Neural dynamics of schema-based associative learning between objects and scenes with SQUID and OPM MEG

About This Grant

Project Summary The goal of this project is to understand how prior knowledge shapes new learning. Research has shown that the congruency between new information and existing schemas can facilitate new associative learning. For example, people are better at remembering the association between a specific object and a specific scene when their categories are semantically related. Although this effect is well-documented behaviorally and linked to certain brain systems, the precise mechanisms by which schema congruency enhances hippocampal and cortical processing during visual associative learning are unknown. Here we explore the fine-grained neural dynamics involved in schema-based learning. We test a prominent mechanistic theory — the representational coherence hypothesis — which proposes that schema congruency benefits associative learning by causing stimuli to have more stable and resonant neural representations. We test this theory with two varieties of magnetoencephalography (MEG), one an established method (SQUID) and the other a next-generation technology (OPM), both of which provide a more time-resolved characterization of object and scene processing than fMRI and better separation and localization of signals to brain regions than EEG. Participants undergoing SQUID (Study 1) or OPM (Study 2) MEG will associate pairs of natural objects and scenes. The object in each pair will be either schema-congruent (SC) or schema-incongruent (SI) with the scene category. Subsequent memory recall for the scene associated with each object will be probed after a 10- minute delay. Pattern classifiers will be trained to decode scene and object categories in a separate localizer task, then applied to the encoding and retrieval trials of SC and SI pairs. We will compare the temporal evolution of classifier evidence for the scene and object in SC and SI pairs and examine the phase coherence of these timeseries, the alignment of the classifier outputs with source-localized hippocampal theta phases, and the interaction between medial prefrontal cortex and hippocampus. This project will yield critical new insights into why some information is easy to learn and how this learning is supported by brain dynamics and systems.