A calming hug: Design and validation of a tactile aid to ease anxiety

Abstract

Anxiety disorders affect approximately one third of people during their lifetimes and are the ninth leading cause of global disability. Current treatments focus on therapy and pharmacological interventions. However, therapy is costly and pharmacological interventions often have undesirable side-effects. Healthy people also regularly suffer periods of anxiety. Therefore, a non-pharmacological, intuitive, home intervention would be complementary to other treatments and beneficial for non-clinical groups. Existing at-home anxiety aids, such as guided meditations, typically employ visual and/or audio stimuli to guide the user into a calmer state. However, the tactile sense has the potential to be a more natural modality to target in an anxiety-calming device. The tactile domain is relatively under-explored, but we suggest that there are manifold physiological and affective qualities of touch that lend it to the task. In this study we demonstrate that haptic technology can offer an enjoyable, effective and widely accessible alternative for easing state anxiety. We describe a novel huggable haptic interface that pneumatically simulates slow breathing. We discuss the development of this interface through a focus group evaluating five prototypes with embedded behaviours ('breathing', 'purring', 'heartbeat' and 'illumination'). Ratings indicated that the 'breathing' prototype was most pleasant to interact with and participants described this prototype as 'calming' and 'soothing', reminding them of a person breathing. This prototype was developed into an ergonomic huggable cushion containing a pneumatic chamber powered by an external pump allowing the cushion to 'breathe'. A mixed-design experiment (n = 129) inducing anxiety through a group mathematics test found that the device was effective at reducing pre-test anxiety compared to a control (no intervention) condition and that this reduction in anxiety was indistinguishable from that of a guided meditation. Our findings highlight the efficacy of this interface, demonstrating that haptic technologies can be effective at easing anxiety. We suggest that the field should be explored in more depth to capture the nuances of different modalities in relation to specific situations and trait characteristics.

Fig 1. Photos of the primary prototypes.

Panel A: Photos of the five primary prototypes presented to participants in the focus group. Panel B: Components of prototype number four (Purring and Breathing), to demonstrate internal components of the cushions; C) outer soft casing, D-F) inner padding, G) Adafruit Gemma circuit board, H) power cable, I) servo motor, J) acrylic disks to hold servo motor.

Fig 2. Participant ratings of the primary prototypes.

Mean and standard deviation (std) of the valence ratings given by 23 of the 24 participants for each cushion. One participants’ data was excluded due to missing values. Confidence intervals of 90% and 95% are included for visual reference calculated using the Cousineau-Morey correction [79]. Participants recorded ratings as integer values between 0 (unpleasant) to 10 (pleasant).

Fig 3. Diagram and photos of the final haptic interface.

Panel A: Diagram of the mechanism driving the haptic sensation; E) Arduino, F) cooling fan, G) transistor, H) power cable to motor, I) motor (hidden) to drive crank, J) crank and slider, K) syringe pump that drives the inflation and deflation of the interface. Panel B: Photos of the interface front and back. Panel C: Inner parts of the interface; L) access zip, M) delrin casing for pneumatic chamber, N) smooth fabric to reduce audible friction, O) pneumatic chamber. Panel D: Photo of the interface being held in a home environment.

Fig 4. Stages of the experiment.

Flow diagram illustrating the nine stages of the experiment. The four anxiety measures T1, T2, T3 and T4 are highlighted. The symbols * and △ label the intervention phases (*) and the anxiety-inducing mathematics test (△), as used in results figures later.

Fig 5. STAI anxiety scores for within-subjects factor.

Comparison of within-subjects factor (time) for mean STAI Anxiety measures at each time T1 (baseline), T2 (post intervention 1, anticipating test), T3 (post test, pre intervention 2) and T4 (post intervention 2), including standard deviation (Std) errorbars, as well as the 90% and 95% confidence intervals (CI). Confidence intervals are calculated using the Cousineau-Morey correction [79] and therefore provide visual reference as to statistically significant changes over the within-subjects factor of time.

Fig 6. STAI anxiety scores against baseline Test-Anxiety scores between T1 and T2.

Relationship between the participants’ change in STAI Anxiety scores and their baseline Test-Anxiety (WTAS) scores—from T1 (baseline) to T2 (post intervention 1, anticipating test). Line of best fit (LSR) shown for each condition using a linear Least Squares Regression model. Shaded regions highlight negative change in STAI, i.e. reduction in anxiety.