Synaptic Theory of Working Memory for Serial Order
Mongillo, G.; Tsodyks, M.
Show abstract
AO_SCPLOWBSTRACTC_SCPLOWPeople effortlessly remember short, but otherwise arbitrary, sequences of familiar stimuli over a brief period of time. This deceptively mundane ability is core for purposeful behavior and cognition. Surprisingly, however, it remains largely unexplained. Here, we propose that both the stimuli and their order of occurrence are encoded by transient synaptic enhancement over multiple time scales. To substantiate this proposal, we extend our previously-proposed synaptic theory of Working Memory (WM) to include synaptic augmentation besides short-term depression and facilitation, consistently with experimental observations. Like facilitation, augmentation builds up with repetitive activation but persists for much longer. We find that the long time scales associated with augmentation naturally lead to the emergence of a "primacy gradient" in the synaptic efficacies that can be used to reconstruct the order of presentation at recall. The novel theory accounts for prominent features of the behavior of humans recalling lists of items, makes testable predictions and, intriguingly, suggests that WM capacity limitations result from a failure in retrieving, rather than storing, information. Taken together, our results advance the understanding of the neuronal mechanisms underpinning the encoding of serial order and offer novel insights into the origin of WM capacity.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Optimal synaptic dynamics for memory maintenance in the presence of noise 96%
- A prefrontal network model operating near steady and oscillatory states links spike desynchronization and synaptic deficits in schizophrenia 96%
- Dynamic control of sequential retrieval speed in networks with heterogeneous learning rules 96%
Similar papers in this journal
Similar papers in this journal
"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.