A short review and primer on electrodermal activity in human computer interaction applications

Chapter (PDF Available) · August 2016with 263 Reads
In book: The Psychophysiology Primer: a guide to methods and a broad review with a focus on human-computer interaction., Chapter: 2, Publisher: Now Publishers
The application of psychophysiology in human-computer interaction is a growing field with significant potential for future smart personalised systems. Working in this emerging field requires comprehension of an array of physiological signals and analysis techniques. One of the most widely used signals is electrodermal activity, or EDA, also known as galvanic skin response or GSR. This signal is commonly used as a proxy for physiological arousal, but recent advances of interpretation and analysis suggest that traditional approaches should be revised. We present a short review on the application of EDA in human-computer interaction. This paper aims to serve as a primer for the novice, enabling rapid familiarisation with the latest core concepts. We put special emphasis on everyday human-computer interface applications to distinguish from the more common clinical or sports uses of psychophysiology. This paper is an extract from a comprehensive review of the entire field of ambulatory psychophysiology, including 12 similar chapters, plus application guidelines and systematic review. Thus any citation should be made using the following reference: B. Cowley, M. Filetti, K. Lukander, J. Torniainen, A. Henelius, L. Ahonen, O. Barral, I. Kosunen, T. Valtonen, M. Huotilainen, N. Ravaja, G. Jacucci. The Psychophysiology Primer: a guide to methods and a broad review with a focus on human-computer interaction. Foundations and Trends in Human-Computer Interaction, vol. 9, no. 3-4, pp. 150-307, 2016.
A short review and primer on electrodermal
activity in human computer interaction
Benjamin Cowley1,2and Jari Torniainen1
1Quantitative Employee unit, Finnish Institute of Occupational Health,
POBox 40, Helsinki, 00250, Finland
2Cognitive Brain Research Unit, Institute of Behavioural Sciences, University of
Helsinki, Helsinki, Finland
Abstract. The application of psychophysiological in human-computer
interaction is a growing field with significant potential for future smart
personalised systems. Working in this emerging field requires compre-
hension of an array of physiological signals and analysis techniques.
One of the most widely used signals is electrodermal activity, or EDA,
also known as galvanic skin response or GSR. This signal is commonly
used as a proxy for physiological arousal, but recent advances of interpre-
tation and analysis suggest that traditional approaches should be revised.
We present a short review on the application of EDA in human-computer
This paper aims to serve as a primer for the novice, enabling rapid fa-
miliarisation with the latest core concepts. We put special emphasis on
everyday human-computer interface applications to distinguish from the
more common clinical or sports uses of psychophysiology.
This paper is an extract from a comprehensive review of the entire field
of ambulatory psychophysiology, including 12 similar chapters, plus ap-
plication guidelines and systematic review. Thus any citation should be
made using the following reference:
B. Cowley, M. Filetti, K. Lukander, J. Torniainen, A. Henelius,
L. Ahonen, O. Barral, I. Kosunen, T. Valtonen, M. Huotilainen,
N. Ravaja, G. Jacucci. The Psychophysiology Primer: a guide to
methods and a broad review with a focus on human-computer in-
teraction. Foundations and Trends in Human-Computer Interac-
tion, vol. 9, no. 3-4, pp. 150–307, 2016.
Keywords: electrodermal activity, psychophysiology, human-computer
interaction, primer, review
1 Introduction
‘Electrodermal activity’ (EDA) is a general term used to describe changes in
the electrical properties of the skin resulting from autonomic nervous system
functions [Dawson et al., 2000]. These fluctuations are caused by activation of
arXiv:1608.06986v2 [cs.HC] 29 Aug 2016
A review and primer for EDA in HCI
sweat glands that are controlled by the sympathetic nervous system, which au-
tonomously regulates the mobilisation of the human body for action. Further-
more, skin conductivity is not influenced by parasympathetic activation. There-
fore, EDA can be considered to act as an indicator of both psychological and
physiological arousal and, by extension, as a measure of cognitive and emotional
activity [Dawson et al., 2000, Boucsein, 2012].
2 Background
EDA has been investigated for well over 100 years, with a number of changes
having occurred in the method and the understanding of the phenomenon. Terms
have changed accordingly, though ‘galvanic skin response’ is still commonly in
use, which can be confusing; instead, one should use the modern terminology, as
outlined in Boucsein et al. [2012]:
“[The] first two letters refer to the method of measurement .. . SP for
skin potential, SR for skin resistance, SC for skin conductance, SZ for
skin impedance, and SY for skin admittance. The third letter refers to
level (L) or response (R)”.
These terms are derived from the methods employed to detect changes in the
electrical properties of the skin, which are the following: the passive measure-
ment of electrical potential difference, or the endosomatic method, and active
exosomatic measurement, wherein either alternating current (AC) or direct cur-
rent (DC) is passed between two electrodes to measure the skin’s conductivity,
the reciprocal of its resistance. In this section, we refer to the latter method, as
it is the more widely used (to our knowledge). For full details on these methods,
especially how to obtain the slightly more complicated SZ and SY terms, see the
work of Boucsein and colleagues.
In the literature, EDA has most often been taken as a measure of arousal
[Bradley, 2000]. Several studies using a picture-viewing paradigm have shown
that EDA is highly correlated with self-reported emotional arousal [Lang et al.,
1993]. That is, arousing pictures of either positive or negative valence result in
increased EDA as compared to low-arousal pictures. This index is affected by
the location of recording, as different skin sites are innervated by different dis-
tributions of nerve bundles, not all of which are involved in emotional responses.
In simple terms, emotional response affects eccrine sweat glands, which are most
densely distributed on the palms and soles, nearly four times more so than on
the forehead, for example. Sixteen recording sites were explored and compared
in a review by van Dooren et al. [2012], which profiled site-wise responsiveness
to emotional inducement (by film clips). Their review illustrates that care must
be taken in the choice of the signal feature to estimate responsiveness. The au-
thors found also that responses did not show full lateral symmetry, so care must
be taken in the decision on which side of the body to record. Picard’s Multiple
Arousal Theory Picard et al. [2015] suggests an explanation: that different brain
areas map to different areas of the body, both contralaterally and ipsilaterally.
Page 2 of 11
A review and primer for EDA in HCI
Fig. 1. The simple emotional circumplex model, with orthogonal bipolar dimensions
of arousal (from alert to lethargic) and valence (pleasant to unpleasant).
EDA is a commonly used physiological measure when one is studying HCI
experiences (see ‘Applications’, below). The arousal models used in HCI studies
are often uni-dimensional and bipolar, and, hence, they can be combined with a
dimension of positive–negative valence to give a circumplex model of emotions,
as highlighted in Figure 1. However, richer models have been proposed, such
as the three-system model of arousal [Backs and Boucsein, 2000]. Indeed, Backs
and Boucsein (p. 6) argued that this might be more appropriate for investigating
the specific sensitivity of physiological effects in HCI. In brief, this model posits
three systems: ‘affect arousal’, ‘effort’, and ‘preparatory activation’, of which only
affect is indexed by EDA. The areas of the CNS that correspond to these systems
are Amygdala, Hippocampus, and Basal Ganglia, respectively. The authors also
provided a review demonstrating the sensitivity of EDA in technology interaction
studies (p. 16).
3 Methods
3.1 EDA Instrumentation
EDA is a well-established recording method, and numerous devices exist for per-
forming laboratory-grade measurements. These devices usually comprise wired
electrodes and often a bulky amplifier, thereby restricting use to controlled en-
vironments. Furthermore, electrodes placed on the hand are often very sensitive
to motion, thereby requiring the hand to stay quite still.
With the recent increase in the quality and popularity of wearable biosensors,
several portable EDA devices have become available. Portability is appealing for
Page 3 of 11
A review and primer for EDA in HCI
both psychological research and clinical use. In psychology, wearable sensors al-
low experiments to take place in more ecologically valid settings [Betella et al.,
2014], while in health care wearable sensors enable continuous physiological mon-
itoring at a relatively low cost [Pantelopoulos and Bourbakis, 2010].
Non-intrusively measuring EDA in a continuous long-term manner is desir-
able for many, quite different fields of research and diagnostics. Popular options
in this regard are wearable EDA sensors, such as the ring-mounted Moodmetric
(Vigofere Ltd., Helsinki, Finland); the wrist-worn E4 (Empatica Inc., Boston,
MA, USA); or the edaMove (movisens GmbH, Karlsruhe, Germany), which com-
bines a wrist-worn amplifier with wired electrodes. A recent study addresses the
comparability of such a wearable sensor to a laboratory-grade device [Torniainen
et al., 2015].
3.2 Recording
EDA measurement registers the inverse of the electrical resistance ‘ohm’ between
two points on the skin – i.e., the conductivity of the skin in that location, ‘mho’.
The recorded EDA signal has two components. The slowly varying tonic com-
ponent of the EDA signal represents the current skin conductance level (SCL)
and can be influenced by external or internal factors such as dryness of the skin
and psychological state. Superimposed on the slow tonic component is a rapidly
changing phasic component, skin conductance response (SCR); see, for example,
Figure 2. The spike-like SCR corresponds to sympathetic arousal, resulting from
an orienting response to either specific environmental stimuli, such as a novel,
unexpected, significant, or aversive stimulus, or non-specific activation, such as
deep breaths and body movements [Boucsein, 2012, Dawson et al., 2000].
Typically, EDA is recorded non-invasively from the surface of the palms and
fingers. Following Boucsein et al. [2012], we recommend recording from the fin-
gers to the extent that this is possible. Fingers provide good signal characteris-
tics, such as amplitude, and responsiveness of the signal to emotional relevance is
well-established. When recording is conducted in situations that demand grasp-
ing actions, which could disturb the sensors, the soles of the feet, or the forehead,
may be used also [van Dooren et al., 2012].
3.3 Preprocessing
In a typical EDA analysis, the acquired signal is preprocessed and then decom-
posed into tonic and phasic components – i.e., SCL and SCRs. The preprocess-
ing is relatively simple: data are down-sampled or low-pass filtered, typically to
<10 Hz. Electrode displacement tends to generate artefacts, represented by sig-
nal discontinuities. These can be detected by a maximum signal-change thresh-
old criterion and handled epoch-wise by rejection or temporal interpolation. For
group analysis, the signal should then be standardised or centred.
Signal decomposition can be performed via a number of methods, depending
on whether stimulus events also have been recorded. If event times are known,
latency-based detection of SCRs can be performed, per Boucsein et al. [2012].
Page 4 of 11
A review and primer for EDA in HCI
Boucsein and colleagues also define the SCL as the signal in the absence of
SCRs; therefore, after SCR detection, SCL can be estimated by subtraction.
However, data-driven methods should be preferred, to minimise errors, because
SCRs do not follow uniformly from events, and events can occur in rapid suc-
cession, causing SCR overlap. A classic example is peak-and-trough detection,
which is achieved by finding zero-derivative points (where the signal is flat). One
can identify SCR features from the trough-to-peak amplitude and latency. This
system tends to be inaccurate for stimulus events that overlap – i.e., that have
a shorter inter-stimulus interval than the recovery time of the phasic peak –
because the amplitude of SCRs begins to sum. See Figure 2 for more details.
Alexander et al. [2005] proposed a method that handles this issue, based on
the deconvolution of the signal to estimate the driver function from sudomotor
nerve activity and the corresponding impulse response function, the latter de-
scribing the temporal profile of each impulse of the phasic driver response and
used as the deconvolution kernel in the decomposition process.
This method is based on standard deconvolution, which does not account
for variations in the SCR shape and can result in a negative driver function
when the SCR has a peaked shape. These problems were addressed by Benedek
and Kaernbach [2010a,b], who introduced two separate solutions: non-negative
deconvolution (NND) and continuous decomposition analysis (CDA)3. Using
NND ensures that any negative component of the driver is transformed to a
positive ‘remainder’, interpreted as the additional phasic component caused by
pore opening. The output of this analysis is depicted in Figure 3. The NND
approach was inspired by the poral valve model of EDA, which suggests that
peaked SCRs result from additional sweat diffusion caused by pore opening, as
illustrated in Figure 1 from Benedek and Kaernbach [2010b]. They state that
“[i]f sweat ducts are filled to their limits, intraductal pressure will cause
a hydraulic-driven diffusion of sweat to the corneum, resulting in a flat
SCR. If intraductal pressure exceeds tissue pressure, the distal part of
the duct and the pore will eventually open, which results in a peaked
CDA takes a different approach, which “abandons the concept of single, dis-
crete responses in favour of a continuous measure of phasic activity” [Benedek
and Kaernbach, 2010a]. The latter is, of course, more plausible in a messy bi-
ological system. The CDA estimate of the phasic driver can take on negative
values, in which case the interpretation is simply that negative values signify
quality issues, either in extraction algorithm parameters or in the original data.
For Benedek and Kaernbach [2010a], estimation is a multi-step optimisa-
tion process using gradient descent to minimise a compound error consisting
of a weighted sum of the negativity and indistinctness of the phasic driver. In-
distinctness describes the sharpness of impulses, and negativity represents the
number of negative values in the phasic driver.
3NND and CDA are implemented as the Ledalab toolbox for Matlab.
Page 5 of 11
A review and primer for EDA in HCI
Fig. 2. Illustration of SCR overlap, reproduced from Alexander et al. [2005], with per-
mission. They explain: “The upper graph shows the smoothed skin conductance signal,
with two groups of three overlapping SCRs. The middle graph shows the commuted
driver signal, which because of its shorter time-constant has six clearly separate peaks.
These separate peaks are used to estimate the individual SCRs shown in the bottom
Page 6 of 11
A review and primer for EDA in HCI
Fig. 3. Screenshot from Ledalab. Top panel: 20 seconds of EDA are shown from a
recording of a continuous-performance task, with inhibit (labelled ‘cor-inb’) and re-
spond (labelled ‘cor-rsp’) targets shown every 2 seconds. Response targets (with the
subject’s responses labelled ‘RESP’) are less frequent in the task so generate a greater
EDA response. The grey area indicates SCL, and the coloured area shows SCRs di-
minishing over time. Bottom panel: Fitting by NND in Ledalab produces an estimate
of the SCRs (‘Driver’, blue) and pore opening components (‘Overshoot’, pink).
3.4 Analysis
For group phasic analyses, the impulse response function generally should be
estimated separately for each participant. The phasic component is then analysed
around selected events (if the phasic component was derived by data-driven
methods as recommended, without reference to the events, there is the added
benefit that a relationship discovered between phasic features and events cannot
be an artefact of the feature extraction method). One can do this either by
averaging the phasic driver or by calculating a set of phasic features and then
performing the analysis in feature space (as in, for example, Khalfa et al. [2002]).
Commonly used phasic features include the number of significant phasic peaks,
the sum of amplitudes of those peaks, the time integral of the phasic response,
and the maximum value of phasic activity.
Page 7 of 11
A review and primer for EDA in HCI
4 Applications
EDA has seen application in a host of areas, from research to clinical practice
and consumer devices. The number of form factors used in such devices remains
relatively limited (they are usually situated on the wrist and fingers), but, as
van Dooren et al. [2012] have shown, there are many options for recording sites.
Therefore, in line with the application, the reader could conceive of implementing
a device in a hat or eyeglasses (to measure forehead EDA), in socks or shoes (to
measure foot sole EDA), or in a wrist-worn strap or other clothing items.
There is an extensive body of literature on EDA applications; here, we cite
only a few examples.
In the area of HCI, EDA is a popular input in helping to classify arousal
(usually referring to ‘affect arousal’ [Backs and Boucsein, 2000]). For example,
Fantato et al. [2013] reported on a na¨ıve Bayes classifier, which was trained to
recognise states of affect arousal from a number of EDA features, on the basis
of validated labelling of arousal levels during work-like tasks. Cross-validation
testing of these tasks achieved an accuracy level above 90%. The system was
tested also by recording of subjects in a computer-game-like learning environ-
ment, where the classifier achieved an accuracy of 69% for predicting the self-
reported emotional arousal of the game. The sensor was the Varioport-ARM
device (Becker Meditec, Karlsruhe, Germany).
Studies have shown more specific effects also. Heiden et al. [2005], studying
work done with a computer mouse, found highly significant differences in EDA
between conditions that differed in the level of task difficulty. Setz et al. [2010]
compared several classifiers in discriminating between work-like tasks with a
baseline cognitive load only and tasks with added stress (considered to be a
form of negative affect arousal). The input consisted of 16 EDA features, and
the researchers’ best-performing classifier (Linear Discriminant Analysis, LDA)
achieved an accuracy of 83%. Their device was an early form of wearable arm-
mounted sensor, lab-built and described in the paper referenced above.
Offering a final example, we focus on an application that is not usually con-
nected with the workplace. Biofeedback is an increasingly popular application
for performance enhancement, and it can be found in such varied contexts as
clinical, occupational, and sports scenarios. In clinical biofeedback, the user is
trained to respond to a given feature of the real-time signal from a physiological
sensor; in this manner, the user can learn to recognise and control the subjec-
tive state that corresponds to the feature. With EDA, the feature that needs
to be classified might be, for example, the number of significant phasic peaks.
In an application, users could learn to recognise the subjective feeling of having
more or fewer phasic peaks, then attempt to control their physiological state
One recreational use of biofeedback involves an affect-based music player, in
which concurrently measured biosignals are used to classify the listener’s emo-
Page 8 of 11
A review and primer for EDA in HCI
tional response as the music is playing. The efficacy of such a system for inducing
target moods has been demonstrated in an ecologically valid office setting, al-
though with only a small sample size, N=10 [van der Zwaag et al., 2013].
O’Connell et al. [2008] demonstrated the ‘Self-Alert Training’ (SAT) system
for EDA biofeedback, to modulate attention via arousal level. This software
was validated with a group of 23 neurologically healthy participants, each of
whom received brief (30–40-minute) biofeedback training sandwiched between
two sustained attention to response task (SART) tests. Half of the participants
were given placebo training. Analysis indicated that the SAT group
significantly reduced their number of commission errors (a measure of re-
sponse inhibition), while the placebo control group did not;
maintained consistent response time variability (RTV – an inverse measure
of sustained attention) after training, whereas the placebo group shows a
significant increase in RTV; and
increased in arousal (SCR amplitude in response to cues) after training, while
the placebo group’s corresponding figures significantly decreased.
The last of these findings indicates that the short training period was enough
to enable participants to counter whatever effects of fatigue and cue exposure
had caused the reduction of arousal in the placebo group. This is important for
the domain of safety-critical operator work in an HCI setting, where the effect
of brief periods of activity to boost vigilance and alertness can be considered a
valuable option for reducing human error. Such systems can now be implemented
at low cost, as sensor devices are becoming robust, lightweight, and wearable,
and interfaces are available for mobile platforms such as smartphones.
5 Conclusion
Electrodermal activity is a reliable, interpretable, and simple-to-use measure
that has seen many applications in various domains. Therefore, it is an excellent
choice for an introduction to the psychophysiological method and a highly suit-
able tool for making inferences about sympathetic nervous system activity. In
addition, EDA aids in providing valuable context for other physiological signals
in multimodal applications.
Page 9 of 11
D M Alexander, C Trengove, P Johnston, T Cooper, J P August, and E Gordon.
Separating individual skin conductance responses in a short interstimulus-
interval paradigm. Journal of neuroscience methods, 146(1):116–23, July 2005.
ISSN 0165-0270. doi: 10.1016/j.jneumeth.2005.02.001.
Richard Backs and Wolfram Boucsein. Engineering Psychophysiology Issues
and Applications. Taylor & Francis, Hoboken, 1 edition, 2000. ISBN
Mathias Benedek and Christian Kaernbach. A continuous measure of phasic
electrodermal activity. Journal of neuroscience methods, 190(1):80–91, June
2010a. ISSN 1872-678X. doi: 10.1016/j.jneumeth.2010.04.028.
Mathias Benedek and Christian Kaernbach. Decomposition of skin conductance
data by means of nonnegative deconvolution. Psychophysiology, 47(4):647–
658, 2010b. doi: 10.1111/j.1469-8986.2009.00972.x.
Alberto Betella, Riccardo Zucca, Ryszard Cetnarski, Alberto Greco, Antonio
Lanat`a, Daniele Mazzei, Alessandro Tognetti, Xerxes D Arsiwalla, Pedro
Omedas, Danilo De Rossi, and Paul F M J Verschure. Inference of human
affective states from psychophysiological measurements extracted under eco-
logically valid conditions. Frontiers in neuroscience, 8:286, January 2014.
ISSN 1662-4548. doi: 10.3389/fnins.2014.00286.
W. Boucsein. Electrodermal activity. Springer Science & Business Media, 2012.
Wolfram Boucsein, Don C Fowles, Sverre Grimnes, Gershon Ben-Shakhar, Wal-
ton T Roth, Michael E Dawson, and Diane L Filion. Publication recommen-
dations for electrodermal measurements. Psychophysiology, 49(8):1017–34,
August 2012. ISSN 1540-5958. doi: 10.1111/j.1469-8986.2012.01384.x.
Margaret M Bradley. Emotion and motivation. In John T. Cacioppo, Louis G
Tassinary, and Gary G. Berntson, editors, Handbook of Psychophysiology,
pages 602–642. Cambridge University Press, New York, NY, 2nd edition, 2000.
Michael E Dawson, Anne M Schell, and Diane L Filion. The electrodermal
system. In John T Cacioppo, Louis G Tassinary, and Gary G Berntson, editors,
Handbook of Psychophysiology. Cambridge University Press, Cambridge UK,
2 edition, 2000.
Martino Fantato, Ben Cowley, and Niklas Ravaja. Arousing learning: a psy-
chophysiological classifier for real-time emotion recognition in technology en-
hanced learning environments. In Paul Cunningham and Miriam Cunningham,
editors, eChallenges e-2013, pages 1–8, Dublin, 2013. IIMC International In-
formation Management Corporation.
M Heiden, E Lyskov, M Djupsj¨obacka, F Hellstr¨om, and A G Crenshaw. Effects
of time pressure and precision demands during computer mouse work on mus-
cle oxygenation and position sense. European journal of applied physiology, 94
(1-2):97–106, May 2005. ISSN 1439-6319. doi: 10.1007/s00421-004-1295-y.
A review and primer for EDA in HCI
St´ephanie Khalfa, Peretz Isabelle, Blondin Jean-Pierre, and Robert Manon.
Event-related skin conductance responses to musical emotions in humans.
Neuroscience letters, 328(2):145–9, August 2002. ISSN 0304-3940.
P J Lang, M K Greenwald, M M Bradley, and A O Hamm. Looking at pictures:
affective, facial, visceral, and behavioral reactions. Psychophysiology, 30(3):
261–273, 1993.
Redmond G O’Connell, Mark A Bellgrove, Paul M Dockree, Adam Lau, Michael
Fitzgerald, and Ian H Robertson. Self-Alert Training: volitional modulation
of autonomic arousal improves sustained attention. Neuropsychologia, 46(5):
1379–90, April 2008. ISSN 0028-3932. doi: 10.1016/j.neuropsychologia.2007.
A. Pantelopoulos and N.G. Bourbakis. A Survey on Wearable Sensor-Based
Systems for Health Monitoring and Prognosis. IEEE Transactions on Sys-
tems, Man, and Cybernetics, Part C (Applications and Reviews), 40(1):1–12,
January 2010. ISSN 1094-6977. doi: 10.1109/TSMCC.2009.2032660.
R. W. Picard, S. Fedor, and Y. Ayzenberg. Multiple Arousal Theory and Daily-
Life Electrodermal Activity Asymmetry. Emotion Review, 8(1):62–75, mar
2015. ISSN 1754-0739. doi: 10.1177/1754073914565517.
C. Setz, B. Arnrich, J. Schumm, R. La Marca, G. Troster, and U. Ehlert.
Discriminating Stress From Cognitive Load Using a Wearable EDA Device.
IEEE Transactions on Information Technology in Biomedicine, 14(2):410–417,
March 2010. ISSN 1089-7771. doi: 10.1109/TITB.2009.2036164.
Jari Torniainen, Benjamin Cowley, Andreas Henelius, Kristian Lukander, and
Satu Pakarinen. Feasibility of an electrodermal activity ring prototype as a
research tool. In 37th Annual International Conference of the IEEE Engineer-
ing in Medicine and Biology Society, Milano, Italy, 2015.
Marjolein D. van der Zwaag, Joris H. Janssen, and Joyce H.D.M. Westerink.
Directing Physiology and Mood through Music: Validation of an Affective
Music Player. IEEE Transactions on Affective Computing, 4(1):57–68, jan
2013. ISSN 1949-3045. doi: 10.1109/T-AFFC.2012.28.
Marieke van Dooren, J J G Gert-Jan de Vries, and Joris H Janssen. Emotional
sweating across the body: comparing 16 different skin conductance measure-
ment locations. Physiology & behavior, 106(2):298–304, May 2012. ISSN 1873-
507X. doi: 10.1016/j.physbeh.2012.01.020.
Page 11 of 11
  • ... The results from our batch analysis of EDA features, accumulated per event type show heightened sympathetic activation during WS events when compared to CS events based on validated physiological indicators. The results of this analysis will enable the CAIE to decide, when to ask clarifying questions in an adaptivewindow based multi-turn interaction as discussed in Interaction Model section.The number of significant phasic peaks and time-integral of phasic peaks are widely used EDA features, wherein a higher number represents stronger sympathetic activation[5]. To perform the IPR analysis, the individual phasic peaks are thresholded to above 5% of the userwise maximum peak-amplitude to mark the significant peaks. ...
    Conference Paper
    Full-text available
    The probabilistic nature of the inferences in a context-aware intelligent environment (CAIE) renders them vulnerable to erroneous decisions resulting in wrong services. Learning to recognize a user’s negative reactions to such wrong services will enable a CAIE to anticipate a service’s appropriateness. We propose a framework for continuous measurement of physiology to infer a user’s negative-emotions arising from receiving wrong services, thereby implementing an implicit-feedback loop in the CAIE system. To induce such negative-emotions, in this paper, we present a virtualreality (VR) based experimental platform while collecting real-time physiological data from ambulatory wearable sensors. Results from the electrodermal activity (EDA) data analysis reveal patterns that correlate with known features of negative-emotions, indicating the possibility to infer service appropriateness from user’s reactions to a service, thereby closing an implicit-feedback loop for the CAIE.
  • Book
    Electrodermal activity is one of the most frequently used psychophysiological evaluations in psychology research. Based on the 1992 edition of this work Electrodermal Activity covers advances in the field since the first publication in 1992. The current volume includes updated information on brain imaging techniques such as PET and fMRI, which provide further insight into the brain mechanisms underlying EDA. In addition, this volume is able to describe more reliably hypotheses that have been successfully tested since the first publication. © Springer Science+Business Media, LLC 2012. All rights reserved.
  • Article
    Using “big data” from sensors worn continuously outside the lab, researchers have observed patterns of objective physiology that challenge some of the long-standing theoretical concepts of emotion and its measurement. One challenge is that emotional arousal, when measured as sympathetic nervous system activation through electrodermal activity, can sometimes differ significantly across the two halves of the upper body. We show that traditional measures on only one side may lead to misjudgment of arousal. This article presents daily life and controlled study data, as well as existing evidence from neuroscience, supporting the influence of multiple emotional substrates in the brain causing innervation on different sides of the body. We describe how a theory of multiple arousals explains the asymmetric EDA findings.
  • Article
    Full-text available
    Compared to standard laboratory protocols, the measurement of psychophysiological signals in real world experiments poses technical and methodological challenges due to external factors that cannot be directly controlled. To address this problem, we propose a hybrid approach based on an immersive and human accessible space called the eXperience Induction Machine (XIM), that incorporates the advantages of a laboratory within a life-like setting. The XIM integrates unobtrusive wearable sensors for the acquisition of psychophysiological signals suitable for ambulatory emotion research. In this paper, we present results from two different studies conducted to validate the XIM as a general-purpose sensing infrastructure for the study of human affective states under ecologically valid conditions. In the first investigation, we recorded and classified signals from subjects exposed to pictorial stimuli corresponding to a range of arousal levels, while they were free to walk and gesticulate. In the second study, we designed an experiment that follows the classical conditioning paradigm, a well-known procedure in the behavioral sciences, with the additional feature that participants were free to move in the physical space, as opposed to similar studies measuring physiological signals in constrained laboratory settings. Our results indicate that, by using our sensing infrastructure, it is indeed possible to infer human event-elicited affective states through measurements of psychophysiological signals under ecological conditions.
  • Article
    This committee was appointed by the SPR Board to provide recommendations for publishing data on electrodermal activity (EDA). They are intended to be a stand-alone source for newcomers and experienced users. A short outline of principles for electrodermal measurement is given, and recommendations from an earlier report (Fowles et al., ) are incorporated. Three fundamental techniques of EDA recording are described: (1) endosomatic recording without the application of an external current, (2) exosomatic recording with direct current (the most widely applied methodology), and (3) exosomatic recording with alternating current-to date infrequently used but a promising future methodology. In addition to EDA recording in laboratories, ambulatory recording has become an emerging technique. Specific problems that come with this recording of EDA in the field are discussed, as are those emerging from recording EDA within a magnetic field (e.g., fMRI). Recommendations for the details that should be mentioned in publications of EDA methods and results are provided.
  • Article
    Full-text available
    Music is important in everyday life, as it provides entertainment and influences our moods. As music is widely available, it is becoming increasingly difficult to select songs to suit our mood. An affective music player can remove this obstacle by taking a desired mood as input and then selecting songs that direct toward that desired mood. In the present study, we validate the concept of an affective music player directing the energy dimension of mood. User models were trained for 10 participants based on skin conductance changes to songs from their own music database. Based on the resulting user models, the songs that most increased or decreased the skin conductance level of the participants were selected to induce either a relatively energized or a calm mood. Experiments were conducted in a real-world office setting. The results showed that a reliable prediction can be made of the impact of a song on skin conductance, that skin conductance and mood can be directed toward an energized or calm state and that skin conductance remains in these states for at least 30 minutes. All in all, this study shows that the concept and models of the affective music player worked in an ecologically valid setting, suggesting the feasibility of using physiological responses in real-life affective computing applications.
  • Conference Paper
    Full-text available
    This paper reports on the PsychoPhysiological Assessment Component which measures the emotional/psychological state of serious game players, to support their competence development within a novel Technology Enhanced Learning (TEL) platform. The authors describe the methods used to build and validate the emotional-state Naive Bayes classifier (NBC). Our approach is tested in two experiments. In the first experiment, participants had to perform emotional- elicitation tasks and we achieved an arousal classification accuracy of ~92%. In the second experiment, participants were asked to play the TEL serious game and we achieved an arousal classification accuracy of ~68%. The results show that our arousal classification system, based on a simple set of sensors, can provide a real- time index of what can be considered a baseline measure of the player’s capacity to learn: in a state of increased arousal/activation, the potential to learn is higher.
  • Article
    Full-text available
    The design and development of wearable biosensor systems for health monitoring has garnered lots of attention in the scientific community and the industry during the last years. Mainly motivated by increasing healthcare costs and propelled by recent technological advances in miniature biosensing devices, smart textiles, microelectronics, and wireless communications, the continuous advance of wearable sensor-based systems will potentially transform the future of healthcare by enabling proactive personal health management and ubiquitous monitoring of a patient's health condition. These systems can comprise various types of small physiological sensors, transmission modules and processing capabilities, and can thus facilitate low-cost wearable unobtrusive solutions for continuous all-day and any-place health, mental and activity status monitoring. This paper attempts to comprehensively review the current research and development on wearable biosensor systems for health monitoring. A variety of system implementations are compared in an approach to identify the technological shortcomings of the current state-of-the-art in wearable biosensor solutions. An emphasis is given to multiparameter physiological sensing system designs, providing reliable vital signs measurements and incorporating real-time decision support for early detection of symptoms or context awareness. In order to evaluate the maturity level of the top current achievements in wearable health-monitoring systems, a set of significant features, that best describe the functionality and the characteristics of the systems, has been selected to derive a thorough study. The aim of this survey is not to criticize, but to serve as a reference for researchers and developers in this scientific area and to provide direction for future research improvements.
  • Article
    Skin conductance (SC) is one of the most commonly used measures in psychophysiological studies involving emotional arousal and is traditionally measured at the fingers or the palms (i.e., the palmar locations) of the hand. Palmar skin conductance recording positions are, however, not always preferred for ambulatory recordings in real-life situations. This study quantifies the responsiveness and similarity with the finger of 16 different recording positions of skin conductance while watching emotional film fragments. Findings indicated foot, fingers and shoulders being most responsive, whereas arm, back, armpit, and thighbone were least responsive. The measurements at the foot were most similar with those of the finger. In contrast, arm, back, and armpit traces differed most from the finger trace. Taken together, foot and shoulders are the best alternatives to the finger for ambulatory measurement of skin conductance to reflect emotional arousal. These findings can help new applications using skin conductance, like automated emotion measurements, to come to fruition.