Interaction between model-based and model-free mechanisms in motor learning.

Area of Science:

• Motor neuroscience
• Cognitive psychology
• Human motor control

Background:

• Motor learning utilizes distinct mechanisms: habitual, model-free (implicit) and strategic, model-based (explicit).
• The influence of prior engagement of one mechanism on subsequent learning, particularly when task demands shift, is not well understood.

Purpose of the Study:

• To investigate how prior exposure to specific motor learning mechanisms (model-free vs. model-based) influences subsequent motor adaptations.
• To determine if these biases persist even when task conditions typically favor the alternative strategy.

Main Methods:

• Participants (n=82) performed reaching movements with varying error magnitudes (small: 15°, large: 30°-60°) in three experiments.
• The study manipulated error size to initially engage either model-free or model-based learning processes.
• Washout phases were included to assess the persistence of learned biases.

Main Results:

• Prior exposure to small errors led to persistent model-free aftereffects and stable reaction times, even with larger errors later.
• Initial large errors induced model-based strategies (flexible reaction times, minimal aftereffects), which persisted with subsequent small errors.
• These mechanistic biases remained even after washout phases, indicating durable imprints.

Conclusions:

• Motor learning is hierarchically shaped by cumulative experiences, with early learning establishing frameworks that constrain future adaptations.
• Prior engagement of either model-free or model-based processes significantly biases subsequent motor learning, prioritizing efficiency over flexibility.
• Findings suggest initiating motor rehabilitation or training with model-based strategies may enhance long-term adaptability.