Towards Voluntary Pupil Control

Training Affective Strategies?

Jan Ehlers, Nikola Bubalo, Markus Loose and Anke Huckauf

General Psychology, Ulm University , 89069 Ulm, Germany

Keywords: Affective Human-Computer Interface, Pupil Size, Biofeedback, Emotions, Voluntary Control.

Abstract: During the past years, increasing attention is being paid to operationalize pupil dynamics for affective

classification (Jacobs, 1996). Thereby it is generally assumed that pupil size displays a genuine impression of

user’s cognitive state but defies any voluntary control (Loewenfeld, 1993). Based on Ekman (2008) we

applied graphical feedback on pupil diameter changes to utilize mechanisms of operant conditioning to

gradually enable voluntary control over pupil size. Participants underwent a training program to exert control

by utilizing affective associations to expand pupil size and relaxation strategies to reduce it. As a result, more

than half of the participants demonstrated to be able to increase pupil sizes relative to baseline recordings.

Training effects did not show up. Furthermore, controlling influence diminishes after about ten seconds.

Intentional increase of sympathetic activity seems to be subject to habituation processes that allow central

inhibition of parasympathetic pathways only over a short period. Beside strategy-based factors, physiological

mechanisms like baseline pupil activity may determine inter-individual differences in exerting voluntary

control. In summary it can be noted that pupil-based communication in HCI extends affective monitoring and

may constitute an active input channel to reliably interfere by means of simple cognitive strategies.

1 INTRODUCTION

In the history of psychological research, pupil size

usually served as a metric of affect or mental effort

(Hess 1972; Janisse, 1974). However, due to the

complexity of the autonomic system, affect is hard to

measure and can’t be predicted with perfect

reliability. This is aggravated by the fact that pupil

size also appears to be influenced by several other

factors like motor responses (Simpson, 1968) or

sensory stimulation (Loewenfeld, 1966). Hyönä

(1995) suggests that the associated dynamics are

related to both, cognitive and affective information. It

is therefore critical to ascribe changes in pupil

diameter exclusively to emotional processing. Still,

controlled studies of the recent past refer to pupil size

as an adequate information channel that provides

insight into the affective state of the user (Partala,

2003).

During the last years, increasing attention is being

paid to operationalize pupil dynamics for affective

classification, even in the traditionally rational

concept of man-machine interaction (Jacobs, 1996).

One main objective behind this inclusion is to reduce

the high amount of accidents and errors in civilian

workplaces that are causally attributed to

impairments resulting from increased stress, sleep

deprivation, cognitive overload or a combination of

these factors (Yu, 2007). Also, the automotive

industry is interested to expand the already good

laboratory registration (Palinko, 2010) on real

environmental conditions. Thereby it is generally

assumed that size and responsiveness of the pupil

display a direct and genuine impression of the user’s

affective and cognitive state but defy any voluntary

control (Hess, 1972; Loewenfeld, 1993).

Classical biofeedback paradigms externalize

covert physiological responses (e.g. heart rate, skin

conductivity or various brain-wave patterns) by

providing visual or auditory online-feedback to allow

strategy-based interference for the purpose of health

preservation or to communicate in affective human-

computer frameworks (Meichenbaum, 1976; Bersak,

2001). Until recently, these parameters were

considered to be an involuntary response; however,

the generation of perceptual awareness led to

successful operant conditioning within all these

physiological functions. Like the aforementioned,

pupil dynamics are regulated by the autonomic

nervous system. Still, the vast majority of studies

Ehlers J., Bubalo N., Loose M. and Huckauf A..

Towards Voluntary Pupil Control - Training Affective Strategies?.

DOI: 10.5220/0005240000050012

In Proceedings of the 2nd International Conference on Physiological Computing Systems (PhyCS-2015), pages 5-12

ISBN: 978-989-758-085-7

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

refer to pupil responses as a passive information

channel that defies voluntary control.

With the present study, the view that pupils

function merely as an uncontrollable expression is put

into question. Initial results of Ekman (2008)

suggested that pupil dilation can be suspect to

training. Based on these findings, we applied

graphical real-time feedback to externalize the covert

muscle response and evaluated strategy-based

cognitive attempts to achieve voluntary control over

pupil dynamics. Preliminary data showed that several

participants were able to control their pupils (Ehlers,

2014).

A key question is whether systematic training

gradually increases the ability to control pupil

dilations. Such an adjustment to imagined scenarios

would challenge the concept of pupillary responses as

passive input variables. The capacity of willful

influencing, however, would make the pupil-based

framework susceptible to basic principles of operant

conditioning. Pupillary information would no longer

constitute a true measure during affective monitoring;

on the contrary: progressively exploring the cause-

and-effect relationship between physiological

parameters (arousal, valence, cognitive load) and

dependent changes of a system’s status may trigger

the user’s need for self-efficacy and open a channel

for strategy-based interventions and voluntary

control.

Pupil Dynamics

Size and responsiveness of the human pupil is at any

time determined by the interplay of two antagonistic

muscle groups, governed by the parasympathetic and

sympathetic nervous system. Compared to the

pupillary sphincter, the dilator muscle exerts a much

smaller force; however, this modulation results in a

dynamic equilibrium of pupillary size. An increase in

sympathetic activity is characteristically

accompanied by central inhibitions of

parasympathetic activity and leads to an enlargement

of pupil diameter. In contrast, low autonomic arousal

usually correlates with a reduction in pupil size

(Lowenstein, 1963).

Accordingly, larger pupil expansions during the

presentation of emotionally negative and positive

sounds compared to neutral conditions can be

observed (Partala & Surakka, 2003). In particular,

they report an onset of pupil dilation at about 400 ms

and a gradual increase up to four seconds after

stimulus onset.

Voluntary Control

There is, as far as we know, only few research on the

possibility of voluntary control of pupil reactions.

Ekman (2008) indicate that pupil behavior can be

influenced intentionally by strategies of affective

regulation or cognitive processing. They report pupil

size changes up to 20 % due to various strategies of

self-induced emotions, cognitive tasks, physical

activity and mild forms of pain.

Preliminary studies of our group also show that

pupil dilation can be voluntarily controlled (Ehlers,

2014). However, we assume the true potential of

pupil size-based mechanisms in HCI to be considered

only when certain aspects of iteration and training are

taken into account. That is, we examined the question

of whether and to what extent voluntary pupil

reactions can be trained. Participants had to consult

various strategies known to have modulatory effects

on pupil behavior, including positive/negative

imaginations and individual relaxation strategies.

During training, graphical real-time feedback was

applied to externalize the covert muscle response and

to enable mechanisms of operant conditioning for

gradually achieving voluntary control over pupil size.

2 METHODS

2.1 Stimuli and Apparatus

Pupil diameter of both eyes was recorded using a SMI

iViewX Hi-Speed Eyetracker working at 500 Hz with

binocular tracking. In order to facilitate control of

pupil dynamics, graphical real-time feedback was

presented on a grey screen (1680x1050 pixels at

60Hz) in a distance of 65cm (lum= 48cd/m2; approx.

5.5°visual angle). The feedback scheme disentangles

task-irrelevant variations (divergences beyond one

standard deviation around baseline) from task-

relevant variations (see Fig. 2). Performance during

baseline calibration was displayed as a static black

(mean) against a grey (standard deviation) circle. The

current pupil size (red circle) was reported back in

real-time.

Real-time feedback was provided according to the

single-value approach as specified in Ehlers (2014)

and Georgi (2014, submitted). This calculation model

is based on findings in Bremner (2012) who studied

the correlation between amplitude and peak velocity

of pupil constriction to the light reflex. Results

showed a mean amplitude of 1.92 mm (SD: 0.39) and

an average peak velocity of 5.65 mm/sec (SD: 1.17).

PhyCS2015-2ndInternationalConferenceonPhysiologicalComputingSystems

Figure 1: Instruction pictograms for self-induced positive

(upper) and negative (lower) imaginations and the

relaxation task (middle).

For the current implementation the reported peak

velocity is used as a limit for detecting (un-)valid

measurements. Converting the specifications to our

sampling rate of 30 Hz leads to an allowable sample-

to-sample change of 0.1883 mm. If the distance

between two values exceeds this range the latest

measurement is substituted by the last valid value.

Blink artefacts are rejected, and a valid pupil

contraction at the outset is detected. Still, the

dynamics subsequent to the first blink are rejected as

well. Pre-tests showed that constantly received

feedback on the basis of untreated values leads to

shaky expansion movements of the feedback circle

that are difficult to handle. As a consequence, the last

two forwarded values are averaged to smooth the

feedback dynamics and ensure increased usability.

2.2 Training Procedure

Every participant passed an introductory session (test

run) to carefully choose for individual task-specific

strategies and to initially test them with regard to the

anticipated effect. Participants were instructed to

improve their voluntary control by utilizing affective

associations to expand pupil size and by consulting

relaxation strategies to reduce it. In the following

weeks, participants underwent a training program

including four consecutive sessions with no more

than four days in between. Accordingly, the current

within-subjected design with repeated measurements

compensates for the comparable few subjects by

sustainable reducing secondary variances; however,

the experimental set-up is still characterized by a low

external validity.

Each training session consisted of three trials; two

auto-suggestion tasks (negative and positive

thoughts) and the relaxation period. Trials were

carried out in randomized order. As Janisse (1974)

suggested, pupil size is linearly related to the intensity

dimension of the stimuli and behaves curvilinearly on

the valence scale with largest expansions at the

negative and positive ends. Participants were

therefore encouraged to utilize affective

autobiographical memories and to maintain a selected

approach during the envisaged training procedure.

The same applied for the individually designed

relaxation strategies. Baseline recording preceded

every trial (Tab. 1).

Table 1: Testing procedure.

Test run

Session

Session (randomized)

Trial 1

(Negative)

Trial 2

(Relaxation)

Trial 3

(Positive)

Trial

Baseline (15 sec) Measurement (30 sec)

Each trial consisted of establishing a baseline and a

subsequent experimental phase. Winn (1993) points

out that the pupil continuously undergoes small

oscillations and that a single „snapshot“ estimate of

its size cannot be accepted as a reliable predictor of

the true mean; it should rather be monitored for a

suitable period. We applied 15 seconds of baseline

recording to enable a reliable estimate of mean pupil

size and standard deviation.

During baseline recording participants were

instructed to avoid any emotionally charged

imaginations. Subsequent measurement time was set

to 30 seconds. Subsequent to baseline recording a

pictogram indicated the respective task (Fig. 1).

Subjects were encouraged to immediately induce

negative/positive imaginations after indicating trial

starts via button-presses. The subjects could therefore

determine the duration of rest periods between single

(randomized) trials to avoid carry-over effects as a

result of previous affective associations.

Figure 2: Feedback scheme (Ehlers, 2014).

TowardsVoluntaryPupilControl-TrainingAffectiveStrategies?

2.3 Participants

A total of ten participants (eight female; mean age:

21.8 years; SD: 2.5) were included in the present

study. One participant’s data had to be excluded due

to technical errors during data acquisition. All

participants were right-handed and had normal or

corrected-to-normal vision. Participants reported no

history of head injury and no neurological or

psychiatric disorder. No participant was taking any

kind of medication during the period of testing.

3 RESULTS

3.1 Effects of Training

Due to the sample size of n=9 and deviations in

sphericity, determination of statistic values followed

the non-parametric paired sample Mann-Whitney-U-

Test for comparing means over the 30 seconds

measuring time.

Figure 3 depicts the grand-averaged signal course

of pupil diameter for all subjects within the first and

last session of training. Participants were instructed to

utilize real-time feedback for operant learning and to

enlarge pupil size using affective associations. Since

arousal (and far less valence) seems to constitute the

influencing variable on pupil enlargement (Partala &

Surakka, 2003) and since we did not observe

differences between positive and negative

imagination trials, we forego the separation of results

and present findings as a mean outcome of affective

interventions in general. Dynamics are pictured

relative to individually determined baselines.

Baseline mean is set to zero; the dotted lines illustrate

limits of the averaged standard deviations.

Figure 3: Grand-averaged course of pupil dynamics during

the attempt to voluntary increase pupil size via self-induced

positive/negative emotions. Data points averaged across all

trials of training session one (blue line) and four (orange

line) (n=9). Signals depict variations from baseline mean

(set to zero; averaged standard deviations dotted).

Compared to baselines, self-induced associations led

to a trend of increased pupil diameters already during

the first session of training but solely within the initial

ten seconds (Fig. 3, blue line; Baseline mean: 5.11,

SD: 0.68; Measurement mean for 30 seconds: 5.49;

SD: 1.16) (T(9) = -1.9; p = .06). Subsequently, pupil

sizes decline and remain on the level of spontaneous

fluctuations. A similar tendency can be observed in

session four (orange line); however, deviations from

baseline do not longer emerge statistically significant.

Thus, an overall effect of training could not be

established.

Figure 4 illustrates the grand-averaged signal

courses during the attempt to voluntary decrease pupil

diameter by utilizing individually designed relaxation

strategies. The blue line depicts mean dynamics

during the first session of training, the orange line

traces the averaged course of session four. Though

self-induced relaxation seems to be associated with a

linear decline of pupil size, values do not fall below

spontaneous variations (dotted lines).

Figure 4: Grand-averaged course of pupil dynamics during

the attempt to voluntary reduce pupil size via individual

designed relaxation strategies. Data points averaged across

all trials of training session one (blue line) and four (orange

line) (n=9). Signals depict variations from baseline mean

(set to zero; averaged standard deviations dotted).

3.2 Grouping of Subjects

Ekman (2008) report large variability between

subjects in the general ability of controlled

interference as well as huge variations in magnitude

of the effect. Crider (1971) report individual

differences with regard to the habituation of skin

conductance responses (SCR) and spontaneous

fluctuations in skin resistance. Both variables have

been used to define a trait called “electrodermal

lability” which exhibits high retest-reliability and

reflects inter-individual differences within a variety

of information processing tasks. Subjects with high

rates of spontaneous fluctuations and/or slow SCR

habituation are referred to as electrodermal “labiles”;

in contrast to electrodermal “stabiles” with only few

‐0,6

‐0,4

‐0,2

0,2

0,4

0,6

0 4 8 12 16 20 24 28

RelativeChanges(mm)

Time(s)

‐0,6

‐0,4

‐0,2

0,2

0,4

0,6

0 4 8 1216202428

Relative Changes (mm)

Time (s)

PhyCS2015-2ndInternationalConferenceonPhysiologicalComputingSystems

fluctuations and/or fast SCR habituation. Individual

distinctions within physiological reactions or

habituation to (endogenous) stimuli correspond to our

basic question and we will hereafter try to apply this

concept on pupillary dynamics.

With regard to the assumptions outlined above,

we inspected the current sample in view of inter-

individual differences. It became apparent that five

out of nine participants were consistently (throughout

all trials of the complete training period) able to

utilize auto-suggestive strategies and voluntary

control pupil dynamics with regard to the anticipated

effect (here: expansion of size beyond one standard

deviation from baseline mean). In contrast, the other

four subjects were either incapable of any intentional

influencing in any session, revealed only a one-time

accidental success, or produced contrary patterns.

Figure 5: Course of pupil dynamics for “poor performing”

(upper illustration) (n=4) and “performing” subjects (lower

illustration) (n=5) during the attempt to increase pupil size

via self-induced positive/negative emotions. Mean values

averaged across all trials of the four training sessions.

Signals depict variations from baseline mean (set to zero;

averaged standard deviations dotted).

Applying the strict criterion of consistent success

(divergences beyond one standard deviation from

baseline in every trial of every sessions) on group

formation results in the separation depicted in Figure

5. The upper illustration outlines the averaged courses

of individual pupil size variations for poor performers

(n=4) during strategy-based intervention. The lower

figure exhibits mean values for participants

successfully exerting voluntary control over pupil

size. As can be inferred, intervention via emotionally

charged imaginations leads to increased pupil sizes

only for the “performers”. However, even an

exclusive analysis of the successful subjects did not

reveal a clear effect of further improvements during

training sessions.

Figure 6 depicts the grand-averaged course of

pupil dynamics for both groups, whereas data points

are averaged across all four training days. Again, due

to increased pupil diameters during the first ten

seconds we encounter significant higher values for

the group of “performers” (blue) (T(9) = -1.96, p =

.05). In both groups pupil sizes decline during the last

two thirds of recording and fall back on baseline

level.

Figure 6: Grand-averaged course of pupil dynamics for

“poor performing” (orange) and “performing” subjects

(blue) during the attempt to increase pupil size via self-

induced positive/negative emotions. Data points averaged

across all four training sessions. Signals depict variations

from baseline mean (set to zero; averaged standard

deviations dotted).

In contrast to group effects recorded during self-

induced emotional associations which result in

increasing pupil size, we did not discover any

performance differences with regard to the attempt to

voluntary decrease pupil diameter. Both groups failed

to exert intentional influence.

3.2.1 Baselines

According to Crider (1971) individually marked

electrodermal characteristics may serve as a

predictive variable for the performance in a variety of

information processing tasks. In accordance, we

investigated spontaneous pupil behaviour during the

15 seconds of baseline recording to explore for group-

specific differences in the underlying physiological

characteristic.

Figure 7 shows the averaged baselines preceding

the measurement of self-induced affective trials for

all participants. As can be seen, the group of

performing subjects (blue) exhibits higher values in

‐1

0 4 8 12 16 20 24 28

RelativeChanges(mm)

‐1

0 4 8 12 16 20 24 28

RelativeChanges(mm)

Time(s)

‐0,5

0,5

0 4 8 1216202428

RelativeChanges(mm)

Time(s)

TowardsVoluntaryPupilControl-TrainingAffectiveStrategies?

pupil diameter compared to the poor performing

participants (red). However, this trend was not of

statistical significance (T(-1,22); p = .28).

Figure 7: Individual courses of baseline data (averaged over

all sessions) preceding the trials of voluntary affective

interference for “performing” (blue) (n=5) and “poor

performing” (red) (n=4) subjects.

4 CONCLUSIONS

Based on findings of bio-feedback on various

physiological variables (Meichenbaum, 1976;

Bersak, 2001), effects of voluntary control over pupil

reactions were assessed in a training study. Five out

of nine participants demonstrated to wilful increase

their pupil sizes relative to baseline recordings.

Training effects did not show up, at least not over the

current period of four weeks and not even within the

well performing subjects. This latter group slightly

tends to exhibit larger pupil sizes already during

baseline recording.

The present study demonstrates that voluntary

controlling pupil responses is possible; at least for

about half of the participants. Probably, this

behaviour requires intuitive feedback on pupil size

changes providing adequate support for control and

facilitating the opportunity of intentional influencing

in pupil-based frameworks. But still, some

participants were not capable of exerting wilful

control, not even after weeklong training. Beside

strategy- or feedback-based factors it seems obvious

to discuss physiological mechanisms of action

determining the observed inter-individual

differences. However, due to the complexity of the

autonomic nervous system that mediates bodily states

of arousal, group distinctions may arise from a

multitude of different reasons. Though we discovered

only a weak trend at this point, poor performing

subjects featured slightly smaller pupil sizes already

during baseline recording. Further investigations with

larger samples may establish this trend. It would,

however, be reasonable to assume that smaller pupil

sizes are accompanied by a decrease in sympathetic

tone which indicates a lower level of arousal. Such a

reduced activity may limit the ability for spontaneous

self-induced changes between affective states. Again,

the current sample does not permit to test for

psychometric constructs, e.g. to consider whether the

outlined physiological characteristic correlates with

reduced trait anxiety. Norris (2007) reports increased

and prolonged deflections in skin conductance to

affective stimuli for subjects who score high on the

neuroticism scale. Such a factor may act as a

confounding psychological variable that could

prevent participants with lower values from utilizing

fear-laden imaginations to the same extent as anxious

subjects.

Partala & Surakka (2003) find significantly larger

pupil size dilations in response to both, negative and

positive stimulation compared to neutral auditory

stimuli. They report pupil dynamics following a

predetermined course with an increase at about 400

ms after stimulus onset and a plateau phase up to four

seconds followed by a slow decrease. Visual

inspection of the signal characteristics within our

performing subjects reveals an almost identical

course, displaying a slope during the first second and

a continuing dilation until about six seconds after

button-press. These similarities emerge despite the

varying methodological concepts. The stimulus-

driven design should provoke a phase-locked reaction

with high degree of inter-trail stability. In contrast,

the current paradigm consulted a response-locked

approach and is subject to blurred effects due to the

individually generated onset of self-induced

autonomic activation. As a consequence, participants

were encountered to immediately induce

negative/positive imaginations after indicating trial

starts via button-presses. However, though a true

starting point of the endogenous response could not

be determined, it can be stated that voluntary pupil

dilations follow a similar temporal sequence

compared to time-locked reactions. Furthermore,

performing subjects achieved twice the relative pupil

resize due to self-induced activation as reported for

stimulus-related changes (Partala, 2003).

Successful participants were able to expand pupil

diameter repeatedly and reliably beyond the range of

spontaneous fluctuations. Still, even for good

performing subjects, the controlling influence

diminishes after a period of about ten seconds. These

findings are in accordance with initial results

provided by Ekman (2008). Intentional increase of

sympathetic activity via self-induced affective

associations seems to be subject to habituation

processes that allow central inhibition of

02468101214

PupilSize(mm)

Time(s)

PhyCS2015-2ndInternationalConferenceonPhysiologicalComputingSystems

parasympathetic pathways only over a short period of

time. Ekman (2008) reports pupil size expansions

over a corresponding time length, even though as a

result of combined strategies (e.g. physical activity,

self-induced pain, focusing gaze). The current results

indicate a comparable outcome to be already achieved

on the sole basis of auto-suggestive strategies.

Naveteur (2005) shows that physiological

reactions to anxiety-related imaginations are not

universal equal across different subjects. Improved

coping strategies may provoke lower activity in

fearful participants compared to less anxious subjects.

With regard to the current issue, fear seemed

particular useful since it evokes the greatest pupil

expansion of all specific emotions (Al-Omar, 2013).

Cacioppo (2000) consistently states that feelings of

anxiety lead to a stronger increase in physiological

activity compared to various forms of happiness.

However, we did not find considerable differences

between self-induced autonomic activation arising

from positive or negative thoughts. This is in line with

reports from Partala & Surakka (2003) who recorded

no deviant pupil courses to stimulation with negative

and positive auditory stimuli.

Still, the unspecific strategies within the current

study may constitute a methodological deficit. During

training of pupil dilations via self-induced negative

imaginations, subjects consulted individualized

autobiographical associations such as fear, grief or

fury. While fear is closely related to sympathetic

activity, grief may involve a decrease of arousal

(Fredrickson, 2000). Reports from our subjects

indicate that, among others, personal mourning

experiences were consulted. It should therefore be

presumed that countervailing effects were produced

within the range of self-induced negative

associations.

In contrast to the findings mentioned above,

results with regard to voluntary pupil constrictions are

unequivocal. A systematic reduction in size could not

be observed. It is conceivable that the demands of a

relaxation task may increment sympathetic arousal

and produce physiological response patterns that

counter attempt parasympathetic inhibition.

Conversely, the feedback of successful pupil dilation

may cause an additional increase of autonomic

excitation and further reinforce the effect. However,

the inability to voluntary constrict pupil size is in

accordance with findings from Loewenfeld (1966)

who studied the effects of various sensory and

psychological stimuli to pupil dynamics and agreed

that none of them caused pupil constriction except

increased light intensity. Biofeedback paradigms

utilizing heart rate variations report similar findings.

Sakakibara (1994) observed that slowing down heart

rate is more difficult to accomplish than learning to

increase cardiac rhythm. Achieving low states of

arousal in laboratory conditions in which participants

have to continually monitor and process feedback

information appears therefore possible only to a

limited extent. However, Laeng (2013) reports

subjects to voluntary adjust their eye’s pupils to

imaginary light; a corresponding instruction may

therefore constitute a more suitable strategy to apply

voluntary pupil constriction.

Against the backdrop of these outstanding issues

it is mandatory to explore and evaluate further pupil

parameters sensitive to task- or individual-specific

differences. Applying the concept of electrodermal

“labiles” and “stabiles” (Criter, 1971) to pupil

dynamics may serve to further validate the depicted

group-specific distinctions and comprise the

opportunity to conceive a similar typology that

determines individual reactions to affective stimuli.

However, preliminary findings within a second test

series challenge a purely physiologic explanation

approach of the observed differences in performance.

Stricter requirements with regard to the adopted

strategy (e.g. no associations of grief during self-

induced negative imaginations) seem to bring out

clearer results and a considerable effect of training.

In summary, it can be noted that pupil-based

communication in human-computer frameworks

extends affective monitoring by far. If the required

conditions for operant learning are available, the

pupil constitutes an active input channel that allows

several users to reliably interfere by means of simple

cognitive strategies. Further studies will demonstrate

whether pupil-based interference solely contaminates

passive user observation or may even be consulted as

an autonomous input option.

ACKNOWLEDGEMENTS

We like to thank Juliane Georgi and Daniel Staiger

for valuable programming assistance and critical

discussions. This research was supported by the

Collaborative Research Center (sfb transregio 62) by

the Deutsche Forschungsgemeinschaft (DFG).

REFERENCES

Al-Omar, D., Al-Wabil, A. and Fawzi, M. (2013). Using

Pupil Size Variation During Visual Emotional

Stimulation in Measuring Affective States of Non

Communicative Individuals. In Stephanidis, C. (Ed.),

TowardsVoluntaryPupilControl-TrainingAffectiveStrategies?

Universal Access in Human-Computer Interaction.

User and Context Diversity (pp. 253-258). Berlin:

Springer.

Bersak, D., McDarby, G., Augenblick, N., McDarby, P.,

McDonnell, D., McDonald, B. and Karkun, R. (2001).

Intelligent Biofeedback Using an Immersive

Competitive Environment. In Abowd, G. D., Brumitt,

B. and Shafer, S. (Eds.), UbiComp 2001-Lecture Note

in Computer Science (Vol. 2201). Heidelberg,

Germany: Springer.

Bremner, F. D. (2012). Pupillometric Evaluation of the

Dynamics of the Pupillary Response to a Brief Light

Stimulus in Healthy Subjects. Investigative

Ophthalmology & Visual Science, 53, 7343-7347.

doi: 10.1167/iovs.12-10881.

Crider, A. and Lunn, R. (1971). Electrodermal Lability as a

Personality Dimension, Journal of Experimental

Research in Personality, 5, 145-150.

Ehlers, J., Georgi, J. and Huckauf, A. (in press). Improving

Voluntary Pupil Size Changes for HCI.

Ekman, I., Poikola, A., Mäkäräinen, M., Takala, T. and

Hämäläinen, P. (2008). Voluntary Pupil Size Change as

Control in Eyes Only Interaction. Proceedings of the

2008

Symposium on Eye Tracking Research &

Applications, 115-118.

Fredrickson, B. L., Mancuso, R. A., Branigan, C., and

Tugade, M. M. (2000). The Undoing Effect of Positive

Emotions. Motivation and Emotion, 24(4), 237-258.

Hess, E. H. (1972). Pupillometrics. In Greenfield, N.S. and

Sternbach, R.A. (Eds.), Handbook of Psychophysiology

(pp. 491-531). New York, NY: Holt, Rinehart &

Winston.

Hyönä, J., Tommola, J. and Alaja, A. M. (1995). Pupil

Dilation as a Measure of Processing Load in

Simulataneous Interpretation and Other Language

Tasks. The Quarterly Journal of Experimental

Psychology, 48(A), 598-612.

Jacob, R. J. K. (1996). The Future of Input Devices. ACM

Computing Surveys, 28, 177-179.

Janisse, M. P. (1974). Pupil Size, Affect and Exposure

Frequency. Social Behavior and Personality, 2, 125-

146.

Laeng, B. and Sulutvedt, U. (2013). The Eye Pupil Adjusts

to Imaginary Light, Psychological Science, 27.

doi: 10.1177/0956797613503556.

Larsen, J. T., Berntson, G. G., Poehlmann, K. M., Ito, T. A.

and Cacioppo, J. T. (2000). The Psychophysiology of

Emotion. In Lewis, M. and Haviland-Jones, J. M.

(Eds.), Handbook of Emotions (2

ed.) (pp.173-191).

New York, NY: Guilford Press.

Loewenfeld, I. E. (1966). Comment on Hess’ Findings.

Survey of Opthalmology, 11, 291-294.

Loewenfeld, I. E. (1993). Anatomy and Physiology. In

Loewenfeld, I. E. and Lowenstein, O. (Eds.), The Pupil:

Anatomy, Physiology, and Clinical Applications (pp.

498-510). Iowa City: Iowa State University Press.

Lowenstein, O., Feinberg, R. and Loewenfeld, I. E. (1963).

Pupillary Movements During Acute and Chronic

Fatigue Investigative Ophthalmology & Visual Science,

2(2), 138-157.

Meichenbaum, D. (1976). Cognitive Factors in

Biofeedback Therapy. Biofeedback and Self-

Regulation, 1(2), 201-216.

Naveteur, J., Buisine, S. and Gruzelier, J. H. (2005). The

Influence of Anxiety on Electrodermal Responses to

Distractors. International Journal of Psychophysiology,

56(3), 261-269.

Norris, C. J., Larsen, J. T. and Cacioppo, J. T. (2007).

Neuroticism Is Associated with Larger and More

Prolonged Electrodermal Responses to Emotionally

Evocative Pictures. Psychophysiology, 44(5), 823-826.

Palinko, O., Kun, A.L., Shyrokov, A. and Heeman, P.

(2010). Proceedings of the 2010

Symposium on Eye-

Tracking Research & Applications, 141-144.

Partala, T. and Surakka, V. (2003). Pupil Size Variation as

an Indication of Affective Processing. International

Journal of Human-Computer Studies, 59, 185-198.

doi: 10.1016/S1071-5819(03)00017-X.

Peavler, W.S. (1974). Pupil Size, Information Overload,

and Performance Differences. Psychophysiology,

11(5), 559-566.

doi: 10.1111/j.1469-8986.1974.tb01114.x.

Sakakibara, M., Takeuchi, S. and Hayano, J. (1994). Effect

of Relaxation Training on Cardiac Parasympathetic

Tone. Psychophysiology, 31(3), 223-228.

Simpson, H. M. and Paivio, A. (1968). Effects on Pupil Size

of Manual and Verbal Indicators of Cognitive Task

Fulfilment. Attention, Perception, & Psychophysics,

3(3), 185-190.

Winn, B., Whitaker, D., Elliott, D. B. and Phillips, N. J.

(1994). Factors Affecting Light-Adapted Pupil Size in

Normal Human Subjects. Investigative Ophthalmology

& Visual Science, 35(3), 1132-1137.

Yu, M., Kautz, M. A., Thomas, M. L., Johnson, D.,

Hotchkiss, E. R. and Russo, M. B. (2007). Operational

Implications of Varying Ambient Light Levels and

Time-Of-Day Effects on Saccadic Velocity and

Pupillary Light Reflex. Ophthalmic and Physiological

Optics, 27(2), 130-141.

doi: 10.1111/j.1475-1313.2006.00450.x.

PhyCS2015-2ndInternationalConferenceonPhysiologicalComputingSystems