Cellular Mechanisms Determine Soma-dendrite Correlation

Abstract

Neurons receive synaptic inputs across spatially separated compartments, and the relationship between dendritic processing and somatic output can shape how information is represented in spikes and bursts. However, it remains unclear how compartmentalization within a single neuron interacts with multiplexing, in which multiple input streams are represented simultaneously in a single spike train. This thesis uses a two-compartment neuron model to study how dendritic and somatic dynamics jointly support multiplexed coding. The model separates dendritic integration from somatic spike generation while allowing interaction in both directions through forward dendrite-to-soma drive and spike-triggered soma-to-dendrite feedback. Compartmentalization is quantified operationally using the Pearson correlation between dendritic and somatic membrane voltages, and multiplexing is quantified by the extent to which two output features, event rate and burst fraction, track two compartment-specific input streams. The results indicate that multiplexing is strongest at an intermediate level of compartmentalization, as measured by soma-dendrite voltage correlation: when this correlation is high, event rate and burst fraction become less specific to their corresponding input streams, whereas when this correlation is low, dendritic and somatic activity are less effectively linked. Finally, the thesis introduces a compact metric to quantify how noise, forward dendrite-to-soma drive, and spike-triggered soma-to-dendrite feedback shape the relationship between compartmentalization and multiplexing. This work offers a perspective on the design principles of neural computation and may inform future modeling of signal processing in complex neural structures.