Metabolic energetics underlying attractors in neural models

Area of Science:

• Computational neuroscience
• Neuroenergetics
• Systems neuroscience

Background:

• Neural population modeling, including neural attractors, aids understanding brain function.
• Cellular-level estimates of neural activity bioenergetic costs (ATP/s/neuron) provide a physiological basis.
• An empirical reference for awake resting brain ATP use is ~2 × 10^9 ATP/s-neuron in mammals.

Purpose of the Study:

• To propose a modeling framework connecting abstract neural population model variables to cellular bioenergetic costs.
• To anchor neural population models, particularly neural attractors, to plausible bioenergetics requirements.

Main Methods:

• Applied an energetics framework to the Wilson-Cowan (WC) model of interacting excitatory (E) and inhibitory (I) neural populations.
• Analyzed steady-state, limit cycle, and sustained activity attractors.
• Quantified energy costs in terms of ATP consumed per second per neuron.

Main Results:

• Energy cost of limit cycles correlates more with firing rate than oscillation frequency.
• Self-sustained firing via recurrent excitation incurs higher energy costs.
• A three-node WC network demonstrated a persistent 'on-switch' behavior with moderate ATP use.

Conclusions:

• The proposed framework bridges neural population modeling and cellular bioenergetics.
• It provides a method for grounding abstract neural models in physiological energy constraints.
• This approach anchors neural attractors to realistic bioenergetic demands, advancing brain function understanding.