Sequence structure organizes items in varied latent states of working memory neural network

In memory experiences, events do not exist independently but are linked with each other via structure-based organization. Structure knowledge largely influences memory behavior, but how it is implemented in the brain remains unknown. Here, we combined magnetoencephalogram (MEG) recordings, computational modeling, and impulse-response approaches to probe the latent states when subjects held a list of items in working memory (WM). We demonstrate that sequence structure reorganizes WM items into distinct latent states, i.e., being reactivated at different latencies, and the reactivation profiles further correlate with recency behavior. In contrast, memorizing the same list of items without sequence requirements disrupts the recency effect and elicits comparable reactivations. Finally, computational modeling reveals a dominant function of high-level representations that characterize the abstract sequence structure, instead of low-level information decaying, in mediating sequence memory. Taken together, sequence structure shapes the way WM items are stored in the brain and essentially influences memory behavior.