210 min) and at the end of the anticipation phase/immediately before the onset of acute stress (0 min). The
measurement period for the thermal infrared imaging (IRI) and Biopac assessments began at 210 min and ended at 15 min relative to stressor onset.
doi:10.1371/journal.pone.0090782.g001

Box-Cox-transformed to account for non-normal distribution and, differential response patterns across multiple variables, a quality
in the case of the cortisol data, concentration-dependent variance which allows them to reliably characterize complex physiological
due to HPA axis dynamics, immunoassay interference and and mental states. So far, they have been applied successfully in
measurement error [43]. Reactivity and rebound scores were diverse research areas ranging from neuroscience to chemistry
calculated using the 210 min baseline, the mean peak (alpha- [44–47].
amylase: 10 min; cortisol: 30 min [CPT], 20 min [TSST]) and the MVPAs were applied separately for CPT and TSST. Four sets
final (alpha-amylase: 20 min; cortisol: 70 min) samples of the of analyses were run. The first analysis tested whether the response
respective time series. Anticipatory increases are consequently patterns of the established stress markers collected throughout
missing for the alpha-amylase and cortisol data. For the mood anticipation, stress and recovery phases of the CPT allowed
ratings, change from baseline calculations used the baseline and predicting which phase of the stress cycle a participant was
the post-anticipation/pre-stress measures at 210 and 0 min. To currently experiencing (MVPA 1). The same analysis was then
conduct multivariate pattern analyses, the calculated difference performed using the response patterns of the thermal imprints
scores were additionally z-standardized. collected throughout all CPT phases (MVPA 2). In a parallel
Univariate ANOVAs. We initially examined which thermal manner, MVPAs 3 and 4 examined whether established stress
imprints and established stress markers were sensitive for change markers or thermal imprints formed characteristic responses
over time in CPT and TSST. One-way repeated measures profiles to the phases of the TSST. Since anticipation scores for
ANOVAs were calculated for the dependent variables (raw or alpha-amylase and cortisol were not collected, MVPAs 1 and 3
logged scores) in each stress test. As within-subject factors, were based on only five established stress markers.
sampling time-points (210 to 20 and 70 min, respectively) were For each MVPA, three separate N-dimensional pattern vectors
entered for alpha-amylase and cortisol, and phases (baseline, were created that contained the N (IR-imaging or established)
anticipation, stress and recovery) for all other dependent variables. variables per participant obtained throughout the three test phases
Violations of the assumption of sphericity were adjusted using the (Figure 2). Using a conservative cross-validation procedure, the
Huynh-Feldt correction. Critical p-values were corrected for resulting pattern vectors of all except one participant were used to
multiple comparisons using the Bonferroni adjustment (a/n) per train a linear support vector machine classifier (i.e., estimate a
stress test. Partial eta-squared (gp2) was used as an effect size decision boundary that separates phase-specific response patterns)
estimate. Significant effects were further investigated using simple with a fixed regularization parameter C = 1 (‘‘training data’’). The
contrasts with the baseline as reference category. classification accuracy of the response patterns as belonging to
Correlation analyses. To examine whether facial thermal either the anticipation, stress or recovery phase was determined
imprints and established stress markers captured similar aspects of depending on which phase the respective response patterns
the stress response, Pearson correlations between the change- actually did belong to. This procedure was repeated 15 times,
sensitive candidates as found in the univariate ANOVAs were always omitting the pattern vectors of another participant from the
calculated for each stress test and phase. As indicators of training. The average percentage of correct classification across
psychophysiological covariance, correlations between the the resulting 15-fold cross-validation steps represented the
change-sensitive candidates and anticipatory mood changes were predictive strength of the multivariate response patterns for the
calculated per stress test and test phase. Critical p-values were three test phases. Using a Wilcoxon test, the average classification
again corrected for multiple comparisons using the Bonferroni accuracy of each analysis (MVPA 1–4) was then statistically tested
adjustment (a/n). against a respective permutation distribution of classification
Multivariate pattern analyses. The next set of analyses accuracies. In so doing, we determined the probability with which
determined the predictive power of our two sets of dependent the average classification accuracy was achieved by chance only.
variables for the three phases of the stress cycle. In detail, we Given three options (phases) to choose from, chance level was at
examined how well the composite multivariate response patterns 33%. Using 1000 repetitions, each permutation distribution was
of either the thermal imprints or the established stress markers obtained by randomly assigning the ‘‘training data’’ to one of the
characterized, and in consequence predicted, a specific test phase. three test phases. The MVPAs were performed using MATLAB
For this purpose we used an advanced multivariate pattern R2012b and the libSVM toolbox (http://www.csie.ntu.edu.tw/
analysis approach that was developed in machine learning. ,cjlin/libsvm/). All remaining tests were performed using the
MVPAs are considered particularly sensitive to the detection of Predictive Analysis Software (PASW) version 17.

PLOS ONE | www.plosone.org 4 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Figure 2. Multivariate pattern analysis of phases in the stress cycle. A Methods. For each subject (S1…15), responses patterns were created
that included the n measures (M) of either established stress markers or thermal imprints of a particular test phase (anticipation, stress, recovery).
Response patterns of all but one subject were used to train a classifier to identify physiological response profiles that are characteristic for each of the
test phases. This was the basis for the subsequent prediction of the response patterns of the remaining subject as belonging to either anticipation,
stress of recovery phase. B Results. Multivariate response patterns of established stress measures obtained for the CPT (MVPA 1, left) reliably predicted
the current test phase of a subject well above chance (56% classification accuracy, p,.001). Patterns of thermal facial imprints in the CPT (MVPA 2,
right), on the other hand, did not significantly predict the correct phase (40% classification accuracy, p = .13). The histograms display the permutation
distribution of classification accuracies achieved by chance (average 33%).
doi:10.1371/journal.pone.0090782.g002

Results and finger temperatures increased during recovery. Heart rate

increased during anticipation and increased further during stress.
Univariate ANOVAs Alpha-amylase levels increased from baseline to 10 min after
For each dependent variable, change sensitivity in CPT and stressor onset. Cortisol levels increased until they reached a peak at
TSST was examined in altogether 13 ANOVAs per stress test 20 min after stressor onset and decreased again thereafter.
(critical p-values were lowered to .004). Both thermal imprints and
established stress markers were sensitive to depict change in CPT Correlation analyses
and TSST (see Table 2 and Figure 3 for the summary statistics and No significant correlations between thermal imprints and
time courses of all variables). In the CPT, nose tip temperature, established stress markers were found in anticipation, stress and
perioral temperature, heart rate and finger temperature showed recovery phases of either the CPT or the TSST (Table 3; based on
significant changes over time. Posthoc tests revealed that none of the maximum number of change-sensitive measures per stress test
the variables changed from baseline to anticipation. Relative to the and test phase, critical p-values were lowered to #.005 for the
baseline, nose tip, perioral and finger temperatures decreased anticipation phase of the TSST, the stress phases of both tests and
during stress and nose tip temperature increased during recovery. to p#.050 and #.017 for the recovery phases of CPT and TSST
Inversely, heart rate increased during stress and decreased during in order to control for multiple comparisons). Even within each
recovery. In the TSST, corrugator temperature, nose tip group of measures, significant associations were scarce. In detail,
temperature, perioral temperature, chin temperature, heart rate, in the CPT, relative-to-baseline changes in nose tip and perioral
finger temperature, alpha-amylase and cortisol showed significant temperatures during acute stress correlated at r = .60 (p = .007). In
changes over time. Posthoc tests revealed that relative to the the TSST, relative-to-baseline changes in corrugator and chin
baseline, corrugator, nose tip, chin and finger temperatures temperatures during anticipation correlated at r = .62 (p = .005),
decreased during anticipation. Chin and finger temperatures changes in heart rate and cortisol during acute stress correlated at
remained significantly decreased during stress. Nose tip, perioral

PLOS ONE | www.plosone.org 5 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Table 2. Univariate statistics (one-way repeated measures ANOVAS with simple contrasts) for raw (logged for alpha-amylase and
cortisol) scores over time in CPT and TSST.

Variable F (df), pa, gp2 F(df), P for post-hoc tests (simple contrasts)b

BL-ANT BL-STR BL-REC

CPT
Facial thermal imprints
FHT 1.60 (3, 42), ..10, .10
CRT 4.57 (3, 42), .009, .23
POBT 0.25 (3, 42), ..30, .02
NTT 13.31 (3, 42), ,.001, .47 6.23 (1, 14), p = .025 (Q) 14.54 (1, 14), p = .002 (q)
POT 7.48 (3, 42), .004, .33 9.61 (1, 14), p = .007 (Q)
CHT 3.11 (3, 42), .033, .17
Established stress markers
HR 8.28 (3, 42), .001, .37 6.71 (1, 14), p = .021 (q) 6.99 (1, 14), p = .019 (Q)
LF 1.75 (3, 42), ..10, .11
HF 1.73 (3, 42), .094, .11
LF/HF 0.44 (3, 42), .30, .03
FT 19.18 (3, 42), ,.001, .56 25.16 (1, 14), p,.001 (Q)
AA 4.03 (2, 28), .014, .21
COR 7.24 (6, 84), .001, .33 BL-20 min: 12.94 (1, 14), p = .003 (q)
BL-30 min: 7.49 (1, 14), p = .015 (q)
TSST
Facial thermal imprints
FHT 4.10 (3, 42), .012, .21
CRT 5.49 (3, 42), .003, .27 7.43 (1, 14), p = .016 (Q)
POBT 1.47 (3, 42), ..10, .09
NTT 16.42 (3, 42), ,.001, .52 8.52 (1, 14), p = .011 (Q) 16.58 (1, 14), p = .001 (q)
POT 15.76 (3, 42), ,.001, .51 14.74 (1, 14), p = .002 (q)
CHT 9.24 (3, 42), ,.001, .38 13.52 (1, 14), p = .002 (Q) 24.63 (1, 14), p,.001 (Q)
Established stress markers
HR 29.11 (3, 42), ,.001, .68 22.92 (1, 14), p,.001 (q) 55.35 (1, 14), p,.001 (q)
LF 3.60 (3, 42),.011, .22
HF 2.28 (3, 42),.067, .15
LF/HF 2.10 (3, 42),.058, .14
FT 46.94 (3, 42), ,.001, .76 44.09 (1, 14), p,.001 (Q) 28.71 (1, 14), p,.001 (Q) 38.51 (1, 14), p,.001 (q)
AA 7.61 (2, 28), .001, .34 BL-10 min: 17.79 (1, 14), p,.001 (q)
COR 17.79 (6, 84), ,.001, .54 BL-10 min: 11.93 (1, 14), p = .004 (q)
BL-20 min: 24.23 (1, 14), p,.001 (q)
BL-30 min: 13.93 (1, 14), p = .002 (q)
BL-40 min: 4.86 (1, 14), p = .044 (q)

Note. AA: alpha-amylase; ANT: anticipation; BL: baseline; CHT: chin temperature; COR: cortisol; CRT: corrugator temperature; FHT: forehead temperature; FT: finger
temperature; HF: high frequency heart rate variability; HR: heart rate; LF: low frequency heart rate variability; LF/HF: ratio of low to high frequency heart rate variability;
NTT: nose tip temperature; POBT: periorbital temperature; POT: perioral temperature; REC: recovery; STR: stress.
a
after multiple comparisons correction p#.004.
b
q: ANT, STR, REC . BL; Q: ANT, STR, REC , BL.
doi:10.1371/journal.pone.0090782.t002

r = .73 (p = .001) and changes in nose tip and perioral tempera- and TSST, and to .025 and .017 for the CPT and TSST recovery
tures during recovery correlated at r = .82 (p,.001). phases). Anxiety was inversely correlated with anticipation-
Psychophysiological covariance (i.e. covariance of stress-induced induced (r = 2.65, p = .003) and acute stress-induced (r = 2.60,
psychological and physiological responses) emerged only for the p = .007) changes in chin temperature. Anger correlated with
thermal imprints and only in the TSST (Table 4; to correct for anticipation-induced changes in corrugator (r = 2.69, p = .002)
multiple comparisons, critical p-values were lowered to .010 for and chin (r = 2.72, p = .001) temperatures.
the anticipation phase of the TSST, the stress phases of both CPT

PLOS ONE | www.plosone.org 6 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Figure 3. Means and ±SEM of all dependent variables (raw or logged scores) during baseline, anticipation, stress and recovery
(sampling time points for alpha-amylase and cortisol) of CPT and TSST.
doi:10.1371/journal.pone.0090782.g003

Multivariate pattern analyses discriminate the phases of the stress cycle and reliably predict
Using multivariate pattern analyses (MVPAs), we investigated whether a participant was currently anticipating a stressor,
the capacities of thermal imprints and established stress markers to exposed to a stressor or in the process of recovering from it. In

PLOS ONE | www.plosone.org 7 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Tables 3. Pearson correlations between the change-sensitive candidates for each test phase in CPT and TSST.

Variable r, pa

CPT
Stress
NTT NTT POT HR FT COR
POT 1 .60, .007 2.26, .172 .34, .098 2.41, .058
HR 1 .30,.140 .10, .357 .47, .032
FT 1 .13, .327 .20, .236
COR 1 .02, .478
NTT 1
Recovery
NTT HR
NTT 1 2.13, .320
HR 1
TSST
Anticipation
NTT CRT CHT HR FT
NTT 1 .02, .469 .14, .198 .02, .473 .49, .028
CRT 1 .62, .005 .28, .160 .11, .227
CHT 1 .22, .216 .03, .460
HR 2.19, .236
FT 1
Stress
CHT HR FT AA COR
CHT 1 .30, .141 2.17, .263 2.39, .068 2.11, 349
HR 1 2.01, .484 .29, .137 .73, .001
FT 1 .17, .262 .01, .496
AA 1 .37, .072
COR 1
Recovery
NTT POT FT
NTT 1 .82, ,.001 .22, .209
POT 1 2.12, .335
FT 1

Note. AA: alpha-amylase; CHT: chin temperature; COR: cortisol; CRT: corrugator temperature; FT: finger temperature; HR: heart rate; NTT: nose tip temperature; POT:
perioral temperature.
a
after multiple comparisons correction, p#.005 for the anticipation phase of the TSST, the stress phases of both tests and p#.050/.017 for the CPT/TSST recovery
phases.
doi:10.1371/journal.pone.0090782.t003

the CPT, patterns of established stress markers were found to While the response patterns of established stress markers allowed
reliably characterize phase-specific stress responses and thus to reliably predict the correct test phase well above chance (56%,
predict test phases well above the chance level of 33% (56% p,.001; MVPA 3), this was not possible for the response patterns
average classification accuracy across participants, p,.001; of thermal imprints (38%, p.0.40; MVPA 4). Taken together,
MVPA 1) (Figure 2). In other words, the established stress markers there was not enough specificity in the response patterns of the
formed unique response profiles for anticipation, stress and thermal imprints to give each phase its discriminable profile.
recovery phases that could be used to correctly predict which
phase of the stress cycle a person was currently exposed to. Discussion
Patterns of thermal imprints in the CPT, on the other hand, did
not allow discriminating and correctly predicting a corresponding The goal of this pilot study was to test the use of thermal IR-
test phase (MVPA 2): The average classification accuracy of 40% imaging in human stress research. Hereby, we compared how
facial thermal imprints and established stress markers performed
was not significantly different from predictions achieved by chance
in two frequently applied laboratory stress tests, the CPT and the
only (p = .13) (Figure 2). This means that the thermal imprints
TSST.
formed fairly similar response profiles for anticipation, stress and
Univariate ANOVAs revealed that indeed most thermal
recovery phases. Parallel results were obtained for the TSST.
imprints changed significantly from baseline in the course of

PLOS ONE | www.plosone.org 8 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

Table 4. Pearson correlations between the change-sensitive candidates and the anticipatory mood changes (anxiety and anger)
per test phase in CPT and TSST.

Variable r, pa

Anxiety Anger

CPT
Stress
NTT .45, .042 .54, .015
POT .23, .194 .28, .151
HR 2.36, .097 2.34, .110
FT 2.45, .042 .24, .185
COR .05, .428 2.01, .491
Recovery
NTT .13, .312 .17, .260
HR .42, .060 2.20, .243
TSST
Anticipation
NTT 2.16, .280 .00, .500
CRT 2.40, .065 2.69, .002
CHT 2.65, .003 2.72, .001
HR 2.13, .315 2.32, .114
FT .15, .277 2.03, .449
Stress
CHT 2.60, .007 2.51, .022
HR 2.32, .115 2.46, .037
FT .11, .340 .17, .262
AA .30, .122 .14, .297
COR 2.03, .455 2.04, .441
Recovery
NTT 2.05, .430 2.02, .466
POT 2.12, .323 2.14, .304
FT 2.03, .461 .04, .447

Note. AA: alpha-amylase; CHT: chin temperature; COR: cortisol; CRT: corrugator temperature; FT: finger temperature; HR: heart rate; NTT: nose tip temperature; POT:
perioral temperature.
a
after multiple comparisons correction, p#.010 for the anticipation phase of the TSST, the stress phases of both tests and p#.025/.017 for the CPT/TSST recovery
phases.
doi:10.1371/journal.pone.0090782.t004

CPT and/or TSST. More specifically, out of six facial thermal of 15 is at the lower end for computing correlations. However, the
imprints, only forehead and periorbital temperatures turned out to lack of association between stress markers is also a well-known
be overall stress-insensitive. Among the remaining markers issue in stress research and has been attributed to the fact that
(corrugator, nose tip, perioral and chin temperatures) the measure different markers originate from different stress-related systems
most frequently yielding significant results in both stress tests was and underlie different time courses [21,23,48]. Accordingly, also
nose tip temperature. Regarding the superior change sensitivity of within each group of measures, significant correlations were
the nasal region, our findings are in line with previous studies in scarce. We conclude that the thermal imprints and established
which infants [31], children [5] and adults [29] were confronted stress markers capture unique aspects of physical and psycholog-
with emotionally arousing stimuli. However, the observed ical stress responses.
insensitivity of the forehead and periorbital regions does not Correlation analyses additionally showed overall weak psycho-
concur with the literature [2,29] and may be specific to the present physiological covariance (i.e. covariance of stress-induced psycho-
context in stress research. Overall, the facial thermal imprints logical and physiological responses). Only corrugator and chin
yielded a similar range of small to medium effect sizes than the temperatures but none of the established stress markers correlated
established stress markers. with stress-induced mood changes. Again, this result may, at least
Correlation analyses revealed that there were no significant partially, be due to our relatively small number of participants but
associations between the change-sensitive thermal imprints and simultaneously reflects a phenomenon known in the literature
established stress markers in either of the tests phases or stress tests. [48250] and has been ascribed to both the time lag between
Importantly, this does not automatically suggest that the thermal psychological and physiological stress responses and general
imprints provide an invalid measure of stress. For one thing, an N difficulties in the assessment of subjective stress responses by self-

PLOS ONE | www.plosone.org 9 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

report methods [50,51]. In this regard, it is especially interesting While nose tip and perioral temperatures are the strongest
that the facial thermal imprints (precisely, corrugator and chin candidates when it comes to predicting change, correlation
temperatures) did capture the experiential nature of stress. analyses show that corrugator and chin temperatures also capture
Possibly, because of evolving at the very surface of our body, the experiential nature of anticipatory and acute stress. In contrast
facial temperature variations are more easily discerned than other, to the established stress markers, the thermal imprints form a non-
internally evolving, physiological processes and consequently exert specific set of measures that respond in a similar way to the
an important influence on our judgments of stress-induced mood different phases of the stress cycle. We conclude that when looking
changes. to characterize complex physiological states over the time course
MVPAs revealed that in contrast to the established autonomic of a given stress cycle, the established autonomous stress markers
stress markers, the thermal imprints were not able to reliably (e.g., heart rate and finger temperature) are more suitable than the
discriminate between anticipation, stress and recovery phases of thermal imprints. If, however, research aims for covert recordings
CPT and TSST. In other words, there was not enough specificity or focusses on special populations which show difficulties in
in their multivariate response patterns to give each phase its complying with the standard instruments of data collection,
discriminable profile. We speculate that the thermal imprints contact-free, non-invasive thermal IR-imaging may be the method
reflect the general, unspecific arousal that underlies the entire of choice. The use of IR-imaging may also be particularly
stress experience and is hence common to all phases of the stress beneficial in the domain of psychophysiological covariance
cycle. The established autonomic stress markers, on the other research. Modern thermal imaging systems can provide adequate
hand, seem to form a more heterogeneous group as they pick up temperature sensitivity and spatial resolution without being
on the specific physiological patterns associated with only dramatically expensive (the cost of an average performance system
anticipation, stress or recovery phases in both stress tests. does not exceed the price of good-quality biosignal recording
Importantly, cortisol as the hallmark endocrine biomarker of systems). Also, the employment of thermal IR-imaging does not
stress was not included in this set of analysis and therefore does not require any time-consuming preparations or special skills. The
contribute to the heterogeneity of the group of established stress
field would profit from the development of software packages
markers.
facilitating data analysis, however. Overall, this technique could be
Several methodological limitations have to be addressed. One
easily integrated in neuro-psycho-physiology experimental settings.
critical point concerns the small sample size of N = 15. Until
Our analyses additionally give an overview of the potency with
replicated in a bigger sample, data have to be considered
which CPT and TSST trigger responses in the different phases of
preliminary. Another critical point is the lack of a control
the stress cycle, and how a large range of stress markers respond
condition to depict the course of normal fluctuations in facial
per stress test and phase. This overview may serve as a useful tool
thermal activity over time. However, two facts speak against the
possibility that we measured normal variations rather than stress- for the targeted selection of the most potent combinations of tests
induced change. For one thing, we found response cycles in most and markers in future research. Both the ANOVA and the
of our dependent variables, i.e. participants started from a correlation results, for example, illustrate that a psychosocial
baseline, responded to the experimental manipulation and then challenge is more qualified than a physical challenge to bring out
returned back to their baseline. Additionally, finding correlations anticipatory stress. In the study of anticipatory stress, focusing on
between thermal imprints and stress-induced psychological the TSST and assessing heart rate and temperature indices would
responses would have been unlikely in the context of normal thus be indicated. Finally, we present a new multivariate analysis
fluctuations in thermal activity. The fact that exclusion criteria approach in behavioral stress research. The claim has been raised
were determined by telephone interview is another limitation. that to accurately represent the multidimensional nature of stress,
Ideally, participants should have been screened for psychological multiple measures across multiple branches of the stress system
issues and somatic conditions by use of a detailed clinical should be assessed [54]. MVPAs provide an elegant method for
diagnostic interview and a medical examination, respectively. exploring complex data sets: Instead of considering the correla-
Lastly, the fact that we used two different time windows for testing tions between single variables, they make predictions based on the
– one in the late morning and one in the afternoon – is not ideal underlying patterns across numerous variables. In the field of stress
given that especially cortisol follows a strong circadian rhythm research, the application of MPVAs to predict vulnerability for
[52,53]. Although it was assured for all participants that testing psychopathology based on various indices of stress sensitivity, for
took place at identical time slots on both days, non-significant example, would constitute an exiting future usage of this
results may nevertheless be due to differences in baseline values technique.
and excitability of the examined physiological systems in the
morning versus afternoon hours. Author Contributions
In summary, this study makes several important contributions to
Conceived and designed the experiments: VE TS. Performed the
the literature. We provide a first-time verification of the usefulness experiments: VE AM JAG DC. Analyzed the data: VE AM DC AT.
of thermal IR-imaging in the classical stress laboratory setting. Contributed reagents/materials/analysis tools: AM TS. Wrote the paper:
Thermal imprints (precisely, corrugator, nose tip, perioral and VE AM JAG AT TS.
chin temperatures) reliably depict stress-induced change over time.

References
1. Shastri D, Merla A, Tsiamyrtzis P, Pavlidis I (2009) Imaging facial signs of 4. Pavlidis I, Tsiamyrtzis P, Shastri D, Wesley A, Zhou Y, et al. (2012) Fast by
neurophysiological responses. IEEE Trans Biomed Eng 56: 477–484. nature - how stress patterns define human experience and performance in
2. Merla A, Romani GL (2007) Thermal signatures of emotional arousal: a dexterous tasks. Sci Rep 2: 305.
functional infrared imaging study. Conf Proc IEEE Eng Med Biol Soc 2007: 5. Ebisch SJ, Aureli T, Bafunno D, Cardone D, Romani GL, et al. (2012) Mother
247–249. and child in synchrony: thermal facial imprints of autonomic contagion. Biol
3. Pavlidis I, Eberhardt NL, Levine JA (2002) Seeing through the face of deception. Psychol 89: 123–129.
Nature 415: 35. 6. Anbar M (2002) Assessment of physiologic and pathologic radiative heat
dissipation using dynamic infrared imaging. Ann N Y Acad Sci 972: 111–118.

PLOS ONE | www.plosone.org 10 March 2014 | Volume 9 | Issue 3 | e90782

 Thermal Infrared Imaging in Human Stress Research

7. Kawasaki H, Kaufman O, Damasio H, Damasio AR, Granner M, et al. (2001) 32. Dowdall N, Pavlidis I, Tsiamyrtzis P (2007) Coalitional tracking. Computer
Single-neuron responses to emotional visual stimuli recorded in human ventral Vision and Image Understanding: 205–219.
prefrontal cortex. Nat Neurosci 4: 15–16. 33. Iriarte J, Urrestarazu E, Valencia M, Alegre M, Malanda A, et al. (2003)
8. Gleeson M (1998) Temperature regulation during exercise. Int J Sports Med 19 Independent component analysis as a tool to eliminate artifacts in EEG: a
Suppl 2: S96–99. quantitative study. J Clin Neurophysiol 20: 249–257.
9. McEwen BS (2000) The neurobiology of stress: from serendipity to clinical 34. Task Force of the European Society for Cardiology TNASoP and Electrophys-
relevance. Brain Res 886: 172–189. iology (1996) Heart Rate Variability: Standards of measurements, physiological
10. Chien S (1967) Role of the sympathetic nervous system in hemorrhage. Physiol interpretation, and clinical use. Circulation 93: 1043–1065.
Rev 47: 214–288. 35. Malik M (1995) Geometrical methods for heart rate variability assessment; Malik
11. Cannon WB (1929) Organization for physiological homeostasis. Physiological M, . Camm AJ, editors. Armonk NY: Futura Publishing Company.
Reviews 9: 399–431. 36. Kistler A, Mariauzouls C, von Berlepsch K (1998) Fingertip temperature as an
12. Haddy FJ, Overbeck HW, Daugherty RM Jr. (1968) Peripheral vascular indicator for sympathetic responses. Int J Psychophysiol 29: 35–41.
resistance. Annu Rev Med 19: 167–194. 37. Nater UM, Rohleder N (2009) Salivary alpha-amylase as a non-invasive
13. Pearce WJ, D’Alecy LG (1980) Hemorrhage-induced cerebral vasoconstriction
biomarker for the sympathetic nervous system: current state of research.
in dogs. Stroke 11: 190–197.
Psychoneuroendocrinology 34: 486–496.
14. Vatner SF (1974) Effects of hemorrhage on regional blood flow distribution in
38. van Stegeren A, Rohleder N, Everaerd W, Wolf OT (2006) Salivary alpha
dogs and primates. J Clin Invest 54: 225–235.
amylase as marker for adrenergic activity during stress: effect of betablockade.
15. Calvin KL, Duffy VG (2007) Development of a facial skin temperature-based
methodology for non-intrusive mental workload measurement. Occupational Psychoneuroendocrinology 31: 137–141.
Ergonomics 7: 83–94. 39. Rohleder N, Nater UM (2009) Determinants of salivary alpha-amylase in
16. Hines EA, Brown GE (1932) A standard stimulus for measuring vasomotor humans and methodological considerations. Psychoneuroendocrinology 34:
reactions: its application in the study of hypertension. Proc Staff Meet Mayo Clin 469–485.
7: 332. 40. Lorentz K, Gutschow B, Renner F (1999) Evaluation of a direct alpha-amylase
17. Kirschbaum C, Pirke KM, Hellhammer DH (1993) The ‘Trier Social Stress assay using 2-chloro-4-nitrophenyl-alpha-D-maltotrioside. Clin Chem Lab Med
Test’—a tool for investigating psychobiological stress responses in a laboratory 37: 1053–1062.
setting. Neuropsychobiology 28: 76–81. 41. Winn-Deen ES, David H, Sigler G, Chavez R (1988) Development of a direct
18. Mourot L, Bouhaddi M, Regnard J (2009) Effects of the cold pressor test on assay for alpha-amylase. Clin Chem 34: 2005–2008.
cardiac autonomic control in normal subjects. Physiol Res 58: 83–91. 42. Dressendorfer RA, Kirschbaum C, Rohde W, Stahl F, Strasburger CJ (1992)
19. Schwabe L, Haddad L, Schachinger H (2008) HPA axis activation by a socially Synthesis of a cortisol-biotin conjugate and evaluation as a tracer in an
evaluated cold-pressor test. Psychoneuroendocrinology 33: 890–895. immunoassay for salivary cortisol measurement. J Steroid Biochem Mol Biol 43:
20. al’Absi M, Petersen KL, Wittmers LE (2002) Adrenocortical and hemodynamic 683–692.
predictors of pain perception in men and women. Pain 96: 197–204. 43. Miller R, Plessow F (2013) Transformation techniques for cross-sectional and
21. Engert V, Vogel S, Efanov SI, Duchesne A, Corbo V, et al. (2011) Investigation longitudinal endocrine data: Application to salivary cortisol concentrations.
into the cross-correlation of salivary cortisol and alpha-amylase responses to Psychoneuroendocrinology 38: 941–946.
psychological stress. Psychoneuroendocrinology 36: 1294–1302. 44. Ivanciuc O, Braun W (2007) Robust quantitative modeling of peptide binding
22. Foley P, Kirschbaum C (2010) Human hypothalamus-pituitary-adrenal axis affinities for MHC molecules using physical-chemical descriptors. Protein Pept
responses to acute psychosocial stress in laboratory settings. Neurosci Biobehav Lett 14: 903–916.
Rev 35: 91–96. 45. Mur M, Bandettini PA, Kriegeskorte N (2009) Revealing representational
23. Schommer NC, Hellhammer DH, Kirschbaum C (2003) Dissociation between content with pattern-information fMRI—an introductory guide. Soc Cogn
reactivity of the hypothalamus-pituitary-adrenal axis and the sympathetic- Affect Neurosci 4: 101–109.
adrenal-medullary system to repeated psychosocial stress. Psychosom Med 65: 46. Pereira F, Mitchell T, Botvinick M (2009) Machine learning classifiers and
450–460. fMRI: a tutorial overview. Neuroimage 45: S199–209.
24. Kajantie E, Phillips DI (2006) The effects of sex and hormonal status on the 47. Misaki M, Kim Y, Bandettini PA, Kriegeskorte N (2010) Comparison of
physiological response to acute psychosocial stress. Psychoneuroendocrinology multivariate classifiers and response normalizations for pattern-information
31: 151–178. fMRI. Neuroimage 53: 103–118.
25. von Baeyer CL, Piira T, Chambers CT, Trapanotto M, Zeltzer LK (2005)
48. Cohen S, Hamrick N, Rodriguez MS, Feldman PJ, Rabin BS, et al. (2000) The
Guidelines for the cold pressor task as an experimental pain stimulus for use with
stability of and intercorrelations among cardiovascular, immune, endocrine, and
children. J Pain 6: 218–227.
psychological reactivity. Ann Behav Med 22: 171–179.
26. Dickerson SS, Kemeny ME (2004) Acute stressors and cortisol responses: a
49. Buchanan TW, al’Absi M, Lovallo WR (1999) Cortisol fluctuates with increases
theoretical integration and synthesis of laboratory research. Psychol Bull 130:
355–391. and decreases in negative affect. Psychoneuroendocrinology 24: 227–241.
27. Kudielka BM, Schommer NC, Hellhammer DH, Kirschbaum C (2004) Acute 50. Schlotz W, Kumsta R, Layes I, Entringer S, Jones A, et al. (2008) Covariance
HPA axis responses, heart rate, and mood changes to psychosocial stress (TSST) between psychological and endocrine responses to pharmacological challenge
in humans at different times of day. Psychoneuroendocrinology 29: 983–992. and psychosocial stress: a question of timing. Psychosom Med 70: 787–796.
28. Rohleder N, Nater UM, Wolf JM, Ehlert U, Kirschbaum C (2004) Psychosocial 51. Hellhammer DH, Wust S, Kudielka BM (2009) Salivary cortisol as a biomarker
stress-induced activation of salivary alpha-amylase: an indicator of sympathetic in stress research. Psychoneuroendocrinology 34: 163–171.
activity? Ann N Y Acad Sci 1032: 258–263. 52. Dallman MF, Bhatnagar S, Viau V (2000) Hypothalamo-pituitary-adrenal axis.
29. Nhan BR, Chau T (2010) Classifying affective states using thermal infrared In: Fink G, editor. Encyclopedia of Stress. San Diego: Academic Press.
imaging of the human face. IEEE Trans Biomed Eng 57: 979–987. 53. Fries E, Dettenborn L, Kirschbaum C (2009) The cortisol awakening response
30. Kuraoka K, Nakamura K (2011) The use of nasal skin temperature (CAR): facts and future directions. Int J Psychophysiol 72: 67–73.
measurements in studying emotion in macaque monkeys. Physiol Behav 102: 54. Lovallo WR, Thomas TL (2000) Stress hormones in psychophysiological
347–355. research: emotional, behavioral, and cognitive implications. In: Cacioppo JT,
31. Nakanishi R, Imai-Matsumura K (2008) Facial skin temperature decreases in Tassinary LG, Bersnton GG, editors. Handbook of Psychophysiology. New
infants with joyful expression. Infant Behav Dev 31: 137–144. York: Cambridge University Press pp. 342–367.

PLOS ONE | www.plosone.org 11 March 2014 | Volume 9 | Issue 3 | e90782