The contributions of brain networks to information processing and learning and memory are classically interpreted within the framework of Hebbian plasticity and the notion that synaptic weights can be modified by specific patterns of activity. However, accumulating evidence over the past decade indicates that synaptic networks are also structurally plastic, and that connectivity is remodelled throughout life, through mechanisms of synapse formation, stabilization and elimination
1. This has led to the concept of structural plasticity, which can encompass a variety of morphological changes that have functional consequences. These include on the one hand structural rearrangements at pre-existing synapses, and on the other hand the formation or loss of synapses, of neuronal processes that form synapses or of neurons. In this Review we focus on plasticity that involves gains and/or losses of synapses. Its key potential implication for learning and memory is to physically alter circuit connectivity, thus providing long-lasting memory traces that can be recruited at subsequent retrieval. Detecting this form of plasticity and relating it to its possible functions poses unique challenges, which are in part due to our still limited understanding of how structure relates to function in the nervous system.We review recent studies that relate the structural plasticity of neuronal circuits to behavioural learning and memory and discuss conceptual and mechanistic advances, as well as future challenges. The studies establish a number of strong links between specific behavioural learning processes and the assembly and loss of specific synapses. Further areas of substantial progress include molecular and cellular mechanisms that regulate synapse dynamics in response to alterations in synaptic activity, the specific spatial distribution of the syn aptic changes among identified neurons and dendrites and the relative roles of excitation and inhibition in regulating structural plasticity.The new findings provide exciting early vistas of how learning and memory may be implemented at the level of structural circuit plasticity. At the same time, they highlight major gaps in our understanding of plasticity regulation at the cellular, circuit and systems levels. Accordingly, achieving a better mechanistic understanding of learning and memory processes is likely to depend on the development of more effective techniques and models to investigate ensembles of identified synapses longitudinally, both functionally and structurally.
Molecular mechanisms of synapse remodellingA remarkable feature of excitatory and inhibitory synapses is their high level of structural variability 2 and the fact that their morphologies and stabilities change over time 3 . This phenomenon is regulated by activity, and the size of spine heads correlates with synaptic strength 4 , presynaptic properties 5 and the long-term stability of the synapse 6 . The morphological characteristics of synapses thus reveal important features of their function and stabilit...