Ahmadian, YasharHolt, Caleb2020-09-242020-09-242020-09-24https://hdl.handle.net/1794/25600Here, we theoretically study how cortical networks’ synaptic connectivity shapes their spiking activity dynamics, and in turn, how dynamics shape the structure of synaptic connectivity. In the first part of this work, we study rhythmic oscillations of neural activity in the primary visual cortex (V1). These oscillations are characterized by power-spectra with peaks frequencies anywhere from 30 to 80 Hz, the gamma band. Gamma peaks shift to higher frequencies when V1 is stimulated by increasing contrasts. Moreover, the peak frequency depends on the local contrast stimulating a V1 sub-network. The local nature of this contrast-dependence would be trivially explained if the long-range intra-cortical connections within V1 were weak. However, experimental observations, such as the suppression of spiking rates with increasing stimulus size, point to those same long-range connections being functionally strong. Here we show that a model of V1 can balance the strength of short- and long-range connections to successfully reproduce observations of gamma peak’s local contrast-dependence as well as “surround suppression” of firing rates. Thus, we demonstrate how cortical network dynamics can be shaped by connectivity structures. Visual information is relayed to V1 from the thalamus. The patterning of thalamocortical connections determines the location of visual space which V1 neurons respond to, their receptive field (RF). Cortical neurons moreover develop particular RF filters from thalamocortical connections. in carnivores and primates, particular features of those filters, such as their orientation, develop smooth maps across the cortex. However, in those same species, the RF filters of neighboring neurons are diverse and heterogeneous. Previous theoretical models of RF filter development predicted that if a smooth map forms for one feature, then RF filters should be homogeneous; failing to predict the observed co-presence. We extend those models by properly considering cortical activity dynamics. We show that in the presence of multiple timescales in cortical dynamics and thalamic inputs, RF filters can develop qualitatively different feature organizations, from smooth maps to heterogeneity. Thus, we demonstrate how cortical network dynamics shape connectivity structures. This dissertation contains previously published co-authored materialen-USAll Rights Reserved.dynamicsgamma oscillationshebbian learningreceptive fieldsstabilized supralinear networkvisual cortexElectronic Thesis or Dissertation