Distinct synaptic plasticity rules operate across dendritic compartments in vivo during learning

Events occurring close in time are often linked in memory, providing an episodic timeline and a framework for those memories. Recent studies suggest that memories acquired close in time are encoded by overlapping neuronal ensembles, but whether dendritic plasticity plays a role in linking memories is unknown. Using activity-dependent labeling and manipulation, as well as longitudinal one- and two-photon imaging of RSC somatic and dendritic compartments, we show that memory linking is not only dependent on ensemble overlap in the retrosplenial cortex, but also on branch-specific dendritic allocation mechanisms. These results demonstrate a causal role for dendritic mechanisms in memory integration and reveal a novel set of rules that govern how linked, and independent memories are allocated to dendritic compartments.

How can neural networks learn to efficiently represent complex and high-dimensional inputs via local plasticity mechanisms? Classical models of representation learning assume that input weights are learned via pairwise Hebbian-like plasticity. Here, we show that pairwise Hebbian-like plasticity only works under specific requirements on neural dynamics and input statistics. To overcome these limitations, we derive from first principles a learning scheme based on voltage-dependent synaptic plasticity rules. Here, inhibition learns to locally balance excitatory input in individual dendritic compartments, and thereby can modulate excitatory synaptic plasticity to learn efficient representations. We demonstrate in simulations that this learning scheme works robustly even for complex, high-dimensional and correlated inputs. It also works in the presence of inhibitory transmission delays, where Hebbian-like plasticity typically fails. Our results draw a direct connection between dendritic excitatory-inhibitory balance and voltage-dependent synaptic plasticity as observed in vivo, and suggest that both are crucial for representation learning.