COMPUTATIONAL INVESTIGATION OF SONGBIRD HVC MICROCIRCUIT FOR PRECISE TIMING
Open Access
- Author:
- Tupikov, Yevhen
- Graduate Program:
- Physics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- September 17, 2019
- Committee Members:
- Dezhe Jin, Dissertation Advisor/Co-Advisor
Dezhe Jin, Committee Chair/Co-Chair
Reka Z Albert, Committee Member
Vincent Henry Crespi, Committee Member
Carina Pamela Curto, Outside Member
Nitin Samarth, Program Head/Chair - Keywords:
- songbird
HVC
polychronous
precise timing
neural circuit
song learning
song development
sequence growth - Abstract:
- Learned sequential behaviors are fascinating brain phenomena, but their underlying neural mechanisms are not well understood. Birdsong is a great model to investigate such behavior, since it is stereotyped and is learned gradually by juvenile songbirds from their tutors. Songbird premotor nucleus HVC (proper name) produces precise bursts of projection neurons during singing and is thought to encode timing in the song. We develop a detailed computational model of zebra finch HVC neural microcircuit, which incorporates all experimentally known features of HVC, including realistic number of neurons, connectivity patterns and axonal conduction delays. We show that a popular model for songbird HVC, a synfire chain, produces strong oscillations in neural dynamics, which is inconsistent with experimental observations. We propose an lternative model, a polychronous network, in which all inputs arrive synchronously to postsynaptic neurons. The proposed network naturally exploits distributed axonal conduction delays and produces neural activity with no significant oscillations and silent gaps, i.e., smooth dynamics. We further explore the role of axonal conduction delays in polychronous network and demonstrate that width of the axonal delay distribution controls the oscillations in network dynamics. Narrow distributions produce networks with prominent oscillations, while wide distributions result in networks with smooth dynamics. The results suggest that distributed axonal delays alone can explain the smoothness of HVC dynamics. Next we develop a biologically realistic model that explains the formation of microcircuit for precise timing in HVC. The model is built on the idea that immature neurons, provided by neurogenesis in HVC during development, are more spontaneously active and become prime targets of self-organizing process via synaptic plasticity. The model predicts that birth order of neurons positively correlates with their burst timing in the formed network. We show that with incorporation of realistic axonal conduction delays, our model produces long polychronous sequences. In contrast, ignoring delays leads to the emergence of synfire chains. The model also reproduces the experimentally observed spatial connectivity profile between projection neurons in HVC. Finally, the model predicts that inhibition plays an important role during formation of HVC microcircuit. Neurons that receive less inhibition are more likely to get incorporated into the network.