Related Experiment Videos
Neurocomputational models of working memory
D Durstewitz1, J K Seamans, T J Sejnowski
1Howard Hughes Medical Institute, Salk Institute for Biological Studies, Computational Neurobiology Laboratory, La Jolla, California 92037, USA.
Nature Neuroscience
|December 29, 2000
Summary
Neural models explain how persistent neural activity in working memory remains stable despite noise. Different mechanisms like recurrent excitation and single-cell bistability are explored for sustained brain function.
Area of Science:
- Neuroscience
- Computational Neuroscience
Background:
- During working memory tasks, single neurons in monkeys exhibit elevated firing rates without external cues.
- This persistent activity is crucial for maintaining information over time.
Purpose of the Study:
- To explore computational models of mechanisms underlying persistent neural activity in working memory.
- To investigate how different neural and synaptic properties contribute to stable information representation.
Main Methods:
- Computational modeling of neural networks and single neurons.
- Simulations exploring recurrent excitation, synfire chains, and single-cell bistability.
Main Results:
- Models demonstrate stable persistent activity in the presence of noise and distractors.
- Synaptic and voltage-gated conductances are shown to be critical for sustained neural firing.
- Neuromodulation's role in the robustness of persistent activity was investigated.
- Models explored the maintenance of novel items and continuous attractor states.
Conclusions:
- Computational models provide insights into the stability and mechanisms of persistent neural activity during working memory.
- Further research is needed to fully capture the diverse neural dynamics involved in working memory.