A theoretical multiscale examination of visual spatial attention in the mouse cerebral cortex

Computational approaches modeling neurons and their networks are essential for untangling the complex interactions underlying brain functions such as visual attention. This thesis aims to construct large-scale biophysically mimetic spiking neural network models of the mouse cortex to examine how visual spatial attention can emerge from neuron networks. To do so, this work is divided into two parts. The first focuses on building large-scale networks strongly informed by biophysical data, while the second part develops how these models can be adapted in the specific context of visual attention by testing network theories and different degrees of biophysical details. As a case study, I implement the network architecture developed in the Synchronous Matching Adaptive Resonance Theory (SMART; Grossberg and Versace, 2008) into our data-driven model. This SMART model is a good example of implementing a theoretical description in explicit wetware that allows consideration of additional biophysical detail, which can, in turn, test the robustness of the original theory in a visual attention context.