Back

Frontotemporal cortex flexibly adapts latent structural representations

Tertikas, G.; Trudel, N.; Klein-Flugge, M.; Hauser, T. U.

2026-06-28 neuroscience
10.64898/2026.06.25.734414 bioRxiv
Show abstract

Humans excel at navigating complex environments by forming abstract structural representations that can be flexibly updated when environments change. Here, we examine how the brain dynamically reconfigures these internal models in response to covert changes in latent hierarchies. Using a novel inference task and fMRI repetition suppression, we find that stable relational knowledge is encoded in medial orbitofrontal cortex (mOFC), while structural changes trigger transient representations across hippocampal and prefrontal regions. Newly inferred associations are first encoded in anterior medial frontal cortex (amFC) and migrate ventrally to mOFC when settling. In contrast, outdated associations transiently engage frontopolar cortex and hippocampus, with the hippocampus, but not frontal areas, retaining a residual memory trace. Notably, the strength of early novel signals in amFC and hippocampus tracks individual differences in behavioural adaptation. Together, these results characterise mechanisms supporting adaptive structural reconfiguration in the human brain, with implications for cognitive inflexibility in psychiatric disorders. SIGNIFICANCE STATEMENTThe brain relies on internal models of hidden environmental structure, yet how these models are revised when the world changes remains unclear. We developed a novel task and neuroimaging approach that allowed us to trace the emergence, maintenance, and dissolution of latent structural representations in humans. Stable representations were encoded in medial orbitofrontal cortex, whereas newly formed and outdated associations followed distinct temporal trajectories across frontal and hippocampal regions. Outdated representations transiently recruited frontopolar cortex but persisted in hippocampus after becoming behaviorally irrelevant. Neural signatures of updating predicted individual differences in adaptation, revealing candidate mechanisms through which cognitive flexibility, and its impairment in psychiatric disorders, may arise.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.