A unified model of short- and long-term plasticity: Effects on network connectivity and information capacity

Description

Iiro Ahokainen and Marja-Leena Linne published this neuroscience modeling study in 2026. Brain synapses change on two clocks at once: a fast, temporary clock and a slow, lasting one. Most models of lasting change ignore the fast clock's influence. The authors fuse the two into a single rule, fitted to recordings from the visual cortex. They then ask what the fusion does inside networks of neurons.

At a single synapse, the fast dynamics reshaped how lasting change depends on firing rates. In recurrent networks, neurons organized themselves into distinct firing-rate clusters, and activity steadied. Adding balancing mechanisms changed how connections correlated across the network. On reservoir computing tests, which gauge a network's information capacity, the fused rule beat the uncoupled version in most tasks. The authors conclude that short-term dynamics are not a detail to skip. They steer how networks wire themselves and how much they can compute.

Catalog record

Source: https://europepmc.org/articles/PMC13581215