BCIwiki (bciwiki.com) — Lightweight EEG electrodes worn around the ear, paired with portable hearing aid research hardware, decoded which talker participants were attending to in both seated and walking conditions, researchers at the Carl von Ossietzky University of Oldenburg in Germany reported in Hearing Research on October 5, 2026 (PMID 42832897; DOI 10.1016/j.heares.2026.109817). Co-authors included Sonova Consumer Hearing GmbH.
Auditory attention decoding is a proposed basis for neuroadaptive hearing technology: it reads how the brain tracks speech to infer which speaker a user is listening to. The authors note its feasibility in naturalistic, mobile settings has been underexplored. Participants performed an attention task seated and while walking, across single-speaker and dual-speaker scenes with cafeteria background noise, using cEEGrids around-the-ear electrode arrays.
The team applied backward (decoding) and forward (encoding) models to continuous speech. Decoding replicated established trends: reconstruction accuracy was highest for a single speaker, followed by the attended voice in the dual-speaker condition, then the ignored voice. The pattern held in both seated and walking contexts.
Post hoc forward modeling, however, revealed an artifactual response in the walking condition that biased reconstruction and prediction accuracy. Further analysis suggested hardware-related interference, possibly from hearing aid cables contacting the electrodes. The artifact was speech-locked and detectable only through the forward model, the authors said.
The researchers conclude that the results show feasibility for a lightweight, unobtrusive EEG system in stationary scenarios, and call for robust hardware integration, real-time validation and user-centered design in future applications.