Spiking neural network model for memorizing sequences with forward and backward recall.

Area of Science:

• Computational Neuroscience
• Cognitive Neuroscience
• Systems Neuroscience

Background:

• Sequential memories are crucial for cognition.
• Existing models often lack detailed neuronal mechanisms for sequence storage and retrieval.
• Neurobiological and psychological evidence suggests oscillatory dynamics play a role in memory.

Purpose of the Study:

• To develop a biologically plausible computational model for memory storage and retrieval of event sequences.
• To investigate the role of oscillatory networks and synaptic plasticity in sequence memory.
• To simulate both forward and backward replay of neural sequences.

Main Methods:

• An oscillatory network of conductance-based Hodgkin-Huxley spiking neurons was designed.
• The network architecture features two layers: a lower layer for event representation and an upper layer for sequence tagging.
• Synaptic plasticity was implemented using spike-timing-dependent plasticity (STDP) and anti-STDP learning rules.

Main Results:

• The model successfully stores and retrieves sequences of events, even with overlapping elements.
• Simulations demonstrated the model's ability to handle complex sequences and difficult cases.
• STDP and anti-STDP rules enabled the simulation of forward and backward replay of neural activity, respectively.

Conclusions:

• The proposed oscillatory network provides a viable model for sequential memory.
• Synaptic plasticity rules like STDP are essential for learning and replaying event sequences.
• The model offers insights into the neural mechanisms underlying memory sequence dynamics.