Computational properties of cortical microcircuit models from the Allen Institute 

Wolfgang Maass, Professor of Computer Science at the Graz University of Technology in Austria

Wolfgang Maass will discuss published results from (Chen et al., 2022, 2023) and ongoing work training the V1 microcircuit model from (Billeh et al., 2020) for visual processing tasks that have commonly been used as behavioural tasks for mice, such as the change detection task and the evidence accumulation task. The trained model achieves a performance that is similar to the published behavioral data. In addition, it exhibits traits in its network dynamics and neural coding that are very similar to those found in-vivo, but unlike those of commonly considered neural network models of the same size. Finally, Maass will discuss results of reverse-engineering of the computations of the cortical microcircuit model that provide hypotheses for refined experimental analyses of in-vivo data.  

Relevant Citations
  
(Billeh et al., 2020) Billeh, Y. N., Cai, B., Gratiy, S. L., Dai, K., Iyer, R., Gouwens, N. W., … & Arkhipov, A. (2020). Systematic integration of structural and functional data into multi-scale models of mouse primary visual cortex. Neuron, 106(3), 388-403.  
  
(Chen et al., 2022) Chen, G., Scherr, F., & Maass, W. (2022). A data-based large-scale model for primary visual cortex enables brain-like robust and versatile visual processing. Science Advances, 8(44), eabq7592.  
  
(Chen et al, 2023) Chen, G., Scherr, F., & Maass, W. (2023). Data-based large-scale models provide a window into the organization of cortical computations. bioRxiv, 2023-04.