The functional role of oscillatory dynamics in neocortical circuits: a computational perspective

Whether neuronal dynamics like oscillations and synchrony are epiphenomena or used for computation is debated. To address this, we simulated recurrent networks of damped harmonic oscillators (HORNs), supported by experimental findings. This approach significantly improved learning speed, noise tolerance, and parameter efficiency over non-oscillatory models. HORNs also replicate key features of natural systems like the cerebral cortex, and adding characteristics like varied oscillation frequencies and network modularity boosts performance without extra parameters. These results reveal the computational power of HORNs, offering insight into previously unclear physiological phenomena, and suggest potential for energy-efficient analog circuits in machine learning.

Zoom link:

https://us06web.zoom.us/j/98286706234?pwd=bm1JUFVYcTJkaVl1VU55L0FiWDRIUT09