Spiking Neural Networks for Simulating Surround Suppression
Kisuk Lee, Sami El-Boustani & Mriganka Sur
Abstract
In this report, the role of somatostatin-expressing (SOM+) interneurons in the generation of cortical surround suppression is studied using a model network composed of conductance-based leaky integrate-and-fire neurons. The model network consisted of three representative types of cortical neurons, i.e., excitatory pyramidal (Pyr) cells, and inhibitory parvalbumin-expressing (PV+) and SOM+ interneurons. The structure of the model network is specified by a simple connectivity rule that reflects the actual connectivity patterns of cortical neurons, which have recently been shown in [1]. This study demonstrates that a simple distance-dependent connectivity rule for SOM+ interneurons, in which a SOM+ model neuron spatially integrates the activities of neighbouring Pyr cells and locally suppresses both the Pyr and PV+ cells, is sufficient to fully reproduce the characteristic size-tuning curves for each neuron types in the model network.
Main Results
Measuring Spreading of Neural Activity based on Network Structure
Kisuk Lee and Prof. Marcus Kaiser
In the spring semester of 2011, I took a graduate-level course entitled 'Computational Neuroscience and Neuroinformatics'. As a term project, I conducted research on measuring spreading of neural activity in human and C.elegans neural networks. Specifically, I adopted two novel measures to assess spreading in cortical networks, i.e. absorption and driftness [2], and applied those measures to both human structural connectivity networks based on diffusion tensor imaging (DTI) data and C.elegans neuronal connectivity networks. Human DTI data was provided by Dr. Marcus Kaiser's Computational Neuroscience module at Seoul National University, and C.elegans data was obtained from Varshney et al. (2011) [3].