Efficient Coding Predicts Synaptic Conductance

Neuroscientist James Stone has published a groundbreaking paper titled 'Efficient Coding Predicts Synaptic Conductance' (arXiv:2603.03347) that demonstrates how biological synapses operate at mathematically optimal efficiency. The research shows synapses naturally maintain conductance values that maximize information transfer (measured in bits) per unit of energy consumed (Joules), falling in line with the efficient coding hypothesis for neural systems. When forced to deviate from these natural values—as shown in earlier work by Harris et al (2015)—efficiency rapidly declines. Stone's model successfully predicts this precise decline curve, bridging biophysics with information theory.

Crucially, the proposed model contains zero free parameters because it's derived directly from the fundamental biophysics of synaptic transmission and Shannon's information theory framework. It builds upon recent findings by Malkin et al (2026) showing synaptic noise is minimized given available energy, establishing a minimal energy boundary. Stone's work demonstrates this boundary is necessary but insufficient; true efficiency requires operating at specific signal-to-noise ratios that maximize bits per Joule. This provides a unified mathematical explanation for why neuronal systems appear evolutionarily tuned for energy-information optimization, with direct implications for designing artificial neural networks and neuromorphic chips that mimic the brain's remarkable efficiency.