Technology for Sensory Regulation: From Noise-Canceling Headphones to Wearables

Introduction: From Sensory Environment Management to Real-Time Self-Regulation

Sensory regulation encompasses strategies designed to modify exposure to stimuli and adjust the physiological or subjective response to them. Noise-canceling headphones act on the acoustic environment, while wearable devices record bodily signals and can provide immediate interventions. This shift moves the focus from the general reduction of stimuli toward the individualized identification of arousal states and their modulation during daily activities. Recent reviews describe a convergence between physiological sensors, artificial intelligence, contextual information, and user feedback (Bolpagni et al., 2024; Taskasaplidis et al., 2024). The evidence, however, distinguishes between enhancing the sensory experience, detecting stress, and producing sustained changes in self-regulation—three outcomes that should not be treated as equivalent.

Auditory Control and Sound Stimulation as Sensory Regulation Strategies

Active noise cancellation, performance, and stress in office environments

Active noise cancellation allows individuals to control their acoustic environment, especially in open-plan offices where irrelevant speech interferes with concentration. In a laboratory experiment with 54 participants, five headphone settings were compared during serial recall and n-back tasks, while heart rate variability was used as a physiological indicator of stress (Radun et al., 2024). No configuration altered performance or heart rate variability. Active noise cancellation without masking sound made the environment seem more pleasant, but the combination of both functions reduced the subjective assessment of interference and distraction compared to not wearing headphones (Radun et al., 2024).

These results are replicated in the conference report based on the same protocol, where active noise cancellation alone also failed to improve cognitive performance or reduce physiological stress (Hongisto et al., 2024). The discrepancy between experience and performance indicates that feeling less bothered does not imply better information processing or a lower autonomic response. For sensory regulation, headphones may offer control and subjective relief, but they should not be presented as a sufficient intervention against the acoustic demands of an office. The benefit depends on the sound configuration and the outcome being assessed; therefore, perceived experience requires a separate measure from attention and physiology.

Auditory hyperreactivity, natural sounds, and musical stimulation

For individuals with auditory hyperreactivity, reducing the overall volume may be an imprecise solution because aversion depends on frequencies, intensities, and individual perceptual profiles. A hybrid active noise-cancellation system, designed based on behavioral and electroencephalographic tests in children with autism, established attenuation targets tailored to their auditory responses; the customized headphones improved ratings of sound perception (Kwong & Choy, 2024). The finding supports adaptation based on auditory profile, although the review does not demonstrate effects on daily stress, school participation, or long-term functioning.

Sound stimulation can be directed toward reducing arousal or replacing aversive stimuli with preferred signals. A meta-analysis found that natural sounds, compared to a quiet environment, produced differences in heart rate, blood pressure, and respiratory rate, but not in perceived stress or oxygen saturation; the evidence was described as inconsistent (Fan & Baharum, 2024). In patients undergoing procedures, a review of 21 trials on technological music interventions reported reductions in anxiety in nine studies and in pain in seven, although it could not clearly distinguish the effects of the device from those of the music program (Park et al., 2024). Another review of 31 studies concluded that the effects of music, noise, white noise, and binaural beats on autonomic measures and cognitive performance were mixed or insufficient (Chee et al., 2024). Therefore, the selection of sound, context, and personal preference must be incorporated into the design before linking an acoustic signal to a presumed general regulatory effect.

Wearables for Detecting and Contextualizing Stress Through Physiological Signals

From biosignal monitoring to personalized and multimodal models

Physiological wearable devices turn regulation into a potentially continuous process: they record signals, estimate changes in activation, and allow these to be linked to specific situations. Reviews identify electrodermal activity and photoplethysmography as common signals for detecting stress, especially when combined with other modalities (Bolpagni et al., 2024). Electrocardiography, skin temperature, movement, posture, speech, and hand tremors are also used, as each signal reflects different dimensions of the stress response (Pinge et al., 2024; Taskasaplidis et al., 2024). This diversity broadens physiological coverage but makes preprocessing, synchronization, and interpretation critical.

Personalization addresses a limitation of the population-based approach: cardiac or electrodermal variations can have different meanings depending on the individual, the activity, and the time of day. A review of personalized models noted frequent use of electrodermal activity and photoplethysmography, along with a trend toward deep learning models, although comparisons with traditional methods remain limited (Bolpagni et al., 2024). In multimodal affective computing, the most common combinations include electrodermal activity and electrocardiography, with convolutional networks and long-term memory networks among the architectures used (Li & Zhang, 2025). The expected advantage of multimodality does not lie in indiscriminately accumulating sensors, but rather in compensating for the incomplete information in each signal.

Detection can be organized as a decision chain that begins with data acquisition and ends with a regulatory action. The following diagram summarizes the components described in reviews of detection and in studies on continuous monitoring (Pinge et al., 2024; Taskasaplidis et al., 2024).

Real-world environments, machine learning, and interpretation of the autonomic nervous system

Validation under everyday conditions alters the interpretation of results obtained in the laboratory. A study using data from 240 subjects collected via electrocardiography, skin temperature, and skin conductance evaluated several models in a real-world setting; the best performance depended on the classification task and whether or not synthetic oversampling was used (Al-Alim et al., 2024). The Random Forest model achieved 98.29% accuracy and an F1 score of 97.89% in binary classification without oversampling, while XGBoost achieved an accuracy and F1 score of 98.98% in three-class classification with oversampling (Al-Alim et al., 2024). These figures describe performance within a specific dataset, not universal clinical accuracy.

Interpreting the autonomic nervous system requires more than an isolated reading of heart rate, heart rate variability, or conductance. A critical review argues that autonomic signals depend on behavioral state, social situation, and environmental context; therefore, their meaning cannot be inferred solely from physiological recordings (Norman et al., 2025). Walking, talking, engaging in physical exertion, or navigating a social interaction can produce partially overlapping patterns. Context-sensitive models must incorporate this information without equating any bodily activation with a psychological label.

The most recent multimodal systems attempt to fuse physiological signals, mental information, and situational variables using adaptive architectures (Myla et al., 2025). The value of these approaches lies in adjusting the contribution of each signal when uncertainty exists, although accuracy demonstrated in an experimental model does not prove validity in independent populations. Research is also advancing toward low-power hardware: a system consisting of a wristband and a chest patch was designed to operate for a week between charges and to facilitate real-time physiological analysis (Brewer et al., 2025). Reduced energy consumption promotes continuous use, but a useful intervention also requires reliable interpretation, understandable communication, and consent regarding the data.

Reliability, Validity, and Usability in the Selection of Wearable Devices

The choice of a device cannot be based solely on the number of sensors or the accuracy claimed by the manufacturer. The Stress in Action Wearables Database compiles technical specifications, recorded signals, availability of raw data, and systematic reviews on reliability, validity, and usability (Schoenmakers et al., 2025). The authors note that studies on usability are scarce, validity statistics are reported inconsistently, and documentation regarding sampling frequency or parameter extraction may be imprecise (Schoenmakers et al., 2025). These shortcomings affect both research and the selection of products for personal use.

Technical quality must be evaluated alongside tolerability. In people with autism, a review of 31 articles found a predominance of solutions based on heart activity using photoplethysmography and electrocardiography, but also small sample sizes, variable conditions, and the need to customize stress thresholds (Cano et al., 2024). Sensory sensitivity can cause a wearable device to be uncomfortable, disrupt behavior, or be abandoned before it produces useful data. Therefore, usability is not a secondary attribute: it determines which signals are obtained, for how long, and in what contexts. Physiological monitoring can only support an intervention if the device remains acceptable to the wearer.

Intervention Wearables for Self-Regulation and Relief from Sensory Overload

Real-time respiratory feedback and psychophysiological regulation

Breathing feedback transforms a physiological measurement into an immediate instruction to modify behavior. The Anti-Stress device detected breathing patterns with up to 99% accuracy, responded in less than 200 milliseconds, and estimated objective indices using heart rate variability (Hou et al., 2025). In an experiment with 60 participants, the group that used the device experienced greater stress reduction than the control group, and the SUS and QUIS scales indicated greater usability than an app without biofeedback (Hou et al., 2025). Participants with higher levels of openness or neuroticism reported greater relief, suggesting that individual characteristics may moderate the response.

The proposed mechanism is behavioral and psychophysiological: the individual observes or receives guidance on their breathing and adjusts their breathing pattern during the exercise. This design may be more effective than a general recommendation to breathe deeply, because it links the instruction to a real-time signal. Nevertheless, the study does not establish the duration of the effect after the device is removed or its effectiveness during specific sensory overload. Biofeedback should be evaluated as a support for self-regulation skills, not as a substitute for contextual understanding or professional care when persistent distress is present.

Deep Pressure, Multimodal Wristbands, and Discreet Supports for People on the Autism Spectrum

Portable interventions can also provide tactile or sensory stimulation directly. Portable Hugs describes a wristband with a mobile app-controlled pneumatic chamber to apply deep pressure, featuring inflation and deflation cycles and an emergency switch (Thomas, 2024). The prototype completed 11 cycles in 15 minutes at a maximum pressure of 1 psi above atmospheric pressure, but the study presents initial performance tests and notes that calibration, safety verification, and preclinical evaluation are still required (Thomas, 2024). Therefore, its plausibility of use does not equate to proven efficacy against sensory overload.

Sereni Smart Bracelet offers a discreet wristband with customizable vibration, ASMR sounds or binaural beats, and light, designed to support self-regulation in people on the autism spectrum (Kiehl, 2025). Its user-centered design addresses the need to reduce stigma and allow the user to select modalities based on their sensory profile. However, the article presents a conceptual framework rather than experimental results on efficacy. The available evidence supports personalization as a design requirement, while reducing crises, improving attention, and enhancing daily autonomy require controlled trials with sufficient follow-up.

Summary: Toward Adaptive Systems That Connect Detection, Context, and Intervention

From isolated devices to closed-loop personalized self-regulation

The convergence of sensors, context, and intervention allows for the definition of a closed-loop system of personalized self-regulation: measure, interpret, offer an action, and verify the response. Reviews on detection already describe the technical components of data acquisition, preprocessing, feature extraction, and classification (Pinge et al., 2024). Research on respiration shows how a signal can be converted into an immediate instruction (Hou et al., 2025), while sensory prototypes extend intervention to include vibration, light, sound, or pressure (Kiehl, 2025; Thomas, 2024). Integration must preserve the user’s ability to accept, modify, or reject the system’s response.

An adaptive system should not interpret autonomic activation without context. The review on real-life monitoring requires considering behavioral state, social environment, and situation before attributing psychological meaning to a signal (Norman et al., 2025). Personalization can establish an individual baseline, adjust thresholds, and select the least intrusive modality, but machine learning must also communicate its uncertainty. Multimodal architectures offer a methodological direction, although their generalizability and interpretability remain limited (Li & Zhang, 2025).

Balancing Physiological Efficacy, Subjective Experience, and User Autonomy

Physiological efficacy is only one dimension of the outcome. In office settings, active noise cancellation improved participants’ assessment of the environment without altering performance or heart rate variability (Radun et al., 2024). In contrast, respiratory feedback reduced stress compared to a control condition and was rated as usable (Hou et al., 2025). These differences show that an intervention can alter the experience without changing physiology, or modify a physiological indicator through guided behavior.

Autonomy requires that the individual understand what is being measured, when a recommendation is triggered, and how to stop it. For autistic users, physical comfort, discretion, and stimulus selection determine continued use (Cano et al., 2024; Kiehl, 2025). A responsible system must prioritize user-controllable adjustments and prevent an automatic classification from superseding the user’s own description of discomfort. Personalization will be acceptable only if it expands regulation options without imposing a single interpretation of the sensory state.

Gaps in the Evidence and Challenges for Everyday Implementation

Ecological validity, noisy measurements, and difficulty in inferring psychological states

The main limitation is the gap between physiological signals and psychological states. Autonomic activity changes with movement, exertion, social interaction, and the environment, so a single measurement does not in itself identify stress, anxiety, or sensory overload (Norman et al., 2025). Real-world studies offer an advantage over controlled protocols, but a model’s performance depends on the dataset, preprocessing, and class definitions (Al-Alim et al., 2024). Multimodality can reduce ambiguities, although it also increases complexity and calibration requirements (Li & Zhang, 2025).

Evidence on intervention presents similar limitations. The breathing system has undergone a trial with 60 participants, while pressure devices and sensory wristbands are in the prototype or conceptualization phases (Hou et al., 2025; Kiehl, 2025; Thomas, 2024). It remains unclear which combination of signal, context, and stimulus works for different sensory profiles. Privacy poses another challenge, as continuous monitoring may reveal sensitive bodily and situational patterns (Bolpagni et al., 2024; Norman et al., 2025).

Need for Longitudinal Studies, Diverse Populations, and Comparable Evaluation Criteria

Future studies require long-term follow-up, diverse samples, and comparators that distinguish the effect of the device from the effect of the intervention. Reviews point to a lack of continuous data in naturalistic settings, limited generalizability, and insufficient interpretation of the models (Li & Zhang, 2025). In autism, small sample sizes and variability in conditions make it difficult to establish thresholds applicable across individuals (Cano et al., 2024). Comparisons should include signal accuracy, contextual validity, dropout rates, comfort, perceived control, and subjective and physiological changes.

Technical selection also requires transparent criteria regarding sampling frequency, raw data, reliability, validity, and usability (Schoenmakers et al., 2025). Without these conditions, a high classification score may mask an uncomfortable device or a model that fails outside its training set. Trials must report when the system recommends an action, what happens if the user rejects it, and whether the intervention maintains benefits after weeks or months.

Conclusion: Conditions for Responsible and Personalized Technological Sensory Regulation

Sensory regulation technology is evolving from acoustic control toward systems that detect physiological states and offer real-time interventions. Noise cancellation improves the experience, but its effects on performance and stress are limited in office settings (Radun et al., 2024). Wearable devices allow for personalized detection using multimodal signals, although the context and data quality determine their interpretation (Bolpagni et al., 2024; Norman et al., 2025). Respiratory feedback has initial experimental evidence, while deep pressure and multimodal wristbands require clinical and real-world validation (Hou et al., 2025; Kiehl, 2025; Thomas, 2024). Responsible implementation must balance measured efficacy, comfort, privacy, user control, and explicit acknowledgment of uncertainty.

References

• Al-Alim, M. A., Mubarak, R. I., Salem, N. M., & Sadek, I. (2024). A machine-learning approach for stress detection using wearable sensors in free-living environments. Computers in Biology and Medicine, 179, 108918. https://doi.org/10.1016/j.compbiomed.2024.108918
• Bolpagni, M., Pardini, S., Dianti, M., & Gabrielli, S. (2024). Personalized stress detection using biosignals from wearables: A scoping review. Sensors, 24(10), 3221. https://doi.org/10.3390/s24103221
• Brewer, A., Garceau, E. L., Baran, I., Dieffenderfer, J. P., Davila, M. I., Roubinov, D. S., Belger, A., & Bozkurt, A. Y. (2025). A wearable system for continuous physiological data collection toward real-time analysis of psychological well-being. In Proceedings of IEEE Sensors/Proceedings of IEEE Sensors … (pp. 1–4). https://doi.org/10.1109/sensors59705.2025.11330808
• Cano, S. P., Cubillos, C., Alfaro, R., Romo, A., García, M., & Moreira, F. (2024). Wearable solutions using physiological signals for stress monitoring in individuals with autism spectrum disorder (ASD): A systematic literature review. Sensors, 24(24), 8137. https://doi.org/10.3390/s24248137
• Chee, Z. J., Chang, C. Y. M., Cheong, J. Y., Malek, F. H. B. A., Hussain, S., de Vries, M., & Bellato, A. (2024). The effects of music and auditory stimulation on autonomic arousal, cognition, and attention: A systematic review. International Journal of Psychophysiology, 199, 112328. https://doi.org/10.1016/j.ijpsycho.2024.112328
• Fan, L., & Baharum, M. R. (2024). The effect of exposure to natural sounds on stress reduction: a systematic review and meta-analysis. Stress, 27(1), 2402519. https://doi.org/10.1080/10253890.2024.2402519
• Hongisto, V. O., Keränen, J. S., & Radun, J. E. (2024). Active noise-canceling headphones do not improve cognitive performance in an office noise environment. In NOISE-CON proceedings (Vol. 270, Issue 2, pp. 9434–9438). https://doi.org/10.3397/in_2024_4248
• Hou, J.-W., Chu, E. T.-H., & Lee, C.-R. (2025). Anti-Stress: A wearable device with real-time breathing feedback for stress relief. IEEE Transactions on Affective Computing, 16(3), 2479–2490. https://doi.org/10.1109/taffc.2025.3564868
• Kiehl, I. T. (2025). The Sereni Smart Bracelet: A Multimodal Sensory Wearable for Autoregulation in Individuals with Autism Spectrum Disorder. RCMOS – Revista Científica Multidisciplinar O Saber, 1(2). https://doi.org/10.51473/rcmos.v1i2.2025.1238
• Kwong, T. C., & Choy, Y. S. (2024). A customized approach to hybrid active noise cancellation system design to address auditory hyperreactivity in children with autism spectrum disorder. In *The Journal of the Acoustical Society of America* (Vol. 156, Issue 4_Supplement, p. A18). Acoustical Society of America. https://doi.org/10.1121/10.0034953
• Li, F., & Zhang, D. (2025). Multimodal physiological signals from wearable sensors for affective computing: A systematic review. Intelligent Sports and Health, 1(4), 210–222. https://doi.org/10.1016/j.ish.2025.10.001
• Myla, J. C., Acharya, T. A., Shree, K. D., Kumar, S. V., R, K., & Kumar, M. V. (2025). Wearable sensor technology for real-time stress, anxiety, and emotional health assessment (pp. 1–6). https://doi.org/10.1109/rcsm67767.2025.11507376
• Norman, G. J., Morgan, E., Raja, S., & Berntson, G. G. (2025). Wearable ANS monitoring in real life: A critical review of context-sensitive interpretation and implications for psychophysiology. Autonomic Neuroscience, 263, 103364. https://doi.org/10.1016/j.autneu.2025.103364
• Park, S., Lee, S., Howard, S., & Yi, J. (2024). Technology-Based music interventions to reduce anxiety and pain among patients undergoing surgery or procedures: Systematic review of the literature. JMIR Mhealth and Uhealth, 12, e48802. https://doi.org/10.2196/48802
• Pinge, A., Gad, V. R., Jaisighani, D., Ghosh, S., & Sen, S. (2024). Detection and monitoring of stress using wearables: a systematic review. Frontiers in Computer Science, 6. https://doi.org/10.3389/fcomp.2024.1478851
• Radun, J. E., Tervahartiala, I.-K., Kontinen, V., Keränen, J. S., & Hongisto, V. O. (2024). Do active noise-canceling headphones influence performance, stress, or experience in an office setting? Building and Environment, 266, 112102. https://doi.org/10.1016/j.buildenv.2024.112102
• Schoenmakers, M., Saygin, M., Sikora, M. T., Vaessen, T., Noordzij, M. L., & de Geus, E. J. C. (2025). Stress in Action Wearables Database: A database of noninvasive wearable monitors with systematic technical, reliability, validity, and usability information. Behavior Research Methods, 57(6), 171. https://doi.org/10.3758/s13428-025-02685-4
• Taskasaplidis, G. V., Fotiadis, D. A., & Bamidis, P. D. (2024). Review of stress detection methods using wearable sensors. IEEE Access, 12, 38219–38246. https://doi.org/10.1109/access.2024.3373010
• Thomas, V. (2024). Portable hugs: A wearable deep pressure application device for sensory overstimulation. https://doi.org/10.18260/1-2–45391