The Neural Infrastructure Related to Conversational Engagement

Sammanfattning

Conversational engagement is a motivating force that compels a speaker to be an interested, contributing and attentive partner in a conversation. This thesis explores the neural infrastructure related to that force and how it is modulated by the engagement of the interlocutor. Conversational engagement, as it is used in this thesis, is expressed through features in the communicative signal usually studied in the fields of pragmatics and conversational analysis. Our research aims to add to the understanding of what makes an interlocutor engaging, from a neurolinguistic point of view. This is an exploration study into which brain areas are sensitive to engagement and how it is modulated by how an experiment operator engages with a research participant. The data is from 28 participants in conversation while undergoing fMRI scans, combined with annotations of the recorded conversation. The annotation identifies how engaged the participant was in the conversation on a speaker-turn-by-turn basis. We use the annotation data to define fMRI events and do statistical parametric mapping and ROI analysis to investigate how engagement level affects brain activity. The results show that regions sensitive to engagement overlap with parts of canonical networks related to language and theory of mind processes. Specifically, the superior and middle temporal gyri, the temporo-parietal junction and surrounding areas are sensitive to engagement. The neural activity elicited by engagement does not increase linearly with the level of engagement, but forms an inverted U-shape; speaking turns done at a low or high level of engagement provoke less overall activity in contrast to turns in which the participant manages the conversation and builds a framework of what their interlocutor understands about the conversation. Such turns were annotated as medium engagement and showed the highest activity. The observations in this thesis illuminate the role of joint epistemic modelling in conversation and the cognitive demand of engaging in a conversation at different levels.