Associative memory of phase-coded spatiotemporal patterns in leaky Integrate and Fire networks

Area of Science:

• Computational Neuroscience
• Neural Dynamics

Background:

• Neural networks exhibit complex collective dynamics.
• Spike-timing-dependent plasticity (STDP) is a key learning mechanism.
• Precise relative spike timing is crucial for neural coding.

Purpose of the Study:

• To investigate the storage and replay of phase-coded spike patterns in a Leaky Integrate and Fire network.
• To explore how network excitability influences working regimes and replay activity.
• To develop a coding scheme where phase, rate, and frequency are dissociable.

Main Methods:

• Utilized a Leaky Integrate and Fire neural network model.
• Employed STDP for storing phase-coded spike patterns in network connectivity.
• Introduced an order parameter to quantify pattern similarity and storage capacity.
• Modulated spiking thresholds to alter oscillation frequency and spike counts.

Main Results:

• Identified distinct working regimes (transient or persistent replay) based on network excitability.
• Demonstrated the ability to selectively replay stored phase-coded patterns at different timescales.
• Showcased a coding scheme where phase relationships are preserved while rate and frequency are modulated.
• Confirmed robustness against noise and neuronal parameter heterogeneity.

Conclusions:

• Leaky Integrate and Fire networks can store and replay precise spike timing patterns as dynamic attractors.
• Network excitability dictates the nature of replay, enabling flexible information processing.
• The proposed coding scheme allows for independent manipulation of phase, rate, and frequency, enhancing neural coding capabilities.
• The retrieval process ensures stable phase relationships and oscillatory frequency even under noisy conditions.