PhD Proposal by Zachary Mobille

Quantitative Biosciences Thesis Proposal

Zachary Mobille

School of Mathematics
Advisor: Dr. Hannah Choi (School of Mathematics)

Open to the Community


Information Coding and Structural Motifs in Spiking Neural Networks

Monday, September 25, 2023, at 2:30pm

In Person Location: Engineered Biosystems Building (EBB) 5029

Zoom Link:

Committee Members:

Dr. Simon Sponberg (School of Physics)

Dr. Samuel Sober (Department of Biomedical Engineering)

Dr. Bilal Haider (Department of Biomedical Engineering)


The relationship between structure and dynamics runs both ways in neural systems. Connectivity can shape the ways in which populations of neurons encode stimuli. The spiking activity of neurons can weaken or strengthen the synaptic coupling that binds them together. The purpose of this thesis is therefore two-fold: 1) to characterize how a ubiquitous feedforward network structure transforms the way in which neurons represent information with action potentials and 2) to understand why certain connectivity motifs are promoted in recurrent, plastic networks of spiking neurons.


I will start by describing the mathematical models that have been developed to study these questions. Our focus is on convergent and divergent network structures at multiple scales. The first half of my thesis focuses primarily on dominantly feedforward network structures with various amounts of convergence and divergence from layer to layer. This architecture is observed in visuomotor pathways, cerebellum-like structures in mammals, and the antennal lobe of insects. Our hypothesis is that convergent pathways transform spike count codes in pre-synaptic populations to spike timing codes in post-synaptic populations, and that this is mediated by a concurrent increase in spike reliability. Population codes and single-neuron codes are quantified by information theoretic measures and decoding analyses. The second half of my thesis is concerned with the evolution of various structural motifs that are embedded in a large recurrent network with plastic connections. We start with purely excitatory networks to study how the dimensionality of a stimulus is related to the dimensionality of convergent/divergent motifs that are up regulated by plasticity. To grasp why these motifs are functionally important, I will perform a partial information decomposition of the transfer entropy relating them. Finally, this analysis will be extended to the more realistic case of balanced networks with multiple inhibitory cell types.