How does the research interest emerge?

Although the neural control of locomotion has been investigated for more than 50 years and enabled brain-machine interface design, most efforts have been devoted to determining the neural mechanisms underlying locomotor control in a single environmental context. However, an unsolved issue is how neurons encode the same type of locomotor control variables in different environmental contexts. Lack of this knowledge has impeded the development of generalizable brain-machine interfaces. While the primary motor cortex relates more directly to motor output, the secondary motor cortex (M2) is considered to encode more complex motor-related information, including movement preparation and various sensory signals linked to specific motor behaviors. Our team focused on how different contexts influence the neural signals of M2 neurons. More specifically, we asked how M2 neurons encode the same type of locomotor control variables in different environmental contexts.

The methods we use

In this study, we established three distinct behavioral contexts: the Y-maze, the running wheel, and the open field. Our objective was to expose mice to varied environmental stimuli, prompting them to employ different movement strategies. For instance, in the Y-maze, although there were no specific rewards, the animals were required to make a choice at the intersection of the three arms. In contrast, the running wheel, considered the simplest context, allowed the animals to run continuously without regard for direction or obstacles. In the open field, the animals had the greatest freedom to explore, with the ability to stop and change direction at will. Although these contexts would elicit different movement patterns, the animal displays some similar locomotor variables, such as: locomotion, rest, movement start and stop, enabling us to further analyze how M2 neurons encode the same variable in different environments.

To ensure that the imaging of neural signals would not influence the animal’s behavior, we performed Ca²⁺ imaging using a head-mounted miniature microscope. This method was effective for imaging a wide range of neurons in freely moving mice. To better evaluate the activity of individual neurons during various locomotion behaviors, we fitted the neural activity around a behavioral event with a second-order polynomial function, based on established methods.

What we found

After screening and classifying 1405 neurons in the superficial layers (layers 2/3) of the M2 region from 8 mice, we first analyzed the locomotion switch signal: locomotion initiation and cessation. Interestingly, we discovered that active M2 neurons before and after locomotion are overrepresented across contexts, suggesting egocentric voluntary control function. And this group of neurons may play a crucial role in regulating the initiation and cessation of self-initiated locomotion and was overrepresented across contexts.

 Similarly, we analyzed locomotion maintenance signal. We found that most M2 neurons exhibited sustained suppression during locomotion and could across different contexts, also suggesting egocentric voluntary control function. However, the smaller populations of locomotion-activated M2 neurons are mostly context-specific, suggesting exocentric navigation functions.

What comes next

Our results offer novel insights into how the M2 region regulates self-initiated locomotor behaviors in both context-dependent and context-independent manners. Although we tested the M2 neural representation in three different contexts, the underlying reasons require further investigation. Additionally, more kinematic parameters should be examined using advanced recording devices, such as those that capture step-cycle and interlimb coordination dynamics, which may elucidate the varying neural responses observed in the M2 region.

We mainly focused on signals related to the initiation and termination of movement in the M2 region. More behavior paradigms could be used to further investigate the cross-context representation of self-initiated locomotion by M2 neurons. For example, obstacle navigation tasks can be employed to investigate context-dependent gait adjustments or to utilize forelimb-hindlimb coordination paradigms (e.g., ladder walking) for more precise control of movement kinematics.

Closing Thoughts

Taken together, this study has identified both context-dependent and context-independent locomotion-related neural signals in M2, which may collectively contribute to the control of self-initiated locomotion behaviors during navigation in different spatial contexts. In addition, our discovery of cross-text stability in M2 neural encoding mechanisms provides biologically inspired insights that may advance the understanding of this fundamental challenge in both neuroscience and AI research.

https://www.nature.com/articles/s42003-025-08131-7