Abstract
In everyday conversation, speakers often convey their intentions indirectly, requiring listeners to infer meaning beyond the literal content of the utterance. For example, the question "Do you know the way to the station?" implies a request such as "Please tell me the way to the station." Although pragmatic inference is generally assumed to support the comprehension of such implicit intentions, the underlying neural mechanisms remain poorly understood. This study investigated the cognitive and neural processes involved in comprehending indirect utterances, using electroencephalography (EEG) recorded while participants listened to spoken dialogues. We manipulated both the contextual explicitness (explicit vs. implicit) and the temporal reference (present intention vs. past experience) of the speaker's implicit intentions. EEG analyses revealed a significant effect of contextual explicitness only in conversations involving past experiences. Specifically, in the implicit context condition relative to the explicit condition, we observed a significant positive deflection in the event-related potential and significant suppression in the θ and β frequency bands of event-related spectral perturbation. The β-band suppression was interpreted as reflecting perspective-taking by the listener. To further investigate the neural mechanisms involved, we analyzed effective connectivity among 28 regions of interest-previously identified in fMRI studies of indirect utterance comprehension-using source-localized EEG data. In the implicit context condition for past-experience conversations, we found a significant increase in information flow to the parahippocampal gyrus, suggesting a role for autobiographical memory retrieval. Multiple regression analyses showed that this connectivity was significantly associated with subscores on the Autism-Spectrum Quotient, particularly the Imagination and Communication subscales-both related to theory of mind (ToM). These findings suggest that autobiographical memory retrieval is guided by second-order ToM processes, enabling listeners to internally simulate the speaker's context retrieval. Our results challenge traditional linguistic models that conceptualize the comprehension of implicit intentions as a stepwise construction of propositional representations. Instead, they support a pragmatic inference as context search model, in which listeners actively search for a context that coherently integrates the indirect utterance with the preceding discourse.