Self-regulation of Slow Cortical Potentials: A New Treatment for Children With Attention-Deficit/Hyperactivity Disorder

Ute Strehl, PhDa, Ulrike Leins, PhDb, Gabriella Goth, MDa, Christoph Klinger, MDa, Thilo Hinterberger, PhDa, Niels Birbaumer, PhDa,c

a Institute of Medical Psychology and Behavioral Neurobiology
b University Hospital for Psychiatry and Psychotherapy, University of Tübingen, Tübingen, Germany
c Human Cortical Physiology, National Institutes of Health, National Institute of Neurological Disorders and Stroke, Bethesda, Maryland


OBJECTIVE.: We investigated the effects of self-regulation of slow corticalpotentials for children with attention-deficit/hyperactivitydisorder. Slow cortical potentials are slow event-related direct-currentshifts of the electroencephalogram. Slow cortical potentialshifts in the electrical negative direction reflect the depolarizationof large cortical cell assemblies, reducing their excitationthreshold. This training aims at regulation of cortical excitationthresholds considered to be impaired in children with attention-deficit/hyperactivitydisorder. Electroencephalographic data from the training andthe 6-month follow-up are reported, as are changes in behaviorand cognition.

METHOD.: Twenty-three children with attention-deficit/hyperactivity disorderaged between 8 and 13 years received 30 sessions of self-regulationtraining of slow cortical potentials in 3 phases of 10 sessionseach. Increasing and decreasing slow cortical potentials atcentral brain regions was fed back visually and auditorily.Transfer trials without feedback were intermixed with feedbacktrials to allow generalization to everyday-life situations.In addition to the neurofeedback sessions, children exercisedduring the third training phase to apply the self-regulationstrategy while doing their homework.

RESULTS.: For the first time, electroencephalographic data during thecourse of slow cortical potential neurofeedback are reported.Measurement before and after the trials showed that childrenwith attention-deficit/hyperactivity disorder learn to regulatenegative slow cortical potentials. After training, significantimprovement in behavior, attention, and IQ score was observed.The behavior ratings included Diagnostic and Statistical Manualof Mental Disorders criteria, number of problems, and socialbehavior at school and were conducted by parents and teachers.The cognitive variables were assessed with the Wechsler IntelligenceScale for Children and with a computerized test battery thatmeasures several components of attention. All changes provedto be stable at 6 months’ follow-up after the end of training.Clinical outcome was predicted by the ability to produce negativepotential shifts in transfer sessions without feedback.

CONCLUSIONS.: According to the guidelines of the efficacy of treatments, theevidence of the efficacy of slow cortical potential feedbackfound in this study reaches level 2: “possibly efficacious.”In the absence of a control group, no causal relationship betweenobserved improvements and the ability to regulate brain activitycan be made. However, it could be shown for the first time thatgood performance in self-regulation predicts clinical outcome.”Good performance” was defined as the ability to produce negativepotential shifts in trials without feedback, because it is knownthat the ability to self-regulate without feedback is impairedin children and adults with attention problems. Additional researchshould focus on the control of unspecific effects, medication,and subtypes to confirm the assumption that slow cortical potentialfeedback is a viable treatment option for attention-deficit/hyperactivitydisorder. Regulation of slow cortical potentials may involvesimilar neurobiological pathways as medical treatment. It issuggested that regulation of frontocentral negative slow corticalpotentials affects the cholinergic-dopaminergic balance andallows children to adapt to task requirements more flexibly.


Key Words: ADHD • biofeedback • neurobehavioral outcome • EEG • electroencephalogram

Abbreviations: ADHD—attention-deficit/hyperactivity disorder • ES—effect size • EEG—electroencephalogram/electroencephalographic • SCP—slow cortical potential • DSM-IV—Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition


Despite the widespread use of stimulant medication for attention-deficit/hyperactivitydisorder (ADHD), there is a strong demand for improving treatmentof ADHD.1 A number of concerns accompany the use of stimulantmedication. Approximately 25% of children’s conditions failto respond favorably to stimulant medication.2 Adverse effectsof stimulant medications include reduced growth,3 sleep disorders,decreased appetite, stomach pain, headache, and, in some cases,tics.4 There is no evidence of long-term efficacy of stimulantsfor ADHD. Results from the Multimodal Treatment Study of ADHD3,5show that effect sizes (ESs) of medication management and ofcombined treatment (medication and behavior therapy) comparedwith behavior therapy and community care 10 months after theend of treatment are small (0.30 for ADHD symptoms and 0.21for oppositional defiant disorder symptoms). Children who discontinuedmedication experienced considerable loss of improvement at follow-up.

Neurofeedback as an additional or alternative treatment is basedon pathophysiological changes that are characteristic of ADHD.Children with ADHD, compared with nonclinical controls, showelectroencephalographic (EEG) slowing in prefrontal regions6and smaller brain volumes, especially in the basal ganglia andcerebellum.7 Since the mid-1970s Lubar and Shouse8 have trainedchildren to regulate their brain states through EEG biofeedbackto reduce the symptoms of ADHD. They provided the first publishedEEG biofeedback (neurofeedback) study and attempted to normalizeEEG patterns: Participants were rewarded for increasing thesensorimotor rhythm (12–14 Hz) at motor brain areas anddecreasing theta frequency (4–7 Hz). In a series of casestudies this method was shown to be successful in improvingEEG spectra during cognitive tasks and promoting performanceon intelligence tests and scores of attention, academic performance,and social behavior.9

Despite these promising results, EEG biofeedback has not beenconsidered a standard therapy for ADHD. Until recently, therewere only a few controlled studies of the efficacy of neurofeedbackfor ADHD, and these studies had methodologic shortcomings suchas lack of controls or inadequate controls, no randomization,and no long-term follow-up.10,11 Although changes in cognitionand behavior are reported to last 10 to 24 months after treatment,12these results are difficult to interpret, because no EEG datawere presented or data were assessed (often posthoc) in clinicalsettings. In 2 recently published controlled studies it wasshown that neurofeedback leads to the same improvements as medication13and that effects of a combined treatment of medication, parentalcounseling, and neurofeedback last after washout of medication.14In contrast, the effects of a combination of medication andparental counseling did not continue after medication washout.

Whereas the rationale for these studies was based on the modificationof oscillatory activity of the brain, there is only 1 studythat aimed at deficiencies observed in event-related EEG activity.Heinrich et al15 were the first to report feedback of slow corticalpotentials (SCPs) for children with ADHD and provided preliminaryevidence for positive behavioral and specific neurophysiologiceffects. SCPs are slow event-related direct-current shifts ofthe EEG, originating from the upper cortical layer.16 They lastfrom 0.3 seconds up to several seconds; they are not oscillatoryin nature but occur as a consequence of external or internalevents. They belong to the family of event-related brain potentials.It has been shown that SCP shifts in the negative directionreflect the depolarization of large cortical cell assemblies,reducing their excitation threshold. In patients with epilepsy,large negative potential shifts have been observed seconds beforea seizure and shifts toward electrical positivity immediatelyafter a seizure.17 In several studies, it was shown that voluntarycontrol of SCPs can be acquired by healthy populations18,19as well as by patients with drug-refractory epilepsy. Suppressionof negative SCP shifts significantly decreased seizures.20

In an earlier study, Rockstroh et al21 compared children withand without attention problems in their ability to voluntarilycontrol SCP. The children with attention problems were ableto modulate SCPs under feedback conditions but were not ableto modulate their SCPs without immediate and continuous feedbackin transfer conditions. In addition, children with attentionproblems had reduced cortical negativity at all electrode positionsin anticipation of a task, suggesting that failure to engagespecific cortical networks contributes to the performance decrement.

Children with attention disorders may be impaired in the regulationof excitation thresholds of the brain. In the study by Heinrichet al,15 less impulsivity errors in the continuous performancetest, less behavioral signs of ADHD (parents’ ratings), anda marked increase in the contingent negative variation was foundafter 25 sessions of SCP feedback. This result was interpretedas an improvement of mobilization of attentional resources anda neurophysiological correlate of improved self-regulatory capacities.However, no EEG data of the training sessions were reported.

We report here EEG data during learning and relate them to theclinical outcome. In addition, changes in behavioral and academicperformance were assessed 6 months after the end of treatment.


Participants were selected according to the following criteria:

  • age between 8 and 13 years;
  • ADHD inattentive or hyperactivetype or combined type accordingto the Diagnostic and StatisticalManual of Mental Disorders,Fourth Edition (DSM-IV);
  • no additionalneurologic disorder; and
  • full-scale IQ >80.


Patients were recruited from the outpatient clinic for psychotherapyat the University of Tübingen and from psychiatric practitioners.Parents and children signed informed-consent forms. ADHD wasassessed with several instruments:

  • Semistructured questionnaireof developmental and health history;
  • DSM-IV questionnairesfor parents and teacher;
  • Eyberg Child Behavior Inventory22;
  • German translation of Conners’ Rating Scale23;
  • Kindl-Questionnairefor Measuring Health-Related Quality ofLife in Children andAdolescents, parents’ and children’s version24;
  • Testbatteriezur Aufmerksamkeitsprüfung, version 1.7,25a computerizedtest battery that measures several componentsof attention;and
  • German version of the Wechsler Intelligence Scale forChildren:Hamburg-Wechsler-Intelligenztest für Kinder.26


With the exception of the above-mentioned questionnaire forassessment of developmental and health history, all instrumentswere used after treatment and at follow-up.

The study was conducted in accordance with the convention ofHelsinki and approved by the local ethics committee of the facultyof medicine.

Neurofeedback: Training of SCPs
During a training session, the participants’ EEGs were recordedat Cz, referred to 2 mastoid electrodes shunted over a 10-k{Omega}resistance. Electrode positions were prepared with a cleaningpaste, and Ag/AgCl electrodes were filled with a conductivepaste (Elefix; Bio-Medical Instruments, Inc, Warren, MI). TheEEG amplifier (EEG 8, Contact Precision Instruments, Cambridge,MA) used a low-pass filter of 40 Hz, and the time constant wasset at 16 seconds. The brain signals were digitized with a samplingrate of 256 Hz. The slow-wave filter consisted of a 500-millisecondinterval moving window. The slow-wave amplitude immediatelybefore the active phase of the trial served as a baseline andwas set to 0 (see Fig 1).

Figure 1
View larger version (19K):
[in this window]
[in a new window]
FIGURE 1 Time course of a trial with baseline, task, and active phase. The curves are mean shifts of SCPs for all trials in 1 “run” (39 trials). The upper line indicates negative shifts; lower line, positive shifts. 

During the active phase, the slow-wave amplitude was calculatedevery 62.5 milliseconds as an average of the preceding 500 milliseconds.The position of the feedback signal (cursor, “ball”) correspondedto the difference between every 500-millisecond amplitude inthe active phase and the amplitude during the baseline. It wascorrected online for eye movements (for additional informationabout signal processing and artifact correction see refs 27and 28). Weber29 proved that respiration did not influence SCP-shifts.

Participants sat in a comfortable chair ~50 inches in front ofa portable computer. As shown in Fig 2, participants saw 2 rectangles(goal boxes) on the top and the bottom of the screen. A highlightedupper rectangle indicated a required SCP shift in the electricalnegative direction. A highlighted lower rectangle indicateda required positive SCP shift. Each trial lasted 8 seconds andwas divided into a 2-second passive phase and a 6-second activephase. Feedback consisted of a small (1-inch diameter) graphicsymbol (“ball”) that moved proportionally to the cortical shiftupward (negativity) or downward (positivity). Ball movementsstarted on the left edge of the screen and moved upward or downwardto the right edge. After successful trials a smiley face appeared(see Fig 2).

Figure 2
View larger version (32K):
[in this window]
[in a new window]
FIGURE 2 Screens: the screen on the left indicates the beginning of a trial, and that on the right indicates the end of a trial. Upper, screen during feedback trials; lower, screen during transfer trials. 

In addition, auditory feedback was given with a high-pitched(negativity trial) and a low-pitched (positivity) tone. A harmoniousjingle was introduced as positive reinforcement if the resultwas correct. As an additional reinforcement at the end of eachsession, the total number of smiley faces was exchanged fortokens. Whenever a certain amount of tokens was accumulated,they were exchanged for small toys, stickers, or other gifts(valued at ~1.50 Euro). Therefore, the number of available toyswas linked to performance.

Each session consisted of 3 to 5 runs, each run comprising 39trials. Trials with required negativity and required positivitywere presented randomly with a 50% probability during the first15 sessions. Thereafter, the proportion between negativity andpositivity tasks was 75% to 25%. To allow generalization toeveryday-life situations, trials with feedback were intermixedwith transfer trials in which no ball movement was shown (seeFig 2, lower). Although no continuous feedback was presentedin transfer trials, the smiley face provided (delayed) informationabout the success. An entire session lasted ~1 hour, includingthe time for preparation.

Participants were instructed that the aim of training was to”speed up their brain” to maintain concentration in situationsthat normally are difficult to attend (listening to somebodyelse, making plans, and sustained mental effort in tasks suchas homework, examinations, etc). The training was introducedas a computer game in which one can score goals by using one’sbrain. No specific instruction was given for how to score points;children were only advised to be attentive to the feedback andto find the most successful mental strategy to move the ballinto the required goal. Because there is no unique cognitivestrategy for the task,30 examples were given that have beenshown to be successful in at least some children. Between runs,therapists asked the subjects to verbalize strategies and encouragedthem to try new strategies or stick to the successful ones.

During a session, the trainer sat in a room next-door to thechild, connected by an intercom and video monitor. The trainerobserved the EEG signals online on a monitor, and a second monitorshowed the child. If necessary, the trainer could interveneby either using a 2-way intercom or joining the child.

Thirty training sessions were subdivided into 3 phases of 10sessions each. As shown in Fig 3, each phase lasted 2 weekswith daily training (5 days per week). The training was scheduledin the afternoon after school classes. Assessment procedures(pretraining/posttraining/6-month follow-up) as well as trainingsessions took place at the same time of the day.

Figure 3
View larger version (21K):
[in this window]
[in a new window]
FIGURE 3 Training schedule. 

Parents whose children were on ADHD medication were asked tomaintain a constant dose and intake. Between each treatmentphase, a 4- to 6-week break allowed the participants to practicethe strategies at home and record their daily practice. At theend of training, a 15 x 5-inch picture of a computer screenwith the ball and goal box (see Fig 2) was given as memory-aidhandout. Participants were instructed to carry it at all timesand use it whenever they needed a cue for the self-regulationstrategy. During the third training phase children exercisedcueing while doing their homework after the end of each trainingsession with the supervision of the trainer. The trainer wasinstructed to guide the child only in using the cue and notto assist in solving the particular cognitive tasks. Trainingand assessment procedures were implemented by either a licensedclinical psychologist or graduate students under the psychologist’ssupervision.

Data Analysis
EEG Data
For each child, mean differences of SCP amplitudes during bothtasks (positivity/negativity) for both conditions (feedback/transfer)were calculated. After testing the normal distribution of data,the difference between the tasks was determined separately foreach assessment point (pretraining, posttraining, follow-up)with an independent-samples t test. Bonferroni correction wasapplied to the levels of significance for multiple comparisons.A repeated-measures analysis of variance (first 2 sessions,last 2 sessions, 2 sessions at follow-up) examined effects oftime, task, and condition. A posthoc paired-samples test comparedmeasurement times separately in the case of a significant resultof analysis of variance. The analysis of variance was correctedwith Greenhouse-Geisser, posthoc tests with Bonferroni correction.

Psychometric Test Data
All data were analyzed with the same statistical procedure (repeated-measuresanalysis of variance) at the 3 assessment points. IQ scoreswere evaluated only twice (pretraining and follow-up) with apaired-samples test.

In addition to P values, ESs for P values of t tests were assessedwith Cohen’s d.31 ESs measure the magnitude of the effect andvary from ≥0.2 (small effect) to 0.5 (medium effect) and ≤0.8(large effect). Cohen’s d is computed as the difference betweenthe means (M1M2) divided by the pooled SD [{sigma}pooled ={surd}[{sigma}12{sigma}22/2]. ESs in analysis of variance (partial {eta}2) estimatethe proportion of variance in the dependent variable that isattributable to each effect.

Effects of Medication
To ensure a real-life clinical sample, children with and withoutmedication were included. As can be seen in Table 1, 5 of 23children used stimulants. To rule out possible effects of medication,an analysis of variance was conducted with a mixed model (2groups: 1 with and 1 without medication with 3 assessment points).Because no differences between groups were found, only datafrom the entire group are reported.


Twenty-five children took part, and all of them completed thetraining. Because 2 children who were not under medication atthe beginning of training were placed on medication after theend of training as a result of problems at school, their datawere excluded from the follow-up. According to their parents’judgment, one of these children showed no more hyperactivityand the other showed no more symptoms of inattention. Therefore,our data were not biased by this change in treatment. Five childrenreceived stimulant medication throughout therapy and follow-upperiods. The mean full-scale IQ score (Wechsler IntelligenceScale for Children) was 103, with a 10-point difference betweenverbal (108) and performance (98) scores.

Regulation of SCPs
Bonferoni-corrected differences between SCP amplitudes for negativityand for positivity tasks were close to significance at the endof training (t42 = 2.133; P = .078; ES = 0.64) for the feedbackcondition and significant at follow-up assessment (sessions32 and 33) for feedback (t40 = 2.749; P = .027; ES = 0.85) aswell as for transfer trials (t40 = 2.814; P = .024; ES = 0.87).As shown in Fig 4 for feedback conditions and Fig 5 for transferconditions, children were not able to produce potential shiftsaccording to the task requirement at the beginning of training.Instead, the potentials were negative when positivities wererequired, and they were negative when a positive shift was askedfor. At the end of training and at the follow-up assessmentthe shifts were as required, and the main effects were causedby responses in the negativity task. A repeated-measures analysisof variance with the factors task and condition for feedbacktrials revealed a significant effect of time (F2,40 = 4; P =.046; ES = 0.17) and of the interaction between time and task(F2,40 = 10.8; P = .001; ES = 0.35). For transfer trials a significanteffect for the interaction between time and task was present(F2,40 = 4,79; P = .016; ES = 0.2). Changes of mean amplitudeswere also calculated with a general linear model (repeated measures)separately for negativity and positivity tasks. Although therewere no significant effects for positivity trials, amplitudesfor negativity trials changed significantly over time (F2 =16.6; P = .000; ES = 0.45) in feedback conditions and in transfertrials (F2 = 5.9; P = .006; ES = 0.23). Posthoc tests revealedthat amplitudes differed significantly between sessions 2 +3 and sessions 29 + 30 (feedback: t21 = 3.5; P = .004; ES =0.93; transfer: t21 = 2.54; P = .038; ES = 0.73) as well asbetween sessions 2 + 3 and follow-up (feedback: t20 = 5.592;P = .000; ES = 1.1; transfer: t20 = 3.399; P = .009; ES = 0.95).

Figure 4
View larger version (16K):
[in this window]
[in a new window]
FIGURE 4 Mean amplitudes in negativity trials and positivity trials with feedback during the first sessions, the last sessions, and during the follow-up assessment. 

Figure 5
View larger version (12K):
[in this window]
[in a new window]
FIGURE 5 Mean amplitudes in negativity trials and positivity trials without feedback (transfer) during the first sessions, the last sessions, and during the follow-up assessment. 

Behavior ratings of parents showed a significant reduction ofproblems as assessed by the Eyberg questionnaire (F2 = 4.478;P = .02; ES = 0.18). A paired-samples test revealed that thechange was observed between baseline and the end of training(P = .018; ES = 0.47). Despite this result, the impact of problemsdid not change according to the parents’ opinion. As shown inFig 6, the number of problems decreased from 149 to 138. Scoresof <127 are considered as normal.

Figure 6
View larger version (10K):
[in this window]
[in a new window]
FIGURE 6 Means and SDs of numbers of Behavior problems (parents’ ratings). a P< .05. The clinical cutoff value of the Eyberg questionnaire is 127. 

Scores of the Conners’ Rating Scale yielded a significant improvement(F2 = 3.98; P = .03; ES = 0.16) that was attributed to the differencebetween pretesting and follow-up (posthoc paired-samples test:t22 = 2.56; P = .054; ES = 0.62). Mean values decreased from53.6 to 46.0 to 42.0 (see Fig 7). Scores <45 are considerednonpathologic.

Figure 7
View larger version (11K):
[in this window]
[in a new window]
FIGURE 7 Means and SDs of Conners’ Rating Scale (parents’ ratings). a P< .05. The clinical cutoff value is 45. 

Parents’ ratings of DSM-IV criteria were close to significancefor inattention (F2 = 3.43; P = .056; ES = 0.14). The changesin diagnosis for the whole group are shown in Table 2.

View this table:
[in this window]
[in a new window]
TABLE 2. Diagnosis Pretraining, Posttraining, and at Follow-up Assessment for 23 Children

Two of 21 children at the end of training and 3 of 19 childrenat the follow-up evaluation no longer fulfilled the diagnosticcriteria for ADHD. Four children with ADHD were diagnosed ashaving ADHD, predominantly inattentive type only, and four werediagnosed as having ADHD, predominantly hyperactive type only.One of 5 children receiving stimulants reduced medication, andanother withdrew from taking the medication. Fisher’s exacttest revealed a significant difference between pretesting andfollow-up in the distribution of children within the diagnosticcategories (ADHD, predominantly inattentive type; ADHD, predominantlyhyperactive type; and below cutoff) (P = .033). The changesbetween a positive ADHD diagnosis to no diagnosis at all frompretesting to follow-up were close to significance (P = .06).

Teachers rated significant improvements in inattention (F2 =4.55; P = .032; ES = 0.19), hyperactivity (F2 = 7.11; P = .003;ES = 0.27), impulsivity (F2 = 4; P = .034; ES = 0.17), and socialbehavior (F2 = 7.1; P = .002; ES = 0.26). No changes were reportedfor the self-worth, emotionality, or academic-achievement scales.Mean scores and SDs are shown in Fig 8. For all subscales scores,<3 is considered nonpathologic.

Figure 8
View larger version (14K):
[in this window]
[in a new window]
FIGURE 8 Teachers’ rating before training, after training, and at the follow-up assessment. a P<.05. 

Posthoc paired-samples tests revealed significant differencesbetween assessment points: for inattention, baseline comparedwith follow-up (t20 = 1.1; P = .048; ES = 0.55), for hyperactivity,baseline compared with end of training (t22 = 2.18; P = .08;ES = 0.28) and with follow-up (t20 = 4.3; P = .000; ES = 0.59),for impulsivity, baseline compared with end of training (t22= 2.97; P = .021; ES = 0.36), and for social behavior, baselinecompared with follow-up (t20 = 3.52; P = .006; ES = 0.64) andend of training compared with follow-up (t20 = 2.56; P = .038;ES = 0.45).

IQ and Attention
Performance IQ scores changed significantly from screening tofollow-up (t22 = –2.76; P = .011; ES = 0.35), whereasfull-scale and verbal IQ changes were not significant.

Measures of attention were assessed with the Testbatterie zurAufmerksamkeitsprüfung. This test evaluates 12 variablesof attention for speed, omissions, and commissions. The datawere aggregated for 7 subtests below the 25th percentile andabove the 75th percentile. As shown in Fig 9, the number ofresults below average was significantly reduced (F2 = 17; P= .000; ES = 0.45). This improvement was observed from baselineto the end of training (t22 = 5.37; P = .000; ES = 0.68) andfrom the end of training to follow-up (t21 = 5; P = .000; ES= 0.72). Test results above average increased significantly(F2 = 8.67; P = .001; ES = 0.29). Here, improvements were observedfrom baseline to the end of training (t22 = –4.25; P =.000; ES = 0.63) and from baseline to follow-up (t21 = –5.05;P = .000; ES = 0.85).

Figure 9
View larger version (28K):
[in this window]
[in a new window]
FIGURE 9 Attention scores below the 25th and above the 75th percentile in pretraining and posttraining and at the follow-up assessment. a P= .000. 

Health-Related Quality of Life
Neither parents nor children showed any changes in their ratingsof health-related quality of life. Compared with mean valuesof healthy children,32 the profile of the patients indicatedthat they were healthy. Self-regulation and Clinical Outcome
To determine if clinical outcome can be ascribed to acquisitionof self-regulation skills, EEG data from training were correlatedwith clinical outcome. Children with at least a 2-point reductionin either hyperactivity or inattention criteria of DSM-IV wereclassified as “improved.” Successful acquisition of self-regulationwas defined on the basis of negativity trials without feedbackduring the third training phase. Means of amplitudes were dividedby the SE for each child, and the median of these means wasused to separate successful from unsuccessful regulators. Thedifference between group means of amplitudes in negativity trialswithout feedback (successful regulators = –5.27 µV;unsuccessful regulators = –0.051 µV) was highlysignificant (t test for independent samples: t12 = –6.58;P = .000). Pearson’s {chi}2 revealed a significant association betweensuccessful self-regulation and clinical improvement at the endof training ({chi}2 = 5.24; degrees of freedom = 1; P = .022). Thisassociation was close to significance at follow-up ({chi}2 = 2.93;degrees of freedom = 1; P = .087).


We report the first, to our knowledge, EEG data during the courseof self-regulation of SCPs. There is clear evidence that childrenlearn to control SCPs. Furthermore, this ability remains stableafter the end of training without booster sessions. The resultsof this study show that children with ADHD are able to learnregulation of slow negative brain potentials (with and withoutfeedback). The assumption that patients with frontal deficitsor lesions and ADHD are not able to self-regulate brain activityrelated to attention was not confirmed.21 Because this earlierstudy did not include as many sessions as our study, the numberof sessions might be an important variable. In addition, thetransfer exercises between training phases and after trainingsessions may have contributed to this result.

As can be seen from Fig 4, children did not produce reliablypositive potentials; even during required positivity all potentialswere negative, although smaller than during required negativity.In the transfer condition (see Fig 5), small positive potentialswere produced, but the difference to baseline did not reachsignificance. In contrast to self-regulation training for patientswith epilepsy,20 the children with ADHD controlled negativepotentials only. This could be the result of the more extendedtraining of negativity in this study. In a comparison betweenyoung (aged 20–28 years) and older (aged 50–64 years)healthy persons, Kotchoubey et al33 found in both groups potentialshifts in positivity trials still negative compared with baseline,although smaller than in negativity trials. Perhaps processingdemands of the task itself prevent subjects from producing largerpositive potentials. Subjects report that producing positivepotential shifts is more difficult and exhausting. Therefore,because they do not need this skill for the treatment of symptomsas in the case of patients with epilepsy, motivation to concentrateon this task might be reduced compared with the negativity task.As in these studies, the children were able to produce electrophysiologicaldifferential responses between the negativity (excitation) andpositivity (inhibition) tasks. The limitation is only that positive(supposedly inhibitory) responses did not reach positive values.Thus, the main goal of the training (self-regulation of excitationthresholds and cued increase of excitatory brain activity) wasachieved.

After such training, parents and teachers report reduction ofbehavioral problems, and test data show improvements in cognitiveperformance. Similar effects have been reported after neurofeedbacktraining in previous studies (eg, refs 13 and 14). In additionwe demonstrated that the improvements are stable 6 months afterthe end of training. ESs of the behavioral changes (between0.14 and 0.64) of attention (between 0.68 and 0.85) and IQ (0.35)are between medium and large. ESs for pretraining/posttrainingdata were not reported in previous studies, which makes a comparisonof outcomes difficult. The Multimodal Treatment Study of ADHD5reports ESs of 0.30 for the difference between the combinedtherapy of medication management and behavior therapy comparedwith behavior therapy and community care for ADHD ratings. Significantdifferences in outcome were caused by medication. In our study,medication did not affect outcome, but group sizes (5 childrenwith medication, 18 without) were rather small.

Although outcomes in attention, IQ, teachers’ ratings, and parents’ratings of Conners’ Rating Scale and number of problems (Eyberg)yielded moderate-to-high ESs, parents’ ratings of DSM-IV criteriashowed only small positive changes at the end of training, whichincreased at follow-up.

This mixed picture of parental judgment may reflect some ofthe problems of using rating scales in the diagnosis of ADHD.34Obviously, parents’ reports differed depending on the scalesthey used. An ES of 0.62 between pretesting and follow-up wasattained for Conners’ Rating Scale. Here, parents have to observetheir child for 3 consecutive days; ratings are given on 8 itemswith scores from 0 (“not at all”) to 3 (“very much”). On theother hand, the rating scale for DSM-IV criteria contains 40items, and parents can agree or disagree with each item. Accordingto the Conners’ Rating Scale, group data are below the clinicalcutoff at follow-up, whereas the categorical judgment with theDSM-IV rating scale yielded much less improvement (ES = 0.14).It may be easier to make a decision that a symptom has weakenedthan that is has disappeared altogether. It is important tonote that the cutoff for the 8 items on Conners’ Rating Scaleis 15 (for a 3-day observation period: 45).

One could further speculate that the small effects assessedwith the DSM-IV scale are related to factors immanent to neurofeedbacktraining. This kind of training may initiate a learning processthat needs time and practice to result in perceivable changesof behavior in complex situations. The same argument might bevalid regarding the result that teachers did not see changesin academic achievement, although they rated attention as improvedand hyperactivity and impulsivity as reduced. Teachers may hesitateto rate academic improvements before the final examinations,which occurred after the 6-month follow-up. A follow-up ≥12 monthsafter the end of training is in progress and should shed lighton this hypothesis.

In the absence of a comparable control group, no conclusionsof causal relationships between improvement in behavior andcognition and the ability to regulate brain activity can bemade. However, it could be shown for the first time that theability to produce potential shifts in negativity trials withoutfeedback predicts clinical outcome. It was our intention toshow in this first experiment that children with ADHD can self-regulatetheir SCPs and that ESs are substantial and comparable to othertypes of treatment. Blinding of therapists and patients is unethicalin most psychiatric, psychological, and even psychopharmacologicstudies. Margraf et al35 demonstrated in a randomized, double-blindcomparison of an antidepressant, a minor tranquilizer, and placebothat the great majority of patients as well as their physicianswere able to rate accurately whether the active drug or a placebohad been given. Thus, even drug studies cannot be double-blindedin many cases because patients and therapists will perceivepositive or negative therapeutic effects. The use of a controlgroup for neurofeedback has comparable limitations: false feedback,for example, is usually detected by patients and leads to adverseeffects.36 Even a waiting-list condition does not control forunspecific effects; in expectation of a therapeutic intervention,hope of success may induce changes in behavior. A control groupwith psychopharmacologic treatment is extremely difficult tocompare with an attention-demanding, highly interactive treatmentsuch as the neurofeedback training used in our study. Consideringthe problems of controlling for unspecific effects, the predictionof clinical outcome by variables of electrophysiology is a viablealternative. The improvements shown are comparable to othereffective treatments such as pharmacologic interventions andbehavioral treatment, as reported above. Assuming that thesestudies controlled for placebo effects, the conclusion thatour new treatment approach showed comparable efficacy to thereports in the literature seems to be acceptable.

The group in this pilot study was rather heterogeneous regardinggender, medication, and diagnosis. A study with more childrenshould clarify these questions. Moreover, more elaborate controlof unspecific effects, medication, gender, and subtypes is neededto confirm the assumption that training to self-regulate SCPsis a viable treatment modality for ADHD. Regulation of SCPsand medication may involve similar neurophysiological and biochemicalpathways; negative potentials of frontal brain areas reflectthe balance between cholinergic and dopaminergic activity.16Improvement of attentional regulation of frontocentral negativeSCPs should affect exactly this balance. With voluntary regulationof SCPs, children may learn to flexibly adjust their cholinergic-dopaminergicbalance to task requirements. We assume that the acquired skillbecomes automatic and, as a motor skill, is preserved withoutexplicit practice. The children use it flexibly, and successrewards and improves the skill, the behavior, and attentionbeyond the end of training.


This work was supported by the Deutsche Forschungsgemeinschaft,the Bundesministerium für Bildung und Forschung, the NationalInstitutes of Health, and the Medical Faculty of the Universityof Tübingen.We thank Nadine Danzer, Sonja Kaller, Georg Kane, Nicola Rumpf,Franziska Schober, and Cornelia Weber for valuable technicalhelp and Tracy Trevorrow for fruitful discussion of the manuscript.


Accepted May 26, 2006.

Address correspondence to Ute Strehl, PhD, Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Gartenstrasse 29, 72074 Tübingen, Germany. E-mail: ute.strehl{at}

The authors have indicated they have no financial relationshipsrelevant to this article to disclose.


1. National Institutes of Health. Diagnosis and treatment of attention deficit hyperactivity disorder. Natl Inst Health Consens Dev Conf Consens Statement. 1998:1–37. Available at: Accessed August 24, 2006

2. DuPaul GJ, Barkley RA, Connor DF. Stimulants. In: Barkley RA, ed. Attention-Deficit Hyperactivity Disorder: A Handbook for Diagnosis and Treatment. New York, NY: Guilford Press; 1998:510–551

3. MTA Cooperative Group. National Institute of Mental Health Multimodal Treatment Study of ADHD follow-up: changes in effectiveness and growth after the end of treatment. Pediatrics. 2004;113:762–769[Abstract/Free Full Text]

4. Chavez N, Hymian SE, Arons BS. Mental health: a report of the Surgeon General. 2002. Available at: Accessed August 24, 2006

5. MTA Cooperative Group. National Institute of Mental Health Multimodal Treatment Study of ADHD follow-up: 24-month outcomes of treatment strategies for attention-deficit/hyperactivity disorder. Pediatrics. 2004;113:754–761[Abstract/Free Full Text]

6. Barry RJ, Johnstone SJ, Clarke AR. A review of electrophysiology in attention-deficit/hyperactivity disorder: II. Event-related potentials. Clin Neurophysiol. 2003;114:184–198[CrossRef][Web of Science][Medline]

7. Castellanos FX, Acosta MT. The neuroanatomy of attention deficit/hyperactivity disorder [in Spanish]. Rev Neurol. 2004;38(suppl 1):S131–S136

8. Lubar JF, Shouse MN. EEG and Behavioral changes in a hyperkinetic child concurrent with training of the sensorimotor rhythm (SMR): a preliminary report. Biofeedback Self Regul. 1976;1:293–306[CrossRef][Web of Science][Medline]

9. Lubar JO, Lubar JF. Electroencephalographic biofeedback of SMR and beta for treatment of attention deficit disorders in a clinical setting. Biofeedback Self Regul. 1984;9:1–23[CrossRef][Web of Science][Medline]

10. Ramirez PM, Desantis D, Opler LA. EEG biofeedback treatment of ADD: a viable alternative to traditional medical intervention? Ann N Y Acad Sci. 2001;931:342–358[Web of Science][Medline]

11. Nash JK. Treatment of attention deficit hyperactivity disorder with neurotherapy. Clin Electroencephalogr. 2000;31:30–37[Web of Science][Medline]

12. Lubar JF. Neurofeedback for the management of attention deficit disorder. In: Schwartz MS, Andrasik F, eds. Biofeedback: A Practitioner’s Guide. New York, NY: Guilford Press; 2003:409–437

13. Fuchs T, Birbaumer N, Lutzenberger W, Gruzelier JH, Kaiser J. Neurofeedback treatment for attention-deficit/hyperactivity disorder in children: a comparison with methylphenidate. Appl Psychophysiol Biofeedback. 2003;28:1–12[CrossRef][Web of Science][Medline]

14. Monastra VJ, Monastra DM, George S. The effects of stimulant therapy, EEG biofeedback, and parenting style on the primary symptoms of attention-deficit/hyperactivity disorder. Appl Psychophysiol Biofeedback. 2002;27:231–249[CrossRef][Web of Science][Medline]

15. Heinrich H, Gevensleben H, Freisleder FJ, Moll GH, Rothenberger A. Training of slow cortical potentials in ADHD: evidence for positive behavioral and neurophysiological effects. Biol Psychiatry. 2004;55:772–775[CrossRef][Web of Science][Medline]

16. Birbaumer N, Elbert T, Canavan AG, Rockstroh B. Slow potentials of the cerebral cortex and behavior. Physiol Rev. 1990;70:1–41[Free Full Text]

17. Ikeda A, Terada K, Mikuni N, et al. Subdural recording of ictal DC shifts in neocortical seizures in humans. Epilepsia. 1996;37:662–674[CrossRef][Web of Science][Medline]

18. Birbaumer N, Elbert T, Rockstroh B, et al. Biofeedback of event-related slow potentials of the brain. Int J Psychology. 1981;16:389–415[CrossRef]

19. Birbaumer N. Slow cortical potentials: plasticity, operant control, and behavioral effects. Neuroscientist. 1999;5:74–78[Abstract/Free Full Text]

20. Kotchoubey B, Strehl U, Uhlmann C, et al. Modification of slow cortical potentials in patients with refractory epilepsy. Epilepsia. 2001;42:406–416[CrossRef][Web of Science][Medline]

21. Rockstroh B, Elbert T, Lutzenberger W, et al. Biofeedback: evaluation and therapy in children with attentional dysfunctions. In: Rothenberger A, ed. Brain and Behavior in Child Psychiatry. Berlin, Germany: Springer; 1990:345–355

22. Eyberg SM, Pincus D. Eyberg Child Behavior Inventory & Sutter-Eyberg Student Behavior Inventory–Revised. Odessa, FL: Psychological Assessment Resources; 1999

23. Conners CK. Conners’ Rating Scales–Revised: Technical Manual. North Tonawande, NY: Multi-Health Systems; 1997

24. Ravens-Sieberer U. The KINDL Questionnaire for Measuring Health Related Quality of Life in Children and Adolescents–Revised Version. In: Schuhmacher J, Klaiberg A, Brähler E, eds. Assessment of Quality of Life and Well-being [in German]. Göttingen, Germany: Hogrefe; 2003:184–188

25. Zimmermann P, Fimm B. Testbatterie zur Aufmerksamkeitsprüfung (TAP). Version 1.7. Herzogenrath, Germany: Psychologische Testsysteme; 2002

26. Tewes U, Rossman P, Schallberger U. Hamburg Wechsler Intelligenztest für Kinder–Dritte Auflage (HAWIK III). Bern, Germany: Huber; 1999

27. Kübler A, Winter S, Birbaumer N. The thought translation device. In: Schwartz S, Andrasik F, eds. Biofeedback. 3rd ed. New York, NY: Guilford Press; 2003:471–481

28. Kotchoubey B, Blankenhorn V, Froscher W, Strehl U, Birbaumer N. Stability of cortical self-regulation in epilepsy patients. Neuroreport. 1997;8:1867–1870[Web of Science][Medline]

29. Weber C. Fact or artefact? Respiration as an artefact in SCP-Biofeedback [Thesis; in German]? Tübingen, Germany: Diplomarbeit am Institut für Psychologie der Eberhard-Karls-Universität Tübingen; 2003

30. Roberts LE, Birbaumer N, Rockstroh B, Lutzenberger W, Elbert T. Self-report during feedback regulation of slow cortical potentials. Psychophysiology. 1989;26:392–403[Web of Science][Medline]

31. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988

32. Ravens-Sieberer U, Bettge S, Erhardt D. Quality of life in children and adolescents: results from a population based survey [in German]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz. 2003;4:340–345[CrossRef]

33. Kotchoubey B, Haisst S, Daum I, Schugens M, Birbaumer N. Learning and self-regulation of slow cortical potentials in older adults. Exp Aging Res. 2000;26:15–35[CrossRef][Web of Science][Medline]

34. Swanson JM, Kraemer HC, Hinshaw SP, et al. Clinical relevance of the primary findings of the MTA: success rates based on severity of ADHD and ODD symptoms at the end of treatment. J Am Acad Child Adolesc Psychiatry. 2001;40:168–179[CrossRef][Web of Science][Medline]

35. Margraf J, Ehlers A, Roth WT, et al. How “blind” are double-blind studies? J Consult Clin Psychol. 1991;59:184–187[CrossRef][Web of Science][Medline]

36. Birbaumer N, Elbert T, Rockstroh B, et al. Clinical-psychological treatment of epileptic seizures: a controlled study. In: Ehlers A, Florin I, Fiegenbaum W, et al, eds. Perspectives and Promises of Clinical Psychology. New York, NY: Plenum Press; 1992:81–96