Abstract
Conversation is a fundamental human behaviour that requires rapid coordination between speaking, listening, and turn-taking, yet datasets capturing its neural dynamics in natural interaction remain scarce. Hyperscanning EEG is particularly valuable for this purpose because it records both interlocutors simultaneously, enabling the study of speaker–listener coupling, response timing, and dyadic coordination during live exchange. Here we present DUET ( D yadic U nderstanding, E EG and T urn-taking), a hyperscanning dataset for studying natural French face-to-face conversation. The dataset comprises recordings from 18 dyads, or 36 French-speaking adults, performing the Diapix collaborative spot-the-difference task across eight 4-minute face-to-face conversation blocks. For each participant, EEG was recorded from 36 participants; most recordings used 64-channel EEG, with one pilot dyad recorded using 32 electrodes. The public release includes raw EEG recordings, precomputed ICA decompositions for reuse in downstream preprocessing as well as various features derived from the audio and manually corrected transcripts.
Full text
1,210 characters
· extracted from
oa-doi-fallback
· click to expand
Abstract
Conversation is a fundamental human behaviour that requires rapid coordination between speaking, listening, and turn-taking, yet datasets capturing its neural dynamics in natural interaction remain scarce. Hyperscanning EEG is particularly valuable for this purpose because it records both interlocutors simultaneously, enabling the study of speaker–listener coupling, response timing, and dyadic coordination during live exchange. Here we present DUET (Dyadic Understanding, EEG and Turn-taking), a hyperscanning dataset for studying natural French face-to-face conversation. The dataset comprises recordings from 18 dyads, or 36 French-speaking adults, performing the Diapix collaborative spot-the-difference task across eight 4-minute face-to-face conversation blocks. For each participant, EEG was recorded from 36 participants; most recordings used 64-channel EEG, with one pilot dyad recorded using 32 electrodes. The public release includes raw EEG recordings, precomputed ICA decompositions for reuse in downstream preprocessing as well as various features derived from the audio and manually corrected transcripts.
Competing Interest Statement
The authors have declared no competing interest.
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.