Review
Neurofeedback treatment of
CONTENTS
Basic rationale of
neurofeedback training
Origins & physiological basis
of neurofeedback in
epilepsy treatment
Clinical research
Current clinical practice
Course of therapy
Costs & reimbursement
Summary & conclusions
Expert commentary
Five-year view
Conflict of interests
References
Affiliations
†
Author for correspondence
Functional MRI Research Center,
Columbia University,
Neurological Institute Box 108,
710 West 168th Street, New York,
NY 10032, USA
Tel.: +1 212 342 0121
Fax: +1 212 342 0855
te2111@columbia.edu
KEYWORDS:
biofeedback,
electroencephalogram operant
conditioning, epilepsy,
neurofeedback, seizure disorders
www.future-drugs.com
epilepsy: from basic rationale to
practical application
Tobias Egner† and M Barry Sterman
The treatment of epilepsy through operant conditioning of the sensorimotor rhythm
electroencephalogram has a 35-year history. Neurophysiological studies have shown that
this phasic oscillation reflects an inhibitory state of the sensorimotor system. Operant
learning of sensory motor rhythm production results in an upregulation of excitation
thresholds within the thalamocortical sensory and motor circuitry, which in turn is
associated with reduced susceptibility to seizures. The clinical benefits derived from this
neurofeedback training protocol, particularly in patients that are nonresponsive to
pharmacotherapy, have been documented in many independent laboratories. Recent
advances in computer technology have resulted in the availability of relatively
inexpensive high-quality equipment for the application of neurofeedback therapy, thus
presenting a viable and promising treatment alternative to the interested clinician.
Expert Rev. Neurotherapeutics 6(2), 247–257 (2006)
Basic rationale of neurofeedback training regulate the targeted EEG parameter(s).
Neurofeedback training aims to teach While the current review emphasizes the
self-regulation of measures of the scalp- importance of operant conditioning, other
recorded electroencephalogram (EEG) via mechanisms may also contribute to successful
biofeedback. For instance, a patient may be learning of EEG self-regulation [1–3].
trained to intentionally up- or down-regu- The clinical rationale behind this type of
late the amplitude of a particular EEG fre- training rests on the EEG’s central role as an
quency component recorded from a specific indicator of a number of neurological and
scalp site. Neurofeedback, also known as psychiatric conditions. Put simply, the goal of
EEG biofeedback, involves the real-time neurofeedback therapy is to convert an
feedback of changes in the EEG to the ‘abnormal’ EEG signature associated with the
trainee participant. This feedback loop typi- patient’s condition to a ‘normal’, healthy EEG
cally consists of decomposing the EEG into topography. Since the abnormal EEG signa-
various frequency components and translat- ture is a reflection of underlying neuropathol-
ing ongoing changes in these components ogy, its normalization presumably indicates
into an audiovisual computer-game-style repair, and would be accompanied by symp-
display, which is presented to the patient. tom amelioration. Note that the one-to-one
Learning takes place by the principles of relationship between brain pathology and
operant conditioning: whenever the patient’s EEG parameters implied here is evident in
EEG signature complies with that desired by epilepsy/seizure disorder, but probably consti-
the therapeutic strategy, the patient is tutes an over-simplification in the context of
‘rewarded’, for example, by scoring points or many other neurological or psychiatric condi-
making a game character move on the tions. In order to guide and quantify the
screen. Through a process of trial and error, course of neurofeedback therapy with respect
the patient gradually develops strategies to to a patient’s EEG features, current practice
modify his or her EEG in order to maximize therefore usually involves ‘quantitative EEG’
reward, thereby learning to intentionally (qEEG) mappings via multi-electrode arrays
10.1586/14737175.6.2.247 © 2006 Future Drugs Ltd ISSN 1473-7175 247
Egner & Sterman
at regular intervals throughout the course of treatment, where
the patients’ qEEG is compared with a qEEG database of an
age-matched healthy control population.
While the causal role of a deficient regulation of
electrocortical activity is self-evident in epilepsy/seizure disor-
der, EEG measures have also been shown to be a promising
indicator of many other conditions, including attention-defi-
cit/hyperactivity disorder (ADHD) [4–10] and depression [11–13].
For a recent review of diagnostic EEG mapping in a large vari-
ety of psychiatric conditions, the interested reader is referred to
Saletu and colleagues [14]. Considering the reliable association
between many clinical symptomatologies and specific EEG
abnormalities, it comes as no surprise that therapeutic neuro-
feedback applications are currently spanning a wide spectrum
of clinical disorders. Furthermore, learned EEG self-regulation
has also shown potential in improving cognitive performance in
healthy individuals [15–17]. The increased interest in (and
acceptance of ) neurofeedback as a feasible clinical tool has been
reflected in recent special issues dedicated to advances in neuro-
feedback research and applications in a number of mainstream
clinical journals (e.g., Clinical Electroencephalography and Child
and Adolescent Psychiatric Clinics of North America). Neurofeed-
back treatment of epilepsy/seizure disorder, the focus of this
review article, constitutes the earliest, best understood and
clinically most well established application of this technique.
Origins & physiological basis of neurofeedback in
epilepsy treatment
The origins of neurofeedback treatment of epilepsy/seizure
disorders can be directly traced to the first systematic demon-
stration of operant EEG conditioning in general [18]. In the
context of sleep research, Sterman and colleagues conducted a
series of studies investigating learned suppression of a previ-
ously rewarded cup-press response for food in cats [19–22]. Dur-
ing learned suppression of this response, the appearance of a
particular EEG rhythm over the sensorimotor cortex emerged
above nonrhythmic, low-voltage background activity. This
rhythm was characterized by a frequency of 12–20 Hz, not
unlike sleep spindle EEG, with a spectral peak around
12–14 Hz, and has been referred to as the ‘sensorimotor
rhythm’ (SMR) [19]. The investigators decided to study this dis-
tinct rhythm directly, attempting to operantly condition the
cats to produce SMR, by making a food reward contingent on
SMR production. Cats learned this feat of EEG self-regulation
with apparent ease, and the behavior associated with SMR pro-
duction was one of corporal immobility, with SMR bursts
regularly preceded by a drop in muscle tone [21,22].
In a twist of extraordinary serendipity, Sterman’s laboratory
was soon afterwards commissioned to establish dose–response
functions of a highly epileptogenic fuel compound. When
employing the cats that had previously taken part in SMR con-
ditioning as experimental animals, these cats were found to dis-
play vastly elevated epileptic seizure thresholds compared with
untrained cats, suggesting that SMR training had somehow
inoculated the cats against experiencing seizures. Subsequently,
248
this research was successfully extrapolated to humans, where it
was repeatedly documented that seizure incidence rates could
be lowered significantly (and even abolished entirely) by SMR
feedback training (see below under Clinical research).
Owing to its close link to intracranial recordings in animals,
the neurogenesis of SMR, and thus the neural mechanisms
underlying its efficacy in treating epilepsy, are fairly well
understood. SMR appears to emanate from the ventrobasal
nuclei (nVB) of the thalamus [23], which are generally con-
cerned with conducting afferent somatosensory information.
During SMR conditioning, nVB firing patterns shift from fast
and nonrhythmic (tonic) discharges to systematic, rhythmic
bursts of discharges [24], which in turn are associated with sup-
pression of somatosensory information passage [25] and reduc-
tion in muscle tone. Upon reduction of afferent somatosensory
input, the nVB cells hyperpolarize. Instead of remaining at a
stable level of inhibition, however, a gradual depolarization
mediated by a slow calcium influx causes the nVB neurons to
discharge a burst of spikes, which are relayed to the sensori-
motor cortex and thalamic reticular nucleus (nRT) neurons.
Stimulation of the latter in turn leads to a γ-aminobutyric acid
(GABA)-ergic inhibition of VB relay cells, thus returning
them to a hyperpolarized state and initiating a new cycle of
slow depolarization. In this way, the interplay between neuro-
nal populations in nVB, nRt and sensorimotor cortex results
in rhythmic thalamocortical volleys and consequent cortical
EEG oscillations (FIGURE 1).
While attenuation of efferent motor and afferent somatosensory
activity can initiate SMR, the oscillatory activity is also largely
influenced by nonspecific cholinergic and monoaminergic neu-
romodulation, which can affect excitability levels both in tha-
lamic relay nuclei and in the cortical areas receiving the relayed
signals. During waking activity, the neuromodulator influences
as well as cortical projections normally keep VB cells depolar-
ized and thus suppress rhythmic bursting patterns, while dur-
ing behavioral stillness, oscillations at SMR frequency may be
observed. As SMR constitutes the dominant resting frequency
of the thalamocortical somatosensory and somatomotor path-
ways, operant training of SMR is assumed to result in improved
control over excitation in this system. Increased excitation
thresholds in turn are thought to underlie the clinical benefits
of SMR training in epilepsy, a condition characterized by
cortical and/or thalamocortical hyper-excitability.
More recently, functional magnetic resonance imaging
(fMRI) studies in human subjects have shown that the SMR
EEG pattern is associated with an increase in metabolic activity
in the striatum of the basal ganglia nuclear complex [N BIR-
BAUMER, PERS. COMM.]. Further, examining fMRI changes in chil-
dren with ADHD who improved significantly in cognitive tests
after SMR neurofeedback training, Lavesque and Beauregard
observed a specific and significant increase in metabolic activity
in the striatum [27]. The striatum, the anterior component of
the basal ganglia, has been characterized anatomically as a sys-
tem of fiber connections that form a loop from cerebral cortex
and back to cerebral cortex [28]. The two major components of
Expert Rev. Neurotherapeutics 6(2), (2006)
 Neurofeedback treatment of epilepsy

thalamus, with subsequent

generation of oscillatory dis-
charge to sensorimotor cortex.
Thus, a significant change in
SI
the organization of motor con-
trol and consequent excitability
is accompanied by a volley of
strong oscillatory discharge to
nRt the cortex with each trained
SMR response.
Findings in the study of
vPL synaptic mechanisms mediating
experience-based neuronal reor-
RN ganization, and thus learning,
provide an appealing theoretical
basis for the clinical effects of
SMR training [33]. Many stud-
SA
ies have shown that strong,
repetitive afferent input to cor-
tical and other forebrain neu-
rons can promote increased syn-
aptic strength in relevant circuits
see [34–37]. These changes, pro-
duced over time by the synthe-
sis and insertion of new excita-
tory transmitter channels in
Figure 1. Overview of neurophysiological changes associated with trained bursts of the 12–15 Hz postsynaptic receptor sites, result
sensorimotor rhythm in the cat. On the efferent side of the motor pathway, an abrupt decrease in muscle tone is in a state known as long-term
seen together with a corresponding decrease in the firing rate of cells in the RN and a suppression of monosynaptic potentiation (LTP), which
stretch reflex excitability (not shown). On the afferent side of the somatosensory pathway, a decrease in the firing rate
increases synaptic sensitivity
of SA cells is accompanied by a shift to an interactive burst-silence pattern in cells of both the VPL and nRt, resulting in
the rhythmic oscillation that can be recorded over SI [26]. and the probability of future
nRT: Thalamic nucleus reticularis; RN: Red nucleus; SA: Somatic afferent; SI: sensory cortex; vPL: Ventroposterolateral. activation. The relevance of this
literature to the present discus-
sion stems from the fact that
the striatum include the putamen/globus pallidus complex and the thalamocortical oscillatory processes discussed above result
the caudate nucleus. The putamen exerts a primary inhibitory in recurrent bursts of strong afferent discharge to the sensori-
influence on the globus pallidus, and receives its input from the motor cortex. This recurrent discharge arrives on the apical
sensorimotor cortex. When excited, the globus pallidus in turn dendrites of cortical pyramidal cells in layer four of this area.
exerts an inhibitory influence on the motor and premotor cor- The unique timing and location of these strong, recurrent
tex via the thalamic loops. If input from the sensorimotor cor- afferent bursts, together with concurrent intracortical inputs
tex is reduced, the globus pallidus is thus activated, and this in to distal portions of these same pyramidal neurons, provide a
turn further increases inhibition of motor cortex output. Sim- potentially ideal milieu for LTP and learning that is consistent
ply put, the striatum has been attributed a role in managing with recent concepts of coincidence detection and synaptic
background motor tone and participating in the planning plasticity [38]. That is, the coincidence of strong thalamic affer-
phase of movements [29,30]. ent input near the pyramidal cell body and back-depolariza-
As mentioned above, there is abundant neurophysiological tion from cognitively based distal excitation of the same cells
evidence for a reorganization in efferent motor function accom- magnifies local depolarization and subsequent LTP.
panying the SMR pattern in the EEG. Consistent with an acti- Thus, the functional changes in sensorimotor pathways
vation of striatal inhibitory mechanisms, these studies have mediating the discrete and recurrent onset of SMR activity in
documented a reduction in background motor tone [20,22,31], a the EEG may be specifically strengthened during feedback
related decrease in stretch reflex excitability [32] and a suppres- training through a progressive potentiation process, resulting in
sion of efferent discharge in voluntary motor pathways [24]. a lasting decrease in sensorimotor excitability and a raised
This convergence of findings suggest that facilitation and/or threshold for seizures. Direct empirical corroboration of this
regulation of the SMR substrate alters motor excitability and proposal represents an important challenge for future research.
sets the stage for reduced proprioceptive afferent input to In conclusion, SMR reflects thalamocortical interactions

www.future-drugs.com 249
Egner & Sterman
affected by the current state of the sensorimotor system as well
as general cortical arousal, and it appears that SMR
neurofeedback efficacy is based on a long-term enhancement of
inhibitory mechanisms in sensorimotor pathways [18,33,39].
It should be noted that a different neurofeedback approach,
based on the measurement of ‘slow cortical potential’ (SCP)
shifts (between positive and negative polarity) has also proved
successful in the treatment of epilepsy [40–42]. Conceptually,
SCP and SMR neurofeedback share the same aim of reducing
cortical excitability. As negative slow potentials are reflective of
lowered excitation thresholds (through depolarization) in the
apical dendrites of cortical pyramidal neurons, while positive
slow potentials represent raised excitation thresholds [43], SCP
training in epileptics is aimed at enabling the patient to volun-
tarily produce cortical inhibition (i.e., positive SCPs) and thus
avoiding seizure onset. Interestingly, Hinterberger and col-
leagues have also found that learned increases in positive SCPs
are associated with increased metabolic activity in the striatum
of the basal ganglia, suggesting a convergent effect of SMR and
SCP training [44]. However, as SCP neurofeedback is not very
commonly employed by clinicians, the current review will focus
exclusively on the more widely used SMR neurofeedback.
Clinical research
Since the first singlecase study, reported over 30 years ago [45], a
fair number of controlled clinical studies, stemming from many
different laboratories, may be argued to have generally pro-
duced consistent data on the efficacy of SMR training in epi-
leptic patients. It is particularly noteworthy that these results
have been achieved in an extremely difficult subgroup of epi-
lepsy patients, those with poorly controlled seizures who had
proven unresponsive to pharmacological treatment. Conversely,
studies have tended to be of small sample size and only rarely
conformed to standard, randomized controlled trial designs
prevalent in pharmacological clinical trials. Here, only a cursory
overview of this clinical research literature will be provided. For
a more detailed treatment the interested reader is referred to a
number of recent reviews [39,46,47].
In initial studies involving relatively small sample sizes and
pretreatment baseline measures as a control condition, on aver-
age 80% of patients trained at enhancing SMR amplitudes were
shown to display significant clinical improvements [48–50]. For
instance, Cott and colleagues reported that 3 months of SMR
training were associated with significantly reduced seizure inci-
dent in five out of seven patients who had previously suffered
from very poorly controlled seizures [51]. These high success
rates of SMR training were further confirmed in investigations
that employed more elaborate control conditions, such as non-
contingent or random feedback [52–55]. In addition, a number of
investigations employing crossover designs have emphasized the
causal and frequency-specific nature of SMR training by docu-
menting symptom reversal when reversing training contingen-
cies [56–58]. In a larger-scale study (n = 24), Lantz and Sterman
employed a double-blind design, with age- and seizure-matched
patients assigned to either a contingent training schedule of
250
enhancing 11–15 Hz SMR activity (while simultaneously
inhibiting slower and higher frequencies), a noncontingent,
‘yoked control’ training schedule or a waiting list control
group [59]. Significant reduction of seizure incidence with a
median of 61% (range 0–100%) was found in the contingent
SMR feedback group only. In the largest study to date, Andrews
and Schonfeld documented clinical improvements from pre-
treatment baseline in 69 out of 83 patients participating in a
mixed relaxation and SMR feedback protocol [60].
In reviewing the data accumulated in these studies, Sterman
found that 82% of 174 participating patients had shown signif-
icantly improved seizure control (defined as a minimum of
50% reduction in seizure incidence), with approximately 5% of
these cases reporting a complete lack of seizures for up to 1 year
subsequent to training cessation [39]. While existing studies
have thus produced overwhelmingly positive results, it has to be
pointed out that additional replications are still highly desira-
ble, particularly because studies have been small and the use
and type of control conditions have been inconsistent. For a
very promising treatment targeting such a serious condition as
epilepsy, the lack of large-scale clinical trials of neurofeedback
training to date is disappointing. A probable reason for this
state of affairs is that neurofeedback research is a very time- and
work-intensive enterprise that has traditionally not received
extensive research funding and has, for obvious reasons, not
been pursued by the pharmaceutical industry.
Current clinical practice
Best practice considerations
Neurofeedback therapy has been greatly aided by advances in
computer technology and software. However, the ease in devel-
opment of new software programs for feedback functionality
and displays, together with the entry into the field of a diverse
group of professionals and semi-professionals, has led to an
unfortunate lack of consensus on methodology and standards
of practice. In turn, this has contributed to reluctance by the
academic and medical communities to endorse the field.
Based on the history outlined above, the practice of neuro-
feedback requires a basic understanding of the principles of
relevant neurophysiology, operant conditioning, neuropathol-
ogy, and, ultimately, neuroplasticity. Further, the unique
opportunity for extended contact with the patient or client
provided by this modality favors skilled clinical insight and
guidance. While some who practice neurofeedback pursue
different models, this review will be restricted to a considera-
tion of practice that adheres to these principles and utilizes
these opportunities.
As pointed out above, the neurofeedback treatment plan
begins with a comprehensive study qEEG. Treatment guid-
ance through qEEG measures represents a relatively recent
development and is not employed universally. Much of the
older research literature did not utilize this tool, and some
clinicians today still prefer to adhere to the use of standard
training protocols that are not informed by multi-electrode
EEG mapping. However, the utility of this tool for
Expert Rev. Neurotherapeutics 6(2), (2006)
 Neurofeedback treatment of epilepsy

assessment and as a dependent variable has been documented
in numerous reports and reviews [40,57,61,62]. It is also demon-
strated clearly in the case study reviewed below. The qEEG
acquisition involves collection of appropriate samples of
EEG data from at least 19 standardized sites over the cerebral
cortex (placement according to the International 10/20 Sys-
tem) during states of rest with eyes closed and open, and
states of task engagement, such as reading and solving math
problems. These data are then digitized and subjected to fre-
quency and amplitude analysis using various versions of spec-
tral transform and database procedures [63–65]. Quantitative
and descriptive data exhibiting the magnitude (a combina-
tion of the incidence of a given frequency and its amplitude
computed for a specific time period, or epoch, as dictated by
spectral analysis) and distribution of relevant frequencies
recorded during these conditions are then displayed as tables
and graphics and compared statistically with group data from
an appropriate normative database. Deviant patterns of fre-
quency and/or magnitude, as well as disturbed interactions
among sites, can be identified through this analysis and used
to direct neurofeedback training. Recent improvements to
this methodology have reduced distortions produced by non-
EEG events (artifacts), biological cycles and certain mathe-
matical and statistical corruptions [66,67]. The importance of
accuracy in this analysis cannot be overemphasized, since
Theta
25
20
Band magnitude (Uv)
these findings will guide the development of neurofeedback
treatment strategies and, in turn, determine the quality of its
application. Follow-up qEEGs after a course of neurofeed-
back training can then provide for an objective assessment of
EEG changes and treatment outcomes.
An example of a statistically significant deviant frequency pat-
tern from a 36-year old epileptic patient is shown in FIGURE 2.
This patient suffered from a documented seizure focus in the
right anterior temporal lobe caused by surgical removal of an
astrocytoma tumor. The topometric display shown is from the
qEEG analysis program of the Sterman–Kaiser Imaging Labora-
tory (SKIL). The graph plots mean spectral magnitudes in the
4–8 Hz τ band across 19 standard sites from a standard EEG
sample with the eyes open, against an age-matched normative
database. Statistically significant and clearly pathological increases
in this frequency can be seen in the lateral frontal and anterior tem-
poral locations on the right (F8 and T4). These findings provided
a rational strategy for neurofeedback treatment, as described below
(see Course of therapy).
Equipment considerations
The hardware and software used in the application of neuro-
feedback training should provide for the collection and evalua-
tion of quality EEG signals needed both for qEEG data acquisi-
tion and valid operant conditioning. A variety of equipment
options is available for qEEG recording,
with a broad range of pricing, depending
on component quality and software
functionality. It is important to note,
however, that the more expensive sys-
4–8 tems have been developed for research
Norm and medical applications and were not
+2SD designed to address the assessment issues
–2SD relevant to neurofeedback. This is, of
course, more a function of software than

hardware. Most systems provide for the

15 collection of valid EEG data but the
interested professional should do their
homework in understanding and inquir-
ing about such important issues as ampli-
10
fier noise, dynamic range, sampling rate
and filter settings.
The objective of the qEEG for the
5 neurofeedback treatment of seizure dis-
orders is to provide for the detection
and frequency/topography characteriza-
tion of relevant EEG pathology. Analy-
0 sis software should thus provide an
Fz F3 F7 F4 F8 T3 C3 C7 C4 T4 T5 P3 P7 P4 T6 accurate indication of localized fre-
quency and amplitude/magnitude devi-
ations, and preferably also address dis-
Figure 2. Topometric plot showing mean eyes-open spectral magnitudes. At standard 10/20 locations
turbances in the functional coordination
in the 4–8 Hz band from a normative database (pink curves) and a patient with localization-specific, partial
complex seizures arising from a right anterior temporal lobe focus (dashed green curves). Database variance of these variables among brain regions,
(2 standard deviations) is shown by dashed blue/purple curves. Statistically significant increases in this as indicated by metrics such as coherence
frequency can be seen at F8 and T4, with the latter exceeding four standard deviations. or comodulation [68,69]. Neurofeedback

www.future-drugs.com 251
Egner & Sterman

is typically not applied with the power transform of provide reward exclusively for the EEG response being
frequency-analyzed EEG data (squaring of magnitudes). reinforced. Novel images or sounds accompanying this
Accordingly, some programs avoid using this transform and response may actually overshadow or block the EEG changes
express data only as magnitude, a measure that also prevents sought and divert the reinforcement effect of the reward [77,78].
the excessive skewing that results from squaring of EEG val- Finally, and perhaps of equal importance to the technical issues
ues, thus improving the validity of parametric statistical anal- discussed above, software programs and training strategies
ysis [70,71]. Further, accurate specification of frequency abnor- should stress exercise rather than entertainment.
malities requires analysis rather than the standardized bands
such as ‘τ’ and ‘α’. These traditional clinical bands overlook Course of therapy
significant individual differences and typically dilute the Typically, neurofeedback treatment for seizure disorders is
accuracy of relevant frequency specification [71,72]. directed by the pattern of EEG pathology detected through
Software used for neurofeedback training should be based on qEEG analysis, in combination with an appreciation for the anti-
the need to apply meaningful operant conditioning procedures epileptic effects of SMR training. Concerning the former, there
to training objectives. Well established scientific literature dic- are many different ways in which seizure pathology may manifest
tates the functional characteristics required for effective operant in the EEG. Further, the EEG can be affected significantly by
conditioning. It is interesting to note, however, that many of any anticonvulsant medications taken by the patient.
the equipment developers who serve the neurofeedback com- Atypical slow or fast EEG patterns may be observed. In
munity have not demonstrated an awareness of, or concern for, some cases these are accompanied by such transients as spike-
this literature. For example, feedback training should ideally be and-wave discharge, sharp waves, or poorly organized, high-
configured to provide for discrete trials, a fundamental element amplitude events termed paroxysms. Knowledge of these EEG
of both classical and operant conditioning [73,74]. Yet many sys- characteristics and complexities is essential for the appropriate
tems offer undifferentiated feedback, with no trial structure and application of neurofeedback treatment. However, despite
more or less continuous rewards. Additionally, empirical data corruption produced by medications, the literature in this
have demonstrated that the latency between response and field indicates that feedback strategies directed to the suppres-
reward must be immediate, or at best less than a quarter of a sion of any or all of these various abnormal patterns, together
second, for meaningful learning to occur [75,76]. Initially, at with the enhancement of central cortical SMR activity, results
least, this requirement was frequently not met by developers of in the most effective therapeutic outcomes [39].
neurofeedback programs and probably still deserves close Often the EEG abnormality disclosed by the qEEG is
attention by the potential customer. Further, trials should rather specific. For example, in the case described above the pathol-
ogy observed was essentially
restricted to the right fronto-
1 2 3 4 5 6 7
temporal area (FIGURE 2),
where the surgical removal of
a tumor resulted in cortical
hyperexcitability and con-
tinuing partial-complex
8 9 10 11 12 13 14
seizures. This patient was in
her mid 30’s and had just
married. She wanted to get
pregnant and was not inter-
15 16 17 18 19 20 21 ested in taking medication
that would place her fetus at
risk. Accordingly, she was
referred for neurofeedback
treatment. Data from the
eyes-open qEEG were
-2 -1 0 1 2 selected for neurofeedback
guidance (FIGURE 3) since this
Figure 3. Activation maps displaying the statistical results comparing this subject with a normative database most approximated the
(as viewed from above, with the nose pointing upwards). Results are shown for each of 21 individual 1 Hz
frequency bins, characterizing the frequency/topography localization of abnormality, in this case during an eyes-open
training condition. How-
recording. The color scale at the bottom shows the normative database variance for spectral magnitudes in these ever, it should be pointed
frequencies, with statistically significance abnormality set at ±2 standard deviations (pink/red and dark blue areas). Note out that a similar pattern
the zone of significantly elevated activity ranging from 5–8 Hz in the right fronto-temporal region. Activity in this of abnormality was seen
region is also increased at other frequencies but not significantly. A significant decrease in slow (4–5 Hz) and faster in all recorded states. The
(20–21 Hz) activity can also be seen in some other cortical areas.
most affected frequency

252 Expert Rev. Neurotherapeutics 6(2), (2006)

 Neurofeedback treatment of epilepsy

range was between 5–8 Hz. Thus, neurofeedback treatmentOther display material can be used for neurofeedback as well.
Puzzles, engaging picture sequences, stop/start video clips, and
involved the rewarded suppression of 5–8 Hz activity in
other action themes have been employed. These materials can be
right lateral frontal and anterior temporal areas, together
used to focus initial attention on the task for some patients, or as
with a facilitation of 12–15 Hz SMR activity at the adjacent
C4 location. occasional alternatives to more simple displays like the one
described here. However, the issue of exercise versus entertainment
Treatment sessions of 1 h duration were provided twice a week
mentioned above is relevant in this regard. The patient should be
for the first 4 weeks and then once a week for the next 28 weeks.
motivated to view the task much like a workout at a gym, and the
The equipment and procedures used conformed to all of the
feedback objectives as a special exercise for the brain. Discrete trials
principles discussed above. The display presented to the patient
with simple but relevant displays achieve this very well.
included two adjacent vertical bars, one dark blue and the other
The patient completed 36 training sessions, during which
light blue in color. She was required to suppress the light blue
there were no significant intervening events that might influence
bar (5–8 Hz activity at T4 and/or F8) below an adjustable
the patient’s condition. Seizures gradually ceased and a post-treat-
threshold line, set initially at approximately 20% below baseline
ment qEEG revealed complete normalization of EEG spectral
level, while increasing the dark blue bar (12–15 Hz activity at
topography (FIGURE 4). Withdrawal from treatment was graded,
C4) above a threshold line set 20% above baseline. If both of
these objectives were achieved simultaneously a counter in the
with subsequent sessions being provided at increasing intervals
top-center of the screen advanced one digit and the unit
over the next several months. Shortly afterwards, the patient
sounded a pleasant tone. The system paused for 2 seconds and
became pregnant and later delivered a pair of healthy twin boys.
the task was repeated. Prior to each in a series of sequential 3
She returned for ‘booster’ sessions at approximately 6 month
min sets the patient specified a numerical goal in the final count
intervals for several years. Except for rare brief auras, this patient
for that set. At appropriate intervals the thresholds were adjusted
has not had another seizure for over 6 years.
to increase the challenge and achieve ‘shaping’ of the desired
This case was rather straightforward, and was not complicated
response pattern. by comorbidities, a long history of failed medications and their
side effects, behavioral or economic adapta-
tions or other disruptive factors. In general,
patients seeking this remedy often present
Rep 1 Rep 4 such complications, a reality that necessi-
Rep 2 Nom tates cooperation with the neurologist and
Rep 3 +/–2.0SD other professionals associated with the case,
Theta 4–8 Hz
30 as well as a competent appreciation for, and
attention to, family and personal dynamics.
But perhaps the most important variable
determining the success of neurofeedback is
the clinician’s ability to instill a motivation
to succeed in the patient. Unlike operant
Band magnitude (Uv)

20
conditioning of behavioral responses, where
the organism has a conscious awareness of
the response leading to primary reward,
with neurofeedback the response is an occult
change in physiology of which the subject
10
has little or no direct experience. The drive
to obtain a symbolic reward results in the
reinforcement that ultimately strengthens
this response. This is where neurofeedback
differs from traditional operant condition-
0 ing, and why clear and meaningful rewards,
Fz F3 F7 F4 F8 T3 C3 C7 C4 T4 T5 P3 P7 P4 T6 combined with proper preparation of the
patient and appropriate clinical skills and
are essential.
Figure 4. Topometric analysis of a patient’s electroencephalogram. As in Figure 2, showing data from
two independent recordings with eyes closed both before and after a course of 36 neurofeedback
treatments. Dashed green curves show database means for the 4–8 Hz band and green curves show two Costs & reimbursement
standard deviations. Data shown as dashed purple and dashed pink curves are from studies carried out Since there is no license required for the
prior to this treatment, while pink and purple curves are data after treatment. The abnormal focal increases delivery of neurofeedback treatment, qual-
of 4–8 Hz activity at T4 and F8 were attenuated to within normal range after treatment. Across this period ifications and competence of therapists
seizures became completely controlled.
may vary widely. However, the discerning

www.future-drugs.com 253
Egner & Sterman
customer, as well as the aspiring therapist, should be aware of
the certification program of the Biofeedback Certification
Institute of America (BCIA) [101]. Thus, costs are also variable,
and will depend upon the credentials and experience of the
provider, as well as regional norms. If the patient is interested
in accountable, scientifically based treatment, an effort should
be made to evaluate the qualifications of the therapist and the
equipment and methods they use. As discussed above, the treat-
ment package may include a comprehensive qEEG mapping,
which adds to the overall cost of the program.
Despite the now well documented efficacy of neurofeed-
back as an adjunct (and sometimes alternative) treatment for
epilepsy and its endorsement by medical organizations, most
insurance providers offer limited coverage for this treatment,
if any at all. Conversely, a possible lifetime of medications,
whether effective or not, as well as expensive neurosurgical
procedures, are both well reimbursed. From a cost-benefit
perspective, however, neurofeedback treatment for seizure
disorders would appear to be a bargain, as long as it is
applied competently.
Summary & conclusions
The neurophysiological rationale as well as the basic and clini-
cal research literature pertaining to the treatment of epi-
lepsy/seizure disorder with SMR neurofeedback have been
reviewed, the longest-running and arguably best established
clinical application of EEG operant conditioning. The basic
research literature provides ample data to support a very
detailed model of the neural generation of SMR as well as the
most likely candidate mechanism underlying its efficacy in epi-
lepsy treatment. The clinical research literature, on the whole,
documents promising benefits from SMR training in a patient
population that has consisted to a large degree of drug non-
responders. However, future large-scale, controlled studies
remain highly desirable.
Skilled practice of neurofeedback requires a good
understanding of the neurophysiology underlying EEG
oscillations, of operant learning mechanisms, and an in-
depth appreciation of the various hardware/software equip-
ment options open to the practitioner. It is suggested that
best clinical practice includes the systematic mapping of
quantitative multi-electrode EEG measures before, during
and after the course of treatment, in comparison to norma-
tive database information. These assessments help to guide
the choice of treatment strategy and document progress in
terms of the physiological normalization of a clinical EEG
signature. The typical treatment course involves at
least weekly training sessions for several months. Unfortu-
nately, adequate reimbursement by insurance companies is
currently scarce.
In conclusion, the research literature reviewed in this article
justifies the assertion that neurofeedback treatment of epi-
lepsy/seizure disorders constitutes a well founded and viable
adjunct treatment (and potentially a long-term alternative) to
anticonvulsant pharmacotherapy.
254
Expert commentary
Therapeutic application of neurofeedback techniques have
been expanding across a variety of neurological and psychiat-
ric conditions in recent years. Increased interest in this non-
pharmacological treatment option has brought with it
enhanced scrutiny (and criticism) regarding the research base
on which claims for its efficacy rest. It is our opinion that the
application of EEG operant conditioning to epilepsy/seizure
disorders constitutes an extremely promising treatment
option, grounded in a solid neurophysiological rationale and
research findings. It is important to point out that most of
the patients who have participated in neurofeedback research
studies and many who seek this treatment today represent
unquestionable failures of pharmacologic or other therapies,
particularly with complex-partial seizure disorders. It is
therefore particularly noteworthy that positive outcomes
have often been obtained in the context of treating this
extremely difficult subpopulation of epilepsy patients. We
view it as unfortunate, therefore, that some professionals still
criticize neurofeedback treatment for the lack of more con-
sistent or successful outcomes, although we acknowledge the
need for additional clinical research. On the contrary, evi-
dence has shown that most of these difficult patients benefit
beyond any chance or placebo outcome, and some do so dra-
matically. Furthermore, in contrast to pharmacological
symptom-management, the learned modulation of
thalamocortical excitability through neurofeedback training
holds the promise of a true long-term recovery from epilepsy.
Considering the common side effects and costs associated
with life-long pharmacotherapy, we do not view neuro-
feedback treatment as a ‘last resort’ option for drug treat-
ment-resistant cases only, but rather as a generally viable
adjunct treatment consideration for any patient suffering
from seizures, which may constitute a long-term alternative
to anticonvulsant medication. In this vein, it has recently
been argued that neurofeeback treatment of seizure disorders
meets the ‘Clinical Guidelines’ criteria of the American
Academy of Child and Adolescent Psychiatry (AACAP) [79],
a recommendation that states that a particular practice
should always be considered by the clinician, but that there
are exceptions to their application. As a comparison exam-
ple, the same criterion is met by stimulant medication treat-
ment for ADHD, a treatment that is abundantly and per-
haps excessively utilized today. While neurofeedback
treatment may initially appear a time- and cost-intensive
undertaking, a more long-term view reveals the potential for
an excellent cost-benefit trade-off. It has to be stressed, how-
ever, that the skilled application of neurofeedback requires a
committed, well trained and motivationally adept professional.
Five-year view
The last 10 years or so have witnessed a substantial renaissance
of research into the basic mechanisms and clinical applications
of neurofeedback training, and a continuous increase in the
number of practitioners, as well as the establishment of a
Expert Rev. Neurotherapeutics 6(2), (2006)
 Neurofeedback treatment of epilepsy

dedicated international society and annual scientific
meetings [102]. In all likelihood, this rapid growth rate of the
field will continue over the next 5 years and beyond and the
continuous advances in hardware and software technology will
result in higher quality and more affordable equipment. Unfet-
tered expansion in a field that is essentially unregulated by pro-
fessional licensing bodies carries with it some obvious dangers
as far as the consistency and quality of patient care are con-
cerned. However, we are optimistic that the clinical success of
the scientifically steeped neurofeedback application to
epilepsy/seizure disorders will garner ever-broader support for
this approach. The currently growing embrace of this
treatment strategy by mainstream scientific journals and pro-
fessional bodies should lead to more appropriate insurance
reimbursement policies in the near future.
Conflict of interests
MB Sterman is co-founder and president of the Ster-
man–Kaiser Imaging Lab, a privately held coorporation. The
images of quantitative EEG analysis shown in this paper were
derived from the software program and database developed
by this company.

References 9 Clarke AR, Barry RJ, McCarthy R, 18 Sterman MB. Physiological origins and
Papers of special note have been highlighted as: Selikowitz M. Children with attention- functional correlates of EEG rhythmic
• of interest deficit/hyperactivity disorder and comorbid activities: implications for self-regulation.
•• of considerable interest oppositional defiant disorder: an EEG Biofeedback Self Regul. 21, 3–33 (1996).
analysis. Psychiatry Res. 111(2–3), 181–190 •• Offers a thorough review of the neural
1 Kotchoubey B, Kübler A, Strehl U, Flor H,
(2002). mechanisms underlying particularly the
Birbaumer N. Can humans perceive their
10 Magee CA, Clarke AR, Barry RJ, generation of the sensorimotor rhythm.
brain states? Conscious Cogn. 11, 98–113
(2002). McCarthy R, Selikowitz M. Examining the 19 Roth SR, Sterman MB, Clemente CC.
diagnostic utility of EEG power measures Comparison of EEG correlates of
2 Lacroix JM. Mechanisms of biofeedback
in children with attention reinforcement, internal inhibition, and
control. In: Consciousness and Self-
deficit/hyperactivity disorder. Clin. sleep. Electroencephalogr. Clin. Neurophysiol.
Regulation (Volume 4). Shapiro D, Schwartz
Neurophysiol. 116(5), 1033–1040 (2005). 23, 509–520 (1967).
GE (Eds). Plenum, NY, USA, 137–162
(1986). 11 Henriques JB, Davidson RJ. Regional brain 20 Sterman MB, Wyrwicka W. EEG
electrical asymmetries discriminate between correlates of sleep: evidence for separate
3 Neumann N, Kübler A, Kaiser J,
previously depressed and healthy control forebrain substrates. Brain Res. 6,
Hinterberger T, Birbaumer N. Perception
subjects. J. Abnorm. Psychol. 99(1), 22–31 143–163 (1967).
of brain states: mental strategies for
(1990). 21 Wyrwicka W, Sterman MB. Instrumental
brain–computer communication.
Neuropsychologia 41, 1028–1036 (2003). 12 Henriques JB, Davidson RJ. Left frontal conditioning of sensorimotor cortex EEG
hypoactivation in depression. J. Abnorm. spindles in the waking cat. Physiol. Behav.
4 Mann CA, Lubar JF, Zimmerman AW,
Psychol. 100(4), 535–545 (1991). 3, 703–707 (1968).
Miller CA, Muenchen RA. Quantitative
13 Miller A, Fox NA, Cohn JF, Forbes EE, • The original demonstration that cats can
analysis of EEG in boys with attention-
Sherrill JT, Kovacs M. Regional be operantly conditioned to produce
deficit-hyperactivity disorder: controlled
patterns of brain activity in adults with sensorimotor rhythm
study with clinical implications. Pediatr.
a history of childhood-onset electroencephalogram.
Neurol. 8(1), 30–36 (1992).
depression: gender differences and 22 Sterman MB, Wyrwicka W, Roth SR.
5 Chabot RJ, Serfontein G. Quantitative
clinical variability. Am. J. Psychiatry Electrophysiological correlates and neural
electroencephalographic profiles of children
159(6), 934–940 (2002). substrates of alimentary behaviour in the
with attention deficit disorder. Biol.
14 Saletu B, Anderer P, Saletu-Zyhlarz GM, cat. Ann. NY Acad. Sci. 157, 723–739
Psychiatry 40(10), 951–963 (1996).
Pascual-Marqui RD. EEG mapping and (1969).
6 Monastra VJ, Lubar JF, Linden M et al.
low-resolution brain electromagnetic 23 Howe RC, Sterman MB. Cortical-
Assessing attention deficit hyperactivity
tomography (LORETA) in diagnosis and subcortical EEG correlates of suppressed
disorder via quantitative
therapy of psychiatric disorders: evidence motor behavior during sleep and waking in
electroencephalography: an initial
for a key-lock principle. Clin. EEG the cat. Electroencephalogr. Clin.
validation study. Neuropsychology
Neurosci. 36(2), 108–115 (2005). Neurophysiol. 32, 681–695 (1972).
13(3), 424–433 (1999).
15 Egner T, Gruzelier JH. Learned self- 24 Harper RM, Sterman MB. Subcortical unit
7 Monastra VJ, Lubar JF, Linden M. The
regulation of EEG frequency components activity during a conditioned 12–14 Hz
development of a quantitative
affects attention and event-related brain sensorimotor EEG rhythm in the cat.
electroencephalographic scanning process
potentials in humans. NeuroReport Fed. Proc. 31, 404 (1972).
for attention deficit-hyperactivity disorder:
12(18), 4155–4160 (2001). 25 Howe RC, Sterman MB. Somatosensory
reliability and validity studies.
Neuropsychology 15(1), 136–144 (2001). 16 Egner T, Gruzelier JH. EEG Biofeedback of system evoked potentials during waking
low beta band components: Frequency- behaviour and sleep in the cat.
8 Barry RJ, Clarke AR, Johnstone SJ. A
specific effects on variables of attention and Electroencephalogr. Clin. Neurophysiol.
review of electrophysiology in attention-
event-related brain potentials. Clin. 34, 605–618 (1973).
deficit/hyperactivity disorder: I. Qualitative
Neurophysiol. 115, 131–139 (2004). 26 Sterman MB, Bowersox SS. Sensorimotor
and quantitative electroencephalography.
Clin. Neurophysiol. 114(2), 171–183 17 Gruzelier J, Egner T. Critical validation EEG rhythmic activity: A functional gate
(2003). studies of neurofeedback. Child Adolesc. mechanism. Sleep 4(4), 408–422 (1981).
Psychiatr. Clin. N. Am. 14, 83–104 (2005).

www.future-drugs.com 255
Egner & Sterman
27 Levesque J, Beauregard M. Effect of
neurofeedback training on the neural
substrates of selective attention in children
with attention-deficit/hyperactivity
disorder: a functional magnetic resonance
imaging study. Neurosci. Lett. (In Press)
(2006).
28 Brodal P. The basal ganglia. In: The Central
Nervous System: Structure and Function.
Oxford University Press, NY, USA,
246–261 (1992).
29 Chevalier G, Deniau JM. Disinhibition as a
basic process in the expression of striatal
functions. Trends Neurosci. 13, 277–280
(1990).
30 DeLong MR. Primate models of movement
disorders of basal ganglia origin. Trends
Neurosci. 13, 281–285 (1990).
31 Chase MH, Harper RM. Somatomotor and
visceromotor correlates of operantly
conditioned 12–14 c/s sensorimotor
cortical activity. Electroencephalogr. Clin.
Neurophysiol. 31, 85–92 (1971).
32 Babb MI, Chase MH. Masseteric and
digastric reflex during conditioned
sensorimotor rhythm. Electroencephalogr.
Clin. Neurophysiol. 36, 357–365 (1974).
33 Sterman MB. Principles of Neurotherapy.
Proceedings of the Annual Conference. Int.
Soc. Neuronal Reg. 13, 23 (2005).
34 Abel T, Lattal KM. Molecular mechanisms
of memory acquisition, consolidation and
retrieval. Curr. Opin. Neurobiol. 11,
180–187 (2001).
35 Malenka RC, Nicoll RA. Long-term
potentiation–a decade of progress? Science
285, 1870–1874 (1999).
36 Soderling TR, Derkach VA. Postsynaptic
protein phosphorylation and LTP. Trends
Neurosci. 23, 75–80 (2000).
37 Walker MP. A refined model of sleep and
the time course of memory formation.
Behav. Brain Sci. 28, 51–64 (2005).
38 Froemke RC, Poo MM, Dan Y. Spike-
timing-dependent synaptic plasticity
depends on dendritic location. Nature
434, 221–225 (2005).
39 Sterman MB. Basic concepts and clinical
findings in the treatment of seizure
disorders with EEG operant
conditioning. Clin. Electroenceph.
31(1), 45–55 (2000).
•• Provides a detailed review of the clinical
research literature regarding sensorimotor
rhythm neurofeedback in the treatment
of epilepsy.
40 Rockstroh B, Elbert T, Birbaumer N et al.
Cortical self-regulation in patients with
epilepsies. Epilepsy Res. 14, 63–72 (1993).
256
41 Kotchoubey B, Strehl U, Holzapfel S,
Blankenhorn V, Froscher W, Birbaumer N.
Negative potential shifts and the prediction
of the outcome of neurofeedback therapy in
epilepsy. Clin. Neurophysiol. 110(4),
683–686 (1999).
42 Kotchoubey B, Strehl U, Uhlmann C et al.
Modification of slow cortical potentials in
patients with refractory epilepsy: a
controlled outcome study. Epilepsia
42, 406–416 (2001).
43 Birbaumer N. Slow cortical potentials: their
origin, meaning, and clinical use. In: Brain
and behaviour - past, present and future.
Boxtel GJM, von Böcker KBE (Eds),
University Press, Tilborg, 25–39 (1997).
• For the interested reader, this provides an
extensive review of issues surrounding the
use of slow cortical potential feedback.
44 Hinterberger T, Veit R, Wilhelm B,
Weiskopf N, Vatine JJ, Birbaumer N.
Neuronal mechanisms underlying control
of a brain–computer interface. Eur. J.
Neurosci. 21, 3169–3181 (2005).
• One of only a few recent publications that
have employed functional magnetic
resonance imaging in order to elucidate
neural correlates of learned
electroencephalogram self-regulation.
45 Sterman MB, Friar L. Suppression of
seizures in an epileptic following
sensorimotor EEG feedback training.
Electroencephalogr. Clin. Neurophysiol.
33, 89–95 (1972).
• The original case study on the use of
sensorimotor rhythm neurofeedback in
human epileptic patients.
46 Monderer RS, Harrison DM, Haut SR.
Neurofeedback and epilepsy. Epilepsy Behav.
3, 214–218 (2002).
• A critical review of the clinical literature.
47 Walker JE, Kozlowski GP. Neurofeedback
treatment of epilepsy. Child Adolesc.
Psychiatr. Clin. N. Am. 14, 163–176
(2005).
48 Sterman MB, MacDonald LR, Stone RK.
Biofeedback training of the sensorimotor
EEG rhythm in man: effects on epilepsy.
Epilepsia 15, 395–417 (1974).
49 Kaplan BJ. Biofeedback in epileptics:
equivocal relationship of reinforced EEG
frequency to seizure reduction. Epilepsia
16, 477–485 (1975).
50 Seifert AR, Lubar JF. Reduction of epileptic
seizures through EEG biofeedback training.
Biol. Psychol. 3, 157–184 (1975).
51 Cott A, Pavloski RP, Black AH. Reducing
epileptic seizures through operant
conditioning of central nervous system
activity: procedural variables. Science
203, 73–75 (1979).
52 Finley WW, Smith HA, Etherton MD.
Reduction of seizures and normalization of
the EEG in a severe epileptic following
sensorimotor biofeedback training:
preliminary study. Biol. Psychiatry
2, 189–203 (1975).
53 Wyler AR, Robbins CA, Dodrill CB. EEG
operant conditioning for control of
epilepsy. Epilepsia 20, 279–286 (1996).
54 Kuhlman WN, Allison T. EEG feedback
training in the treatment of epilepsy: some
questions and some answers. Pav. J. Biol.
Sci. 12(2), 112–122 (1978).
55 Quy RJ, Hutt SJ, Forrest S. Sensorimotor
rhythm feedback training and epilepsy:
some methodological and conceptual
issues. Biol. Psychol. 9, 129–149 (1979).
56 Lubar JF, Bahler WW. Behavioral
management of epileptic seizures following
EEG biofeedback training of the
sensorimotor rhythm. Biofeedback Self
Regul. 7, 77–104 (1976).
57 Sterman MB, MacDonald LR. Effects of
central cortical EEG feedback training on
incidence of poorly controlled seizures.
Epilepsia 19, 207–222 (1978).
58 Lubar JF, Shabsin HS, Natelson SE et al.
EEG operant conditioning in intractible
epileptics. Arch. Neurol. 38, 700–704
(1981).
59 Lantz D, Sterman MB. Neuropsychological
assessment of subjects with uncontrolled
epilepsy: effects of EEG biofeedback
training. Epilepsia 29(2), 163–171 (1988).
•• The largest double-blind controlled
clinical trial on sensorimotor rhythm
training and epilepsy to date.
60 Andrews DJ, Schonfeld WH. Predicitve factors
for controlling seizures using a behavioral
approach. Seizure 1(2), 111–116 (1992).
61 Sterman MB, Shouse MN. Quantitative
analysis of training, sleep EEG and clinical
response to EEG operant conditioning in
epileptics. Electroenceph. Clin. Neurophysiol.
49, 558–576 (1980).
62 Whitsett SF, Lubar JF, Holder GS. A
double-blind investigation of the
relationship between seizure activity and
the sleep EEG following EEG biofeedback
training. Biofeedback Self Regul. 7, 193–209
(1982).
63 Etevenon P. Applications and perspectives
of EEG cartography. In: Topographic
mapping of brain electrical activity. Duffy
FH (Ed.). Butterworths, MA, USA,
113–141 (1986).
64 Johnstone J, Gunkelman J, Lunt J. Clinical
database development: characterization of
EEG phenotypes. Clin. EEG Neurosci.
36, 99–107 (2005).
Expert Rev. Neurotherapeutics 6(2), (2006)

65 Lorensen TD, Dickson P. Quantitative EEG
normative databases: a comparative
investigation. J. Neurotherapy 8, 53–68 (2004).
66 Kaiser DA, Sterman MB. Automatic
artifact detection, overlapping windows,
and state transitions. J. Neurotherapy
4(3), 85–92 (2006).
67 Kaiser DA, Sterman MB. Correcting
sampling bias of tapering windows. Int. J.
Psychophysiol. (In Press) (2005).
68 Sterman MB, Kaiser DA. Comodulation: a
new QEEG analysis metric for assessment
of structural and functional disorders of the
CNS. J. Neurotherapy 4(3), 73–83 (2001).
69 Thatcher RW. Cyclic cortical
reorganization during early childhood.
Brain Cogn. 20, 24–50 (1992).
70 Sterman MB, Mann CA, Kaiser DA,
Suyenobu BY. Multiband topographic EEG
analysis of a simulated visuomotor aviation
task. Int. J. Psychophysiol. 16, 49–56 (1994).
71 Kaiser DA. QEEG: state of the art or state
of confusion. J. Neurotherapy
4, 57–75 (2001).
72 Klimesch W, Schimke H, Pfurtscheller G.
Alpha frequency, cognitive load and
memory performance. Brain Topogr.
5, 241–251 (1993).
www.future-drugs.com
Neurofeedback treatment of epilepsy
73 Brogden WJ. Animal studies of learning. • A recent useful introduction to
In: Handbook of Experimental Psychology. neurofeedback and other new perspectives
Stevens SS (Ed.). John Wiley & Sons, NY, on the treatment of various neurological
USA, 568–612 (1951). and psychiatric conditions.
74 Ferster CB, Skinner BF. In: Schedules of
Reinforcement. Appleton-Century-Crofts, Websites
NY, USA. (1957). 101 The Biofeedback Certification Institute of
75 Felsinger JM, Gladstone AL, Yamaguchi America (BCIA).
HG, Hull CL. Reaction latency (StR) as a www.bcia.org/
function of the number of reinforcements. 102 International Society for
J. Exp. Psychol. 37, 214–228 (1947). Neuronal Regulation.
76 Grice GR. The relation of secondary www.isnr.org/index.html
reinforcement to delayed reward in visual
discrimination learning. J. Exp. Psychol. Affiliations
38, 1–16 (1948).
• Tobias Egner, PhD
77 Pearce JM, Hall G. Overshadowing the Post-doctoral Research Fellow, Functional MRI
instrumental conditioning of a lever-press Research Center, Columbia University,
response by a more valid predictor of Neurological Institute Box 108, 710 West 168th
reinforcement. J. Exp. Psychol. Anim. Behav. Street, New York, NY 10032, USA
Process. 4, 356–367 (1978). Tel.: +1 212 342 0121
Fax: +1 212 342 0855
78 Williams BA. Associative competition in
te2111@columbia.edu
operant conditioning: blocking the
response-reinforcer association. Psychon. • M Barry Sterman, PhD
Professor Emeritus, Department of Neurobiology,
Bull. Rev. 6, 618–623 (1999).
UCLA, 315 S. Beverly Drive, Suite 409, Beverly
79 Hirshberg LM, Chiu S, Frazier JA. Emerging Hills, CA 90212, USA
brain-based interventions for children and Tel.: +1 310 273 6379
adolescents: overview and clinical perspective. Child Fax: +1 310 273 6340
Adolesc. Psychiatr. Clin. N. Am. 14, 1–19 (2005). msterman@ucla.edu.
257