Auditory EEG viability for Brain Passage Retrieval

Determine whether electroencephalography signals recorded during auditory stimulus presentation can serve as effective query representations for Brain Passage Retrieval, enabling retrieval for voice-based interfaces and for users with visual impairments.

Background

Brain Passage Retrieval (BPR) maps EEG signals directly to dense passage representations, bypassing intermediate text decoding. Prior BPR work has relied exclusively on visual EEG recorded during reading, leaving the applicability of auditory EEG untested despite the relevance of audio-centric contexts such as conversational search and accessibility needs for visually impaired users.

Establishing whether auditory EEG can yield effective retrieval performance is critical for broadening BPR beyond visual stimuli and for supporting inclusive brain-machine interfaces that operate in voice-based environments or in the absence of visual input.

References

However, existing BPR research exclusively uses visual stimuli, leaving critical questions unanswered: Can auditory EEG enable effective retrieval for voice-based interfaces and visually impaired users? Can training on combined EEG datasets from different sensory modalities improve performance despite severe data scarcity?

— Auditory Brain Passage Retrieval: Cross-Sensory EEG Training for Neural Information Retrieval  (2601.14001 - McGuire et al., 20 Jan 2026) in Abstract (page 1)