A new emerging research study by Imhof, Raschle, Wenderoth, Meissner, and colleagues explores a deceptively simple but clinically intriguing question: if people learn to volitionally regulate pupil-linked arousal, does that change how their body and subjective experience respond to emotional stimuli? The answer, at least from this proof-of-concept study, is nuanced—but promising.

The authors build on earlier work showing that healthy participants can learn to upregulate and downregulate pupil size using real-time feedback. Under controlled lighting conditions, pupil dynamics are closely linked to central arousal systems, especially the locus coeruleus–noradrenergic system. That matters because the locus coeruleus is deeply implicated in vigilance, stress responsivity, autonomic regulation, and the kind of hyperarousal that shows up across anxiety and stress-related conditions.

Biofeedback is a method that helps individuals gain voluntary control over physiological processes using real-time information about those processes. Neurofeedback is a subtype focused specifically on brain activity, most often EEG. Pupil-based biofeedback sits in an especially interesting middle ground: it is not EEG neurofeedback, but it may provide a more direct behavioral window into arousal regulation than many conventional psychophysiological tools.

What makes this paper particularly relevant for clinicians is that it moves beyond the usual question of whether participants can change pupil size. Instead, it asks whether training this skill alters emotional responding when negative stimuli are encountered. That shift—from isolated self-regulation to emotionally meaningful context—is exactly where biofeedback begins to matter clinically. If arousal regulation can transfer into emotionally provocative situations, even modestly, that opens the door to new interventions for hyperreactivity, autonomic overactivation, and stress sensitivity.

Methods

The study enrolled 25 healthy adults, with 23 completing the full protocol. Participants were free of neurological and psychiatric disorders, were not taking centrally acting medication, and had normal or corrected-to-normal vision. The design involved three pupil-based biofeedback training sessions followed by a fourth session combining self-regulation with emotional sound exposure.

During the first three sessions, participants learned to volitionally increase or decrease pupil size using mental strategies while receiving real-time pupil biofeedback. The paper builds on the group’s prior work linking this process to activity changes in the locus coeruleus and related arousal-regulating structures. Although the present article does not re-establish those imaging findings directly, it uses this previously acquired skill as the platform for the experimental emotion task.

In the fourth session, participants completed trials involving one of three conditions: upregulation, downregulation, or a non-regulatory control task. They were exposed to 60 negative and 60 neutral sounds drawn from the IADS-2 database. Trials were organized in pseudorandomized blocks of five.

Each trial followed a structured sequence. Participants first received an instruction, then completed a 3-second baseline phase while counting backward from 100 in steps of four to stabilize mental state and reduce premature strategy use. This was followed by a 2-second pre-sound modulation phase, during which participants either began upregulating, began downregulating, or continued the counting task in the control condition. Next came a 6-second sound presentation phase, during which they continued regulating or not regulating, depending on condition. After each sound, participants rated affect intensity, arousal, and valence on continuous visual analogue scales. They also received post-trial performance feedback indicating whether their pupil modulation was successful.

Physiological recordings included continuous pupil diameter and electrocardiography. The primary subjective outcome was affect intensity. Pupil analyses examined both pre-sound self-regulation success and sound-evoked dilation responses. Heart-rate analyses focused on relative changes during sound presentation compared with the pre-sound modulation phase.

Results

The first result is straightforward and important: participants improved at pupil self-regulation over the course of training, but this improvement was driven mainly by downregulation. The interaction between session and self-regulation was significant, with downregulation improving from session 1 to session 3, while upregulation did not show a comparable training effect.

The primary behavioral finding was more restrained. Negative sounds, unsurprisingly, produced stronger affect ratings than neutral sounds. However, online self-regulation condition—upregulation, downregulation, or non-regulatory control—did not significantly alter affect intensity ratings at the group level. In other words, simply entering a trial in an upregulated or downregulated pupil state did not immediately and reliably change how intense the sounds felt across participants.

The more interesting signal emerged in the individual-differences analyses. Greater improvement in downregulation training predicted lower affect intensity in response to negative sounds during downregulation trials, and also during non-regulatory control trials. Upregulation training gain did not predict stronger affective responses. This suggests that what mattered was not merely attempting downregulation in the moment, but how well participants had actually learned it over time.

Physiologically, both negative and neutral sounds elicited pupil dilation, with larger dilation to negative sounds. Self-regulation also significantly influenced pupil responses: both upregulation and downregulation were associated with larger sound-evoked pupil dilation than non-regulatory control trials. This effect did not depend on whether the sound was negative or neutral. The authors interpret this cautiously as possibly reflecting regulatory effort rather than simple emotional amplification.

Heart-rate findings moved in a somewhat different direction. Downregulation was associated with significantly greater heart-rate deceleration during sound presentation compared with both upregulation and non-regulatory control. There was no significant main effect of sound type on heart-rate change. This pattern is consistent with stronger parasympathetic dominance during pupil downregulation.

Discussion

This study does something I really appreciate: it tests transfer. Not transfer in the grand clinical sense just yet, but a first experimental version of it. The authors are no longer asking whether participants can make a bar move, enlarge a pupil, or follow a feedback signal. They are asking whether a learned arousal-regulation skill still matters when something emotionally salient enters the environment. That is a much more clinically relevant question.

What the paper clearly shows is that pupil-based biofeedback can train a self-regulation skill, especially for downregulation, and that individuals who learn that skill better appear less affectively impacted by negative sounds. What it does not show is that a single episode of online regulation reliably reduces self-reported affect in the moment across the whole sample. That distinction matters. The results support a learning-dependent effect more than an immediate state effect.

The divergence between subjective and physiological findings is also interesting. Pupil dilation increased during both upregulation and downregulation relative to control during sound presentation, while heart rate decelerated more strongly during downregulation. At first glance, that may seem contradictory. It is probably not. Pupil responses may be indexing a blend of emotional reactivity and regulatory effort, whereas heart-rate slowing may reflect a shift toward parasympathetic dominance or enhanced attentional control. In other words, “more pupil response” in this paradigm may not mean “more distress.” It may mean “more active regulation.”

For clinicians working with hyperarousal, this matters a great deal. Many clients do not look calmer when they are beginning to regulate; early regulation can be effortful, metabolically active, and only partially reflected in subjective reports. A patient may still feel activated while physiological organization is already beginning to shift. This study fits that kind of psychophysiological complexity rather well.

There are also good reasons to remain cautious. The sample was small and healthy, not clinical. The paper is best understood as mechanistic and translational rather than therapeutic. The design also differs from many classic emotion-regulation paradigms because participants started regulating before sound onset. That is useful for testing whether preconfigured arousal states change responding, but it is not the same as asking someone to regulate after an emotional response has already been triggered. Those are related but distinct regulatory challenges.

Another limitation is that the authors infer locus coeruleus involvement from earlier validation work rather than measuring it directly here. That is reasonable scientifically, but it means this paper should not be overread as direct evidence that locus-coeruleus modulation caused the emotional effects in this sample. It is more accurate to say the findings are consistent with that broader framework.

Clinically, though, the translational potential is real. A pupil-based intervention could be attractive in populations where explicit cognitive reappraisal is difficult, fatiguing, poorly tolerated, or developmentally mismatched. It is noninvasive, immediate, and tied to a physiological signal that may be closer to arousal regulation than many verbal techniques. It could plausibly serve as a bridge between autonomic biofeedback and more explicit self-regulation training.

For referring professionals, the paper offers a refreshing reminder that emotional regulation is not only a top-down cognitive act. It is also a bodily skill. For biofeedback clinicians, it suggests that the pupil may be a clinically useful target beyond laboratory curiosity. For neurofeedback practitioners, it raises an especially compelling question: should some clients first learn to regulate arousal physiology before being asked to modulate EEG patterns in emotionally loaded or high-demand contexts?

My broader read is that this is not yet a treatment paper. It is a translational proof-of-principle paper with meaningful physiological signals and a modest but clinically interesting behavioral pattern. Those are the kinds of papers that often matter later, once protocol refinement catches up.

Brendan’s perspective

What excites me most about this paper is not just the result pattern, though that is certainly interesting. It is the doorway it opens. If the earlier validation work is right—and I think it is compelling enough to take seriously—then pupil-based biofeedback gives us a practical handle on a system that EEG neurofeedback has always cared about indirectly but struggled to reach directly: the locus coeruleus and the broader brainstem arousal machinery.

That is a big deal.

EEG neurofeedback is incredibly useful, but it has limits that good clinicians should acknowledge without apology. Scalp EEG is excellent for tracking cortical rhythms and network-level expressions of state. It can tell us a great deal about vigilance, inhibition, instability, cortical slowing, inefficient activation, and state shifts across distributed systems. But when the clinical story lives deep in the brainstem—hypervigilance, exaggerated orienting, chronic threat-readiness, autonomic overcoupling, exaggerated sensory responsivity—we are often training cortical signatures of that problem, not the subcortical engine itself.

That does not make EEG neurofeedback weak. It makes it honest.

In practical terms, many of us have seen cases where EEG neurofeedback helps attention, sleep, emotional range, or cognitive stability, yet the client still has that “always on” quality in the background. The body is listening for danger. The pupils are reactive. The startle response is quick. The system is organized around readiness rather than ease. In those clients, I suspect we are often working around the locus coeruleus more than through it.

So when a modality appears that may provide voluntary access to pupil-linked arousal regulation—one that prior work has connected to locus-coeruleus-centered circuitry—I pay attention. Not because it replaces EEG neurofeedback, but because it may finally complement it in exactly the place where scalp-based training has the hardest time being precise.

This is where I think the field may be standing at the beginning of something genuinely new.

If pupil biofeedback continues to hold up, we may be looking at a new modality class that slots beautifully into neurofeedback practice. Not a gimmick. Not a wellness add-on. A true bridge modality between autonomic biofeedback and brain-based self-regulation.

Here is how I imagine that integration clinically.

For the highly activated client—the one with chronic anxiety, sensory defensiveness, trauma-related hyperreactivity, panic-proneness, or relentless cognitive-emotional overcoupling—I would be very tempted to begin with arousal training rather than jumping straight into classic EEG optimization. In some cases, that might mean starting with pupil downregulation work, paced breathing, and perhaps HRV biofeedback to build state access first. Once the client can find a physiological “gear shift,” EEG neurofeedback may become more efficient because the brain is no longer trying to learn under constant threat bias.

Then, when moving into EEG work, I would think less in terms of one magical protocol and more in terms of sequencing. If the presenting picture is hyperarousal with poor inhibitory control, SMR-oriented work around Cz or the sensorimotor strip may make sense for stabilizing behavioral inhibition and reducing excessive reactivity. If the clinical picture includes anxious overactivation, rumination, or a chronically “busy” frontal system, some clinicians may consider carefully designed fast-wave inhibition or alpha-stabilization approaches depending on the assessment picture. If the person cannot settle into internal safety at all, posterior alpha support or eyes-closed alpha-theta work might eventually become useful—but probably not too early, and not without adequate titration.

That last point matters. Protocols should never be chosen because they sound elegant in a paragraph. They should be chosen because the intake, physiology, symptoms, and response patterns all line up. Pupil training does not remove the need for individualization. If anything, it makes the need for individualization even more obvious.

What I like here is the conceptual division of labor. Pupil-based training may help us work closer to the arousal generator. EEG neurofeedback may then help shape how cortical networks regulate, interpret, inhibit, and generalize that arousal. One is not superior to the other. They may simply be acting at different levels of the same hierarchy.

That also helps explain why some clients plateau in EEG work. Sometimes the cortex is trying very hard to regulate a system whose baseline operating point remains too high. In those cases, adding a modality that helps the organism reduce tonic alerting may improve outcomes not by replacing EEG, but by making EEG learning more accessible. To put it bluntly: a less alarmed brain learns better.

I can also imagine the reverse sequence in some cases. A client may first need enough cortical stability through EEG neurofeedback to tolerate interoceptive awareness and deliberate arousal work. For someone dissociative, extremely fragile, or easily overwhelmed by bodily signals, starting with pupil biofeedback may be too activating or too effortful. So again, this is not dogma. It is an expanded toolkit.

The most exciting future, in my view, is multimodal. Imagine a treatment arc where we combine pupil-based training for locus-coeruleus-linked arousal access, HRV biofeedback for autonomic flexibility, respiration training for vagal support, and EEG neurofeedback for cortical regulation and transfer into task-relevant states. That begins to look less like isolated techniques and more like a coordinated neuroscience of self-regulation.

And clinically, that is how real people present anyway. They do not come in with “an EEG problem” or “a biofeedback problem.” They come in with layered dysregulation across brain, body, attention, emotion, and context.

Of course, this paper does not prove clinical efficacy, and we should not pretend it does. The sample is healthy, the design is early-stage, and the subjective effects were subtler than the physiological ones. Good. That is how serious new modalities usually begin: not with grand promises, but with plausible mechanisms and interesting early transfer signals.

Still, I find this line of work deeply encouraging. For years, neurofeedback clinicians have worked with the suspicion that some of our most difficult cases are limited not by cortical trainability, but by unresolved arousal bias deeper in the system. Pupil-based biofeedback may give us a way to engage that layer more directly.

If that continues to be supported, then yes—I think we may be witnessing the beginning of a genuinely important new modality, one that could make neurofeedback outcomes broader, faster, and more durable when thoughtfully integrated into expert care.

Conclusion

This study offers an elegant early test of whether pupil-linked arousal self-regulation matters in emotionally relevant moments. In healthy adults, pupil-based biofeedback improved self-regulation skills—especially downregulation—and those who learned downregulation more successfully reported less intense affective reactions to negative sounds. At the same time, the physiological picture was richer than any simple “calmer versus not calmer” story: self-regulation increased pupil responses during sound presentation, likely reflecting regulatory effort, while downregulation promoted stronger heart-rate deceleration, suggesting a more parasympathetically organized state.

That combination is clinically interesting. It suggests that effective regulation may not always look subjectively or physiologically simple in the moment. Sometimes the body is working. Sometimes the signal of success is not immediate emotional relief, but improved flexibility and a more adaptive autonomic response profile.

For now, this remains a proof-of-concept in a healthy sample, not a clinical efficacy trial. Still, it gives us a thoughtful new direction. If pupil-based biofeedback can help people build better control over arousal before or during emotionally challenging situations, it may eventually become a valuable tool for treating hyperarousal, stress sensitivity, and emotionally driven physiological reactivity. That is a future well worth following.

References

Imhof, J., Raschle, N. M., Wenderoth, N., & Meissner, S. N., et al. (2026). Pupil-based arousal self-regulation: Impact on physiological and affective responses to emotional stimuli. Translational Psychiatry. https://doi.org/10.1038/s41398-026-03937-3