← Return to the Publications page

Inscopix Publications

Self-organization of songbird neural sequences during social isolation


Authors: Emily L Mackevicius, Shijie Gu, Natalia I Denisenko, Michale S Fe
Publication: eLife
Date: May 30, 2023
Link to article: https://elifesciences.org/articles/77262


Emily L. Mackevicius et al., using the Inscopix nVista tool, investigated whether the formation of sequences in the high vocal center (HVC) of songbirds depends on prior exposure to a tutor song. They found that even in birds isolated from a tutor, HVC network activity can develop long repeatable sequences. However, these sequences in isolated birds exhibited differences compared to typical birds, showing less reliability and a weaker correlation with vocal output. The researchers had previously hypothesized that tutor experience is necessary for the formation of HVC sequences, but their new findings revealed that latent sequences exist even before tutoring. This suggests that simple Hebbian learning mechanisms, combined with spontaneous activity within or driven by inputs to HVC, may contribute to the emergence of sequences in isolated birds. The discovery of latent sequences suggests a separation between processes for building a stable representation of sequential moments and processes for associating actions with each state. It aligns with reinforcement learning models of song learning, where HVC sequences remain stable during song changes. These results contrast with other models like the “inverse model,” indicating that stable latent sequences are influential in song learning. The presence of latent structure in the brain prior to exposure to relevant experiences suggests the utilization of pre-existing sequential patterns to facilitate rapid learning. The study also raises questions about potential variations in latent sequences among species with different song characteristics, which could be influenced by genetic factors or the timing of feedback from respiratory and auditory centers.Overall, the ability of brains to generate complex learned behavior may depend on the intrinsic developmental formation of appropriate latent dynamics in motor and sensory circuits. Further research is needed to explore the underlying mechanisms and implications of latent sequences in brain learning processes.

Scroll to Top