Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Predictive processing enhances neural alignment of word meaning during dyadic interactions
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Mia Steffen1,2, Emilia Kerr1,2, Daria Goriachun1,2, Benjamin Morillon1,3, Kristof Strijkers1,2; 1Aix-Marseille University, 2CNRS, 3Inserm
Context. Recent research adopts dyadic approaches to study the cognitive neuroscience of language, showing that interlocutors align at both linguistic and neural levels during interaction. While linguistic alignment concerns the phenomenon where interlocutors converge on each other’s language use, neural alignment is the observation that a speaker and listener’s brains become correlated during communication. Yet, whether both these phenomena are related and which mechanisms underlie both types of alignment remains unclear. Objective. This study tests whether predictive processing constitutes a mechanism linking both linguistic and neural alignment. The idea here is that because we try to predict what our partner is about to say, the same representations can become activated in the speaker and listener’s mind, resulting in alignment of their brain activity and language behaviour. Methodology. We employ EEG hyperscanning in an interactive paradigm where two participants in a dyad play a simple semantic association game with animal or tool names. A trial starts with a sentence context that is either highly predictable (e.g., ‘Man’s best friend is a …’) or unpredictable (e.g., ‘Outside there is a …’) with respect to an upcoming picture. One interlocutor names the picture (e.g., ‘dog’), while the other produces a semantic associate (e.g., ‘cat)’; roles alternate between the interlocutors on a trial-by-trial basis. Behavioural data and Inter Brain Synchrony (IBS) of the lexico-semantic contrast (animal versus tool words) are analysed depending on contextual predictability. Preliminary Results (N = 26 dyads). At the behavioural level, predictable contexts facilitated speech production for both interlocutors, with a greater facilitation effect for the respondent. At the neural level, when contrasting interactions about animals versus tools, significant IBS appeared in theta and beta frequency ranges when preceded by predictable contexts. This effect emerged during the interactive gap; that is, the time window between interlocutors' speech productions. Conclusion. In this study we observed that prediction elicited meaning-specific neural alignment of words in the theta and beta bands between interlocutors. We propose that the theta coupling may highlight the prediction of syllabic and lexical information, while the beta coupling may reflect a predictive top-down signal. Most importantly for present purposes, contextual predictability modulated both these synchrony patterns and the speed of the naming behaviour between the interlocutors, suggesting that predictive processing can shape the neural alignment of linguistic information during conversation.
Topic Areas: Language Production, Speech Perception