Associative synaptic plasticity creates dynamic persistent activity
Wakhloo, A. J.; Clark, D. G.; Abbott, L.
Show abstract
In biological neural circuits, the dynamics of neurons and synapses are tightly coupled. We study the consequences of this coupling and show that it enables a novel form of working memory. In recurrent neural network models with ongoing Hebbian plasticity, we find that following oscillatory stimulation, neurons continue to oscillate long after the input is removed. This creates a dynamic form of memory that has no explicit storage or retrieval phases and that requires no prior knowledge of the input. We trace the mechanism of these "persistent oscillations" to an interaction between neurons and synapses that creates complex outlier eigenvalues of the connectivity matrix. This is shown both in simulation and analytically. We leverage this mechanistic understanding to generate persistent oscillations with prespecified dynamics, creating a dynamic analog of a classical Hopfield network. Our work demonstrates that coupling neuronal and synaptic dynamics enables novel forms of computation.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Modularity-dependent storage of dynamic spiking patterns: bridging micro- and mesoscopic representations 98%
- Exponential trajectories, cell size fluctuations and the adder property in bacteria follow from simple chemical dynamics and division control 96%
- Temporal stimulus segmentation by reinforcement learning in populations of spiking neurons 96%
Similar papers in this journal
- Switches to rhythmic brain activity lead to a plasticity-induced reset in synaptic weights 97%
- From space to time: Spatial inhomogeneities lead to the emergence of spatio-temporal activity sequences in spiking neuronal networks 97%
- Tuning network dynamics from criticality to the chaotic balanced state 97%
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.