Physics

Sociology

Neuroscience

Population

Systems neuroscience

Central nervous system

Biology

Algorithm

Curse of dimensionality

Dynamics (music)

Demography

Computational neuroscience

Artificial intelligence

Management

Economics

Acoustics

Oligodendrocyte

Myelin

Models of neural computation

Computer science

Computation

Task (project management)

Quantum mechanics

Mechanism (biology)

Artificial neural network

Nature Neuroscience

Neural computations are currently investigated using two separate approaches: sorting neurons into functional subpopulations 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 on neuroscience tasks, here we show that the dimensionality of the dynamics and subpopulation structure play fundamentally complementary roles. Although various tasks can be implemented by increasing the dimensionality in networks with fully random population structure, flexible input–output mappings instead require a non-random population structure that can be described in terms of multiple subpopulations. Our analyses revealed that such a subpopulation structure enables flexible computations through a mechanism based on gain-controlled modulations that flexibly shape the collective dynamics. Our results lead to task-specific predictions for the structure of neural selectivity, for inactivation experiments and for the implication of different neurons in multi-tasking. Neural computations are envisioned as arising from either distinct function subpopulations or distributed collective dynamics. Dubreuil and Valente et al. examined recurrent neural networks trained on various cognitive tasks and found that a mixed-selective yet non-random subpopulation structure enabled flexible responding through gain-modulated latent dynamics.

The role of population structure in computations through neural dynamics