Search Papers | Poster Sessions | All Posters
Poster B142 in Poster Session B - Thursday, August 8, 2024, 1:30 – 3:30 pm, Johnson Ice Rink
Electrophysiological neural mechanisms for domain-general intelligence
Runhao Lu1 (), Nadene Dermody1, John Duncan1, Alexandra Woolgar1; 1MRC Cognition and Brain Sciences Unit, University of Cambridge
Understanding the mechanisms supporting domain-general intelligence is crucial for both cognitive neuroscience and artificial intelligence. While human fMRI studies have identified a frontoparietal multiple-demand network that contribute to multiple tasks, it is largely unknown whether the human brain supports multiple tasks with common electrophysiological responses. Here, we recorded magnetoencephalography and electroencephalography (MEG/EEG) signals while participants completed three different cognitive tasks with different content (alphanumeric vs. colour stimuli) and cognitive demand (easy vs. hard). After separating the oscillatory and the aperiodic components of the electrophysiological signals, we used multivariate pattern analysis (MVPA) to decode task demand for each subtask. We found that both oscillatory and aperiodic components could decode task demand for all six subtasks. Aperiodic broadband power showed the strongest generalisability on coding task demand across different subtasks. Source estimation results showed distinct spatial patterns for domain-general oscillatory and aperiodic components, with the aperiodic broadband power overlapping with the frontoparietal multiple-demand network. Our findings suggested the existence of oscillatory and aperiodic electrophysiological mechanisms in support of human domain-general cognition, which provides a novel way to understand how domain-general intelligence arises and might inspire relevant research in the fields of neuroscience and artificial intelligence.
Keywords: General intelligence Generalisability Neural oscillations Aperiodic activity