Spikes as perturbations of resonant neural circuits: an RLC framework with testable predictions

Area of Science:

• Computational neuroscience
• Electrophysiology
• Systems neuroscience

Background:

• Current computational neuron models often simplify subthreshold membrane dynamics to a leaky RC circuit, neglecting resonant properties.
• Experimental data reveal excitable membranes possess band-pass impedance profiles with tunable resonant peaks, unrepresentable by first-order RC models.

Purpose of the Study:

• To develop a novel computational framework, the spike-as-perturbation model, using a parallel RLC membrane circuit.
• To investigate how this RLC model captures transient dynamics and state variables missed by traditional RC models.
• To demonstrate a primitive computation (phase-based temporal discrimination) enabled by this model and its network-level implications.

Main Methods:

• Developed a spike-as-perturbation framework with an equivalent parallel RLC membrane model.
• Analyzed the biological grounding and validity domain of the RLC reduction.
• Demonstrated phase-based temporal discrimination using a spike-timing readout mechanism.
• Compared RLC model performance against a matched first-order RC model.

Main Results:

• The RLC model reveals that post-perturbation ringdowns encode circuit identity and timing information.
• RLC model exhibits a longer sensitivity half-life (44 ms) compared to RC decay (17 ms) for specific parameters.
• Identified the quality factor (Q) as a key neuromodulatory control variable for the membrane's transient memory horizon.
• Demonstrated readout reliability in quiescent states, suppressed under high-conductance bombardment.

Conclusions:

• The parallel RLC neuron model offers a more biologically realistic representation of subthreshold dynamics and computation.
• This framework provides a mechanism for temporal discrimination and short-term memory, modulated by parameters like Q.
• The model's predictions are falsifiable across cellular, network, decoding, and population levels, suggesting potential for in vivo relevance in specific states.