No Effects of Successful Bidirectional SMR Feedback Training on Objective and Subjective Sleep in Healthy Subjects

Abstract

There is a growing interest in the application of psychophysiological signals in more applied settings. Unidirectional sensory motor rhythm-training (SMR) has demonstrated consistent effects on sleep. In this study the main aim was to analyze to what extent participants could gain voluntary control over sleep-related parameters and secondarily to assess possible influences of this training on sleep metrics. Bidirectional training of SMR as well as heart rate variability (HRV) was used to assess the feasibility of training these parameters as possible brain computer interfaces (BCI) signals, and assess effects normally associated with unidirectional SMR training such as the influence on objective and subjective sleep parameters. Participants (n = 26) received between 11 and 21 training sessions during 7 weeks in which they received feedback on their personalized threshold for either SMR or HRV activity, for both up- and down regulation. During a pre- and post-test a sleep log was kept and participants used a wrist actigraph. Participants were asked to take an afternoon nap on the first day at the testing facility. During napping, sleep spindles were assessed as well as self-reported sleep measures of the nap. Although the training demonstrated successful learning to increase and decrease SMR and HRV activity, no effects were found of bidirectional training on sleep spindles, actigraphy, sleep diaries, and self-reported sleep quality. As such it is concluded that bidirectional SMR and HRV training can be safely used as a BCI and participants were able to improve their control over physiological signals with bidirectional training, whereas the application of bidirectional SMR and HRV training did not lead to significant changes of sleep quality in this healthy population.

Full text

Vol.:(0123456789)
1 3
Appl Psychophysiol Biofeedback
DOI 10.1007/s10484-017-9384-y
No Effects ofSuccessful Bidirectional SMR Feedback Training
onObjective andSubjective Sleep inHealthy Subjects
OlafBinsch1· EllenS.Wilschut1· MartijnArns2,5· CharelleBottenheft1·
PierreJ.L.Valk1· EricH.G.J.M.Vermetten3,4
© Springer Science+Business Media, LLC 2017
either SMR or HRV activity, for both up- and down regula-
tion. During a pre- and post-test a sleep log was kept and
participants used a wrist actigraph. Participants were asked
to take an afternoon nap on the first day at the testing facility.
During napping, sleep spindles were assessed as well as self-
reported sleep measures of the nap. Although the training
demonstrated successful learning to increase and decrease
SMR and HRV activity, no effects were found of bidirec-
tional training on sleep spindles, actigraphy, sleep diaries,
and self-reported sleep quality. As such it is concluded that
bidirectional SMR and HRV training can be safely used as a
BCI and participants were able to improve their control over
physiological signals with bidirectional training, whereas the
application of bidirectional SMR and HRV training did not
lead to significant changes of sleep quality in this healthy
population.
Keywords Sleep· Military· BCI· Biofeedback·
Neurofeedback· Training· Heart rate variability
Introduction
Neurofeedback (NFB) has been in use since the 1960s in
clinical settings for revalidation purposes, more recently,
due to technical improvements and new computer technol-
ogy, it is geared towards non-clinical domains for preven-
tion purposes or improvements of performances in applied
domains (Hammond 2007). For that reason, there is a
need for more solid theoretical and methodological sound
experimentation and scientific evidence to test and validate
the NFB applications as reliable methods in order to help
advance this field (Gruzelier etal. 2006; Gruzelier 2014).
Therefore, the underlying motivation to conduct the cur-
rent study was to test the most beneficial NFB method on
Abstract There is a growing interest in the application of
psychophysiological signals in more applied settings. Unidi-
rectional sensory motor rhythm-training (SMR) has demon-
strated consistent effects on sleep. In this study the main aim
was to analyze to what extent participants could gain volun-
tary control over sleep-related parameters and secondarily to
assess possible influences of this training on sleep metrics.
Bidirectional training of SMR as well as heart rate vari-
ability (HRV) was used to assess the feasibility of training
these parameters as possible brain computer interfaces (BCI)
signals, and assess effects normally associated with unidirec-
tional SMR training such as the influence on objective and
subjective sleep parameters. Participants (n = 26) received
between 11 and 21 training sessions during 7 weeks in which
they received feedback on their personalized threshold for
* Olaf Binsch
olaf.binsch@tno.nl
Ellen S. Wilschut
ellen.wilschut@tno.nl
Martijn Arns
martijn@brainclinics.com
Pierre J. L. Valk
pierre.valk@tno.nl
Eric H. G. J. M. Vermetten
e.vermetten@lumc.nl
1 Netherlands Organisation forApplied Scientific Research
(TNO), Kampweg 5, 3769ZGSoesterberg, TheNetherlands
2 Research Institute Brainclinics, Nijmegen, TheNetherlands
3 Ministry ofDefense, Central Military Hospital, Utrecht,
TheNetherlands
4 University ofLeiden, Leiden, TheNetherlands
5 Department ofExperimental Psychology, Utrecht University,
Utrecht, TheNetherlands

Appl Psychophysiol Biofeedback
1 3
a sufficiently measurable variable such as sleep quality, in
order to develop an adequate self-paced intervention in the
domain of high risk professions. Military, police officers and
fire fighters are often forced to operate in high risk environ-
ments and are, therefore, exposed to a significant amount of
stressors during their professional career (e.g., Binsch etal.
2015). Insufficient quality and duration of sleep during mis-
sions and shift work can lead to sleep disorders and insomnia
(e.g., Peterson etal. 2008), which is associated with negative
daytime performance (e.g., Morin etal. 2006). Therefore,
it is important to decrease or reduce insomnia symptoms
personnel in high risk professions.
One of the most promising approaches that may increase
sleep quality are NFB applications involving the training of
the sensory motor rhythm (SMR) that have been studied in
multiple settings. Clinically, unidirectional SMR enhance-
ment training has been reported to have effects in ADHD
(Monastra etal. 2005; Arns etal. 2009; Lofthouse etal.
2012) and epilepsy (Sterman 2000; Tan etal. 2009). SMR
feedback training has also improved cognitive performance
in healthy subjects; Egner and Gruzelier (2001) showed a
positive effect of SMR training on attention using a continu-
ous performance task. Furthermore, Vernon etal. (2003)
showed that SMR feedback training is associated with a
slight improvement of working memory and attentional pro-
cessing. A number of studies have shown that SMR training
also transfers to the sleeping state, this was first reported by
Sterman etal. (1969) using cats as subjects. In particular,
SMR frequencies overlap with dominant oscillations in stage
two sleep (i.e. stage of light sleep), called sleep spindles,
which have the same topography and frequency as the SMR
rhythm trained during wakefulness. Furthermore, it has
been demonstrated that SMR enhancement neurofeedback
induces and increases the occurrence of sleep spindles (Ster-
man etal. 1969; Hoedlmoser etal. 2008) as well as improves
sleep parameters such as decreased sleep onset latency and
increased sleep duration (Cortoos etal. 2010; Hoedlmoser
etal. 2008; Arns etal. 2014). SMR training was applied
to help patients with insomnia (Cortoos etal. 2010; Hoe-
dlmoser etal. 2008). Also, in a healthy population, results
showed it was possible to gain control over the SMR fre-
quency with positive effects on quality of sleep; Hoedlmoser
etal. (2008) demonstrated that after 10 SMR training ses-
sions their participants showed an improvement in declara-
tive learning, an increased number of sleep spindles during
stage 2 sleep and a reduced sleep onset latency. Furthermore,
it was recently hypothesized that activating and deactivat-
ing the reticular–thalamocortical–cortical sleep spindle
circuitry would increase the synaptic strengths within that
network (Arns etal. 2014, 2015). As a result the probability
of future activation of this network would increase (Sterman
and Egner 2006; Arns and Kenemans 2013), which explains
the increased sleep density during sleep.
A second application that involves feedback of physi-
ological activity is in the field of brain–computer interfac-
ing (BCI). BCI is well-known as a technique based on the
interaction between the brain and a device (i.e. computers).
More specifically, the BCI technique uses electrophysiologi-
cal signals extracted from the brain that enables the user to
direct multidirectional and multidimensional activity, such
as control of a cursor, a computerized language or speech
program, and/or an motorized wheelchair without muscu-
lar activity or overt speech. Such control can be beneficial
for patients with severe motor disabilities. For example,
Birbaumer etal. (1999, 2007) combined the BCI technique
with SMR–NFB and developed and validated a spelling
device [thought translation device (TTD)] for patients suf-
fering from amyotrophic lateral sclerosis (ALS) to improve
their ability to interact with their social environment. The
BCI–TTD bidirectional interface only required to select a
certain topic from a list and to learn how to up- and down-
regulate a courser in order to perform (i.e., select and edit)
the communication that matches the corresponding intent
of the individual through the route of electroencephalogra-
phy (EEG). Other studies demonstrated that patients with
spinal cord injuries and healthy users were able to pro-
vide point-to-point movements in motorized wheelchairs
(Wolpaw and McFarland 2004), or participants learned to
control helicopter flights in 3-dimensional space (Royer
etal. 2010) by applying BCI multidirectional trainings to
increase and decrease the amplitude of SMR (see also Yuan
and He 2014 for an overview). As such, BCI bidirectional
training methods are beneficial as the user get enhanced to
control alternately the de- and increase of brain signal output
and simultaneously also inhibiting brain activity (Ancoli and
Kamiya 1978; Vernon etal. 2009). A bidirectional training
which incorporates both enhancement and suppression may
also enable a user to obtain a greater degree of voluntary
control in less time.
Besides the positive effects of SMR–NFB, research has
shown that increasing heart rate variability (HRV) using
biofeedback (BFB) can have a positive effect on mental and
psychological health, including a decrease in depression,
anxiety, post-traumatic stress disorder (PTSS), medically
unexplained syndromes, high blood pressure and an increase
in lung function (Lehrer etal. 2000). Furthermore, Hansen
etal. (2009) found a positive correlation of HRV perfor-
mance on cognitive tasks and self-regulatory control with
positive implications for the military domain. In summary,
HRV–BFB has been applied and examined for many differ-
ent clinical uses, including in patients with major depres-
sive disorder (Hassett etal. 2007), hypertension (Del Pozo
etal. 2004) and PTSD (Zucker etal. 2009), as well as in
healthy people (Lehrer etal. 2003). However, the results of
these studies are mixed. In patients with depression, clini-
cal symptoms decreased in the BFB group (Hassett etal.

Appl Psychophysiol Biofeedback
1 3
2007), whereas BFB for people with PTSD showed similar
results compared to other, less intense relaxation techniques
(Zucker etal. 2009).
As there is an indication that SMR neurofeedback has
effects on objective and subjective sleep parameters in both
healthy populations (Hoedlmoser etal. 2008) as well as in
clinical populations (Cortoos etal. 2010; Arns etal. 2014),
in this study we aimed to investigate BCI bidirectional
NFB–SMR training (i) to investigate how well volunteers
could control this activity, and (ii) to quantify the effects on
sleep to further explore if clinical effects are also obtained
with bidirectional as opposed to unidirectional SMR train-
ing. As a control variable BCI HRV–BFB training was
used, as it was expected that volunteers could also learn to
gain bidirectional control over HRV, and this training was
expected to have more non-specific effects e.g. increase par-
asympathetic activity (Lehrer etal. 2000), thereby increas-
ing the quality of sleep through relaxation. Concerning the
bidirectional BCI training we hypothesized that this could
be an efficient method that result in increased sleep spindle
density and therefore in improved sleep in a healthy military
population.
Method
Participants
A total of 62 participants, all military working at the Dutch
Ministry of Defense, were invited for a screening consist-
ing of the Holland Sleep Disorders Questionnaire (HSDQ;
Kerkhof etal. 2013). 20 participants were excluded because
they had a score above 2 on the HSDQ, suggesting a possible
sleep disorder (mean score = 1.4, SD = 0.3; Kerkhof, 2013).
The remaining 42 participants were matched and assigned to
either NFB or BFB group. The assignment to the groups was
based on gender, age, and HSDQ score. Of the remaining 42
participants, 16 participants were further excluded from the
analysis because they were not able to attend the predefined
number of 10 required training sessions (see Hoedlmoser
etal. 2008), due to their preparation for a military mission
in Africa. In total 26 participants (8 female) remained, aged
between 21 and 52years (M = 32.46, SD = 8.90). Of those 26
participants, 12 belonged to the HRV–BFB group and 14 to
the SMR–NFB group. The study’s protocol was approved by
the Ethics Committee (TCPE) of the Dutch Research Insti-
tute for Applied Sciences (TNO).
Design
The treatment consisted of 7 weeks of feedback training,
with one to three training sessions per week. To assess the
effects of this training on sleep quality, pre- and post-tests
were conducted including EEG measurements during an
afternoon nap from 12:30 to 14:30 p.m. and several ques-
tionnaires out to asses subjectively perceived quality of
sleep. In addition to these assessments, participants kept a
sleep journal and wore an Actiwatch (Actiwatch Sleep &
Activity Software V 5.32, Cambridge Neurotechnology) for
the duration of 1 week, after the pre- and post-test.
Apparatus andMaterials
The training sessions were conducted using Brainquiry PET
EEG 4.0 (four channels) NFB equipment (Brainquiry B.V).
The software was programmed in BioExplorer (CyberEvo-
lution, Inc.).1 Disposable electrodes were placed on EEG
locations C3 and C4, referenced behind the left ear, on the
mastoid. In addition, ECG electrodes were placed on the
sternum and left clavicle (Ruehland etal. 2011). The sam-
pling rate was 200Hz and signals were low pass filtered at
1Hz and high pass filtered at 41Hz. In the SMR group, the
power in the SMR frequency band (12–15Hz) was calcu-
lated with Butterworth filter (Bianchi and Sorrentino 2007).
For the HRV group, ECG signals were low pass filtered at
5Hz and high pass filtered at 50Hz to calculate HRV. Dur-
ing training, the thresholds were adjusted (SMR: steps of
0.5µV and HRV: steps of 2.5V). Both, SMR and HRV
feedback was displayed on a 19″ computer screen (LCD
Screen, 1920 × 1200 WUXGA Matte Wide, Dell©) to the
participant. The exact processing and real-time artifact han-
dling is described in detail in Kleinnijenhuis etal. (2008).
During the pre- and post-tests to assess sleep parameters
the PET system was used again to measure only EEG with
four channels, placed on locations C3, C4, F4, O2 which
were referenced to the left mastoid (these channels were
chosen in agreement with the AASM guidelines for poly-
somnography). Subjective sleepiness was assessed with the
Pittsburgh Sleep Quality Index (PSQI; Buysse etal. 1989)
and the Stanford Sleepiness Scale (SSS; Hoddes etal. 1973).
The Groninger Sleep Quality Scale (GSQS; Mulder-Hajon-
ides van der Meulen etal. 1980) was applied to measure the
subjective perceived sleep after the pre- and post-test nap-
ping. Following the week of the pre- and post-test, a sleep
diary was kept which included the GSQS and other items
providing an indication of the total sleep time (TST), sleep
onset latency (SOL), wake after sleep onset (WS), time in
bed (TB) and sleep efficiency (TST/TB). Actigraphy was
used to objectively assess sleep characteristics, total sleep
time, activity, and fragmentation (Ancoli-Israel etal. 2003).
1 For a more detailed explanation of the bidirectional screens, the
exact processing and real-time artefact handling that were applied
during the feedback training sessions see Kleinnijenhuis etal. 2008;
Spronk etal. 2010.

Appl Psychophysiol Biofeedback
1 3
Procedure
During the pre-test, participants received the briefing and
signed the informed consent. The SSS and PSQI were rated.
EEG measurements were collected using the PET system
during an afternoon nap. After a nap of 120min sleep, again
the SSS and the GSQI were rated. After the pre-test, the
selected participants had to fill-out a sleep diary every day
for 1 week and they wore the Actiwatch. The participants
were instructed to press the button on the Actiwatch every
time they went to bed or went out of bed. The post-test pro-
cedure was similar to the pre-test, and ended with a debrief-
ing of the treatment phase.
During feedback training sessions, the participants of
both groups were trained to attain, and endure in, a certain
range level of SMR or HRV activity, respectively. Each ses-
sion started with the application of electrodes on the scalp
and chest of the participant. The first time, participants
were instructed on the functions of the various elements of
the feedback window and the task requirements. Follow-
ing Kober etal. (2013) who found that a ‘Just do it’ task
instruction is most effective, a general explanation of BFB-
and NFB was given; participants were not provided with a
directed strategy to control the HRV or SMR. Participants
were only told that up-regulation is associated with relaxa-
tion, and down-regulation is associated with effort.
The sessions lasted 1 h, including 25min of preparation,
24min of effective training and ca. 11min of removing
the electrodes and debriefing. The timeline of the experi-
ment was scheduled equally for each participant. That is, the
experiment lasted 10 weeks for each participant and started
with the pre-test arranged in the initial 2 weeks. This period
was followed by 7 weeks of training sessions, and ended
with the post-test planned in the last week. Participants who
started the experiment with the pre-test in the first week,
started the training sessions in the following week, and
ended the experiment also in the first days of the post-test
week. The maximum latency time between pre-test and the
initial training session lasted 7 days (including days of the
weekend), and lasted a maximum of 6 days between the last
training session and the post-test.
Feedback Training Task
Each training session consisted of four runs with a duration
of 6min (Fig.1). During a run, 45 trials were presented;
trials for the up-regulation and the down-regulation were
mixed pseudo-randomly (Kleinnijenhuis etal. 2008; Spronk
etal. 2010). In summary, (also see Fig.1) a blue bar filled up
slowly either upwards or downwards to indicate the direc-
tion of the trial. The feedback signal of the participants was
shown in real-time on the screen i.e. either SMR or HRV in
the form of a yellow bar on the feedback screen and a trial
lasted 7s in which the participant had to regulate the signal
in the desired direction. The inter-trail interval varied ran-
domly between 1.5 and 3s. The participant had to try and
cross the threshold line that was indicated (red and green) for
at least 300ms to complete the trial successfully. If the trial
was successful, visual feedback (smiley was shown and per-
centage of successful up- or down regulation was updated)
and auditory feedback (sounds played over headphone) were
provided. If the participant was unable to exceed the thresh-
old line for more than 300ms, the trial was unsuccessful and
the percentage correct was lowered accordingly. The thresh-
olds used for positive feedback were based on the achieved
performance during each run. If, in one training session,
three of the four runs scored 50% or higher, the threshold
was increased one step upwards for the next training ses-
sion. The thresholds were adjusted separately for upward/
downward regulating e.g. a participant who achieved better
scores in relaxation could have a higher threshold for upward
regulation than for downward regulation. However, if a par-
ticipant scored lower than 50% in six of the eight runs (over
two training sessions) the threshold was decreased one step
downward. Apart from these adaptations, the threshold val-
ues remained unchanged for the following training session.
In order to keep participants motivated, every 2 weeks a
result list was published with top 10 performers. The point
system was based on the combination of the number of runs
and the number of levels the participants were able to regu-
late up or down. Participants deserved one point when the
threshold was increased one step after a session. A deduction
of one point occurred when the threshold was decreased one
Fig. 1 Screenshot of the feedback training application as was pre-
sented to the participant. A Smiley and sound indicated a success-
ful trial in which the yellow bar (either SMR or HRV activity) was
beyond the threshold (green/red horizontal lines, 60Hz update rate)
for at least 300ms. The numbers above and below on the screen rep-
resent the percentage of successful trials for up- and down regulation,
respectively. These numbers were updated after every trial. Training
was bidirectional and trials for up- and down-regulation were mixed
randomly. (Color figure online)

Appl Psychophysiol Biofeedback
1 3
step after a session. These points were added up to a score
for all sessions. Because the SMR and HRV training groups
had different threshold values, the top performers of the two
groups were placed alternately on the list.
Analysis
Statistical analyses were performed using Statistical Pack-
age for the Social Sciences (SPSS, IBM® statistic software
for Windows version 22.0.0). Please note that due to the
operational setting of the military population, the number
of completed trainings sessions varied between 11 and 21.
In line with Hoedlmoser etal. (2008), only the datasets of
participants that attended a minimum of 10 training ses-
sion were analyzed. Twenty-six participants were able to
attend the minimum of 10 or more sessions. All data of
these participants were used for the analysis, including the
data of all completed training sessions, the sleep data of
the 2 × 120min during the pre- and post-test, and all sub-
jective data to analyze sleep quality. Kolmogorov–Smirnov
tests were applied to test for normality of the distribution
of the data. For all analyses, the significance level was set
to p < .05.
Training Effect
Prior to the analyses of the training effect a training effect
score was calculated per participant by using a standard
z-transformation procedure. That is, the maximum obtained
training level (α), the baseline level (β) at which the train-
ing started, the number of all completed training sessions
(ω) and the number of steps (σ) that the participant needed
to achieve α. These factors were used to examine the ratio
(i.e. training effect score) between the level that the par-
ticipant have achieved and the level that the participant
could have achieve transferred into percentage (α − β)/
((ω − 1) × σ) × 100%. Consequently, the training effect data
were analyzed by using a paired samples T-test for each
group (SMR–NFB, HRV–BFB) separately on the training
scores obtained during up- (relaxation) and down- (effort)
regulation.
Pre‑ andPost‑test
Sleep spindles were derived from the EEG during the
pre- and post-test naps. They were determined automati-
cally (13–15Hz at C4) in the selected periods of sleep and
verified by visual inspection. More specifically, sleep was
defined as periods were alpha or theta power decreased rela-
tively to the beginning of the nap. For the detection of the
spindles a spindle threshold of 3µV was used and the dura-
tion of the spindle should be between 0.5 and 2s (Piantoni
etal. 2013). From this data the total number of spindles and
spindles per minute were derived. Artefacts were removed
manually from the raw data during a semi-automatic pro-
cess, i.e. after visual inspection for abnormalities (i.e., spin-
dle threshold of 3µV and spindle duration between 0.5 and
2s; see also Piantoni etal. 2013) of the previous automati-
cally detected data through the analyses of relative alpha
(7.5–13.0Hz) and theta power (3.5–7.5Hz). To compare
the sleep spindle/quality data between pre- and post-test
generalized linear model (GLM) repeated measures analy-
sis of variance (ANOVA) for both groups separately were
conducted using a 2 test (pre-, post-) × 2 regulation (up-,
down-) design. Greenhouse-Geisser correction was used in
case of non-sphericity of the data. Pair-wise comparisons
using Bonferroni correction (Kinnear and Gray 2000) were
made to identify specific mean differences when appropri-
ate. Partial eta squared (ηp
2) assessed the explained variance
in the ANOVA models. Subjective ratings were analyzed
using non-parametric techniques (i.e. Wilcoxon Matched-
Pairs Signed-Ranks, Mann–Whitney U) with a within-sub-
ject factor (pre-test, post-test) and a between-subject factor
(SMR–NFB vs. HRV–BFB).
Results
Training Effect
Participants in the HRV–BFB group completed on average
15 sessions (min. = 13, max. = 21, median = 14 sessions).
Figure2 shows the average correct achieved trials during up-
regulation (i.e., relaxation, in %; black bars) and the average
correct achieved trials during down-regulation (i.e., effort,
in %; gray bars) of the HRV–BFB group.2
In addition, the average maximum obtained training level
during up-regulation (relaxation) is presented by the green
line and the average maximum obtained training level dur-
ing down-regulation (effort) is presented by the red line. As
such, Fig.2 indicate that the HRV–BFB group was not able
to achieve correct trails for both up- and down-regulation
2 Figure 2 show all 18 HRV–BFB and Fig. 3 all 21 SMR–NFB
training sessions for the sake of completeness. As stated earlier, due
to the operational setting of the military population, the number of
completed trainings sessions varied between 11 and 21 per partici-
pant. Therefore, starting from training session 11 for the SMR-NFB
and 13 for the HRV-BFB groups the bars and lines in Figs.2 and 3
show average data assessed from a decreasing number of participants.
Note, the number of completed training sessions were not different
between the groups, t(24) = 1.38, p = .181. Next, in both Figs.2 and 3
the average percentage of successful trials and achieved training level
for down-regulation (effort; grey bars and red line, respectively) were
converted into negative numbers to show the results for up- (relaxa-
tion) and down- (effort) regulation for both groups in only two Fig-
ures.

Appl Psychophysiol Biofeedback
1 3
above or below the given threshold of 50%, respectively.
Therefore, participants of this group were also not able
to achieve higher or lower training levels during up- and
down-regulation. The paired samples T-test on the training
effect scores of the HRV–BFB group confirmed no achieved
training performance as no significant difference was found
between up-and down regulation, t (11) = − .479, p = .641.
Participants of the SMR–NFB group completed on average
17 training sessions (min. = 11, max. = 21, median = 18 ses-
sions; Fig.3). In line with Figs.2, 3 also shows the average
correct achieved trials during up-regulation (i.e., relaxation,
in %; black bars) and the average correct achieved trials
during down-regulation (i.e., effort, in %; gray bars) of the
SMR–NFB group. Also Fig.3 shows the average maximum
obtained training level during up-regulation (relaxation;
green line) and the average maximum obtained training level
during down-regulation (effort; red line). Thus, Fig.3 shows
that the SMR–NFB group obtained on average continuously
correct trails above the 50% threshold during up-regulation
(relaxation); hence, participants achieved also higher train-
ing levels.
During down-regulation (effort) the SMR–NFB group
obtained on average successful trails above the 50% thresh-
old during the first 12 training sessions followed by unvaried
percentages of successful trials around the 50% threshold
and a steady training level. The difference between up- and
down regulation of the SMR–NFB group was significant as
revealed by the paired samples T-test on the training effect
scores, t (13) = 8.382, p < .001. The mean difference between
the training effect scores indicate that participants in the
SMR–NFB group were able to up-regulate (relax) much
more as their training scores were much higher, M = 81.21,
SD = 20.10, compared to when they tried to down-regulate,
M = 38.93, SD = 18.19 (see also Table1).
Sleep Spindles
The total nap duration that was analyzed after artifact rejec-
tion, ranged between 52 and 66min and the total number
of sleep spindles ranged between 718 and 1076. During the
pre-test nap the mean number of spindles per minute was
16,9 (SD = 2.1) for SMR and 22.3 (SD = 3.1) for HRV. After
Fig. 2 Training performance
in percentage successful trials
per training session (1–18) and
corresponding achieved training
levels in steps for the HRV–
BFB group
Fig. 3 Training performance
in percentage successful trials
per training session (1–21) and
corresponding achieved training
levels in steps for the SMR–
NFB group

Appl Psychophysiol Biofeedback
1 3
the training session the mean number of spindles was 17.4
(SD = 2.6) for SRM and 22.8 (SD = 3.9) for HRV. The 2 test
(pre-, post-) × 2 group (SMR–NFB, HRV–BFB) repeated
measures ANOVA on the number of sleep spindles per min-
ute revealed no significant main effect for test, F(1, 24) = 0.2,
p = .90, ηp
2 = 0.06, no significant main effect for group, F(1,
24) = 0.8, p = .38, ηp
2 = 0.13, nor an interaction between test
and group (F(1, 24) = 0.9, p > .36, ηp
2 = 0.14). In addition,
the same ANOVA design on the total number of sleep spin-
dles revealed also no significant main effect for test, F(1,
24) = 0.6, p = .46, ηp
2 = .11 no significant effect for group,
F(1, 24) = 2.4, p = .14, ηp
2 = 0.24 and also no interaction
between test and group (F(1, 24) = 0.5, p = .47, ηp
2 = .10).
Sleep Diaries andActigraphy
Participants reported that they on average went to bed at
23:35hand woke up at 07:16h during the pre-test week.
During the post-test week participants went to bed on aver-
age at 23:25h and woke up at 07:13h. Mean sleep onset
latency was 13min during the pre-test and 12min dur-
ing the post-test period. During the pre-test period par-
ticipants had a sleep efficiency between 78.63 and 96.48%
and during the post-test between 74.23 and 97.89%. Three
2 test (pre-, post-) × 2 group (SMR–NFB, HRV–BFB)
repeated measures ANOVA on sleep onset latencies, time
of going to bed and wake up time revealed no significant
main effects for test, Fs(1, 19) < .72, ps > .41, no signifi-
cant main effects for group, Fs(1, 24) < 1.89, ps > .19, and
no interactions between test and group (Fs(1, 24) < 1.13,
ps > .26). If anything, the two 2 test (pre-, post-) × 2 group
(SMR–NFB, HRV–BFB) repeated measures ANOVA on
‘minutes awake during night’ and ‘times awake during
night’ both revealed significant main effects for group,
F(1, 24) = 10.08, p = .005; ηp
2 = .35, and F(1, 24) = 4.72,
p = .04; ηp
2 = 0.20, respectively. The effects for test
were not significant (Fs(1, 24) < 0.5, ps > .52), and no
interactions between test and group were found (Fs(1,
24) < 0.73, ps > .40). Post hoc pair-wise comparisons on
the significant group effects revealed that participants in
the SMR–NFB group reported that they were on aver-
age 8min longer awake when they woke up (p = .006),
and on average 0.5 times more awake during the night
(p = .043) compared to the HRV–BFB group. Also, the
ANOVA’s conducted on actigraphy parameters concern-
ing actual sleep time, immobility percentage and moving
minutes revealed no significant main effects for test and
group (Fs(1, 24) < 1.2, ps > .43), nor interactions (Fs(1,
24) < 0.71, ps > .34). However, a Pearson product-moment
correlation conducted in order to assess the relationship
between the sleep diaries and actigraphy revealed a posi-
tive correlation for the pre-test (r = .55, p = .027) and post-
test (r = .63, p = .009).
PSQI
During the pre-test, the PSQI median for participants in
the SRM–NFB group was 4.0 (range 2.0–8.0) and for the
participants in the HRV–BFB group the median was also
4.0 (range 1.0–8.0). In comparison with the post-test, the
PSQI median for participants in NFB group was 3.5 (range
1.0–8.0) and for the participant in the BFB group the median
was 3.0 (range 2.0–5.0). Non-parametric tests showed no
significant effects on the PSQI from pre- to post-test and no
significant effects between groups.
GSQS
The analysis showed no statistical significant differences of
the subjective quality of sleep from pre- to post-test for the
SMR group. Significant improved subjective quality of sleep
ratings were found during the post-test compared to the pre-
test for the BFB group (N = 12, z = −2.395; p = .017). There
were no significant differences between the groups.
SSS Before andAfter Nap
Significant higher sleepiness scores were found after the
nap in comparison with sleepiness scores before the nap,
only during the post-test for both the SMR group (N = 14,
z = −3.357; p = .001) and BFB group (N = 12, z = −2.489;
p = .013). The analysis showed no significant group differ-
ences in sleepiness levels for both the pre-test and post-test.
Table 1 Mean training effect scores in percentage with standard deviations (SD) of successful training for both up- (relaxation) and down-
(effort) regulation during the training sessions and the corresponding mean training level in steps for both groups
Group Up-regulation (relaxation) Down-regulation (effort)
Training level Training effect Training level Training effect
Mean SD Mean SD Mean SD Mean SD
SMR–NFB 7.39 steps 2.00 81.21% 20.10 3.86 steps .80 38.93% 18.16
HRV–BFB 9.17 steps 2.89 11.84% 8.15 9.79 steps 4.58 13.09% 11.83

Appl Psychophysiol Biofeedback
1 3
Discussion
In this study, we investigated the feasibility of bidirectional
SMR neurofeedback and HRV biofeedback training applica-
tion and investigated (i) if participants were able to gain con-
trol over these parameters, and (ii) to investigate the effect
of bidirectional training as opposed to unidirectional SMR
training on sleep. Participants were able to learn to control
their signals over the course of 10–21 training sessions. The
neurofeedback (NFB) group showed that they were better
able to up-regulate their SMR signal, than to down regulate
their signal. In the biofeedback (BFB) group, this unidirec-
tional preference effect was not found. For the BFB group,
it was more difficult to achieve a higher threshold compared
to the NFB group, it also seemed that the participants in
the BFB group reached a ceiling effect earlier in the train-
ing. This might be explained by the differences in threshold
step sizes, which, due to their nature of the physiological
signals, cannot be made identical. Another explanation for
the differences found in the learning curves was discussed by
Vernon etal. (2009). They stated that there are natural limits
to the increase and decrease of heart and brain activity in a
certain frequency, and that it is unlikely that such activity
can be increased ad infinitum. As an example, they relate to
evidence that has indicated that alpha power NFB training
cannot enhance alpha beyond that level seen at rest with eyes
closed. Overall, all participants showed an increase in their
scores on the feedback task showing that they are all able
(to some degree) to learn to consciously control their physi-
ological signals. This study also showed that it was feasible
to execute the training in a military setting.
The sleep spindles derived from the EEG during pre- and
post-training naps did not show any effect of the bidirec-
tional training on sleep quality nor on sleep spindle density
for both groups. Furthermore, analyses of sleep diaries at
home, revealed no post-training differences on self-reported
sleep variables. There was a minor effect when participants
in the NFB group during the post-test 0.5 times woke up
more often during the night therefore being awake for 8min
longer than the BFB group. However, no differences were
found between the pre- and post-test, which also indicates
that participants were not awake more often and/or longer
due to training effects.
An improvement on subjective quality of sleep ratings
was found on the GSQS after BFB (pre Md = 1.8 and post
Md = 1.5). Under normal conditions—an unrestricted and
undisturbed night’s sleep—a score of 1–2 was found, so
these averages stay within a normal range (Meijman etal.
1990). Results of the PSQI questionnaire, showed no sig-
nificant improvement after BFB/NFB training. Before the
training, both groups were, according to the PSQI, good
sleepers. This may explain the lack of effect of the feedback
training on sleep quality. For the post-training nap, higher
sleepiness scores were found for both groups after the nap
in comparison with sleepiness scores before the nap. This
effect could be contributed to sleep inertia.
The present study has various strengths and limitations.
The automated procedure for calculating the sleep spindle
density could have limitations, since sleep spindles were not
scored by a certified polysomnographer and the number of
sleep spindle reported seem to be relatively high. On the
other hand, this method has been used and published before,
and it was identical for both groups. If expected system-
atic changes had occurred, they would become prevalent
because of the within-subject comparison. Concerning the
statistical power of the results, the sample size is similar in
a comparable study (Cortoos etal. 2010), and even larger
compared to another study with a similar design (Hoedl-
moser etal. 2008). A power analyses was conducted on the
effect size derived from Hoedlmoser study (2008). These
analyses revealed a minimal sample size of 10 participants
per group for the current experiment. Although, the current
study had a higher sample size, the selected group of healthy
participants were not sensitive for the treatment because
they were good sleepers, indicated by low effect sizes of
the sleep spindle data. Furthermore, the current study used
on average 15 and 17 training sessions, while a number of
comparable unidirectional studies only used ten trainings
sessions to indicate effects of NFB (Hoedlmoser etal. 2008;
Gruzelier etal. 2006; Gruzelier 2014; Schabus etal. 2014).
This effort was specifically taken to ensure the bidirectional
feedback training was long enough to be effective, in this
study the bidirectional training was a new aspect. This could
be the reason that no effects were observed on sleep. How-
ever, in earlier studies that applied bidirectional protocols,
it was shown that participants obtained a greater degree of
conscious control (Ancoli and Kamiya 1978; Vernon etal.
2009). Another addition within the current study was to use
a ranking top ten list with the purpose to motivate the partic-
ipants. Due to the ranking list, and the announcement of top
performers frequently, in combination with the competitive
spirit among the participants, the participants stayed moti-
vated and involved throughout the training sessions.
For future studies, a next step in the usage of bidirec-
tional training could focus on selection of a more sensi-
tive group within the population e.g. with sleep disorders.
Also, in the NFB group there was a tendency for less drop
outs and more learning effects during the trial, making
it more suitable for a self-paced intervention training.
Participants in the NFB group were more motivated to
train because they were aware of their improvements when
reaching a higher threshold, which in fact they did achieve
more often than the BFB group. Continuous developments
in the hardware of physiological measurements could also
decrease the intrusiveness of this training on daily activi-
ties e.g. dry electrodes would make it more acceptable

Appl Psychophysiol Biofeedback
1 3
to train during working hours. Finally, further research
should aim to make feedback training more context spe-
cific and attractive; the repetitive nature of the training that
is needed to achieve results can also become quite boring
(Tables2, 3).
Summarizing, our findings illustrate that bidirectional
SMR training is feasible i.e. high number of training rep-
etitions and that extensive training of SMR was success-
ful i.e. revealed training effects: participants were able
to actively control their SMR frequency bidirectionally.
However, opposed to studies using unidirectional uptrain-
ing of SMR, bidirectional SMR neurofeedback had no
effects on sleep in this study, demonstrating that bidirec-
tional training of SMR had no clinical effects. Possibly,
the bidirectional training does not result in the same neu-
roplastic changes seen with unidirectional training, due
to the constant changing contingencies (i.e. up- vs. down
required), which can be desirable in BCI applications.
Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict
of interest.
Ethical Approval All procedures performed in studies involving
human participants were in accordance with the ethical standards of the
institutional research committee and with the 1964 Helsinki declaration
and its later amendments or comparable ethical standards.
Informed Consent Informed consent was obtained from all indi-
vidual participants included in the study.
References
Ancoli, S., & Kamiya, J. (1978). Methodological issues in alpha bio-
feedback training. Biofeedback and Self‑Regulation, 3, 155–183.
Ancoli-Israel, S., Cole, R., Alessi, C., Chambers, M., Moorcroft, M.,
& Pollak, C. P. (2003). The role of actigraphy in the study of sleep
and circadian rhythms. Sleep, 26, 342–392.
Arns, M., de Ridder, S., Strehl, U., Breteler, M., & Coenen, A. (2009).
Efficacy of neurofeedback treatment in ADHD: The effects on
inattention, impulsivity and hyperactivity: A meta-analysis. Clini‑
cal EEG and Neuroscience, 40, 180–189.
Arns, M., Feddema, I., & Kenemans, J. L. (2014). Differential effects
of theta/beta and SMR neurofeedback in ADHD on sleep onset
latency. Frontiers in Human Neuroscience, 8, 1019.
Arns, M., Heinrich, H., Ros, T., Rothenberger, A., & Strehl, U. (2015).
Editorial: Neurofeedback in ADHD. Frontiers in Human Neuro‑
science. https://doi.org/10.3389/fnhum.2015.00602.
Arns, M., & Kenemans, J. L. (2013). Neurofeedback in ADHD and
insomnia: Vigilance stabilization through sleep spindles and cir-
cadian networks. Neuroscience and Biobehavioral Reviews, 44,
183–194.
Bianchi, G., & Sorrentino, R. (2007). Electronic filter simulation &
design. New York: McGraw-Hill Professional, pp.17–20.
Binsch, O., Banko, K., Veenstra, B. J., & Valk, P. J. L. (2015). Examin-
ing the relationship between mental, physical and organizational
factors associated with attrition during maritime forces training.
Journal of Strength & Conditioning Research, 29, 187–191.
Table 2 Sleep quality
descriptives Pre-test Post-test
SMR–NFB HRV–BFB SMR–NFB HRV–BFB
Mean SD Mean SD Mean SD Mean SD
Time to bed (hh:mm) 23:27 00:54 23:41 00:34 23:13 00:51 23:35 00:47
Wake up time (hh:mm) 07:14 00:30 07:19 00:44 07:23 00:48 07:04 00:22
Sleep onset latency (min) 16.4 16.8 10.3 6.5 16.2 14.3 9.1 5.6
Sleep efficiency (%) 89.02 5.95 90.48 3.61 87.60 6.92 92.22 3.51
Times awake .86 .65 .47 .42 .89 .91 .25 .49
Minutes awake (min) 7.2 9.3 3.1 4.6 10.4 16.2 2.5 2.9
GSQS sleep diary (score) 1.93 1.24 1.79 1.30 2.19 1.72 1.49 .84
Table 3 Median scores (MdS)
and ranges (minimum and
maximum value) of PSQI,
GSQS, SSS before sleep session
and SSS after sleep session
for each group and time of
measurement
Pre-test Post-test
SMR–NFB HRV–BFB SMR–NFB HRV–BFB
MdS Range MdS Range MdS Range MdS Range
PSQI 4.0 2.0–8.0 4.0 1.0–9.0 3.5 1.0–8.0 4.0 2.0–5.0
GSQS 3.0 0.0–8.0 2.0 0.0–4.0 3.0 0.0–9.0 4.0 1.0–5.0
SSS before sleep session 3.0 1.0–4.0 3.0 1.0–4.0 2.0 1.0–3.0 2.0 1.0–3.0
SSS after sleep session 3.0 2.0–4.0 3.0 2.0–4.0 3.0 1.0–4.0 3.0 2.0–4.0

Appl Psychophysiol Biofeedback
1 3
Birbaumer, N., & Cohen, L. (2007). Brain-computer-interfaces (BCI):
Communication and restoration of movement in paralysis. Journal
of Physiology, 579, 621–636.
Birbaumer, N., Ghanayim, N., Hinterberger, T., Iversen, I., Kotchoubey,
B., Kübler, A., etal. (1999). A spelling device for the paralysed.
Nature, 398, 297–298.
Buysse, D. J., Reynolds, C. F. III, Monk, T. H., Berman, S. R., & Kup-
fer, D. J. (1989). The Pittsburgh Sleep Quality Index: A new instru-
ment for psychiatric practice and research. Psychiatry Research,
28, 193–213. https://doi.org/10.1016/0165-1781(89)90047-4.
Cortoos, A., De Valck, E., Arns, M., Breteler, M. H., & Cluydts, R.
(2010). An exploratory study on the effects of tele-neurofeedback
and tele-biofeedback on objective and subjective sleep in patients
with primary insomnia. Applied Psychophysiology and Biofeed‑
back, 35, 125–134.
Del Pozo, J. M., Gevirtz, R. N., Scher, B., & Guarneria, E. (2004).
Biofeedback treatment increases heart rate variability in patients
with known coronary artery disease. American Heart Journal,
147, G1-G6.
Egner, T., & Gruzelier, J. H. (2001). Learned self-regulation of EEG
frequency components affects attention and event-related brain
potentials in humans. NeuroReport, 12, 4155–4159. https://doi.
org/10.1097/00001756-200112210-00058.
Gruzelier, J. H. (2014). EEG-neurofeedback for optimising per-
formance. In: A review of cognitive and affective outcome in
healthy participants. Neuroscience and Biobehavioral Reviews,
44, 124–141.
Gruzelier, J. H., Egner, T., & Vernon, D. (2006). Validating the efficacy
of neurofeedback for optimising performance. Progress in Brain
Research, 159, 421–431.
Hammond, D. C. (2007). Neurofeedback for the enhancement of ath-
letic performance and physical balance. The Journal of the Ameri‑
can Board of Sport Psychology, 1, 1–9.
Hansen, A. L., Johnsen, B. H., & Thayer, J. F. (2009). Relationship
between heart rate variability and cognitive function during
threat of shock. Anxiety, Stress, & Coping, 22, 77–89. https://doi.
org/10.1080/10615800802272251.
Hassett, A. L., Radvanski, D. C., Vaschillo, E. G., Vaschillo, B., Sigal,
L. H., & Karavidas, M. K. (2007). A pilot study of heart rate
variability (HRV) biofeedback in patients with fibromyalgia.
Applied Psychophysiology and Biofeedback, 32, 1–10. https://
doi.org/10.1007/s10484-006-9028-0.
Hoddes, E., Zarcone, V., Smythe, H., Philips, R., & Dement, W. C.
(1973). Quantification of sleepiness: A new approach. Psychophys‑
iology, 10, 431–436. https://doi.org/10.1111/j.1469-8986.1973.
tb00801.x.
Hoedlmoser, K., Pecherstorfer, T., Gruber, G., Anderer, P., Doppel-
mayr, M., Klimesch, W., & Schabus, M. (2008). Instrumental
conditioning of human sensorimotor rhythm (12–15 hz) and
its impact on sleep as well as declarative learning. Sleep, 31,
1401–1408.
Kerkhof, G. A., Brouwer, A., Rijsman, R. M., Schimsheimer, R. J., &
van Kasteel, V. (2013). Holland sleep disorders questionnaire:
A new sleep disorders questionnaire based on the international
classification of sleep disorders-2. Journal of Sleep Research, 22,
104–107. https://doi.org/10.1111/j.1365-2869.2012.01041.x.
Kinnear, P. R., & Gray, C. D. (2000). SPSS for Windows made simple.
Hove: Psychology Press.
Kleinnijenhuis, M., Arns, M. W., Spronk, D. B., Breteler, M. H. M.,
& Duysens, J. E. J. (2008). Comparison of discrete-trial based
SMR and SCP training and the interrelationship between SCP and
SMR networks: Implications for brain-computer interfaces and
neurofeedback. Journal of Neurotherapy, 11, 19–35.
Kober, S. E., Witte, M., Ninaus, M., Neuper, C., & Wood, G. (2013).
Learning to modulate one’s own brain activity: The effect of
spontaneous mental strategies. Frontiers in Human Neuroscience,
7, 695. doi:https://doi.org/10.3389/fnhum.2013.00695.
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. https://doi.org/10.1023/A:1009554825745.
Lehrer, P. M., Vaschillo, E., Vaschillo, B., Lu, S. E., Eckberg, D. L.,
Edelberg, R., Shih, W. J., Lin, Y., Kuusela, T. A., Tahvanainen,
K. U., & Hamer, R. M. (2003). Heart rate variability biofeedback
increases baroreflex gain and peak expiratory flow. Psychosomatic
Medicine, 65, 796–805.
Lofthouse, N., Arnold, L. E., Hersch, S., Hurt, E., & DeBeus, R.
(2012). A review of neu-rofeedback treatment for pediatric
ADHD. Journal of Attention Disorders, 16, 351–372.
Meijman, T. F., Thunnissen, M. J., & De Vries-Griever, A. G. H.
(1990). The after-effects of a prolonged period of day-sleep on
subjective sleep quality. Work and Stress, 4, 65–70.
Monastra, V. J., Lynn, S., Linden, M., Lubar, J. F., Gruzelier, J., &
LaVaque, T. J. (2005). Electroencephalograpic biofeedback in
the treatment of attention-deficit/hyperactivity disorder. Applied
Psychophysiology & Biofeedback, 30, 95–111.
Morin, C. M., LeBlance, M., Daley, M., Gregoire, J. P., & Merette, C.
(2006). Epidemiology of insomnia: Prevalence, self-help treat-
ments, consultations, and determinants of help-seeking behav-
iors. Sleep Medicine, 7, 123–130. https://doi.org/10.1016/j.
sleep.2005.08.008.
Mulder-Hajonides van der Meulen W. R. E. H., Wijnberg, J. R., Hol-
landers, J. J., DeDiana, I., & Hoofdakker, R. (1980). Measurement
of subjective sleep quality. Fifth European Congress on Sleep
Research, Amsterdam.
Peterson, A. L., Goodie, M. J. L., Satterfield, W. A., & Brim, W. L.
(2008). Sleep disturbance during military deployment. Military
Medicine, 173, 230–235.
Piantoni, G., Poil, S.-S., Linkenkaer-Hansen, K., Verweij, I. M.,
Ramautar, J. R., Van Someren, E. J. W., & Van Der Werf, Y.
D. (2013). Individual differences in white matter diffusion affect
sleep oscillations. Journal of Neuroscience, 33, 227–233.
Royer, A.,S., Doud, A.,J., Rose, M.,L., & He, B. (2010). EEG control
of a virtual helicopter in 3-dimensional space using intelligent
control strategies. IEEE Transactions on Neural Systems and
Rehabilitation Engineering, 18, 581–589.
Ruehland, W. R., O’Donoghue, F. J., Pierce, R. J., Thornton, A. T.,
Singh, P., Copland, J. M., Stevens, B., & Rochford, P. D. (2011).
The 2007 AASM recommendations for EEG electrode placement
in polysomnography: Impact on sleep and cortical arousal scoring.
Sleep, 1, 34, 73–81.
Schabus, M., Heib, D. P. J., Lechinger, J., Griessenberger, H., Kli-
mesch, W., Pawlizki, A., Kunz, A. B., Sterman, B. M., & Hoe-
dlmoser, K. (2014). Enhancing sleep quality and memory in
insomnia using instrumental sensorimotor rhythm conditioning.
Biological Psychology, 95, 126–134.
Spronk, D., Kleinnijenhuis, M., Luijtelaar, G., & Arns, M. (2010).
Discrete-trial SCP and GSR training and the interrelationship
between central and peripheral arousal. Journal of Neurotherapy,
14, 217–228.
Sterman, M. B. (2000). Basic concepts and clinical findings in the
treatment of seizure disorders with EEG operant conditioning.
Clinical EEG (Electroencephalography), 31, 45–55.
Sterman, M. B., & Egner, T. (2006). Foundation and practice of neuro-
feedback for the treatment of epilepsy. Applied Psychophysiology
and Biofeedback, 31, 21–35.
Sterman, M. B., Wyrwicka, W., & Roth, S. R. (1969). Electrophysi-
ological correlates andneural substrates of alimentary behaviour
in the cat. Annals of the New York Academy of Science, 157,
723–739.

Appl Psychophysiol Biofeedback
1 3
Tan, G., Thornby, J., Hammond, D. C., Strehl, U., Canady, B.,
Arnemann, K., & Kaiser, D. A. (2009). Meta-analysis of EEG
biofeedback in treating epilepsy. Clinical EEG and Neuroscience,
40, 173–173.
Vernon, D., Dempster, T., Bazanova, O., Rutterford, N., Pasqualini,
M., & Andersen, S. (2009). Alpha neurofeedback training for per-
formance enhancement: Reviewing the methodology. Journal of
Neurotherapy, 13, 214–227.
Vernon, D., Egner, T., Cooper, N., Compton, T., Neilands, C., Sheri,
A., & Gruzelier, J. (2003). The effect of training distinct neuro-
feedback protocols onaspects of cognitive performance. Interna‑
tional Journal of Psychophysiology, 47, 75–86.
Wolpaw, J. R., & McFarland, D. J. (2004). Control of a two-dimen-
sional movement signal by a noninvasive brain-computer interface
in humans. Proceedings of the National Academy of Sciences of
the U. S. A. 101, 17849–17854.
Yuan, H., & He, B. (2014). Brain-computer interfaces using sensorimo-
tor rhythms: Current state and future perspectives. IEEE Transac‑
tions on Biomedical Engineering, 61, 1425–1435.
Zucker, T. L., Samuelson, K. W., Muench, F., Greenberg, M. A., &
Gevirtz, R. N. (2009). The effects of respiratory sinus arrhyth-
mia biofeedback on heart rate variability and posttraumatic stress
disorder symptoms: A pilot study. Applied Psychophysiology &
Biofeedback, 34, 135–143.