Architectural Principles and Predictive Modeling of the Mammalian Connectome

Although mammalian brains show a massive, 5 orders of magnitude variation in weight, they present common processing features, implying that cortical computing is built on a scalable architecture. We hypothesize the existence of network architectural organizational principles in the mammalian brain, critical for efficient and hierarchically modular information processing. Extending our empirical, consistent tract-tracing network databases, we confirm the validity of the Exponential Distance Rule (EDR) in the macaque cortex, showing that the EDR is an architectural network invariant. We have also developed and cross-validated novel, machine learning based imputation algorithms, suitable for dense interareal networks, exploiting the weighted, directed and the spatially embedded nature of these networks. As we show, these algorithms can efficiently be used to guide further tract-tracing experiments, for example by identifying potential injection targets that would generate the largest information gain, after every new injection.
References: R. Gamanut et al. Neuron 97, 698-715 (2018), H.R. Noori et al. PLoS Biol., 15(7), e2002612 (2017); N.T. Markov et al. Science 342(6158), 1238406 (2013)