We follow the idea that a critical resource for human skill learning derives from a neural organisation of skill encoding that is both hierarchical and modular. In a modular hierarchy, components of skilful behaviour, such as individual movements of a motor skill, are thought to be merged in representational units, coding, e.g. for independent spatial and temporal patterns of behaviour (Diedrichsen and Kornysheva, 2015). Instead of selecting each movement individually in an effortful, time-consuming and resource-absorbing process, entire patterns can be selected for automatised execution as a unit. Thus, a modular hierarchy has the potential to liberate cognitive resources, enhance behavioural efficiency, and enable transfer through flexible recombination of patterns.
Here, we will test critical behavioural and neurophysiological predictions of this model, examine how a modular hierarchy provides a key resource for human skill learning within and beyond the motor system, and investigate how allocation of this resource is regulated, and can be optimised through a metacognitive intervention. We will combine behavioural evidence of automatisation, transfer, and performance monitoring during motor and perceptual skill learning with time-resolved multivariate decoding of human magnetoencephalography (MEG) and electroencephalography (EEG). Our plan for later funding periods is to investigate, and optimise through our intervention, the efficiency with which the potential to encode skills in a modular hierarchy is exploited at different points throughout the lifespan.

Conceptual framework of the neural resource examined in project C03, together with our readouts and interventions. a) Early during learning, intermediate task goals, e.g. individual finger movements in a sequence, are selected for execution individually in a resource-absorbing and time-consuming process. b) During skill learning, individual movements become merged in representational units at an intermediate level in the hierarchy (green), coding, e.g. for temporal (pink arrow) or spatial (blue arrow) patterns of movements. These patterns can be selected for execution as a whole and executed via their mapping onto motor primitives (blue level). Control is thus partially delegated to this intermediate level, and behaviour becomes automatised. Modularity within this hierarchy, e.g. independent coding of temporal vs. spatial patterns at an intermediate level, allows for flexible recombination of intermediate-level skill representations to form novel behaviours. SMA (supplemental motor area), M1 (primary motor cortex), PMC (pre-motor cortex).