Motor cortex (M1) exhibits a rich repertoire of neuronal activities to support the generation of complex movements. Although recent neuronal-network models capture many qualitative aspects of M1 dynamics, they can generate only a few distinct movements. Additionally, it is unclear how M1 efficiently controls movements over a wide range of shapes and speeds. We demonstrate that modulation of neuronal input–output gains in recurrent neuronal-network models with fixed architecture can dramatically reorganize neuronal activity and thus downstream muscle outputs. Consistent with the observation of diffuse neuromodulatory projections to M1, a relatively small number of modulatory control units provide sufficient flexibility to adjust high-dimensional network activity using a simple reward-based learning rule. Furthermore, it is possible to assemble novel movements from previously learned primitives, and one can separately change movement speed while preserving movement shape. Our results provide a new perspective on the role of modulatory systems in controlling recurrent cortical activity.

Motor cortex (M1) exhibits a rich repertoire of activities to support the generation of complex movements. Although recent neuronal-network models capture many qualitative aspects of M1 dynamics, they can generate only a few distinct movements. Additionally, it is unclear how M1 efficiently controls movements over a wide range of shapes and speeds. We demonstrate that simple modulation of neuronal input-output gains in recurrent neuronalnetwork models with fixed architecture can dramatically reorganize neuronal activity and thus downstream muscle outputs. Consistent with the observation of diffuse neuromodulatory projections to M1, we show that a relatively small number of modulatory control units provide sufficient flexibility to adjust high-dimensional network activity using a simple reward-based learning rule. Furthermore, it is possible to assemble novel movements from previously learned primitives, and one can separately change movement speed while preserving movement shape. Our results provide a new perspective on the role of modulatory systems in controlling recurrent cortical activity.

Working memory involves the short-term maintenance of information and is critical in many tasks1. The neural circuit mechanisms underlying this information maintenance are thought to rely on persistent activities2,3 resulting from attractor dynamics4,5. However, how information is loaded into working memory for subsequent maintenance remains poorly understood. A pervasive assumption is that information loading requires inputs that are similar to the persistent activities expressed during maintenance6–9. Here, we show through mathematical analysis and numerical simulations that optimal inputs are instead largely orthogonal to persistent activities and naturally generate the rich transient dynamics characteristic of prefrontal cortex (PFC) during working memory10–17. By analysing recordings from monkeys performing a memory-guided saccade task, and using a novel, theoretically principled metric, we show that PFC exhibits the hallmarks of optimal information loading. Our theory unifies previous, seemingly conflicting theories of memory maintenance based on attractor8,9,18–20 or purely sequential dynamics21–23, and reveals a normative principle underlying the widely observed phenomenon of dynamic coding in PFC10–12,15–17,24. These results suggest that optimal information loading may be a key component of attractor dynamics characterising various cognitive functions and cortical areas, including long-term memory25,26 and navigation27,28 in the hippocampus, and decision making in the PFC19,20,29,30.