Can HRV Biofeedback Sharpen Working Memory?

A new emerging review by e Silva, Wöllner, and Norte (2026) takes on a surprisingly underexplored question: can heart rate variability biofeedback do more than calm the nervous system? More specifically, can it sharpen working memory?

That matters because most of the HRV-biofeedback literature has lived, quite reasonably, in the worlds of anxiety, depression, stress physiology, and emotional regulation. Those are important domains, and the evidence base there is much deeper. But this review tilts the conversation toward cognition—particularly working memory, one of the central executive functions supporting reasoning, goal maintenance, learning, and adaptive performance under pressure.

Biofeedback refers to the use of real-time physiological signals—such as heart rate, respiration, muscle tension, or skin conductance—to help individuals learn voluntary self-regulation. Neurofeedback is a subtype of biofeedback focused specifically on brain activity, most commonly EEG. In HRV biofeedback, the signal of interest is the moment-to-moment variation between heartbeats, usually trained through slow breathing near an individual’s resonance frequency in order to strengthen autonomic flexibility and baroreflex functioning.

What makes this review especially interesting for clinicians and performance-oriented professionals is its shift in emphasis. Rather than asking whether HRV-BFB reduces distress, the paper asks whether better autonomic regulation may support better cognitive regulation. That is a plausible and clinically appealing idea, particularly under the neurovisceral integration model, where prefrontal control, vagal regulation, and goal-directed behavior are deeply intertwined.

The authors do not oversell the case. Their conclusion is appropriately cautious: HRV-BFB shows promise, but current evidence remains preliminary, heterogeneous, and methodologically uneven. Still, the review gives us something valuable—a clearer map of where the cognitive promise may be real, where it remains uncertain, and what future trials need to do better.

Methods

This paper was conducted as a systematic review following PRISMA guidelines and was preregistered in PROSPERO, which is already a welcome signal of methodological seriousness. The authors searched Web of Science, PubMed, and PsycINFO in August 2025, using a broad search strategy that combined HRV biofeedback terms with executive-function and cognition terms. Although the search cast a wide net across executive domains, the final inclusion criteria were deliberately narrow: adult human studies, published in English in peer-reviewed journals, that used HRV biofeedback and included pre- and post-intervention working-memory outcomes.

From 59 identified records, 10 studies met inclusion criteria. That alone tells us something important. Despite the popularity of HRV-BFB, the number of studies directly examining working memory remains quite small.

The included studies were notably heterogeneous. Seven used randomized controlled designs, while three were non-randomized pre-post intervention studies. Samples ranged from healthy young adults to older adults, combat veterans with PTSD, and individuals with severe brain injury. Sample sizes were often modest.

The HRV-BFB protocols also varied substantially. Some studies used a single brief session lasting only 6 to 30 minutes, while others used multi-session interventions over days or weeks. One study provided 10 sessions of 60 minutes each. Several included home practice, which is clinically relevant but also adds variability in adherence and dose.

Most protocols trained slow breathing near resonance frequency, generally around 4.5 to 6.5 breaths per minute, aiming to maximize baroreflex engagement and HRV amplitude. Equipment varied across studies and included systems such as emWave, emWavePRO, HeartMath, BIOPAC, CorSense, StressEraser, and ActiveTwo.

Working memory was assessed with a wide range of instruments: N-back tasks, backward digit span, list sorting, Trail Making Test Part B, modified Stroop variants, and even verbal fluency in one study. This matters because these tasks do not all tap working memory with equal precision. High-load N-back conditions and backward digit span are arguably more defensible working-memory measures than verbal fluency or mixed executive tests.

The authors also extracted physiological outcomes, including RMSSD, SDNN, HF power, LF power, LF/HF ratio, RSA, and cardiac coherence. This allowed them to compare whether autonomic modulation was more consistent than cognitive change—which, as it turns out, it was.

Results

The headline result is encouraging, but far from definitive. Five of the 10 included studies reported significant improvement in working memory after HRV-BFB, while the remaining studies found either null results or effects that were partial, task-specific, or difficult to interpret.

Some of the more positive findings came from populations where cognitive control may be under strain. Ginsberg and colleagues reported improved backward digit span in combat veterans with PTSD after four weeks of training. Bahameish and Stockman found improved correct responses on a 2-back task in healthy adults following a brief HRV-BFB intervention. Schlatter and colleagues reported improved 3-back accuracy in one experimental context, suggesting that higher-load working-memory tasks may sometimes be more sensitive to training effects than simpler conditions.

But the positive pattern was not universal. Raaijmakers et al. found no significant working-memory benefit on a 2-back task in a triple-blind randomized study with sham biofeedback. Nashiro et al. did not find significant improvement on a list-sorting working-memory task after five weeks of training in younger and older adults. In another study, improvements were observed in forward digit span but not backward digit span, which suggests gains in short-term retention may not necessarily translate into stronger executive manipulation of information.

One of the most consistent findings was physiological rather than cognitive. The vast majority of studies reported increased HRV-related indices after intervention, including measures such as RMSSD, SDNN, coherence, RSA, or low-frequency power in the resonance range. In other words, HRV-BFB generally seemed to do what it is supposed to do physiologically.

That distinction is important. Improved autonomic modulation did not automatically produce improved working memory. In one study, working-memory gains were not mediated by cardiac vagal tone changes, which complicates simple mechanistic stories. It suggests that HRV-BFB may influence cognition through indirect routes—or that some observed cognitive benefits reflect expectancy, attentional engagement, stress reduction during testing, or task-specific factors rather than a clean vagal-to-cognition pathway.

The review’s risk-of-bias findings also temper enthusiasm. Among the seven randomized studies, only two were judged low risk overall; three raised some concerns; and two were judged high risk. The non-randomized studies carried considerable risk of bias, especially due to lack of controls, confounding, and small samples. This means the signal is intriguing, but the confidence interval around that signal is still fairly wide.

Discussion

This review lands in a clinically interesting and scientifically awkward place: promising enough to stay excited, but not yet strong enough to make sweeping claims.

What the paper shows, at minimum, is that HRV biofeedback can reliably shift physiology in the expected direction. Across most of the included studies, participants increased HRV-related indices after training. That is not trivial. It means the intervention appears capable of modifying autonomic regulation across a range of populations and settings.

What remains less certain is whether those physiological changes translate into working-memory enhancement in a robust and generalizable way. The answer, for now, is: sometimes, under some conditions, in some populations. That is not a disappointing conclusion. It is actually a useful one.

For clinicians, the practical implication is that HRV-BFB may be most worth considering when working memory is being degraded by dysregulation rather than by a primary structural or neurodegenerative process. The more plausible use case is not “turn anyone into a cognitive super-performer,” but rather “reduce the physiological interference that disrupts cognitive performance.” That distinction matters. A stressed, hyperaroused, trauma-exposed, or cognitively overloaded nervous system may benefit more from improved autonomic flexibility than a well-regulated healthy adult already performing near ceiling.

That reading fits the pattern in the review. Studies involving veterans with PTSD or stressful laboratory conditions appeared somewhat more promising than trials in broadly healthy populations. It also fits the broader psychophysiology literature, where vagally mediated HRV is often associated with attentional control, emotional regulation, and prefrontal efficiency rather than with raw intelligence or broad cognitive upgrading.

There is also a measurement issue here that deserves more attention than it often gets. “Working memory” is not one thing. A backward digit span, a 3-back task, a list-sorting task, and a modified Stroop paradigm may overlap conceptually, but they are not interchangeable. Some tasks are more dependent on updating, some on interference control, some on processing speed, and some on attentional stability. When a literature is already heterogeneous in dose, population, and protocol, using loosely comparable outcome measures makes interpretation even harder.

Another major issue is intervention dose. A six-minute session and a five-week training protocol are not the same intervention just because both are called HRV biofeedback. From a learning perspective, they likely recruit different mechanisms. Brief sessions may produce state effects—temporary improvements related to arousal regulation, attentional settling, or reduced noise in the system. Multi-session protocols may be more likely to produce trait-level changes in self-regulation, baroreflex engagement, or task readiness. If future studies do not separate acute from longitudinal effects, the field will keep mixing apples and cardiac oscillations.

Methodological quality remains the biggest barrier to confident interpretation. Blinding is hard in biofeedback research, but not impossible, and the review rightly highlights the importance of sham-controlled designs. The triple-blind study by Raaijmakers et al. is especially informative because it failed to find the sort of clean positive effect that might be expected if the mechanism were strong and universal. That does not invalidate the field. It simply reminds us that expectancy, motivation, therapist contact, and repeated testing can all influence performance.

For referring professionals and cognitive-performance clinicians, the current takeaway is measured optimism. HRV-BFB may have a place in performance-oriented work, especially where stress-reactivity, attentional instability, or physiological rigidity appear to be constraining cognition. But it should not yet be marketed as a validated working-memory intervention in the same way it is often discussed for stress regulation or autonomic training.

For neurofeedback professionals, this review is relevant even though it is not an EEG paper. It reinforces a principle many of us already see in practice: cognition does not float above physiology. Working memory depends on state regulation, and state regulation is not only cortical. Breathing, baroreflex dynamics, vagal tone, interoceptive awareness, and emotional load all shape the conditions under which executive control can actually function. That does not mean HRV-BFB replaces EEG neurofeedback. It suggests that, in some clients, autonomic training may improve readiness for neurofeedback, enhance session stability, or reduce the physiological “noise floor” that interferes with cognitive control.

Mechanistically, the authors are careful not to reduce everything to one pathway, and that is wise. The review discusses plausible routes involving baroreflex stimulation, allostatic load reduction, improved attentional control, and reduced mind-wandering. Those are reasonable hypotheses, but they are not yet settled. In fact, one of the most valuable contributions of this review is that it resists the temptation to confuse plausibility with proof.

Brendan’s perspective

1. HRV biofeedback as a gateway to better cognitive readiness before EEG neurofeedback

One of the clearest clinical uses of this literature is not replacing EEG neurofeedback, but preparing people for it. Some clients arrive dysregulated enough that asking for stable attention, introspection, and consistent engagement with EEG feedback is already a big ask. In those cases, HRV biofeedback may serve as a very practical entry point. If the person can learn to reduce physiological noise, improve breathing regularity, and increase autonomic flexibility, they may be better able to benefit from later cortical training. In other words, sometimes the brain is not unwilling; the system is simply too busy firefighting.

This is especially relevant when working memory problems appear state-sensitive rather than trait-fixed. A client whose cognition collapses under pressure may not need a “memory protocol” first. They may need a nervous system that can tolerate cognitive load without tipping into hyperarousal or mental fragmentation. From an EEG perspective, I would think here about pairing HRV-BFB with later work on SMR stability, frontal midline regulation, or individualized attention protocols once the person can actually sustain the training state.

2. When “cognitive enhancement” is really state regulation in disguise

I suspect many real-world gains labeled as cognitive enhancement are, in fact, regulation gains. That is not a criticism; it is often the more clinically meaningful truth. A person may perform better on working-memory tasks not because HRV biofeedback has directly upgraded executive circuitry in some isolated way, but because they are less physiologically taxed, less distractible, and less consumed by internal noise. The cognition looks sharper because the system is less burdened.

That distinction matters for expectation-setting. I would be cautious about promising HRV biofeedback as a pure nootropic tool for healthy, already well-regulated adults. But I would be much more optimistic when the client is running cognition through a layer of chronic stress, trauma load, somatic tension, sleep disruption, or performance anxiety. In those cases, improving regulation may be the shortest path to improving mental performance. Not flashy, perhaps—but clinically elegant.

3. Building hybrid protocols: HRV biofeedback plus EEG neurofeedback for attention and working memory

This review makes me think in sequences, not silos. A hybrid protocol can be especially appealing when working memory difficulties sit at the intersection of arousal dysregulation and attentional instability. HRV biofeedback can help build interoceptive awareness, breath control, and autonomic flexibility, while EEG neurofeedback can then target more specific patterns related to vigilance, impulsivity, sustained attention, or cognitive control. The order matters. In some clients, HRV-BFB may come first to stabilize the platform; in others, it may be layered alongside EEG as a transfer-support tool.

Practically, I could imagine using HRV work between EEG blocks, before sessions, or as home practice to reinforce state access. For EEG targets, this would depend entirely on the person, but clinically relevant areas might include central SMR training for stability and inhibition, frontal beta/low beta work when underactivation and poor task engagement are central issues, or alpha-based downshifting when excess arousal is degrading performance. The lesson is not that everyone needs the same combination. The lesson is that autonomic and cortical training can be complementary rather than competitive.

4. What EEG neurofeedback professionals can learn from the HRV literature

The HRV literature is a helpful corrective for neurofeedback professionals because it reminds us that not every cognitive problem begins with a cortical protocol choice. Sometimes what looks like inattention, poor updating, or weak executive control is partly a body-state problem. Breathing patterns, vagal regulation, cardiovascular flexibility, and interoceptive accuracy can all influence whether a person is even available for learning. That perspective can make our EEG work better, because it broadens what we are actually assessing.

It also reinforces an old but useful principle: self-regulation is multi-level. We do not train a disembodied cortex. We train people. The more our protocols reflect the interaction among physiology, perception, context, and brain state, the more likely we are to get transfer. For some clients, that may mean integrating paced breathing cues, HRV checks, or arousal-monitoring into the treatment plan rather than assuming that EEG alone has to do all the heavy lifting.

5. Working memory is not "just cortical"

This, to me, is the biggest conceptual gift of the paper. Working memory is often discussed as though it lives neatly in dorsolateral prefrontal cortex and nowhere else. But in practice, working memory depends on a whole regulatory ecology: arousal level, emotional interference, bodily safety signals, respiratory rhythm, sustained attention, and the capacity to resist internal distraction. That does not make working memory any less cortical. It makes it more biologically honest.

For neurofeedback clinicians, this argues for protocol individualization rather than theory tribalism. Some clients will indeed need direct EEG work around frontal control networks, task engagement, or stability across sites and states. Others may need autonomic regulation first, or alongside EEG, because the bottleneck is not purely neural in the narrow sense. So I come away from this review encouraged. Not because it proves HRV biofeedback “boosts cognition” in a broad consumer-wellness sense, but because it supports a richer model of performance—one where better thinking often begins with better regulation.

Conclusion

So, can HRV biofeedback sharpen cognition?

The latest review suggests that it might—but not in a simple, universal, or fully proven way.

The strongest conclusion at present is that HRV-BFB reliably improves physiological regulation, while its effects on working memory remain mixed and context-dependent. That is still an important finding. For professionals interested in cognition, this review broadens the conversation around HRV beyond anxiety, mood, and stress-management. It points toward a more performance-oriented application, especially in cases where working memory is vulnerable to dysregulation, overload, or chronic stress physiology.

At the same time, the field needs better trials: larger samples, more rigorous sham conditions, more consistent working-memory measures, and clearer separation between acute state effects and durable training effects. Until then, HRV-BFB is best viewed as a promising cognitive-support tool rather than a settled cognitive enhancer.

That may sound modest, but it is actually a very usable conclusion. In psychophysiology, better regulation often comes before better performance. If HRV biofeedback helps stabilize the system, it may prove to be an increasingly valuable part of how we support cognition—not instead of more targeted cognitive or neurofeedback interventions, but alongside them.

References

e Silva, F. R. da C., Wöllner, E. P. F., & Norte, C. E. (2026). Effects of heart rate variability biofeedback on working memory: A systematic review. International Journal of Psychophysiology, 225, 113381. https://doi.org/10.1016/j.ijpsycho.2026.113381