A new emerging research with novel insights by Glaubig and colleagues reviews a part of neurofeedback that is often discussed clinically but less often parsed carefully in the pediatric literature: slow-oscillation neurofeedback. Rather than grouping all slow EEG work together, the review separates three related but distinct methods—slow cortical potential (SCP), infra-low frequency (ILF), and infra-slow fluctuation (ISF) neurofeedback—across children and adolescents with ADHD, autism spectrum disorder (ASD), epilepsy, tic disorders, and eating-related concerns.

Biofeedback is a method that helps individuals learn to regulate physiological processes by providing real-time information about signals such as heart rate, respiration, muscle tension, skin conductance, or temperature. Neurofeedback is a specialized form of biofeedback in which real-time brain activity, most commonly measured with EEG, is translated into visual, auditory, tactile, or combined feedback so the individual can learn to modify neural activity.

In this review, SCP neurofeedback refers to training of very slow EEG voltage shifts, typically below 0.5 Hz, that are interpreted as changes in cortical excitability. Negative SCP shifts are generally associated with increased cortical excitability and preparation for cognitive or behavioral action; positive shifts are associated with reduced excitability, inhibition, or rest. SCP training is usually explicit and task-based: the child is asked to intentionally produce activation or inhibition while feedback indicates the direction of the shift.

ILF neurofeedback targets even slower EEG activity below approximately 0.1 Hz. It is usually framed less as training a discrete frequency band and more as supporting arousal regulation, emotional stability, autonomic balance, and whole-brain state stabilization. A central clinical concept is the optimal response frequency, the individualized infra-low frequency at which the person appears to show the most favorable regulatory response.

ISF neurofeedback targets infra-slow fluctuations in the 0.01–0.1 Hz range. These rhythms are thought to relate to brain state transitions, resting-state connectivity, and large-scale functional integration. Importantly, ISF activity may reflect not only neuronal activity but also glial, vascular, metabolic, and autonomic contributions. That complexity is not a weakness by default—but it does mean interpretation requires technical humility.

The review's central message is promising but cautious: slow-oscillation neurofeedback may support pediatric self-regulation, especially in ADHD, but the three methods differ substantially in mechanism, measurement, feedback timing, and evidentiary maturity.

Methods

Glaubig and colleagues conducted a structured narrative review of peer-reviewed literature on slow-oscillation neurofeedback in pediatric clinical populations. Searches were performed in PubMed and Google Scholar through May 2025. The search terms included "Neurofeedback," "Slow," "Low," "Infra-Low-Frequency," and "Infra-slow," each paired with "Children." The authors excluded duplicate records, review papers, studies with misleading terminology, and one study involving only healthy participants. The final review included 36 original research articles or case reports, organized by diagnostic category, participant characteristics, study design, neurofeedback modality, and reported clinical change.

The 36 included studies were unevenly distributed across both diagnostic categories and protocol families. ADHD accounted for 24 of the 36 — including 18 randomized controlled trials or RCT-derived analyses, one comparative controlled study, four uncontrolled pre/post designs, and one qualitative study — and was almost entirely represented by SCP protocols, with the lone exception of one large uncontrolled pre/post ILF report. ASD accounted for seven studies, more evenly spread across the three protocol families: three RCTs and two clinical trials (predominantly SCP), plus two case reports (one ILF, one ISF). Epilepsy contributed two studies (one RCT and one case report, with SCP and infra-slow approaches both represented), tic disorders contributed two (a case report and a clinical case series, both ILF), and eating-related concerns contributed one ILF controlled pilot in pediatric obesity. Across the full set, this works out to roughly 22 RCT or RCT-derived articles, two clinical trials, five case reports or case series, four uncontrolled pre/post studies, one comparative study, one qualitative study, and one pilot study.

A caveat worth foregrounding: these counts refer to published articles, not independent trials. Several ADHD and ASD papers are follow-up analyses, secondary analyses, or reports from overlapping cohorts — the Gevensleben group generated four publications from a single RCT cohort; the Hasslinger group generated two from another. The "18 RCTs in ADHD" headline is therefore better read as 18 RCT-labeled publications drawn from a smaller number of underlying trials — perhaps 11 or 12 independent studies, depending on how one counts. The signal is real, but it is concentrated in fewer hands than the raw counts suggest.

The three protocol families also differ meaningfully in study design. SCP studies most often use unipolar montages over central sites such as Cz with mastoid or earlobe references, although bipolar placements appear when the goal is to train functional relationships between cortical sites. ILF studies use unipolar or bipolar montages, including cross-hemispheric placements such as T3–T4, and are typically organized around the optimal response frequency rather than a fixed band. ISF studies typically use full-band EEG recording with bipolar montages. Across all three families, the review highlights a shared theme: slow-oscillation training is not simply about "more" or "less" of a frequency band, but about modulating the regulatory architecture that supports attention, inhibition, arousal, sleep, emotional balance, and behavioral control.

Results

The clearest way to read the results is by method rather than by assuming that all slow-oscillation neurofeedback belongs to one family.

SCP neurofeedback

SCP neurofeedback carried the strongest evidence base. Most of the pediatric SCP literature focused on ADHD, where studies examined whether children could learn to regulate cortical excitability and whether that learning translated into improvements in attention, hyperactivity, impulsivity, or executive functioning.

Early work suggested that children with ADHD could learn to regulate SCPs and that clinical improvements in attention, cognition, and behavior could remain stable at six-month follow-up. Importantly, several studies found that outcomes were related to learning performance: children who successfully acquired SCP self-regulation tended to show greater symptom improvement. This learner/non-learner distinction is one of the review's most clinically useful findings.

Randomized controlled trials provided a mixed but clinically meaningful picture. Some studies found that SCP neurofeedback improved ADHD symptoms compared with active control conditions, including electromyography biofeedback or computerized attention training. Other trials found improvements in both neurofeedback and comparison groups, without clear superiority for SCP. Several studies comparing SCP neurofeedback with theta/beta training, self-management, working memory training, or cognitive training found broadly comparable outcomes. This makes interpretation delicate. The evidence supports the feasibility and potential benefit of SCP neurofeedback in ADHD, but it does not uniformly establish protocol-specific superiority across all studies.

Predictor studies added an important layer. Older age, stronger executive functioning, higher IQ, stable methylphenidate medication, and neurophysiological markers such as contingent negative variation and cue-related P3 activity were associated with better learning or better outcomes in some studies. Again, the question becomes less "Does SCP work?" and more "Which children can learn this form of regulation, and what supports improve that learning?"

SCP neurofeedback was also studied in ASD and epilepsy. In ASD, SCP studies reported possible improvements in affective functioning, impulsivity, cognitive flexibility, and emotional regulation, though several effects were not clearly specific to neurofeedback compared with standard treatment or counseling. One Bayesian analysis found treatment-specific improvements in ASD-related parameters after SCP training. In focal epilepsy, an RCT compared SMR neurofeedback, SCP neurofeedback, and sham across 44 adolescents (Morales-Quezada et al. 2019). SMR emerged as the cleaner story for seizure reduction; SCP showed only limited sustained benefit, and the broader quality-of-life gains were partly attributed to placebo.

ILF neurofeedback

ILF neurofeedback was represented by fewer studies, but the reviewed domains were broader. In ADHD, one large pre/post study reported improvement after approximately 39 sessions. In ASD, ILF evidence included a randomized controlled trial suggesting improved inhibitory control in high-functioning adolescents and a case report describing reductions in physical symptoms, sleep disturbance, and some behavioral symptoms in a young child.

ILF was also the main slow-oscillation method represented in tic disorders. One case report and one clinical case series described reductions in tic severity and improved functioning. In pediatric obesity, a controlled pilot study reported subjective improvements in appetite regulation, attention, and sleep, although the neurofeedback group lost less weight than the control group. That finding is a useful reminder that subjective regulatory gains and primary medical outcomes do not always move in the same direction.

Taken together, ILF findings are clinically intriguing but not yet definitive. The method may be especially relevant for arousal instability, sleep disruption, sensory regulation, emotional reactivity, and tic-related self-regulation. However, the evidence base remains constrained by uncontrolled designs, case reports, heterogeneous implementation, and limited replication.

ISF neurofeedback

ISF neurofeedback had the smallest pediatric evidence base. The review included a small ASD case report (Smith et al. 2016) and a case series of three pediatric epilepsy patients undergoing ISF NF (Legarda et al. 2011). Reported changes included improvements in sleep, headaches, medication use, seizure control, and broad functioning. These observations are encouraging, but they cannot establish efficacy.

The major issue with ISF is not that it lacks clinical plausibility; rather, the method sits at the edge of what is technically and mechanistically easy to interpret. Infra-slow activity overlaps with neural, glial, vascular, metabolic, and autonomic processes, and it requires careful recording methods. At present, ISF should be described as promising and experimental in pediatric clinical populations.

Across methods, the result pattern is encouraging but uneven: SCP neurofeedback has the most pediatric evidence, especially for ADHD; ILF is emerging across a wider range of regulation-related clinical concerns; and ISF remains preliminary, with case-based evidence only.

Discussion

This review is useful because it slows down our thinking about slow neurofeedback. SCP, ILF, and ISF are sometimes discussed as if they are variations on the same intervention, but clinically and technically they are quite different.

SCP neurofeedback is the most research-ready of the three. It has clearer task structure, clearer outcome logic, and a longer pediatric evidence base, especially in ADHD. The child is generally trained to intentionally shift cortical excitability in one direction or another. That makes SCP attractive for attentional control, inhibitory control, and preparation for action. It also makes SCP dependent on developmental capacities: understanding the task, sustaining motivation, experimenting with strategies, tolerating frustration, and eventually transferring the skill without immediate feedback.

ILF neurofeedback asks a different clinical question. Instead of training a child to produce a discrete activation or inhibition shift during a task, ILF is usually framed around stabilizing state regulation. That may explain why the reported applications cluster around sleep, arousal, impulsivity, sensory dysregulation, tics, and emotional reactivity. These are clinically meaningful targets, but the evidence remains less mature. The optimal response frequency model is appealing because it fits real-world individualization, yet it also creates major research challenges. When the protocol is individualized session by session, standardization becomes harder, and replication becomes more demanding.

ISF neurofeedback extends this complexity further. It aims at ultra-slow activity related to large-scale brain state transitions and functional integration. That is fascinating, especially in pediatric populations where regulation problems often involve shifting states: waking to sleeping, calm to reactive, focused to distracted, socially engaged to withdrawn. But the method still needs much stronger empirical support. Case reports are useful for generating hypotheses; they are not enough to guide broad clinical claims.

For referring professionals, this review is most useful as a quantitative anchor for what to ask. When a colleague recommends SCP neurofeedback for pediatric ADHD, the supporting evidence base — flawed but real — is now visible and can be discussed honestly. When a colleague recommends ILF or ISF for the same indication, the right clinical question is which peer-reviewed pediatric trials they are leaning on; this review now lets you ask that question with concrete backing. Families deserve to hear that "slow-oscillation neurofeedback" is not a single recommendation but at minimum three different ones, with very different levels of supporting evidence behind each.

For clients and families, the most practical message is that neurofeedback is a learning process, not a passive treatment applied to the brain. This is especially true for SCP training. The review repeatedly points toward individual learning capacity as a meaningful predictor of outcome. Children who learn to regulate the target signal tend to improve more. Children who do not learn may still benefit from structure, therapeutic attention, reward, expectancy, or complementary supports, but we should not assume that the intended neurophysiological mechanism has occurred.

For neurofeedback professionals, this raises a clinical responsibility: measure learning, watch transfer, and adapt. A child who cannot yet regulate SCPs may need preparatory work in breathing, HRV biofeedback, interoceptive awareness, sleep stabilization, parent-supported routines, or shorter shaping steps. A child with sensory sensitivity may need feedback adjusted before we interpret poor performance as non-response. A child with unstable arousal may not be ready for cognitively demanding SCP work and may require a more state-stabilizing approach first.

The methodological limitations are substantial. The reviewed studies vary widely in sample size, control condition, outcome measure, number of sessions, diagnostic profile, and protocol implementation. Some evidence comes from randomized controlled trials; other evidence comes from uncontrolled clinical trials, retrospective analyses, case reports, or case series. This creates a familiar tension in neurofeedback research: clinical practice often advances through individualized, adaptive protocols, while research demands standardization, blinding, and reproducibility. Both needs are legitimate. The field will not mature by abandoning individualization, but it also cannot rely indefinitely on small uncontrolled reports.

The best reading of this review is neither hype nor dismissal. It shows a field with genuine signal, especially in SCP neurofeedback for ADHD, but also a field that needs larger trials, better controls, clearer reporting, and stronger mechanistic work. Slow waves may matter. Now the task is to determine exactly when, how, and for whom they matter most.

Brendan's perspective

This review lands in an area where I think neurofeedback clinicians need to be both curious and a little bit stubborn about precision. "Slow-oscillation neurofeedback" sounds like a tidy category, but SCP, ILF, and ISF are not interchangeable. They may all live below the familiar EEG frequency bands, but they are doing very different clinical and technical things.

Let's start with SCP neurofeedback, because it has the best pediatric evidence here and also the clearest conceptual tension. SCP training is often described as neurofeedback, and it is, but it is not neurofeedback in quite the same way as reinforcing SMR at Cz or inhibiting high beta at Fz. SCP feedback is slow. The signal unfolds over seconds. The participant is usually asked to intentionally generate a shift toward activation or inhibition. That means SCP training leans heavily on conscious self-regulation: strategy use, anticipation, effort, imagery, task engagement, and transfer practice.

That is not a flaw. In fact, for many children with ADHD, learning to intentionally prepare, inhibit, and shift state may be exactly the point. But it does change the learning model. Classic operant conditioning depends strongly on temporal contiguity: the reward needs to follow the target behavior closely enough for the nervous system to connect action and consequence. With SCP training, the relevant change is slow, and the feedback is necessarily delayed or temporally smeared compared with faster EEG protocols. That likely weakens the pure operant-conditioning component. The child may still learn, but much of the learning may be mediated through explicit strategy, cognitive control, motivation, therapist coaching, and repeated task practice.

In other words, SCP may be less like "the brain is rewarded instantly for producing a target rhythm" and more like "the child learns to recognize and intentionally shape a preparatory brain state using delayed physiological information." That distinction matters clinically. It means SCP training may be better suited to children who can understand the task, sustain effort, tolerate delayed reward, and experiment with internal strategies. It also explains why age, executive functioning, IQ, and learner status show up as relevant moderators. SCP is probably not simply something that happens to the child's brain; it is something the child learns to do.

From a clinical standpoint, I would therefore approach SCP as a structured self-regulation protocol. Cz-based training, CNV-related preparation, activation/inhibition trials, and transfer runs all make sense when the goal is attentional control, response preparation, and inhibitory regulation. But I would not assume that every child with ADHD is ready for that on day one. Some children may need shaping through easier feedback first: HRV biofeedback, respiration pacing, skin conductance awareness, or simpler EEG protocols with more immediate reinforcement. SCP may be powerful, but it asks quite a lot from the learner.

ILF and ISF raise a different set of questions. Clinically, I understand why these approaches are attractive. Many children we see are not struggling only with "attention" in a narrow cognitive sense. They are struggling with state regulation: falling asleep, waking up, recovering after frustration, shifting out of threat, tolerating sensory input, stopping tics, returning to baseline after conflict, or maintaining enough arousal to engage. ILF and ISF are appealing because they speak the language of state stability rather than symptom boxes.

Before I go further on what those approaches might offer clinically, I want to land a number that this review buries inside its Figure 2 but that deserves to be stated plainly. In the pediatric ADHD column, twenty-three of the twenty-four studies are SCP. One is an uncontrolled pre/post ILF report. Zero are ISF. That is the published peer-reviewed pediatric ADHD literature on slow-oscillation neurofeedback as of May 2025. It is wildly out of step with how confidently ILF and ISF are currently marketed for pediatric ADHD specifically — sometimes positioned as the more modern, more individualized, somehow-more-advanced choice for the very condition where the trial evidence supporting them is essentially absent. That is overreach in the most literal evidentiary sense: the marketing has run ahead of what published trials in the target population actually support. I want to be careful here. I am not arguing that ILF or ISF have no clinical value or no future, and I am not arguing against their use as part of carefully framed clinical exploration. I am arguing something more specific. For pediatric ADHD, anyone presenting them as evidence-based protocols of first choice is presenting something the published pediatric literature does not support. We are exploring this as an adjunct, here is the case-level evidence we have, here is what we will measure is a clinical conversation. This is the modern way to treat pediatric ADHD is a sales conversation in clinical clothing. We should be able to tell the difference.

But the technical complexities are also real. Measuring very slow EEG signals is not the same as measuring alpha or beta amplitude. Once we move below 0.1 Hz—and especially into direct-current or near-direct-current territory—we are entering a world where the signal may include many contributors: cortical postsynaptic potentials, glial activity, vascular dynamics, blood oxygenation relationships, autonomic rhythms, skin potentials, sweat gland activity, movement artifact, electrode polarization, impedance drift, thermal drift, amplifier behavior, reference effects, and filtering choices. That does not mean the signal is meaningless. It means the signal is physiologically rich and technically fragile.

This is where I think our language needs to be careful. When we train ILF or ISF, are we training "brainwaves"? Partly, perhaps. But we may also be interacting with slow regulatory dynamics that sit at the boundary of brain, body, vasculature, and autonomic state. That may actually be the clinical magic. A child's nervous system does not regulate attention separately from sleep, autonomic tone, sensory load, and emotional safety. The brain is not floating in a jar; it is metabolically, vascularly, and autonomically embedded. Slow signals may capture some of that embedded regulation better than faster frequency bands.

Still, clinical usefulness does not excuse sloppy interpretation. If a child sleeps better after ILF training, that is meaningful. If tics reduce, that is meaningful. If a family reports fewer meltdowns, that is meaningful. But we should not immediately conclude that we have normalized a specific neural generator unless the study design and recording method support that claim. ILF and ISF need better pediatric trials, clearer reporting of amplifiers and filters, electrode preparation standards, artifact handling, adverse-event monitoring, and follow-up. We also need mechanistic studies that can separate neural effects from peripheral physiological effects—or, perhaps more realistically, model how those systems interact.

In practice, I would place these approaches on a continuum. SCP is the most explicit and cognitive: useful when the child can participate in intentional self-regulation. ILF is more state-oriented: potentially useful when arousal instability, sleep, sensory dysregulation, or emotional volatility dominate the picture. ISF is the most experimental and system-level: fascinating, but requiring the most technical humility and the least marketing enthusiasm.

For pediatric neurofeedback, the review reinforces two clinical habits I trust deeply. First, train the child in front of you, not the diagnosis. Second, track whether the intended learning is actually happening. For SCP, that means examining self-regulation and transfer, not just symptom checklists. For ILF and ISF, it means tracking sleep, arousal, affective stability, sensory tolerance, tic frequency, medication changes, and day-to-day functioning with enough rigor that clinical impressions do not outrun the data.

My positive take is this: slow-oscillation neurofeedback may help us think more deeply about regulation. Not just attention. Not just inhibition. Regulation. The ability to shift, stabilize, prepare, recover, and rest. That is a beautiful target in pediatric care. But the slower the signal, the more careful we need to be — because in that slowness, we may be listening not only to the brain, but to the whole nervous system whispering through the EEG.

Conclusion

Glaubig and colleagues offer a timely synthesis of slow-oscillation neurofeedback in pediatric clinical care. The review suggests that SCP neurofeedback has the strongest current evidence base, particularly for ADHD, where several studies report meaningful improvements in attention, behavior, and self-regulation. At the same time, effects are not uniform, and clinical gains appear closely tied to whether the child actually learns to regulate the targeted neural activity.

For ASD, epilepsy, tic disorders, and eating-related concerns, the evidence is earlier and more tentative. ILF and ISF protocols are conceptually intriguing because they aim at deeper regulatory dynamics — arousal stabilization, sleep, state transitions, and large-scale integration — but the available pediatric literature remains too limited for strong clinical conclusions, and considerably thinner than current marketing for these methods would suggest, especially in pediatric ADHD.

The take-home is that slow-oscillation neurofeedback deserves a serious place in pediatric neurofeedback research and a thoughtful, honestly-framed place in clinical practice. Used carefully, individualized appropriately, presented to families with the evidence base named accurately, and integrated within broader care, it may help children build regulation from the bottom up — one slow shift at a time.

References

Glaubig, L., Azza, Y., Beber, S., Silbernagl, P., Barradas, I., Peer, A., & Tschiesner, R. (2026). Slow-oscillation neurofeedback: A narrative review on clinical efficacy in pediatric settings. Behavioral Sciences, 16(3), 337. https://doi.org/10.3390/bs16030337