Learning of Precise Spike Times with Homeostatic Membrane Potential Dependent Synaptic Plasticity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Synaptic Plasticity

Background:

• Precise neuronal firing patterns are crucial for brain functions like sensory processing and motor control.
• Current synaptic plasticity rules often lack the temporal precision needed for learning specific spike times.
• Existing supervised learning models for spike timing lack biological plausibility.

Purpose of the Study:

• To propose a novel, local, and unsupervised synaptic plasticity mechanism for learning temporal activity patterns.
• To introduce Membrane Potential Dependent Plasticity (MPDP) as a biologically plausible model for synaptic adaptation.
• To demonstrate MPDP's ability to learn and store spatio-temporal spike associations.

Main Methods:

• Developed a plasticity rule based on the requirement of a balanced membrane potential.
• Utilized postsynaptic voltage as the primary signal for synaptic change, termed MPDP.
• Investigated the interplay of MPDP with spike after-hyperpolarization to achieve spike-timing sensitivity.

Main Results:

• MPDP reproduces Hebbian Spike-Timing-Dependent Plasticity for inhibitory synapses.
• MPDP can distinguish between relevant (teacher) and irrelevant (spurious) spikes, providing a basis for comparing actual and target activity.
• The rule demonstrates high storage capacity for spike associations and robust memory retrieval under noisy conditions.

Conclusions:

• MPDP offers a biophysically plausible mechanism for unsupervised learning of temporal target activity patterns.
• The sensitivity to subthreshold membrane potential allows for robust learning and memory recall.
• This mechanism provides a neuronal basis for precise temporal computations in the brain.