Codependent plasticity underlies cortical memory strengthening

The brain continuously encodes new information as memories. At the synaptic level, memory formation is enabled by adjusting connections between activated excitatory neurons in the cortex, with mirroring inhibitory activity ensuring stable storage. Particularly, synaptic plasticity occurs as memory reactivation takes place during sleep, contributing to increased memory stability. However, the mechanism by which reactivation mediates plasticity to balance network currents for stable memory storage remains unclear. Here, we investigated whether a newly proposed form of codependent plasticity could explain such reactivation-mediated memory strengthening in a spiking neural network model. This form of plasticity models how synapse-specific activity depends on both excitatory and inhibitory network currents, thereby capturing the dependence of memory strength on the degree of reactivation. In our model, learnt representations emerged as groups of excitatory synapses organised into distinct cell assemblies that were stabilised by matched inhibition. Periods of disinhibition induced excitatory plasticity, allowing for the potentiation of synapses within cell assemblies with high levels of reactivation. Conversely, cell assemblies with low levels of reactivation were weakened, leading to a gradual activity decrease. Over longer timescales, inhibitory synapses co-tuned themselves to match the shape of the established excitatory profile, where firing patterns and synaptic strength depended on the level of reactivation, faithfully capturing experimental observations. In summary, our model suggests that codependent plasticity is a plausible mechanism to support reactivation-mediated memory strengthening in the cortex and sheds light on neural- and synaptic-level dynamics reflected in data acquired from humans, even when they lack high spatiotemporal resolution.