Complementary roles of dimensionality and population structure in neural computations

Neural computations are currently investigated using two competing approaches: sorting neurons into functional classes, or examining the low-dimensional dynamics of collective activity. Whether and how these two aspects interact to shape computations is currently unclear. Using a novel approach to extract computational mechanisms from networks trained with machine-learning tools on neuroscience tasks, here we show that the dimensionality of the dynamics and cell-class structure play fundamentally complementary roles. While various tasks can be implemented by increasing the dimensionality in networks consisting of a single global population, flexible input-output mappings instead required networks to be organized into several sub-populations. Our analyses revealed that the subpopulation structure enabled flexible computations through a mechanism based on gain-controlled modulations that flexibly shape the dynamical landscape of collective dynamics. Our results lead to task-specific predictions for the structure of neural selectivity and inactivation experiments.