Xia Liang

This study compares how causal knowledge is represented in human brain and large language models through reactivation patterns and finds similar intrinsic manifold structures.

This paper models spontaneous thought transitions using a foraging-like framework, showing human narratives follow value-guided memory patch dynamics, with similar patterns in large language models.

This study uses energy landscape modeling on fMRI data to identify brain states and their transitions during movie watching, revealing thalamic modulation of neural dynamics.

This paper identifies a core brain network with high connectivity and variability as potential neural correlates of consciousness, showing it is modulated during propofol-induced unconsciousness.

The paper investigates how phonological features of spoken word-forms modulate auditory N1 repetition-attenuation, finding that low-frequency pseudo words show greater attenuation than other conditions.