Summary: New research explains how motor memory is consolidated during sleep.
Source: UCSF
When a player like Steph Curry takes a free throw, the smooth, automatic motion relies on motor memory. Researchers at the University of California, San Francisco (UCSF) now describe how the brain consolidates motor memory during sleep, turning practiced actions into subconscious skills.
Published in Nature on December 14, 2022, the study reveals that sleep helps the brain review a range of practiced attempts and reinforces the successful patterns. In practical terms, the brain sifts through many prior free throws and retains the sequences of motion that consistently work, while down-weighting or discarding those that do not.
As a result, once a motor routine is consolidated, a player can execute the movement with high accuracy without conscious effort.
“Even elite athletes make mistakes, and their performance reflects both predictable errors and predictable successes,” said Karunesh Ganguly, MD, Ph.D., professor of neurology and member of the UCSF Weill Institute for Neurosciences. “Motor memory isn’t about perfection; it’s about capturing stable patterns. If the errors follow a predictable pattern from day to day, the brain locks that pattern into memory.”
Ganguly and his team found that this “locking in” happens through coordinated activity across several brain regions during deep, restorative non-REM sleep. Far from a simple one-region process, consolidation involves complex dialogue between the hippocampus, the prefrontal cortex (PFC) and primary motor cortex (M1).
Sleep provides a valuable window for this process because the waking brain tends to foreground recent failures. During non-REM sleep, the brain can more evenly review both successes and mistakes and selectively strengthen the successful patterns, Ganguly explained.
How the brain reviews motor skills during sleep
Early theories emphasized a single role for the motor cortex in learning motor skills. This study paints a more nuanced picture. To investigate, researchers trained rats to reach for food pellets and then recorded neural activity during non-REM sleep across the hippocampus, PFC and motor cortex.
They observed two broad stages of consolidation. In a fast-learning phase, the prefrontal cortex coordinated with the hippocampus, helping represent movement relative to space and explore the variety of actions generated during practice. This phase appeared to support rapid adjustments and variability in the motor system’s internal model.
In a subsequent slow-learning phase, the PFC engaged with both the hippocampus and motor cortex to evaluate outcomes. Reward-related signals amplified successful movement patterns and suppressed those associated with failure. Over time, as neural activity across these regions synchronized, hippocampal influence waned and motor representations stabilized in the cortex—forming durable motor memory.
During initial learning, the animals’ neural signals were noisy and variable. As performance improved—reaching roughly 70 percent success—the patterns became more consistent. At that point the brain began to ignore occasional errors and maintained the consolidated motor plan as long as success levels remained stable. In other words, the brain accepts a predictable level of error and stops updating the motor program.
Because motor learning is built on a mental model of the physical environment—gravity, spatial cues and body dynamics—skills trained in one context do not automatically transfer to a very different environment. “If the physical context changed dramatically, such as placing an athlete in an entirely different gravity or spatial world, their established motor patterns might not work as well at first,” Ganguly noted.
How to change or break a motor habit
The study also sheds light on how entrenched motor habits can be altered. It is possible to unlearn or update a skill, but doing so generally requires perturbing the situation enough to produce new errors and variability.
When researchers introduced a modest change in the rats’ task, performance briefly worsened and neural activity grew noisier—signals that the brain had re-entered a phase of exploration. The animals did not need to relearn the task from scratch; instead they reverted to a “breaking point” and relearned from that adjusted state.
Because complex motor routines are a sequence of actions chained together in time, modifying a single initiating action can reset the sequence and enable relearning. For example, Ganguly suggested that a basketball player who normally bounces the ball twice before a shot might retrain by changing that routine—bouncing once or three times—to create a fresh context for the brain to form a new motor pattern.
About this memory and sleep research news
Author: Press Office
Source: UCSF
Contact: Press Office – UCSF
Image: The image is in the public domain
Original Research: Open access.
“Cortical–hippocampal coupling during manifold exploration in motor cortex” by Jaekyung Kim et al. Nature
Abstract
Cortical–hippocampal coupling during manifold exploration in motor cortex
Systems consolidation describes how new memories initially dependent on the hippocampus become integrated into cortical networks over time. How hippocampal–cortical interactions evolve during this process, and how cortical representations stabilize in parallel, remains unclear.
Using a skill-learning task, the study tracked cross-area coupling during non-REM sleep while measuring representational stability in primary motor cortex (M1). The results show two distinguishable stages of processing marked by precise coupling patterns among hippocampus, prefrontal cortex and M1.
Early post-training sleep showed increased coupling between hippocampal sharp-wave ripples and M1 slow oscillations, linked to rapid learning and variability in M1’s low-dimensional manifold. As animals stabilized performance, there was a sharp rise in prefrontal–M1 slow oscillation coupling that predicted a drop in hippocampal–M1 coupling, consistent with hippocampal disengagement and a transition to a second consolidation stage.
Crucially, altering task parameters to induce new exploration re-engaged hippocampal–M1 coupling, demonstrating that hippocampal–cortical dialogue can be dynamically reactivated during adaptation and relearning.