Neurofeedback Learning: From Brainwaves to Lived Change

This post explores new emerging research with novel insights from Kerson, Sherlin, and Davelaar’s 2025 paper on neurofeedback, biofeedback, and basic learning theory. The authors revisit and update the classic 2011 framework on how neurofeedback works, bringing in newer perspectives from reinforcement learning, placebo research, and the neurobiology of dopamine and serotonin.

In simple terms, biofeedback and neurofeedback are methods that turn your internal physiology into information you can see or hear in real time, so you can gradually learn to change it. Sensors measure signals like brainwaves, heart rate variability, muscle activity, or skin conductance, and these signals are transformed into feedback – a game that moves, a tone that shifts, or a movie that plays more smoothly when your body or brain is heading in the desired direction.

What makes this paper so valuable is that it links these very practical training tools back to fundamental learning principles. It reminds us that brains don’t change just because we plug someone into an EEG or heart-rate sensor. They change when feedback is timely, specific, and meaningful enough for the nervous system to reorganise itself.

The authors also widen the lens beyond just EEG neurofeedback. They frame electrodermal, heart rate variability, surface EMG, and EEG-based approaches within a unified learning model, and they emphasise the often-undervalued role of subjective experience and the client–clinician relationship in driving long-term change.

Methods

Although this article appears in a scientific journal, it is best understood as a conceptual and integrative review rather than a traditional methods-and-results study. The authors’ "method" is to take decades of psychophysiology, learning theory, and neurofeedback practice and organise them into a practical map of how biofeedback and neurofeedback seem to work in real people over time.

They begin by walking through the main psychophysiological systems that clinicians actually train: electrodermal activity, heart rate variability, surface EMG, and EEG. Each modality is described less as a laboratory instrument and more as a window into how the nervous system maintains stability under stress.

Electrodermal activity is framed as a direct readout of sympathetic arousal – the way the skin’s conductance rises and falls with emotional and cognitive demands. Heart rate variability is treated as a marker of how flexibly the heart–brain–breathing loop can adapt, highlighting the role of the baroreflex and resonance breathing in restoring balance. Surface EMG becomes a tool for seeing how the motor system "holds on" through tension, guarding, and habitual postures, while EEG neurofeedback is situated as a way of nudging large-scale neural networks and oscillatory patterns toward more efficient, less noisy functioning.

From there, the paper turns to reinforcement and timing as the real engine under the hood. Instead of detailing a specific experimental protocol, the authors unpack how different ways of delivering feedback – continuous versus intermittent, tightly contingent versus partially random – are likely to feel and function for the learner. The emphasis is on how the brain interprets reward signals over time, and how even imperfect feedback can still shape behaviour and physiology if it lands often enough in the right window.

The article also weaves in neurochemical and computational perspectives, not to add complexity for its own sake, but to show how dopamine, serotonin, and reinforcement-learning models help make sense of what clinicians see every day: some clients change quickly, others slowly; some find a reliable internal "click" with the training, others stay in their head trying to make it work.

Finally, the multistage learning model is presented as a way of tying all of this together. Instead of a single curve of improvement, learning is described as unfolding across several overlapping layers: fast, within-session pattern shifts; between-session consolidation; emerging interoceptive awareness; and, over longer time scales, deliberate exploration of mental strategies. This structure sets the stage for the more detailed analysis that follows.

Results

In a review like this, the most important "results" are the clarified ideas and organising principles that emerge from putting many lines of evidence side by side. Several stand out.

One is the rethinking of sham and placebo conditions. The authors show that so-called placebo neurofeedback is rarely a true null condition: if feedback is even partially contingent on physiology, the brain still receives enough correct signals to learn something, albeit more slowly and with more noise. This goes a long way toward explaining why some controlled trials find only modest differences between active and sham conditions, even when clinicians see meaningful change in practice.

Another central insight concerns timing. The article synthesises work on dopaminergic "teaching signals" and feedback-related brain responses to argue that there is a sweet spot, on the order of a few hundred milliseconds, in which feedback is most effective. When the reward arrives close enough to the underlying physiological event, learning is sharpened; when it is too delayed or too jittery, the association between what the person is doing and what the system is rewarding becomes blurred.

The paper also reframes placebo and expectation effects as part of the learning process rather than contaminants to be stripped away. Belief in the training, rapport with the clinician, and the meaning a person assigns to the feedback all feed into how their nervous system updates its models of safety, control, and predictability. This does not mean that neurofeedback or biofeedback are "just placebo"; instead, it highlights that expectation and relationship are among the channels through which physiological learning is expressed.

Perhaps the most clinically useful contribution is the way the multistage learning model is fleshed out. Early, largely implicit changes in spectral power and autonomic balance are seen as laying the groundwork for later, more explicit interoceptive awareness and strategy use. The familiar observation that good learners often report simply feeling different – more settled, more present – while non-learners describe constant effort and trying is given a theoretical home.

Finally, the article cautions that not all protocols calling themselves biofeedback or neurofeedback actually honour these learning principles. When feedback is poorly timed, contingencies are unclear, or artefacts are inadvertently rewarded, studies may end up testing the limits of bad training design rather than the potential of the modality itself. This critical note sets up an important bridge to clinical reality, where practitioners must constantly balance theoretical ideals with the constraints of real people, real time, and real-world complexity.

Discussion

This updated framework paints a rich picture of neurofeedback and biofeedback as multi-layered learning processes, rather than simple “brain training gadgets.” Because this is a conceptual and integrative paper rather than a single experimental trial, the most important outcomes are clarified mechanisms, corrected misunderstandings, and updated models that reshape how we think about these methods. Instead of asking whether neurofeedback or biofeedback “work” in a yes-or-no way, the article invites us to ask how they work, for whom, and under what learning conditions.

One of the clearest threads running through the review is that sham or placebo conditions, timing effects, and relational factors are not side issues; they are central to how the nervous system actually learns. The nuanced treatment of random reinforcement helps explain why some controlled trials see only small gaps between active and sham conditions, even when clinicians observe meaningful changes in real-world practice. If sham protocols still deliver some correct rewards in roughly the right time window, they will still drive learning – just more slowly and less precisely.

For people seeking help, one of the most reassuring messages here is that learning to self-regulate is not all-or-nothing. The paper shows that different systems – brain oscillations, autonomic balance, interoceptive awareness, and conscious strategy use – each adapt on their own timescales. Early sessions might bring subtle shifts in heart rate variability or EEG activity that are not yet consciously noticeable. With time, these shifts can crystallise into a distinct felt sense: the client begins to recognise what “this regulated state” feels like in their own body, and can find it more easily outside the office.

For those considering neurofeedback or biofeedback as part of a broader treatment plan, another important theme is complementarity rather than competition. Expectancy, belief, and the therapeutic relationship are not contaminants to be scrubbed out of research designs; they are among the channels through which physiological learning is expressed. Relaxation skills, attentional control, and breathing practices can be woven into a feedback-based protocol without diluting its scientific basis. In fact, the multistage learning model suggests that these elements may be essential for supporting the transition from early spectral and autonomic shifts to later interoceptive awareness and strategy use.

On the professional side, the paper carries a gentle but firm critique: fidelity to learning theory is not optional. If reinforcement timing is off, thresholds are static and poorly calibrated, or artefacts are rewarded as often as genuine signal, the intervention may still look like neurofeedback but function very differently under the hood. The reminder to pay close attention to schedules of reinforcement, post-reinforcement synchronisation, and threshold adjustment is particularly relevant in busy clinics where software defaults can be tempting shortcuts. In this sense, some “negative” trials may tell us more about weak implementation than about the underlying potential of the modality.

The multistage learning framework offers a practical lens for structuring care. In the early, striatal-learning phase, the clinician’s main job is to make the feedback loop clean and rewarding: good signal quality, clear contingencies, and a game or display that the client finds engaging. In the thalamic consolidation phase between sessions, psychoeducation and realistic expectations help clients trust that change is unfolding even when they do not feel dramatic shifts yet.

As interoceptive homeostasis emerges, structured reflection becomes central. Asking questions such as “What did it feel like when the game was going well?” or “How did your breathing and body change when the bars went up?” helps clients label and anchor new internal states. This subjective language then becomes a bridge between in-session training and daily life: “Can you find that same ‘loose, heavy, steady’ feeling you noticed at the end of last session while you are sitting in traffic or before a performance?”

Finally, in the neurocognitive foraging stage, experimentation with mental strategies is not a side-quest; it is part of the learning trajectory. The client might discover that imagery, counting, body scanning, or a particular breathing pattern reliably nudges the feedback in the right direction. Here, the clinician’s task is to support exploration without turning sessions into over-effortful “trying,” which, as Jacobson reminded us almost a century ago, is often the opposite of genuine relaxation.

Interpretively, this paper is also a quiet defence of neurofeedback and biofeedback against overly simplistic placebo critiques. Expectancy, belief, and relationship clearly influence outcomes, but they do so through neurobiological channels that intertwine with operant learning, dopamine and serotonin signalling, and network-level brain changes. Rather than arguing about whether effects are “real” or “placebo,” the framework invites us to ask a more interesting question: How can we design protocols that harness expectation, interoception, and reinforcement together to support robust, generalisable change?

Brendan’s perspective

What I appreciate most about this paper is how clearly it tries to stand with one foot in each world: the messy, relational, improvisational reality of clinical work, and the structured, controlled, sometimes rigid world of research. It feels less like an abstract defence of neurofeedback and biofeedback and more like a translator between these two cultures.

In a research setting, the pressure is always toward standardisation: fixed protocols, minimal therapist contact, tidy reinforcement schedules that look good on paper. In a clinic, the pressure runs the other way: you are constantly adjusting in response to a child who is tired today, a parent who is overwhelmed, an adult whose trauma history makes "just relax and watch the screen" a non-trivial request. This article offers a language that honours both realities. It says, in effect: here are the learning principles that matter most; here is why they matter; and here is how to think about them while still leaving room for human nuance.

From a neurofeedback perspective, that bridge is incredibly useful. Take a fairly standard protocol for ADHD: training sensorimotor rhythm around 12–15 Hz at C3, Cz, or C4, while inhibiting excessive theta and high beta. A strictly research-driven approach might insist on identical settings for every participant to preserve internal validity. Clinical reality, however, quickly teaches you that one child needs more reward density to stay engaged, another becomes overstimulated by certain games, and a third relaxes only when a parent is present in the room.

What this paper helps clarify is which adjustments are noise and which are principled. Adjusting thresholds to keep reward rates in a sweet spot, cleaning artefacts so that movement and muscle tension are not accidentally reinforced, and tuning feedback timing to feel "snappy" to the nervous system – these are not protocol violations; they are implementations of learning theory. At the same time, it reminds us to be cautious about drifting so far into idiosyncrasy that we lose sight of what is actually being trained.

The same bridge appears in multimodal work. In practice, many of us combine EEG neurofeedback with heart rate variability training and sometimes EMG or electrodermal feedback, especially when anxiety or hyperarousal are prominent. A research purist might view this as hopelessly confounded. But in the therapy room, it often makes sense: breathing at a resonance-like rhythm while down-training high beta at Fz or Pz can give the client a richer, more embodied experience of calm control. The framework in this article – particularly the emphasis on multi-stage learning and interoceptive homeostasis – helps justify these choices without abandoning scientific discipline.

It also gives a more generous reading of placebo and relational effects, which again sits right at the junction of research and practice. Rather than treating the clinician’s warmth, consistency, and enthusiasm as unwanted contamination, the paper frames them as part of the reinforcement landscape. When I sit with a teenager who has "failed" multiple medications and school interventions, and I show them – patiently, over weeks – that their brain can learn, that small changes count, that their efforts are visible and valued, I am not undermining the protocol. I am participating in the learning system the protocol is designed to engage.

At the same time, the article’s criticism of poorly designed or loosely described protocols is a healthy reminder not to hide behind "clinical art" when our setup could be tighter. Latencies, filter settings, reference choices, reward/inhibit ratios – these matter. If a study shows no difference between real and sham, one question is whether neurofeedback "doesn’t work"; another, equally important question is whether the implementation was faithful to what experienced clinicians know is necessary for learning. The paper leans hard into this second question, offering a more nuanced standard of what counts as a fair test.

For day-to-day practice, I come away with a renewed commitment to a kind of disciplined flexibility. The discipline is in respecting the core learning principles: clear contingencies, well-timed feedback, thoughtful reinforcement schedules, and attention to interoceptive experience. The flexibility is in how we embody those principles with each person: choosing electrode sites and frequency bands that fit their presentation, adjusting reward density and task difficulty, blending in breathing or cognitive strategies, and pacing the work to match their nervous system’s capacity.

And for research, the message is equally constructive: if we want trials that genuinely speak to clinical reality, we need designs that capture the relational, expectancy-driven, and multi-modal nature of how neurofeedback and biofeedback are actually delivered, while still keeping a clear grip on the learning mechanisms at play. This paper does not solve that tension, but it maps it honestly – and that, in itself, is an important step toward studies that clinicians recognise as reflecting the world they work in.

Conclusion

Kerson, Sherlin, and Davelaar’s updated framework brings neurofeedback and biofeedback back to their roots in learning theory, while embracing newer insights from reinforcement learning models, neuromodulator research, and phenomenology. It reminds us that change does not come from fancy equipment alone, but from well-designed contingencies, accurately timed feedback, and the gradual emergence of a reliable felt sense of regulation.

For clinicians, this means paying as much attention to reinforcement schedules, feedback timing, and thresholding as to protocol choice and hardware. For clients, it offers hope that self-regulation can become a lived skill rather than something that only happens in front of a screen. And for the field as a whole, it offers a roadmap for research that honours both the technical and human dimensions of psychophysiological learning.

In the end, neurofeedback and biofeedback are at their best when they help people discover that their brains and bodies are not fixed problems to be managed, but adaptive systems that can learn, with the right guidance, to move toward healthier and more flexible patterns.

References

Kerson, C., Sherlin, L. H., & Davelaar, E. J. (2025). Neurofeedback, biofeedback, and basic learning theory: Revisiting the 2011 conceptual framework. Applied Psychophysiology and Biofeedback. https://doi.org/10.1007/s10484-025-09756-4