Content uploaded by Wayne E Derman
Author content
All content in this area was uploaded by Wayne E Derman on Nov 15, 2017
Content may be subject to copyright.
The Effect of Short Duration Heart Rate Variability (HRV) Biofeedback on
Cognitive Performance During Laboratory Induced Cognitive Stress
GABRIELL E. PRINSLOO
1
*, H. G. LAURIE RAUCH
1
, MICHAEL I. LAMBERT
1
,
FREDERICK MUENCH
2
, TIMOTHY D. NOAKES
1
and WAYNE E. DERMAN
1
1
MRC/UCT Research Unit for Exercise Science and Sports Medicine, University of Cape Town, Cape Town, South Africa
2
The National Center on Addiction and Substance Abuse (CASA), Columbia University, New York, USA
Summary: The aim of this study was to examine the effect of 10 minutes of heart rate variability (HRV) biofeedback on cognitive
performance and affect scores during induced stress. Eighteen healthy male volunteers (aged 23–41 years) exposed to work-
related stress, were randomised into an HRV biofeedback intervention (BIO) and a comparative intervention group (COM).
Subjects completed a modified Stroop task, which included having to mentally count 18 white squares randomly presented between
colour words, before and after a 10-minute intervention. Subjects also completed questionnaires to rate their anxiety. BIO subjects
improved their reaction times and consistency of responses, and made fewer mistakes in counting squares during the modified
Stroop task. They also felt more relaxed, less anxious and less sleepy than the COM subjects. In conclusion our results suggest that
short duration HRV biofeedback is associated with improved cognitive performance while concurrently aiding relaxation.
Copyright #2010 John Wiley & Sons, Ltd.
Recently, numerous publications of biofeedback and
neurofeedback training have demonstrated efficacy in the
management of a wide range of medical and psychological
disorders (Egner, Zech, & Gruzelier, 2004; Gunkelman JD &
Johnstone J, 2005; Hassett et al., 2007; Lehrer, Vaschillo, &
Vaschillo, 2000; Lehrer et al., 2003; Moss, 2004; Nolan et al.,
2005). Neurofeedback training generally requires costly
electroencephalographic (EEG) equipment, however bio-
feedback training including the use of heart rate variability
(HRV) measuring devices, provide a more economical
alternative.
HRV is regulated by neural input from both the
parasympathetic and sympathetic divisions of the autonomic
nervous system. Generally the parasympathetic division has
the predominant effect on HRV by either increased
parasympathetic input or parasympathetic withdrawal
(Berntson et al., 1997; Grasso, Schena, Gulli, & Cevese,
1997). Adverse chronic psychological stress induces states of
low HRV (Delaney & Brodie, 2000; Hjortskov, Rissen,
Blangsted, Fallentin, Lundberg, & Sogaard, 2004; Lucini, Di
Fede, Parati, & Pagani, 2005; Madden & Savard, 1995) and
decreases the high frequency (HF) component of HRV
indicating decreased vagal activity (decreased parasympa-
thetic activation) (Bernardi et al., 2000; Delaney & Brodie,
2000; Dishman, Nakamura, Garcia, Thompson, Dunn, &
Blair, 2000).
Furthermore, chronic psychological stress also impairs
general cognitive function (Kirschbaum, Wolf, May,
Wippich, & Hellhammer, 1996; Lupien et al., 1997; Ohman,
Nordin, Bergdahl, Slunga, & Stigsdotter, 2007) specifically
including impaired set shifting (Orem, Petrac, & Bedwell,
2008). Both acute (Payne, Nadel, Allen, Thomas, & Jacobs,
2002) and chronic (Seeman, McEwen, Singer, Albert, &
Rowe, 1997) stress result in impaired memory related to an
increase in plasma cortisol concentrations (Elzinga, Bakker,
& Bremner, 2005; Kirschbaum et al., 1996; Lupien et al.,
1997). Stress impairs working memory linked to high
cortisol concentrations (Oei, Everaerd, Elzinga, van Well, &
Bermond, 2006) and has also been associated with greater
error-related brain activity (Hajcak, McDonald, & Simons,
2003).
On the other hand, greater HRV has been associated with
increased executive functioning including faster reaction
times and more correct responses to cognitive tasks (Hansen,
Johnsen, Sollers, Stenvik, & Thayer, 2004; Hansen, Johnsen,
& Thayer, 2003). The finding that increased HRV may be
associated with improved cognitive performance, illustrates
the importance of techniques that could counter the vagal
lowering effects of chronic stress.
The StressEraser is a handheld portable HRV biofeedback
device (StressEraser
TM
, Helicor, USA) which is registered as
a 510 (k) exempt, Class II medical device (FDA) indicated
for use in relaxation, relaxation training and stress reduction.
This device has been well described by Muench (2008).
Studies using the StressEraser have shown that 20 minutes of
HRV biofeedback daily has resulted in decreased anxiety
(Reiner, 2008) and been more effective at reducing anxiety
than a concentrative biofeedback control (Muench, 2008).
Therefore, the aim of this study was to examine the acute
effect of 10 minutes of HRV biofeedback using the Helicor
StressEraser on cognitive performance during induced stress
in the form of a modified Stroop task.
METHODS
Participants
Eighteen male volunteers aged between 23 and 41 years who
were employed in senior managerial positions were recruited
for this study. To be included in this study, volunteers had to
have been exposed to work related stress and subjectively
rate their own perception of life stress as high. Their life
stress was measured using the trait component of a
Spielberger State-Trait Anxiety Inventory (STAIT) which
they completed online. A previous clinical diagnosis of
Applied Cognitive Psychology,Appl. Cognit. Psychol. (2010)
Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/acp.1750
*Correspondence to: Dr Gabriell E. Prinsloo, MRC/UCT Research Unit for
Exercise Science and Sports Medicine, Department of Human Biology,
Faculty of Health Sciences, University of Cape Town, PO Box 115, New-
lands 7725, Cape Town, South Africa. E-mail: Gabriellprinsloo@gmail.com
Copyright #2010 John Wiley & Sons, Ltd.
anxiety related disorders disqualified participation in the
study.
Further, exclusion criteria included: previously diagnosed
cardiac or psychiatric disorders, current use of psychotropic or
heart rate altering medications or current use of stimulants
or recreational drugs, as well as volunteers practicing
regular meditation techniques. Volunteers were asked about
their exercise practices as well as breathing and meditation
techniques as these would have had the greatest impact on the
outcome of the intervention. Volunteers who were technically
unable to use the biofeedback device or perform the modified
Stroop task were excluded from the trial. This applied to two
subjects, one of whom had chronic low circulation and so the
device was unable to obtain a clear pulse reading. The other
subject was excluded as on completion of the familiarisation
Stroop task he still did not understand what to do and continued
to confuse the response keys, and therefore could not respond
effectively. Subjects were screened via email. On arrival before
the training session a brief history was taken by a medical
doctor, to confirm the responses.
The sample size was limited to 18 people as the study also
included an EEG component to be reported subsequently and
as a result of the lack of availability of stressed men in senior
managerial positions. Despite the small sample size, the
difference in cognitive performance which was the main
outcome variable which the study was powered to detect was
significant, indicating that the sample size was adequate. We
elected to test only men in an attempt to limit heterogeneity
because men and women have different cortisol responses to
stress (Zorawski, Blanding, Kuhn, & LaBar, 2006) as well as
differences in cognitive performance in response to stress
(Stark et al., 2006) and differences in memory (Seeman et al.,
1997; Wolf, Schommer, Hellhammer, McEwen, & Kirsch-
baum, 2001).
Volunteers were instructed not to eat a heavy meal, ingest
caffeine or alcohol, or exercise within 4 hours before arriving
at the laboratory. Compliance was checked before the onset
of each testing session.
The study protocol was approved by the Research and
Ethics Committee of the University of Cape Town (Rec ref:
296/2005) in accordance with the Declaration of Helsinki.
All subjects signed informed consent prior to participation in
the study.
Subjects were then randomly assigned to either an HRV
biofeedback (BIO) group or a comparative (COM) group
using a process of stratified randomisation. The initial group
of subjects was matched into two groups based on age and
thereafter subjects were randomly assigned to either group.
All subjects were informed that they would be using a
biofeedback device to aid relaxation. We decided to use a
single intervention and comparative group design, as a
crossover design may have confounded the interpretation in
the event of there being any carry over from the intervention.
At recruitment all subjects completed the trait component of
a Spielberger State-Trait Anxiety Inventory (STAIT) online.
Training and familiarisation protocol
All subjects underwent a single standardised training session
during the week prior to the start of their experimental trials.
A hand-held mobile HRV biofeedback device (StressEr-
aser
TM
, Helicor, USA) validated by Heilman, Handelman,
Lewis, and Porges (2008) was used for both the training and
the testing sessions in the BIO group. An infrared emitter and
sensor incorporated into the device into which the subject
placed their index finger, measured the heart rate on a beat to
beat basis. As Muench (2008) explains, the device measures
the real-time interbeat-interval (IBI) of the heart using finger
photoplethysmography. The IBI data are transformed and
displayed as an RSA wave on an LCD screen, allowing users
to see the real-time fluctuations of their pulse rates. Using the
RSA wave, users are guided to find their optimal slow
respiration rates and to maintain a cognitive focus so that
real-time heart rates and respirations covary in a perfect
phase relationship. To achieve this, the subjects were
instructed to inhale until the RSA wave reached its peak
and exhale until the wave started to rise again.
The device rewards users with points based on the
wavelength for each RSA cycle. If the wavelength meets a
certain threshold (10 seconds), users are given 1 point
marked by three vertical squares. Two vertical squares
receive .5 point and one vertical square receives no credit.
The goal is to accumulate continuous points during the
session. To assist users in obtaining points, the device
anticipates the peak of the RSA wave based on its slope and
marks the peak with a triangle. The peak of the wave
indicates the moment heart rate deceleration is to begin,
indicating the parasympathetic response. Users are
instructed to begin their exhale when the triangle appears.
They are instructed to extend their exhale for as long as
possible until the wave begins to rise again. Although the
device offers points based on RSA wavelength, users were
instructed to maximise RSA amplitude simply by following
the RSA wave (e.g. inhale until the wave stopped rising and
exhale until the wave stops falling).
This is not technically resonance frequency breathing as
time domain analysis was used rather than frequency
analysis; however it does functionally facilitate a similar
breathing frequency. As the average respiratory frequency of
the BIO group during the intervention was 0.10 0.01 Hz
while the COM groups was 0.25 0.05 Hz, it can be
suggested that the BIO group was functionally achieving a
similar frequency as targeted in traditional HRV biofeedback
using frequency analysis. Resonance frequency breathing
differs between individuals and ranges from 4.5 to 7 breaths
per minute. Breathing at this frequency results in large
increases in HRVand baroreflex gain (Lehrer et al., 2003) as
well as optimising respiratory efficiency (Giardino, Chan, &
Borson, 2004). In addition to the R–R wave, a cumulative
score on the display screen enabled them to monitor the
impact of their breathing on their HRV.
The device used in the COM group was also manufactured
by Helicor (USA). It appeared identical to the BIO device but
made use of a different algorithm to display a wave on the
screen. The algorithm that generated this wave was derived
from the subject’s heart rate measured by the sensor, divided
by 2 plus a random number which ranged between 0 and 25%
of the heart rate/2 value. This result was then smoothed by
averaging the calculation over 5 seconds. Subjects were
informed that the wave represented their blood density and
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
G. E. Prinsloo et al.
were instructed to watch the wave whilst releasing stressful
thoughts. They were not instructed to alter the wave in any
way.
After subjects had received their respective instruction,
they completed a formal 10-minute biofeedback session with
either the BIO or COM device and their scores were
recorded. A score of 30 or more on the BIO device was
indicative of a successful session. Baseline blood pressure
(BP) and heart rates (HR) were recorded at the training
session using an automated blood pressure monitor (model
HEM-705CP, Omron, Illinois, USA, validated by O’Brien,
Mee, Atkins, and Thomas (1996)).
Experimental trial
The time line for the experimental trial is shown in Figure 1.
On arrival at the laboratory, subjects completed the state
component of a Spielberger State-Trait Anxiety Inventory
(STAIS) and the Smiths Relaxation States Inventory 3
(SRSI3) and were reminded how to use their respective
devices. Blood pressure and heart rate were recorded after
the questionnaires were completed. Subjects then underwent
a full familiarisation modified Stroop task lasting 5 minutes
and 24 seconds. After the familiarisation test, electrodes and
transducers were applied to the subjects and connected to a
Biopac MP150WSW (Biopac Systems, Goleta, CA, USA) to
record electrocardiogram (ECG) and respiratory rates (Rr).
The Biopac MP system is a laboratory-based physiological
monitoring system which is commonly used as a research
tool, and has been cited in numerous publications (Heilman
et al., 2008). Subjects were seated throughout the testing and
were instructed to move as little as possible. Leads were
taped down to prevent interference.
Subjects were instructed to relax with their eyes closed for
5 minutes (Rest 1) during which time measurements of
baseline ECG and respiratory rate were recorded. Immedi-
ately after the baseline period and without moving, subjects
were instructed to open their eyes and complete the pre-
intervention Stroop task (Stroop 1) which lasted 5 minutes
and 24 seconds in duration.
Subjects then completed a 10-minute intervention with
either the BIO or COM device. After the intervention,
subjects completed a further 5-minute rest period (Rest 2)
with their eyes closed before completing a post-intervention
Stroop task (Stroop 2). ECG and respiratory rate were
recorded throughout the testing. BP and HR were recorded
after the intervention, after Rest 2 and after Stroop 2. It was
not feasible to record BP more frequently because it may
have interfered with the flow of the trial.
After the final HR and BP measurements, subjects
completed post-testing STAI-S and SRSI3 questionnaires.
In addition, subjects were also asked to rate the subjective
Figure 1. Time line for the experimental trial. Abbreviations: min, minute; s, second
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
The effect of heart rate variability biofeedback
efficacy of the intervention as well as feelings of sleepiness
using a visual analogue scale (VAS) from 1 to 10 with 1 being
‘not sleepy at all’ and 10 being ‘very sleepy’.
Modified Stroop Task
We modified the original Stroop Task (Stroop, 1935) by
computerising so that subjects responded by pressing keypad
buttons instead of responding verbally. In addition, we added
a working memory component. The colour word component
was previously validated by Rauch, John, St Clair Gibson,
Noakes, and Vaughan (2005) as follows. Twelve female
subjects (students aged 18–25) repeated a Stroop Task of
about 5-minute duration which consisted of only incongruent
cues. Each Task contained 72 colour words with each word
being displayed for 400 ms followed by a 3.5-second
response time. Five Tasks were repeated one after another
with a 1-minute break in between each Task. Between 7 and
10 days later the subjects repeated the 5 5 minute Stroop
Tasks at the same time of day (day 2).
In the current study, the modified Stroop task entailed the
individual presentation of cues (2 cm in height), in the centre
of a computer monitor on a black background. Cues consisted
of colour-words (red, blue, green and yellow) and single white
squares. Cues appeared every 3 seconds and were displayed
for 400 ms after which they were replaced by a black screen
which lasted for 2600 ms and which constituted the response
period. The four colour words were presented in five different
colour inks: red, blue, green, yellow and grey. The colour
words were either presented in grey ink or in a colour ink
incongruent with the meaning of the word, i.e. red in blue
colour, or green in yellow colour, but never red in red colour or
blue in blue colour, etc. The subjects were asked to respond as
quickly and as accurately as possible by pressing one of four
response buttons to indicate either the colour of the word (if
the word was written in colour ink), or the word itself (if the
word was written in grey ink). Subjects were not required to
respond to the white squares other than counting the
cumulative total number of squares throughout the test. They
were instructed to report the total number of squares at the end
of each Stroop. This value was compared between Stroops 1
and 2. They were not told how many squares there were in the
test, only that they were randomly generated. Counting
squares tests the updating of working memory, while
responding to the colour words tests inhibition of prepotent
responses together with the delicate balancing of the speed-
accuracy trade-off (Forstmann et al., 2008).
Subjects used the index and middle fingers of each hand to
press the response buttons (1 for red and 2 for blue with the
left hand and 7 for green and 8 for yellow with the right hand)
on an 8-button response box.
In total 108 cues were randomly presented, 18 incongruent
colour words in each of the four colours (72 incongruent
words in total); 18 grey words and 18 white squares. The use
of the grey words ensured that subjects had to read and
recognise the colour words rather than just noticing the
colours, thereby invoking the Stroop effect (Stroop, 1935).
Data recording
ECG activity was recorded from three electrodes (Blue
Sensor, Ambu, Denmark) placed in positions representing
Eindhoven’s triangle namely, subclavicular bilaterally and
over the left anterior superior iliac crest. The skin surface was
cleaned and gently abraded with an alcohol swab before
electrodes were attached. Electrode cables were taped down
to prevent movement artefact. The three electrodes were
connected to the Biopac ECG amplifier set to band-pass filter
between 0.5 and 35 Hz and a sampling frequency of 1000 Hz.
ECG recordings were analysed with AcqKnowledge for
MacIntosh OS X (version 3.9.0). This software used a
modified Pan and Tompkins algorithm to detect QRS
complexes. The filtered ECG recording tachograms were
then visually inspected to determine the correct recognition
of QRS complexes and T waves. Missed and ectopic beats
were corrected by either adding or spacing beats. This
procedure was only necessary in two of the 90 ECG
recordings and only affected a total of five heartbeats.
Only after each tachogram showed no spurious beats were
the data analysed using HRV analysis software from the
Biomedical Signal Analysis Group (Department of Applied
Physics, University of Kuopio, Finland). Data were
transformed using autoregressive (AR) analysis with an
AR model order of 25 into LF (0.04–0.15 Hz) and HF (0.15–
0.4 Hz) components (Task Force, 1996; Yildiz & Ider, 2006).
We calculated the frequency domain parameters including
low frequency (LF) power and high frequency (HF) power.
The HRV power spectrum analysis during the intervention
was conducted from minute 1 to minute 6 of the 10-minute
session. This provided a 5-minute period as the standard time
period of comparison in keeping with the recommendation
that the optimal length of recording for short-term recordings
is 5 minutes (Task Force, 1996). Minute 1 to minute 6 was
selected to allow the subjects 1 minute to equilibrate their
respective physiological states after having just completed a
Stroop task. In addition, this allowed time for the subjects to
achieve their optimal breathing frequency. HRV was
measured during the intervention to provide evidence that
the biofeedback device was working as expected. The HRV
power values in ms
2
were natural log transformed to adjust
for the unequal variance.
The respiratory rate per minute was measured via a force
transducer fixed to a belt placed around the chest wall.
Subjects were asked to expel the air from their lungs when
the transducer belt was first fitted and then instructed to
breathe normally. The chest transducer was connected to an
amplifier with a low-pass 10Hz filter. Respiratory frequen-
cies (RF), i.e. breaths per second, were calculated from the
respiratory rate.
Statistical analysis
Differences between blood pressure and heart rate data as
well as STAIS and SRSI3 questionnaires, and reaction time
between Stroops 1 and 2 were analysed using repeated
measures analysis of variance. Specifically, differences in the
main effects (group and time) and the interaction of group X
time were determined. A Tukey’s post hoc test was used to
determine specific differences in the event of there being a
significant main effect or interaction.
Subject characteristics, HRV power values during the
intervention and differences in STAIT and VAS scores
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
G. E. Prinsloo et al.
between the BIO and COM group were analysed using an
independent t-test as there was only a single time period. In
addition an independent t-test was used to analyse the
percentage improvement in reaction time and percentage
improvement in standard deviation of reaction between
Stroops 1 and 2. This last analysis was done in view of the
wide range of standard deviation as a result of the different
responses to different colours. We believe that this is an
important finding and so have included it; however it is
important to interpret the percentage improvement results
with caution at the risk of making a type 1 error.
Levene’s test of homogeneity revealed that the respiratory
data and mistakes made in responding to cues during the
Stroop test were of unequal variance, and thus non
parametric tests were used to determine differences with
respect to the outcome measures. Accordingly the Fried-
man’s analysis of variance for repeated measures was used to
determine differences in values between tests. A Wilcoxon
test was used to locate the specific differences when the
overall value was significant and a Mann–Whitney test was
used to compare values between groups.
All data are described as means standard deviation (SD).
Ap-value of <0.05 was considered to be statistically
significant.
RESULTS
Subject characteristics
The BIO and COM groups were similar in age (33 6vs.
34 6 years, respectively), body mass index (28.7 6.8 vs.
27.2 4.5 kg/m
2
, respectively) and adherence to exercise
training (3 1vs.31 units respectively) measured on a
scale of 1–5 (1 ¼less than once a month, 5 ¼every day).
Questionnaires
Psychometric data from the STAIT, STAIS, SRSI3 and VAS
questionnaires are shown in Table 1. The STAIT scores
(48 6vs.447 points; BIO vs. COM), were not different
between groups. Subjective measures of STAIS showed a
time effect as they were significantly lower after the protocol
in both BIO and COM groups (F
1,16
¼23.05, p<0.001),
however there was no group or interaction effect. The Smith
Relaxation States Inventory 3 (SRSI3) total scores were also
not different between groups prior to the testing protocol but
were significantly lower after the protocol in both BIO and
COM groups (F
1,16
¼10.72, p<0.001), with subjects in the
BIO group being significantly more relaxed after the testing
protocol than those in the COM group (65.9 12.3 vs.
53.9 13.0 points, BIO vs. COM, F
1,16
¼6.79, p<0.05).
Subjects in the BIO group felt less sleepy during the second
rest than those in the COM group as shown by the VAS score
(5 3vs.72 points respectively, p<0.05). The remainder
of the VAS scores were not different between groups.
Heart rate and blood pressure
Cardiovascular parameters measured before the test, after the
intervention, after the second rest and after the test are shown
in Table 2. Data were missing from two subjects in the BIO
group for this analysis as a result of technical problems.
There were no differences with regards to systolic or
diastolic blood pressure throughout testing and there was no
reduction in blood pressure after testing. Heart rate was also
not different between the BIO and COM groups throughout
testing. However in both groups, the heart rate after the test
was lower than before the test ( p<0.01) (Table 2).
Blood pressure recording were borderline high. The
increase in blood pressure occurred mainly in the systolic
values, with nine of the subjects having high systolic values
and only three having very slightly high diastolic values.
While stress is frequently associated with hypertension
(Lucini et al., 2005), situational anxiety results in a specific
increase in systolic blood pressure. The subjects were
anticipating performing challenging stressful tasks alongside
various physiological measures neither of which they had
experienced previously, giving rise to a situational anxiety.
Respiratory rate
We have chosen to display the respiratory data using both
respiratory rate and respiratory frequency (Table 3) as
respiratory rate is more easily applicable to clinical
situations, while the respiratory frequency parameter is
beneficial in understanding the HRV data. Respiratory data
were missing for one of the BIO subjects as a result of
equipment failure.
The respiratory rate was lower in the BIO group compared
to the COM group during both the rest 1 period ( p<0.05)
and the rest 2 periods ( p<0.01). However, there were no
differences in respiratory rate during the modified Stroop
task either before or after intervention both between and
within groups. Respiratory rate in the BIO subjects was
20 4vs.204 breaths per minute, while in the COM
group it was 19 4vs.194 breaths per minute.
The respiratory rate during the intervention was signifi-
cantly lower in the BIO vs. the COM group (6 1vs.153
breaths per minute respectively, p<0.001). The respiratory
rate of the BIO group during the intervention was also
Table 1. Results of the questionnaires used at recruitment and
throughout the experimental trial in the BIO (N¼9) and COM
(N¼9) groups
Test BIO score COM score
STAIT 48 6447
STAIS —pre-test 41 94114
STAIS —post-test 30 7
32 9
SRSI3 —pre-test 51.5 9.6 50 12.4
SRSI3 —post-test 65.9 12.3
,
53.9 13.0
VAS —helpful 7 162
VAS —sleepy after 5 minutes 4 363
VAS —sleepy after 10 minutes 5 372
VAS —sleepy during rest 5 3
#
72
#
BIO, biofeedback group; COM, comparative group; STAIT, Spielberger
Trait Questionnaire; STAIS, Spielberger State Questionnaire; SRSI3, Smith
Relaxation States Inventory 3; VAS, Visual analogue scale.
p<0.05 BIO SRSI3 vs. COM SRSI3 (interaction effect).
p<0.001 BIO SRSI3 post-test vs. BIO SRSI3 pre-test, COM SRSI3 post-
test vs. COM SRSI3 pre-test, BIO STAIS post-test vs. BIO STAIS pre-test,
COM STAIS post-test vs. COM STAIS pre-test.
#
p<0.05 BIO VAS sleepy during rest vs. COM VAS sleepy during rest.
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
The effect of heart rate variability biofeedback
significantly lower than during rest 1 ( p<0.05) as well as
rest 2 ( p<0.05), which together with the lower respiratory
rate during the intervention illustrate the effectiveness of the
intervention. In this regard, the respiratory frequencies of all
nine subjects in the BIO group were within the LF band
(0.10 0.01 Hz, with a range of 0.08–0.13 Hz; LF
band ¼0.04–0.15 Hz). However the respiratory frequencies
of all nine COM subjects were within the HF band
(0.25 0.05 Hz, with a range of 0.17–0.32 Hz; HF
band ¼0.15–0.4 Hz). As a result the 5 minute natural log
transformed total power values were significantly greater in
the BIO compared to the COM group (7.52 0.90 vs.
6.20 0.95 ms
2
/Hz respectively, p<0.01) and more
specifically the LF power values were significantly greater
in the BIO compared to the COM (7.48 0.91 vs.
5.91 0.88 ms
2
/Hz, respectively, p<0.005).
Cognitive performance
Subjects were instructed to respond as fast and as accurately
as possible. During any decision making task subjects have to
balance speed with accuracy (known as the speed-accuracy
trade-off) and studies have shown that when subjects focus
more on responding as fast as possible they make more
mistakes, thereby sacrificing accuracy for speed (Franzon &
Hugdahl, 1987; Rauch et al., 2005).
The data from one subject in the COM group were
excluded as the mistakes that he made during Stroop 2 were
more than three standard deviations greater than the mean
while all of the other subjects were within 1 standard
deviation. Furthermore, he increased his reaction time more
than any other subject and in some colours more than two
standard deviations from the mean. This suggests that he was
not as focused on minimising mistakes as the other subjects
were, and thus was not performing optimally in the Stroop
tests. All his Stroop data were excluded as the changes in
reaction time would have an impact on changes in his
accuracy, since subjects balance speed and accuracy
(Forstmann et al., 2008). The data from two other subjects
(one from each group) were missing as a result of failure of
the recording equipment.
Mistakes made
A detailed analysis of the mistakes made by the subjects in
the BIO and COM groups during the modified Stroop tasks
before and after intervention is shown in Figure 2.
There were no differences in the total mistakes made in
either group in identifying the 108 cues over time. By
differentiating the mistakes made on identification of colour
words from the mistakes made while counting the total
number of squares we were able to determine that word
mistakes were not significantly different pre vs. post-
intervention in either group. However, there was a significant
difference in mistakes made in responding to squares. After
the intervention no subject in the BIO group missed any
squares (1 1vs. 0 squares missed, Stroop 1 vs. Stroop 2,
p<0.05), while subjects in the COM group missed as many
squares after intervention as they did before intervention
(1 1vs.01 squares missed, Stroop 1 vs. Stroop 2).
Table 2. Blood pressure and heart rate measurements from once off recordings throughout the experimental trial in the BIO (N¼7) and COM
(N¼9) groups
BIO COM
BP (mmHg) HR (b/min) BP (mmHg) HR (b/min)
Pre-test 141/83 12/10 71 10 143/81 13/9 70 11
Post-intervention 136/81 13/11 67 7 133/82 10/11 66 10
Post-rest 2 135/82 18/13 67 9 138/83 9/8 63 9
Post-test 140/83 14/12 66 9
140/83 22/8 63 10
BIO, biofeedback group; COM, comparative group; BP, blood pressure (systolic/diastolic); HR, heart rate; mmHg, millimetres mercury; b/min, beats per minute.
Time effect
p<0.01 BIO pre-test vs. BIO post-test, COM pre-test vs. COM post-test.
Table 3. Respiratory rate and respiratory frequency throughout the experimental trial in the BIO (N¼8) and COM (N¼9) groups
BIO COM
Rr (b/min) RF (b/s) Rr (b/min) RF (b/s)
Rest 1 10 4
0.17 0.06
15 4
0.25 0.07
Stroop 1 20 4 0.34 0.07 19 4 0.32 0.07
Intervention 6 1
,a
0.10 0.01
,a
15 3
0.25 0.05
Rest 2 10 5
,b
0.17 0.09
,b
15 3
0.26 0.04
Stroop 2 20 4 0.34 0.06 19 4 0.31 0.07
BIO, biofeedback group; COM, comparative group; Rr, respiratory rate; RF, respiratory frequency; b/min, breaths per minute; b/s, breaths per second)
Time effect
a
p<0.05 BIO intervention Rr vs. BIO rest 1 Rr, BIO intervention RF vs. BIO rest 1 RF.
b
p<0.05 BIO rest 2 Rr vs. BIO intervention Rr, BIO rest 2 RF vs. BIO intervention RF.
Group effect
p<0.05 BIO rest 1 Rr vs. COM rest 1 Rr, BIO rest 1 RF vs. COM rest 1 RF.
p<0.001 BIO intervention Rr vs. COM intervention Rr, BIO intervention RF vs. COM intervention RF.
p<0.01 BIO rest 2 Rr vs. COM rest 2 Rr, Bio rest 2 RF vs. COM rest 2 RF.
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
G. E. Prinsloo et al.
Reaction time
Figure 3 shows each subjects average percentage improve-
ment in reaction time to all colours and Figure 4 shows each
subjects average percentage improvement in reaction time to
the grey colours. ANOVA showed a time effect with
improved reaction time to all colours (F
1,13
¼12.54,
p<0.01) and to grey colours (F
1,13
¼5.19, p<0.05), as
well as an interaction effect with the BIO group improving
more than the COM group in response to grey colours (BIO
group: 1.30 0.21s vs. 1.14 0.21s, pre vs. post; COM
group: 1.26 0.17s vs. 1.27 0.23s, pre vs. post; F
1,13
¼
7.32, p <0.05). There was no group effect.
When looking at the percentage improvement in reaction
time, the BIO group improved significantly more than the
COM group in their reaction time when responding to grey
colours ( p<0.05) as well as to all colours ( p<0.05).
It is interesting to note that all eight of the BIO subjects
improved their reaction time more than 5%, while only three
of the COM subjects improved their reaction time by more
than 5%. Furthermore seven of the BIO subjects had tighter
standard deviation values in their reaction times compared to
only two of the COM groups.
Figure 5 shows the percentage change in the standard
deviation of the reaction time between Stroops 1 and 2. The
BIO group significantly decreased the standard deviation of
their reaction times to all colours compared to the COM
group ( p<0.05).
DISCUSSION
The key finding in this study was that the use of a short
duration HRV biofeedback intervention resulted in improved
cognitive performance during a modified Stroop task. This
effect was not seen after the comparative intervention.
Importantly, during the intervention there were significant
physiological differences between groups as seen by the
decreased respiratory rate and increased HRV in the BIO
compared to COM subjects. Improvements in cognitive
performance were evidenced by increased speed and
consistency of reaction time after the intervention as well
as fewer mistakes made in counting the number of square
cues. This suggests that the improvements in cognitive
performance resulted from the slowed respiratory rate and
resultant increased HRV.
Figure 3. Individual subject’s average percentage improvement in reaction time to all colours during the first and second modified Stroop tests
in the BIO and COM groups. Abbreviations: BIO, biofeedback group; COM, comparative group. # p<0.05. BIO mean percentege
improvement in reaction time vs. COM mean percentage improvement in reaction time
Figure 2. The percentage of correct responses to squares cues (a),
word cues (b) and total cues (c) during the first and second modified
Stroop tests in the BIO and COM groups. Abbreviations: BIO,
biofeedback group; COM, comparative group. # p<0.05 BIO
square mistakes Stroop 2 vs. BIO square mistakes Stroop 1
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
The effect of heart rate variability biofeedback
Factors influencing cognitive performance and mistakes
made are complex. The modified Stroop Task provided a
complex cognitive challenge which included three elements
of executive functioning; updating of working memory;
mental set shifting; and inhibition of prepotent responses
(Miyake, Friedman, Emerson, Witzki, Howerter, & Wager,
2000). Counting squares entails updating information in
working memory; mental set shifting occurs when subjects
shift from counting squares to responding to the colour
words; and suppressing prepotent responses (inhibition)
occurs when reacting to the colour rather than to the word.
Both groups improved their reaction time between Stroops
1 and 2. This improvement is unlikely to be due to the
learning effect of the Stroop. A previously completed
validation trial showed a learning effect between the first and
second modified Stroop Task but not between the second and
third (Rauch et al., 2005). For this reason we used a
familiarisation Stroop Task as the first Task, so that there
would have been little learning effect between Stroops 1 and
2. ANOVA showed that the BIO group improved signifi-
cantly more than the COM group, reinforcing that this
improvement did not result from a learning effect.
The standard deviation of the reaction time to cues in the
BIO group decreased after the intervention while the
standard deviation of the COM group increased. This shows
that the BIO groups were more consistent in their responses
during the post-Stroop task than during the pre-task.
These findings suggests that the BIO subjects were more
focused, while the COM subjects may have allowed their
attention to wander somewhat more during the post-
compared to the pre-Stroop task.
Remarkably, while there were no differences in word
mistakes between groups, none of the BIO subjects missed
any of the squares after the intervention resulting in a
Figure 4. Individual subject’s average percentage improvement in reaction time to Grey colours during the first and second modified Stroop
tests in the BIO and COM groups. Abbreviations: BIO, biofeedback group; COM, comparative group. # p<0.05. BIO mean percentege
improvement in reaction time vs. COM mean percentage improvement in reaction time
Figure 5. Individual subject’s average percentage change in the standard deviation if their reaction time to all colours during the first and
second modified Stroop tests in the BIO and COM groups. Abbreviations: BIO, biofeedback group; COM, comparative group. # p<0.05. BIO
mean percentage change in the standard deviation of reaction time vs. COM mean percentage change in the standard deviation of reaction time
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
G. E. Prinsloo et al.
significant improvement. The COM subjects missed as many
squares after the intervention as before. This strongly
suggests working memory enhancement in the BIO, but not
in the COM subjects.
In a previous validation trial Rauch et al. (2005) showed
that some subjects were more intent on obtaining a faster
reaction time while others seemed to be more concerned with
accuracy and so sacrificed speed. Subjects consistently made
a similar amount of mistakes—whether high or low—in
response to a modified Stroop task (including only
incongruent colour words and no white squares) on two
separate days. Rauch et al. (2005) found that there was an
inverse relationship between the number of mistakes made
and the reaction time. The group of subjects that made more
mistakes increased their reaction time more from day 1 to
day 2, while the conservative approach resulted in no
improvements in reaction time but fewer mistakes. These
subjects could generally tell how many mistakes they made,
but they did not have a clear idea about their reaction time
(unpublished observations).
In this regard, Mayes and co-workers found that 7–9-year
old children, who were cocaine exposed prenatally, reacted
significantly slower than non-drug-exposed matched con-
trols during a Stroop task, but they did not make more
mistakes on the colour words (2% vs. 3%). Thus, even though
the cortical processing speed of the cocaine-exposed children
was impaired, they merely slowed their reaction time to keep
their mistakes in line with controls (Mayes, Molfese, Key, &
Hunter, 2005).
We also found Stroop word mistake percentages in the 1–
3% range. While most of our subjects increased their reaction
time in the post-intervention modified Stroop task (except for
two COM subjects), they made similar amounts of word
mistakes pre vs. post-intervention. This suggests that, in line
with the above findings, our subjects were willing to sacrifice
reaction time up to a point, i.e. to keep their mistakes below
3% throughout. This indicates that there is a range of
mistakes, remarkably similar throughout the above studies,
that subjects are comfortable with; and that reaction times are
adjusted to accommodate this mistake range.
However it is important to note that the subjects in the BIO
group increased their reaction time more than the COM
group, while not increasing the amounts of mistakes made,
indicating improved cognitive performance.
Despite the fact that the control subjects were given
meaningless data, they perceived that the control interven-
tion was as effective as the HRV biofeedback intervention in
aiding relaxation. This was indicated by similar VAS scores
when subjects were asked to rate the efficacy of the
intervention in helping them to relax on a scale of 1–10 at the
conclusion of the experimental trial.
The effectiveness of both interventions in aiding
relaxation and reducing anxiety is seen by the significant
improvement in the SRSI3 and the Spielberger State total
scores after testing in both groups. However, the BIO group
SRSI3 score improved significantly more than the COM
group indicating that HRV biofeedback was more effective
than the control intervention in aiding relaxation. A further
difference was that subjects in the COM group felt
significantly more sleepy 5 minutes after the intervention
than the BIO subjects. This suggests that, while both groups
felt more relaxed after the intervention, the BIO group felt
both more relaxed and more alert which explains the
improvement in the Stroop results.
With regards to the COM intervention, the focus of
attention on a single variable has been shown to facilitate the
relaxation response and reduce skeletal muscle tension
(Benson, 1993). There is thus physiological benefit in
focusing attention on a single object, which was evident in
the COM group by positive changes in HR, VAS and STAIS
and SRSI3 scores. What this study adds, is that there seems to
be a greater benefit in focused attention together with slow
breathing as seen in the BIO group.
It is of interest to note that the above cognitive
improvements were witnessed after only a 10-minute
intervention. We were specifically interested in the acute
effect of 10 minutes of biofeedback, as this is the period
recommended by the manufacturer for clinical use. The
minimum recommended time per session is 5 minutes with a
goal to accumulate a total of 20 minutes per day (Muench,
2008). The intention is to create a device that could be used
easily for short periods of time as needed throughout the day.
Importantly, the results of this study perhaps fail to
appreciate the magnitude of the effect which might become
evident with longer duration intervention or indeed regular
training with the intervention. This study looked at one
method of inducing slow breathing. Other methods could
have similar effects.
The main limitation of this study was the small sample size,
as it would have been better to have had more statistical power,
however despite the small sample size there were still
significant results. We decided to use a single intervention and
comparative group design, as a crossover design may have
confounded the interpretation in the event of there being any
carry over from the intervention; however it would be useful to
redesign the study to include a crossover design effectively.
In conclusion, we found that the use of short duration HRV
biofeedback intervention resulted in improved cognitive
performance. Reaction time was improved and more
consistent, and there was a reduction in mistakes made in
counting squares during a modified Stroop task. We found
that subjects felt more relaxed but were also less sleepy and
better able to focus during short duration HRV biofeedback
intervention. Future research exploring the effects of
extended use of a short duration HRV biofeedback
intervention, as well as studies on women is warranted.
ACKNOWLEDGEMENTS
The authors thank Helicor for providing funding for this
study. They receive no remuneration for merchandise sold
and therefore have no vested interest in the outcome of the
study.
REFERENCES
Benson, H. (1993). The relaxation response. In Goleman, D. & Gurin J.
(Eds.), Mind/body medicine: How to use your mind for better health.
Yonkers, NY: Consumer Reports Books.
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
The effect of heart rate variability biofeedback
Bernardi, L., Wdowczyk-Szulc, J., Valenti, C., Castoldi, S., Passino, C.,
Spadacini, G., et al. (2000). Effects of controlled breathing, mental
activity and mental stress with or without verbalization on heart rate
variability. Journal of the American College of Cardiology,35, 1462–
1469.
Berntson, G. G., Bigger, J. T., Jr, Eckberg, D. L., Grossman, P., Kaufmann,
P. G., Malik, M., et al. (1997). Heart rate variability: Origins, methods,
and interpretive caveats. Psychophysiology,34, 623–648.
Delaney, J. P., & Brodie, D. A. (2000). Effects of short-term psychological
stress on the time and frequency domains of heart-rate variability.
Perceptual and Motor Skills,91, 515–524.
Dishman, R. K., Nakamura, Y., Garcia, M. E., Thompson, R. W., Dunn,
A. L., & Blair, S. N. (2000). Heart rate variability, trait anxiety, and
perceived stress among physically fit men and women. International
Journal of Psychophysiology,37, 121–133.
Egner, T., Zech, T. F., & Gruzelier, J. H. (2004). The effects of neurofeed-
back training on the spectral topography of the electroencephalogram.
Clinical Neurophysiology,115, 2452–2460.
Elzinga, B. M., Bakker, A., & Bremner, J. D. (2005). Stress-induced cortisol
elevations are associated with impaired delayed, but not immediate recall.
Psychiatry Research,134, 211–223.
Forstmann, B. U., Dutilth, G., Brown, S., Neumann, J., von Cramon, D. Y.,
Ridderinkhof, K. R., et al. (2008). Straitum and pre-SMA facilitate
decision-making under time pressure. The Proceedings of the National
Academy of Sciences of the United States of America,105, 17538–17542.
Franzon, M., & Hugdahl, K. (1987). Effects of speed vs. accuracy in vocal
reaction time to visual half-field presentations of incongruent (Stroop)
color-words. Cortex,23, 615–629.
Giardino, N. D., Chan, L., & Borson, S. (2004). Combined heart rate
variability and pulse oximetry biofeedback for chronic obstructive pul-
monary disease: Preliminary findings. Applied Psychophysiology and
Biofeedback,29, 121–133.
Grasso, R., Schena, F., Gulli, G., & Cevese, A. (1997). Does low-frequency
variability of heart period reflect a specific parasympathetic mechanism?
Journal of the Autonomic Nervous System,63, 30–38.
Gunkelman, J. D. & Johnstone, J. (2005). Neurofeedback and the brain.
Journal of Adult Development,12, 93–98.
Hajcak, G., McDonald, N., & Simons, R. F. (2003). Anxiety and error-
related brain activity. Biological Psychology,64, 77–90.
Hansen, A. L., Johnsen, B. H., Sollers, J. J. III, Stenvik, K., & Thayer, J. F.
(2004). Heart rate variability and its relation to prefrontal cognitive
function: The effects of training and detraining. European Journal of
Applied Physiology,93, 263–272.
Hansen, A., Johnsen, B., & Thayer, J. (2003). Vagal influence on working
memory and sustained attention. International Journal of Psychophysiol-
ogy,48, 263–274.
Hassett, A. L., Radvanski, D. C., Vaschillo, E. G., Vaschillo, B., Sigal, L. H.,
Karavidas, M. K., et al. (2007). A pilot study of the efficacy of heart rate
variability (HRV) biofeedback in patients with fibromyalgia. Applied
Psychophysiology and Biofeedback,32, 1–10.
Heilman, K. J., Handelman, M., Lewis, G., & Porges, S. W. (2008).
Accuracy of the StressEraser in the detection of cardiac rhythms. Applied
Psychophysiology and Biofeedback,33, 83–89.
Hjortskov, N., Rissen, D., Blangsted, A. K., Fallentin, N., Lundberg, U., &
Sogaard, K. (2004). The effect of mental stress on heart rate variability
and blood pressure during computer work. European Journal of Applied
Physiology,92, 84–89.
Kirschbaum, C., Wolf, O. T., May, M., Wippich, W., & Hellhammer, D. H.
(1996). Stress- and treatment-induced elevations of cortisol levels associ-
ated with impaired declarative memory in healthy adults. Life Sciences,
58, 1475–1483.
Lehrer, P. M., Vaschillo, E., & Vaschillo, B. (2000). Resonant frequency
biofeedback training to increase cardiac variability: Rationale and manual
for training. Applied Psychophysiology and Biofeedback,25, 177–191.
Lehrer, P. M., Vaschillo, E., Vaschillo, B., Lu, S. E., Eckberg, D. L.,
Edelberg, R., et al. (2003). Heart rate variability biofeedback increases
baroreflex gain and peak expiratory flow. Psychosomatic Medicine,65,
796–805.
Lucini, D., Di Fede, G., Parati, G., & Pagani, M. (2005). Impact of chronic
psychosocial stress on autonomic cardiovascular regulation in otherwise
healthy subjects. Hypertension,46, 1201–1206.
Lupien, S. J., Gaudreau, S., Tchiteya, B. M., Maheu, F., Sharma, S., Nair,
N. P., et al. (1997). Stress-induced declarative memory impairment in
healthy elderly subjects: Relationship to cortisol reactivity. Journal of
Clinical Endocrinology and Metabolism,82, 2070–2075.
Madden, K., & Savard, G. K. (1995). Effects of mental state on heart rate and
blood pressure variability in men and women. Clinical Physiology,15,
557–569.
Mayes, L. C., Molfese, D. L., Key, A. P. F., & Hunter, N. C. (2005). Event-
related potentials in cocaine-exposed children during a Stroop task.
Neurotoxicology and Teratology,27, 797–813.
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., &
Wager, T. D. (2000). The unity and diversity of executive functions and
their contributions to complex ‘Frontal Lobe’ tasks: A latent variable
analysis. Cognitive Psychology,41, 49–100.
Moss, D. (2004). Heart rate variability (HRV) biofeedback. Psychophysiol-
ogy Today,1, 4–11.
Muench, F. (2008). The portable stress eraser heart rate variability biofeed-
back device: Background and research. Biofeedback,36, 35–39.
Nolan, R. P., Kamath, M. V., Floras, J. S., Stanley, J., Pang, C., Picton, P.,
et al. (2005). Heart rate variability biofeedback as a behavioral neuro-
cardiac intervention to enhance vagal heart rate control. American Heart
Journal,149, 1137.e1–1137.e7.
O’Brien, E., Mee, F., Atkins, N., & Thomas, M. (1996). Evaluation of three
devices for self-measurement of blood pressure according to the revised
British Hypertension Society Protocol: The Omron HEM-705CP, Philips
HP5332, and Nissei DS-175. Blood Pressure Monitoring,1, 55–61.
Oei, N. Y., Everaerd, W. T., Elzinga, B. M., van Well, S., & Bermond, B.
(2006). Psychosocial stress impairs working memory at high loads:
An association with cortisol levels and memory retrieval. Stress,9,
133–141.
Ohman, L., Nordin, S., Bergdahl, J., Slunga, B. L., & Stigsdotter, N. A.
(2007). Cognitive function in outpatients with perceived chronic stress.
Scandinavian Journal of Work, Environment and Health,33, 223–232.
Orem, D. M., Petrac, D. C., & Bedwell, J. S. (2008). Chronic self-perceived
stress and set-shifting performance in undergraduate students. Stress,11,
73–78.
Payne, J. D., Nadel, L., Allen, J. J., Thomas, K. G., & Jacobs, W. J. (2002).
The effects of experimentally induced stress on false recognition. Mem-
ory,10, 1–6.
Rauch, H. G., John, L., St Clair Gibson, A., Noakes, T., & Vaughan, C.
(2005). Validation study of EEG responses during a modified Stroop
colour word test. Proceedings of the Physiology Society of Southern
Africa.
Reiner, R. (2008). Integrating a portable biofeedback device into clinical
practice for patients with anxiety disorders: Results of a pilot study.
Applied Psychophysiology and Biofeedback,33, 55–61.
Seeman, T. E., McEwen, B. S., Singer, B. H., Albert, M. S., & Rowe, J. W.
(1997). Increase in urinary cortisol excretion and memory declines:
MacArthur studies of successful aging. Journal of Clinical Endocrin-
ology and Metabolism,82, 2458–2465.
Stark, R., Wolf, O. T., Tabbert, K., Kagerer, S., Zimmermann, M., Kirsch, P.,
et al. (2006). Influence of the stress hormone cortisol on fear conditioning
in humans: Evidence for sex differences in the response of the prefrontal
cortex. Neuroimage,32, 1290–1298.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions.
Journal of Experimental Psychology: Human Perception and Perform-
ance,18, 643–662.
Task Force of the European Society of Cardiology and the North American
Society of Pacing and Electrophysiology. (1996). Heart rate variability.
Standards of measurement, physiological interpretation, and clinical use.
European Heart Journal,17, 354–381.
Wolf, O. T., Schommer, N. C., Hellhammer, D. H., McEwen, B. S., &
Kirschbaum, C. (2001). The relationship between stress induced cortisol
levels and memory differs between men and women. Psychoneuroendo-
crinology,26, 711–720.
Yildiz, M., & Ider, Y. Z. (2006). Model based and experimental investigation
of respiratory effect on the HRV power spectrum. Physiological Measure-
ment,27, 973–988.
Zorawski, M., Blanding, N. Q., Kuhn, C. M., & LaBar, K. S. (2006). Effects
of stress and sex on acquisition and consolidation of human fear con-
ditioning. Learning and Memory,13, 441–450.
Copyright #2010 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. (2010)
G. E. Prinsloo et al.