Testing the Cortical Column Conjecture
Presidential Lectures are free public colloquia centered on four main themes: Biology, Physics, Mathematics and Computer Science, and Neuroscience and Autism Science. These curated, high-level scientific talks feature leading scientists and mathematicians and are intended to foster discourse and drive discovery among the broader NYC-area research community. We invite those interested in the topic to join us for this weekly lecture series.
Many contemporary theories of neural information processing suggest that the neocortex employs algorithms composed of repeated instances of a limited set of computing primitives. There is a recognized need for tools that interrogate the structure of the cortical microcircuits believed to embody these primitives. The cortical column conjecture suggests that neurons in the neocortex are connected in a graph that exhibits motifs representing repeated processing modules. Carey Priebe and his collaborators will present a notional demonstration of how statistical inference on graphs can inform our understanding of cortical computing.
By modeling the cortical graph as a hierarchical stochastic block model (HSBM), with induced subgraphs, which are themselves independent stochastic block models, a natural question is to estimate the extent to which identified subgraphs share common structure. This will require addressing the problem of identifying candidate subgraphs, and of determining the impact of imperfect subgraph identification on subsequent inference. The application of this connectomics theory and the associated methods will be demonstrated via a bio-inspired, large-scale simulation study.
If this lecture is videotaped, it will be posted here after production.