Multi-area activity in mouse motor cortex associated with one- and two-handed oromanual dexterity

By recording neural activity in mice performing uni- and bimanual food handling, this study reveals that forelimb motor cortices (fl-M1 and fl-M2) encode distinct information about hand laterality and usage for bimanual coordination, whereas the lateral oral and manual (LOM) area maintains invariant activity focused on oromanual ingestion parameters.

The Gist

Imagine your brain's motor cortex as a bustling, high-tech control room for your body. In this control room, there are different departments (brain areas) responsible for moving your limbs. Scientists have long known how these departments work when you use just one hand (like typing or waving). But what happens when you use both hands together, or when you switch hands? Does the control room reorganize, or does it keep running the same way?

This study, conducted on mice, explores exactly that. The researchers watched mice pick up and eat food, a task they can do with one paw, the other paw, or both paws at once. They listened to the "whispers" (neural spikes) of three specific departments in the mouse's brain to see how they changed their chatter based on how the mouse was eating.

Here is the breakdown of what they found, using some everyday analogies:

The Three Departments

The researchers focused on three specific areas in the mouse's brain:

1. fl-M1 & fl-M2 (The "Hand Specialists"): These are the primary and secondary motor areas dedicated to the front limbs (paws). Think of them as the specialized mechanics in a car shop who know exactly how to tune the left engine and the right engine separately.
2. LOM (The "Face-Hand Coordinator"): This is a unique area that controls both the mouth (for eating) and the hands. Think of this as the conductor of an orchestra who cares less about which specific instrument is playing, and more about whether the music is flowing smoothly toward the audience.

The Experiment: The "Hand Blocker"

To test these areas, the scientists put the mice in a harness and used a clever 3D-printed gadget to block one paw at a time.

• Scenario A: Block the left paw (mouse must use the right).
• Scenario B: Block the right paw (mouse must use the left).
• Scenario C: No blocks (mouse uses both).

They recorded the brain activity while the mice grabbed sunflower seeds and brought them to their mouths.

The Big Discovery: Two Different Styles of Control

1. The Hand Specialists (fl-M1 & fl-M2): The "Strict Managers"

When the mice used their paws, the "Hand Specialist" departments changed their behavior significantly depending on which paw was moving and how many paws were moving.

• The Analogy: Imagine a manager who has two different playbooks. If you use your right hand, they pull out the "Right Hand Playbook." If you use your left hand, they pull out the "Left Hand Playbook." If you use both hands, they pull out a completely different "Two-Hand Playbook."
• What this means: These brain areas are very specific. They keep separate, detailed information about what each hand is doing. This allows the brain to coordinate complex, two-handed tasks (like holding a seed with one paw and peeling it with the other) by keeping the instructions for each hand distinct.

2. The Face-Hand Coordinator (LOM): The "Chill Conductor"

In contrast, the LOM area was remarkably consistent. It didn't care much if the mouse was using the left paw, the right paw, or both.

• The Analogy: Imagine a conductor who only cares about the destination of the music. As long as the food is moving toward the mouth, the conductor keeps the same rhythm and volume. Whether the food is coming from the left, the right, or both sides, the conductor's signal remains the same: "Bring it to the mouth!"
• What this means: This area is focused on the goal (getting food to the mouth) rather than the specific tool (which hand is doing it). It encodes the relationship between the hand and the face, regardless of which hand is moving.

Why Does This Matter?

Think of it like a GPS system:

• The Hand Specialists (fl-M1/2) are like the detailed turn-by-turn navigation. They need to know exactly which lane you are in (left or right hand) and if you are driving a single car or a truck with a trailer (one or two hands) to give you precise instructions.
• The Face-Hand Coordinator (LOM) is like the "Destination" button. It just knows you are heading to the restaurant (the mouth). It doesn't care if you drove there alone or with a passenger; it just keeps the route focused on the goal.

The Takeaway

This study shows that our brains (and mice brains) are incredibly flexible. They don't just have one "move hand" button. Instead, they have different layers of control:

1. Specific layers that keep track of exactly which limb is moving to ensure precision and coordination.
2. Goal-oriented layers that focus on the big picture (getting food to the mouth) and stay consistent no matter how you achieve it.

This helps explain how we can effortlessly switch between using one hand, the other, or both, without our brains getting confused. It's a beautiful dance between specific mechanics and a unified goal.

Technical Summary

1. Problem Statement

While cortical dynamics during goal-directed hand movements are well-studied in primates (particularly regarding contralateral control), there is a significant gap in understanding how rodent motor cortex activity adapts when the ipsilateral hand, contralateral hand, or both hands are used simultaneously.

• Knowledge Gap: Previous rodent studies have examined bilateral encoding in isolation but lack a comparative analysis of how spiking activity in multiple motor cortical areas changes when the same complex task is performed unimanually (one-handed) versus bimanually (two-handed).
• Context: Most existing rodent studies use constrained, low-dimensional tasks (e.g., lever pressing). This study aims to address this using a natural, ethological behavior (food handling) that involves complex, high-dimensional kinematics and flexible coordination modes.

2. Methodology

The researchers employed a combination of high-speed 3D kinematics, electrophysiology, and computational modeling in head-fixed mice.

• Behavioral Paradigm:

  • Task: Mice were trained to handle food (sunflower seeds or pellets) while head-fixed.
  • Manipulation: A custom 3D-printed apparatus with rotating "hand blockers" was used to force mice into three specific conditions: Ipsilateral unimanual, Contralateral unimanual, and Bimanual handling.
  • Modes: The behavior was segmented into two distinct modes: holding/chewing (static holding, jaw movement) and oromanual/ingestion (coordinated hand-to-mouth transport and manipulation).

• Kinematic Tracking:

  • Imaging: Stereoscopic high-speed video (1000 fps) using a prism/mirror setup.
  • Analysis: DeepLabCut and Anipose were used to reconstruct 3D trajectories of the nose and digits (specifically the third digit, D3). Key metrics included hand-to-nose distance and hand-to-hand distance.

• Electrophysiology:

  • Recording: 32- or 64-channel linear silicon probes were implanted in three specific motor cortical areas:
    1. fl-M1: Primary forelimb motor cortex.
    2. fl-M2: Secondary forelimb motor cortex.
    3. LOM (Lateral Oral and Manual): A region previously characterized as tongue-jaw but implicated in hand-face coordination.
  • Units: On average, 75 active units (single and multi-units) were recorded per session.

• Computational Analysis:

  • Unit-Level: Bootstrap tests to determine firing rate significance during "transport-to-mouth" events. Preference indices ($PI = \frac{A-B}{A+B}$) were calculated to quantify laterality (ipsi vs. contra) and manuality (uni vs. bi) bias.
  • Population-Level: Principal Component Analysis (PCA) was used to assess dimensionality and subspace alignment across conditions.
  • Correlation Structure: Pairwise correlation matrices were compared between conditions to detect reorganization of population activity.
  • Decoding: General Linear Models (GLMs) with ridge regularization were trained to decode 3D hand kinematics. Generalization was tested by training on one condition (e.g., left hand) and testing on another (e.g., right hand or bimanual).

3. Key Contributions

• Naturalistic Paradigm: Successfully adapted a complex, ethological behavior (food handling) to a controlled experimental setting where mice can flexibly switch between uni- and bimanual modes.
• Multi-Area Comparison: Provided the first direct comparison of how fl-M1, fl-M2, and LOM encode movement parameters across different laterality and manuality conditions within the same task.
• Decoupling Laterality and Manuality: Demonstrated that these two variables are encoded differently across cortical areas, challenging the view of a uniform motor map.

4. Key Results

A. Single-Unit Activity

• fl-M1 and fl-M2: Activity was highly dependent on both laterality and manuality.
  • A significant portion of units responded selectively to specific conditions (e.g., only contralateral unimanual or only bimanual).
  • The largest group of responsive units in fl-M1 preferred contralateral unimanual movements (22.8%), followed by bimanual-only (20.7%).
  • Units showed strong biases toward contralateral and unimanual preferences.
• LOM: Activity was largely invariant to laterality and manuality.
  • The largest category of units responded significantly during all three conditions (ipsilateral, contralateral, and bimanual).
  • LOM units showed minimal preference for specific hands or the number of hands used.

B. Population Dynamics (PCA)

• Dimensionality: LOM exhibited lower-dimensional activity (higher variance explained by the top PC) compared to fl-M1 and fl-M2.
• Subspace Alignment:
  • fl-M1/fl-M2: The neural subspaces for different conditions (ipsi vs. contra, uni vs. bi) were distinct. Projecting data from one condition into the subspace of another resulted in significant trajectory deviations and reduced magnitude.
  • LOM: The subspaces were highly aligned across conditions. Unimanual and bimanual activity followed nearly identical rotational trajectories within the same low-dimensional space.

C. Population Correlation Structure

• fl-M1/fl-M2: The pairwise correlation structure between neurons reorganized significantly depending on which hands were moving.
• LOM: The correlation structure remained preserved (correlation coefficient ~0.89) across all conditions, indicating a stable population code regardless of the effector configuration.

D. Kinematic Decoding & Generalization

• Decoding Accuracy: Kinematics could be decoded from all areas, with LOM showing the highest accuracy, followed by fl-M2 and fl-M1.
• Generalization (Invariance):
  • fl-M1/fl-M2: Decoders trained on one hand or unimanual condition failed to generalize well to other conditions (average accuracy drop of -23% to -35%).
  • LOM: Decoders showed high generalization; a model trained on one condition predicted movements in other conditions with high accuracy (average drop only -7%), indicating a condition-invariant mapping from neural activity to kinematics.

5. Significance and Conclusion

The study proposes a functional model for the division of labor in the mouse motor cortex during complex dexterity:

1. fl-M1 and fl-M2 (Forelimb Specification): These areas maintain separable information about the specific limb and the coordination mode (uni- vs. bimanual). This specificity likely facilitates the precise control and coordination required for bimanual tasks, allowing the brain to distinguish between the movements of the left and right limbs.
2. LOM (Oromanual Coordination): This area encodes the proximity of the hand(s) to the mouth and the state of ingestion, regardless of which hand is used or if one or two hands are involved. Its invariance suggests it acts as a higher-order coordinator for the integration of hand and face movements, focusing on the goal (getting food to the mouth) rather than the specific effector configuration.

Implications:

• The findings suggest that motor cortex is not a monolithic map but a collection of areas with distinct computational roles: some dedicated to effector-specific control (M1/M2) and others to task-goal invariant coordination (LOM).
• This challenges the strict "premotor vs. primary" hierarchy often assumed in rodents, suggesting instead flexible, recurrent interactions where M1 and M2 share similar coding properties for this specific task.
• The results provide a framework for understanding how the brain manages the trade-off between specificity (controlling individual limbs) and invariance (achieving a common goal via different means).