READ / RESEARCH · VIBROTACTILE BREATHING AND EVERYDAY USE

This paper tests a wearable that uses gentle vibrations on the abdomen to guide breathing. People practise matching their breathing to a personalised rhythm, then use it during a stressful task. The study found a modest reduction in self-reported anxiety, but following the rhythm became harder under stress. For my project, the key question is whether a discreet cue can help without becoming another task to manage.

Why I looked at this

My sticky note identifies the core interaction: two tactors attached to the abdomen deliver personalised vibrations that participants can synchronise their breathing with. A tactor is a small component that produces a touch sensation; here, it vibrates.

This interests me because a breathing cue could be delivered without asking the traveller to watch a screen or listen to instructions. The paper also examines engagement with the technology, which is especially relevant when someone is already dealing with airport demands.

How the technology works

The Personalizable, Inconspicuous Vibrotactile (PIV) pacer uses a weaker vibration to indicate inhalation and a stronger vibration to indicate exhalation. The cycle repeats at a personalised pace, generally between five and nine breaths per minute in this study.

Participants’ breathing during a meditation stage informed the pacing. Research assistants helped adjust the vibration settings, and participants practised matching their breathing to the cues before the stressful task.

This is an active breathing aid. The wearer has to notice the cue, understand which phase it represents and coordinate their breathing with it. The device does not automatically change someone’s breathing merely by being worn.

What the researchers tested

The analysis included 97 participants, with 44 in the treatment group and 53 in the control group. Participants completed two rounds of a timed word-association task designed to create cognitive stress. The treatment group received the breathing intervention during the second stressor.

The researchers measured anxiety using the brief STAI-6 questionnaire, alongside ratings of positive and negative feelings. Comparing the two post-stressor measurements, the treatment group showed a modest anxiety reduction, with a reported within-group effect size of d = .33. The control group changed very little.

The study found no significant intervention effect on the broader positive- and negative-affect measures. I would therefore describe the result as evidence of a limited anxiety benefit in this laboratory task, rather than a general improvement in emotional wellbeing.

The most useful finding: support can require effort

Participants found synchronising their breathing with the vibrations harder than simply noticing the vibrations or distinguishing the two phases. Synchronisation also became more difficult when the stressor was introduced.

This is particularly relevant to my project. A rhythm that feels easy during a quiet introduction may become difficult to follow while navigating, processing an announcement or dealing with a change of plan.

Successful delivery of a cue does not mean the person can comfortably use it. Testing under realistic demands matters as much as testing in a calm setting.

The paper’s exploratory analysis also links engagement with differences in response. Wanting to turn the vibrations off was an important feature in its predictive model. This supports investigating acceptability, but it does not establish a reliable method for deciding who will benefit.

Where biosensing fits

The study collected physiological signals, including breathing and electrodermal activity. However, this does not mean the pacer continuously interpreted stress and adapted its rhythm in real time. Personalisation and physiological measurement are different from an autonomous adaptive system.

The machine-learning analysis explored differences between participants after data collection. It was not an implemented controller, and its reported held-out predictive performance was poor. I would not use it as evidence that AI can already select the right breathing response for an individual.

For my project, a separate possibility is that contextualised physiological changes could prompt an offer of a familiar breathing rhythm. The user would choose whether to start it, retain control over the settings and be able to stop immediately.

Limitations for my intended audience

Participants were recruited through university pools, and eligibility excluded people with a history of neurological or psychological disorders, among other conditions. This was not a study of autistic adults or sensory overload.

The task lasted four minutes and involved timed word problems. It does not reproduce the sensory, social and physical demands of an airport journey. The laboratory setup also used taped abdominal tactors and noise-cancelling headphones, so discreet everyday wear remains a design question.

What I am taking forward

I am keeping personalised tactile breathing cues as a possible response worth investigating. This paper provides more direct evidence about anxiety during a task than a device study concerned only with producing sensations, while still leaving important questions about audience and context.

My priority would be to explore whether the cue feels supportive or demanding, how people want to start and stop it, and when they would prefer no cue at all. A manual breathing aid would be an important comparison before adding biosensing-assisted prompts.

A useful breathing aid needs to fit the person’s available attention. The interaction should remain optional, understandable and easy to leave.

Source

Miri, P. et al. (2020). ‘Evaluating a Personalizable, Inconspicuous Vibrotactile (PIV) Breathing Pacer for In-the-Moment Affect Regulation’. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems.

Key sections: Method and Participants; State Affect Regulation Measures; User-Technology Engagement; Model Selection Procedure; Discussion.