CORNET - Integrating top-down and bottom-up processing in the marmoset and macaque cortex

点击翻译按钮获取中文摘要

The Kennedy-Knoblauch lab has established an extensive database of inter-areal pathways in the macaque cortex. This database enabled the development of a predictive, large-scale model of the cortex with numerous interesting features, including hierarchical organization (Markov et al., 2013b; Song et al., 2014). The Fries team is internationally known for developing cutting-edge technology for large-scale electrophysiology, leading to important findings on the interplay between neuronal dynamics and structural connectivity (Fries, 2009). Recently, the Fries team has developed dynamic causal models (DCMs) for looking at asymmetries in feedforward and feedback pathways in relation to predictive coding (Bastos et al., 2015a; Bastos et al., 2012). The teams of Fries and Kennedy-Knoblauch have collaborated to compare hierarchies in macaque visual cortex derived from structural and functional data (Bastos et al., 2015b). They showed that directed influences constrain a functional hierarchy with critical features of the structural hierarchy, while exhibiting task dependent dynamics. The present project will extend and deepen this collaboration, by combining our complementary skills to integrate large-scale structural and functional datasets in the smooth-brained marmoset. We will use identical retrograde tracer technology from our earlier work in macaques to derive a weighted and directed connectivity matrix for the marmoset cortex. Tracers will be injected in widely distributed sites of the marmoset cortex, with a particular focus on visual areas. The physiology of these areas will be characterized using high-density electrocorticography (hdECog, 200 electrodes/cm2). Tracer experiments will be carried out in animals that have undergone hdECog recording so as to co-register electrophysiological and anatomical maps. Data will be used to construct DCMs of the mechanics of visual predictive coding at an unprecedented level of detail. Developing a Bayesian framework for structural network completion will augment our anatomical data. We will improve existing algorithms for data completion and the estimation of uncertainty by incorporating distance and weight in our procedures. These procedures will be of critical importance for network completion in marmoset, macaque and mouse. These developments will allow us to refine and extend our existing macaque database, by increasing our cortical matrix from 91 to 131 areas. Additionally, refining our existing weights will enable flexible parcellation and improve specificity for correlation to electrophysiology and imaging data. The present proposal will investigate large-scale structural models in macaque, marmoset and mouse providing insight into how the EDR model scales with brain size. The weight-distance relationships in these three species will allow us to determine the scaling features of cortex and will have important consequences for extrapolating our present findings to the large human brain.