The Departure of Skill Memories from Motor Cortex: Deeper Directions for Neuroscience

You probably have certain skills that I don’t.  Each of us, having spent enough time practicing something new, can become an expert.  A simple, ubiquitous example is driving a car with a manual transmission.  The precise sequence and timing of controlling the clutch, giving gas, and shifting the gears are challenging to coordinate when initially learning to drive a stick shift.  But eventually, the precision with which we can perform these sequences of movements is impressive.  Specifically, the errors we make when learning, indicated by gear grinding and stalling, are reduced with repeated practice.  Skill learning is relatively easy to train both in humans and in animals.  It therefore serves as a great way to study how the brain forms memories and uses those memories in the future.

The unique skills we learn belong to a class of memories called procedural memories.  These are the skilled movements that we refine with practice, like the precise sequences of movements we use to drive a manual car.  Early psychologists in the beginning of the 20th century acknowledged that procedural, or ‘muscle,’ memories were different from what we usually refer to when we think of the word memory.  These are the episodic memories, which we use to travel back in time to remember our past experiences, and semantic memories, which contain the facts we know about the world.  It wasn’t until the 1960s that neuroscientists first performed experiments to demonstrate that brain areas containing procedural memories are different from those used to retrieve episodic and semantic memories.  Despite over 50 years of work, neuroscientists still do not fully understand how new movements become part of our skill set.  New findings published in Neuron by Risa Kawai in the Ölveczky laboratory at Harvard University elegantly remind us that procedural memories may not function as we expected.  The paper also provides interesting new avenues for studying how we remember and later use our hard-earned skills.

We know from centuries of study that brain areas that control movement are widely distributed throughout the central nervous system.  Despite abundant evidence for this, more traditional ideas that the motor cortex must be fundamentally involved in controlling our movements have deep roots and widespread acceptance.  We usually think of the motor cortex, which sits atop your brain, as our body’s command center, controlling how our brain stem and spinal cord move the muscles of our limbs.  According to this model, when we learn a new skill, the correct patterns of movements should be organized and stored as procedural memories in the motor cortex.  It would also mean that damage to the motor cortex would leave us completely paralyzed, and any skills that we had already learned and perfected prior to losing the motor cortex would be entirely lost.  Kawai et al. developed a simple task using rats to test the hypothesis that the motor cortex is required for both learning and storing procedural memories.


To test this hypothesis, rats learned to press a little lever two times in a row to receive a juice reward.  To receive the biggest drop possible, the rats had to time their presses so they occurred 700 milliseconds apart.  Rats are good at pressing levers and naturally want to press much faster than this, so they need to learn to withhold the second press just a little longer.  It takes rats about a month of practice to learn to time their movements well.  The authors emphasized that the rats did not need to control their paw movements at all; instead, this was more of a shoulder muscle-dependent action, like pulling the gearshift.


To determine how the motor cortex is involved in learning and storing this skill, the experimenters divided the rats into two groups.  One group had their motor cortex removed before they started learning the task.  It is important to remember that the rats are still able to move around without a motor cortex.  In fact, they generally appear normal.  A second group was trained for a month on the pressing task before having their motor cortices removed.  These two groups would test whether the motor cortex was necessary for learning the task and whether it was necessary for expressing the skill after it was well practiced.  Unsurprisingly, the rats could not learn the task without a motor cortex.  No matter how long they spent practicing, their movements would never stabilize to the correct latency, but stayed at the faster interval that rats are more inclined to naturally produce.  However, they could still press the lever.  They just were not able to reorganize their behavior to create a new skill of pressing at a new rate.

By contrast, the rats which had their motor cortices removed after learning the skill were still able to perform the movements perfectly well.  Using high-speed cameras, the authors were able to show that the movements were identical before and after the surgery.  This demonstrates that the procedural memory for this skill is not stored in the motor cortex.  This transfer of information could be a mechanism by which the motor cortex frees up resources to focus on learning new movement patterns while not needing to store previously learned skills.  The authors also propose a model by which the motor cortex is more important for helping other motor control areas of the brain reorganize their activity to produce new patterns of movements that the rats already knew how to perform.  This concept is reflected in the observation that rats in the first group were still able to press the lever without a motor cortex.  They just could not change the time interval between their two presses no matter how much they practiced.


While this study was performed in rats, the results could be different in other animals.  For example, in primates, the motor cortex may play a larger role in learned movements, particularly skills that rely on precise movement of the hand.  Nevertheless, this study is a simple and convincing demonstration that the motor cortex is not the source of all of the skilled movements we make.  The big question remaining is, “where exactly do procedural memories go?”  Positioned underneath the cortex are a great number of mysterious brain structures that are prime candidates for the storage and expression of these skills.  Among these, the basal ganglia, thalamus, cerebellum, and brainstem are thought to be more directly involved in controlling patterns of movements via the spinal cord.  Our understanding of the functions of those areas in procedural memory is murky.  These regions receive less attention from researchers, mostly because their deeper positioning prevents experimenters from easily recording their activity.  It is possible that neuroscience has remained fairly biased in thinking that higher brain areas like the motor cortex are always important for behavior.  The above findings strongly suggest that our investigations of procedural memories should extend to deeper brain areas.


Written by Konstantin Bakhurin.



Kawai, Risa, Timothy Markman, Rajesh Poddar, Raymond Ko, Antoniu L. Fantana, Ashesh K. Dhawale, Adam R. Kampff, and Bence P. Ölveczky. “Motor Cortex Is Required for Learning but Not for Executing a Motor Skill.” Neuron: 800-12. doi: 10.1016/j.neuron.2015.03.024

Images from Kawai et al., 2015, and made by Konstantin Bakhurin and Jooyeun Lee.


Knowing Neurons is an award-winning neuroscience education and outreach website that was created by young neuroscientists. The global team members at Knowing Neurons explain complicated ideas about the brain and mind clearly and accurately using powerful images, infographics, and animations to enhance written content. With an extensive social media presence, Knowing Neurons has become an important science communication outlet and resource for both students and teachers.

2 thoughts on “The Departure of Skill Memories from Motor Cortex: Deeper Directions for Neuroscience

  • October 7, 2015 at 10:09 am

    You will want to study the work of Paul Cizek which debunks “cognitivism” and zeros in on the millisecond processes behind behavior and “choice.” Also, “The part of the brain responsible for seeing is more powerful than previously believed. In fact, the visual cortex can essentially make decisions just like the brain’s traditional “higher level” areas”

    “That is one sense in which our study is counterintuitive and surprising,” Brascamp said. “The part of the brain that is responsible for seeing, for the apparently ‘simple’ act of generating the picture in our mind’s eye, turns out to have the ability to do something akin to choosing, as it actively switches between different interpretations of the visual input without any help from traditional ‘higher level’ areas of the brain.”

  • August 11, 2019 at 9:13 pm

    Your blog is big and there is a lot of good information! Thanks

Comments are closed.