We built an inconspicuous breathing pacer system that applies vibrations onto the body with which a user may synchronize their breathing to bring down arousal in high-stress situations. We tested the system on neurotypical adults and found that the treatment group had reduced anxiety in the presence of a cognitive stressor (d = .33, p = .004) compared to the control group.
The study took place in three stages. In Stage One, participants were recruited and checked for eligibility; a total of 97 participants (61 females, 36 male) took part in the study. After consenting to the study on-site, each participant completed surveys, such as Body Sensation surveys and the Big Five Inventory. Next, the participant’s baseline physiology was measured and answered a questionnaire asking how positive or negative he or she felt at the time. Each participant also took the STAI-6, a six question survey measuring state anxiety. Finally, the participant practiced a mindful, slow-paced breathing exercise to estimate BPM.
In Stage Two, the vibrations’ pattern, frequency, amplitude were customized to each participant based on his/her baseline BPM and breathing ratio (BR). This was done so that the vibrations are comfortable to each user. Once the participant was satisfied in the pattern, he/she synchronized their breathing with the vibrations and completed the same questionnaire documenting how positive or negative he or she felt while breathing with the pacer. Physiology was measured during the synchronization phase. The participants completed the STAI-6 at the end of this phase.
In Stage Three, the participants were introduced to the creativity task: a timed word puzzle. After three practice problems, the participant was asked how positive or negative he/she felt (Pre-stressor) and took the STAI-6. Next, two tests with 27-word puzzles each were administered; physiology was measured for two minutes before and two minutes after each test. Two minutes after the first test, participants were given those same questionnaires about how positive and negative they were feeling (Post-stressor 1) and the STAI-6. In the second test, the experimental group was administered their personalized vibrations. They were told: “Please allow your breathing to conform with the sensation of the pacer if you can, whenever you can.” The control group received no vibrations for the second test. After, each group was given the same questionnaire of how positive and negative they were feeling and the STAI-6 (Post-stressor 2). Finally, the treatment and control groups filled out remaining open-ended questions about their experience. Participants were then debriefed and compensated accordingly. Please note that all participants took the STAI-6 test, a six question survey that measures state anxiety at five different conditions:
—Baseline (to capture baseline physiology at the end of Stage One)
—Synchronization to Personalized Vibrations (at the end of Stage Two)
—Pre-Stressor 1 (just before taking the first test of word puzzles in Stage Three)
—Post Stressor 1 (just after taking the first test of word puzzles in Stage Three)
—Post Stressor 2 (just after taking the second test of word puzzles in Stage Three)
We predicted that the control and treatment groups would show approximately equal levels of anxiety at post-stressor 1, and that the treatment group would show reduced anxiety at post-stressor 2. We observed the interaction effect F(1,9) = 5.55, eta squared = .055, p = .02 and the main effect effect of condition, F(1,94) = 6.01, eta squared = .060, p = .01. What we found was that the treatment group showed less anxiety than the control group at post-stressor 2 compared to post-stressor 1 (treatment group mean difference post-stressor 1 post-stressor 2 = 3.26, d = .33, p = .004), but the control group did not showed reduced anxiety at post-stressor 2 compared to post-stressor 1 (control group mean difference post-stressor 1 vs 2 = .06 , d = .05, p = .93 ).
In this figure, you can see the distribution of the STAI-6 scores across the two conditions (Post-stressor 1 and Post-stressor 2) and the two groups (treatment and control). Again, these distributions show that the treatment group showed less anxiety than the control group when comparing STAI-6 scores at Post-stressor 1 and at Post-stressor 2.
In this linear mixed model with random intercepts, you can see the STAI-6 scores for every participant, separated for the control and treatment groups. It is important to note that since the treatment group received vibrations in the block with the second stressor (the second test of word puzzles) while the control did not, we are most interested in comparing STAI-6 scores between Post-stressor 1 and Post-stressor 2. The red lines are from a linear-mixed model which incorporate the individual difference as random effect. In the Treatment Group, you can see the red lines slope downwards, revealing a drop in STAI-6 scores from Post-stressor 1 to Post-stressor 2.