How do animals store learned behaviours in their neuronal networks and retrieve them when performing those behaviours? It is widely believed that synapses, or connections between neurons, are the memory substrate for learned behaviours. Under this assumption, learning involves the formation, elimination, or tuning of these synapses, enabling the brain to utilise the stored connectivity pattern later. The sum total of these synaptic connections is called the connectome.
Despite this belief, we still lack a clear understanding of how behaviour is encoded in connectivity patterns, even for the simplest behaviours.
Using the zebra finch songbird as a model, we explore how song memories are stored and retrieved from brain circuits. These birds can perform songs as adults that they practiced as juveniles, much like how humans learn language.
To map these brain circuits at sufficient resolution to identify potential memory-forming synapses, we use high-throughput 3D electron microscopy. This process generates vast amounts of image data, far more than could be manually examined. Therefore, we employ advanced deep learning techniques to infer the connectomic map, allowing artificial neural networks to reconstruct the real ones.
Our ultimate aim is to understand how the learned zebra finch song is encoded in synaptic wiring patterns, linking specific behaviours to their underlying connectome.