Motion-Steered Noise Management in Hearing Aids Reduces Listening Effort and Stress Indexed by Pupil Size and Heart...

Head and body movements influence real-world hearing outcomes, yet they are often overlooked in clinical evaluations of hearing aids. This reduces ecological validity and limits assessments of motion-based features. To overcome this, the present study introduces a novel experimental approach that incorporates natural human behavior during listening, utilizing virtual reality (VR) in an anechoic chamber equipped with a spherical 64-channel loudspeaker array and a naturalistic two-stage listening task. We hypothesized that heart rate, obtained unobtrusively with a wrist-worn device, and pupillometry would be sensitive to differences in listening difficulty and hearing aid technology.

Participants (N = 32) freely moved their head and body orientation to complete two listening tasks with a natural transition between them. First, they located a target talker among 15 source locations, then comprehended a 33-sec news clip from the target location. Task difficulty was manipulated by introducing either two or four interfering talkers at various spatial locations. Beat-to-beat heart rate was recorded, and listening stress was calculated as the relative change in beats per minute based on the number of interfering talkers. In addition, pupil dilation was recorded with head-mounted cameras in the VR headset to estimate sustained listening effort. Head and gaze direction were tracked throughout the experiment. Participants used two different hearing aid models, both featuring noise management systems based on Deep Neural Networks. However, only one model utilized head movement information from an embedded accelerometer to adjust noise management in response to detected motion activity.

Speech comprehension performance decreased, and heart rate and pupil dilation increased in the four interfering talkers condition compared with the two interfering talkers condition. Comparing the two hearing aid models, speech comprehension was similar, but heart rate and pupil dilation were significantly higher with the model that did not incorporate head movement information in its noise management. We also found that pupil dilation and heart rate were significantly and positively associated across trials within participants.

VR combined with unobtrusive heart rate monitoring is an effective method for evaluating hearing aids in naturalistic conditions. This approach allows for head movements and changes to body orientation, which are necessary for assessing hearing aid features that steer by motion detection. Relative heart rate was sensitive to both task difficulty and hearing aid technology, aligning with changes in pupil dilation and suggesting that motion-steering of hearing aid noise management can benefit speech comprehension following a localization task by reducing listening stress and effort.