Individual differences drive social hierarchies in male mouse societies

Using a novel semi-naturalistic platform called NoSeMaze, this study demonstrates that social hierarchies in male mice are stable, non-despotic structures driven by consistent individual behavioral and cognitive traits that persist across changing group compositions.

The Gist

Imagine a bustling, high-tech apartment building where 10 mice live together. In this building, there are no walls between the rooms, just a central hallway with two narrow tunnels connecting the living quarters to a kitchen and a playground.

This is the NoSeMaze (Non-invasive Sensor-rich Maze), a fancy, open-source "smart home" built by scientists to watch mice live their lives without ever bothering them.

Here is the story of what they discovered, explained simply:

1. The Setup: A Mouse City with a Twist

Usually, when scientists study mouse hierarchies (who is the boss and who is the underling), they put two mice in a tiny tube and force them to fight. It's like forcing two strangers into an elevator and seeing who pushes the other out. It's stressful and doesn't feel like real life.

The NoSeMaze is different. It's a semi-natural city where mice can eat, sleep, and play whenever they want. The only rule is: to get water, they have to run through the tunnels. Because the tunnels are narrow, sometimes two mice try to enter from opposite ends at the same time. One has to back down. This is a Tube Competition.

The scientists also tracked Chasing: when one mouse actively runs after another through the tunnel.

2. The Big Question: Are Mice Born Bosses, or Do They Become Bosses?

Scientists have always wondered: Is a mouse's social rank (its place in the pecking order) something it carries inside its personality, like a permanent tattoo? Or is it just a temporary thing that happens because of who it happens to be living with right now?

To find out, the scientists did something clever. They took their 79 mice and kept reshuffling the "roommates."

• Round 1: Mice A, B, and C live together.
• Round 2: The scientists mix them up. Now Mouse A lives with Mice X, Y, and Z.
• Round 3: Mix them up again.

If Mouse A was the "Boss" in Round 1, was it still the "Boss" in Round 2 and 3, even though it was living with totally different neighbors?

3. The Discovery: The "Internal GPS" of Personality

The Answer: Yes.
Just like a human might be naturally assertive or shy regardless of who their coworkers are, the mice showed stable personalities.

• If a mouse was a high-ranking "Boss" in one group, it was likely to be a high-ranking "Boss" in the next group, even with new neighbors.
• If a mouse was a "Chaser" (someone who likes to run after others), it kept chasing, no matter who the group was.

This suggests that social status isn't just a reaction to the current crowd; it's an internalized trait. The mice have a "social GPS" that tells them where they stand, and that GPS stays calibrated even when the map changes.

4. The Two Types of "Bossing Around"

The study found two very different ways mice establish their place in the world:

A. The "Tube Test" (Incidental Dominance)
This is the passive stuff. Two mice bump into each other in a tunnel. One backs down. Over time, this builds a ladder of who is on top.

• Analogy: This is like a quiet office where people know who the boss is because of who gets the corner office. It's a stable, structural hierarchy.

B. The "Chase" (Active Negotiation)
This is active stuff. One mouse decides to run after another.

• The Surprise: Chasing wasn't just the Boss bullying the weakest mouse. Instead, chasing happened mostly between the top-tier mice.
• Analogy: Imagine a high school. The "King" and the "Queen" aren't fighting the freshmen; they are constantly checking in on each other, maybe playfully shoving or racing, to make sure they are still the top dogs.
• The Context: In groups where the hierarchy was messy (nobody knew who was boss), the top mice chased more. It was like they were shouting, "I'm the boss!" to clarify the rules. In groups where the hierarchy was clear, they chased less because everyone already knew the score.

5. Brains vs. Brawls: Are Smart Mice Also Bosses?

The scientists also gave the mice a puzzle to solve to get water (a smell game). They measured how impulsive or patient the mice were.

• The Result: Being good at the puzzle (smart, patient, or impulsive) had almost nothing to do with being the social Boss.
• Analogy: Think of a sports team. You can have the star quarterback (the social boss) who is a bit impulsive, and the quiet strategist (the smart learner) who is great at the game but doesn't care about leading the team. In this mouse world, social skills and cognitive skills are two different engines. One doesn't necessarily drive the other.

6. The "Heavyweight" Factor

The scientists also looked at body weight.

• Heavier mice were less likely to lose a tube fight or get chased.
• Analogy: Being big is like wearing a suit of armor. It doesn't make you a bully who starts fights, but it makes other mice think, "Nah, I'm not messing with that one," so they leave you alone. It's a passive shield, not an active sword.

The Takeaway

This paper tells us that individuality is real and stable. Even in a group of animals that changes its members constantly, each mouse carries its own unique "social fingerprint."

• Some are naturally calm and cautious.
• Some are impulsive and aggressive.
• Some are the "Bosses" who maintain their status across different groups.
• Some are the "Chasers" who constantly negotiate their rank with other high-rollers.

The NoSeMaze allowed scientists to see that social status is a mix of who you are (your personality) and where you are (your group), but "who you are" matters a lot more than we thought. It's a blueprint for understanding resilience: knowing your place in the world is a stable part of your identity, not just a temporary reaction to your surroundings.

Technical Summary

1. Problem Statement

Social hierarchies are fundamental to group organization in many species, yet the mechanisms driving their formation and stability remain unclear. Previous research faces several limitations:

• Reductionist Paradigms: Most studies assess social and cognitive behaviors in isolation, often in artificial, short-term assays (e.g., forced tube tests) that lack ecological validity and fail to capture moment-to-moment interactions.
• Context Dependency vs. Internalization: It is debated whether social rank is purely an emergent property of current group dynamics or if it reflects stable, internalized individual traits that persist across changing social contexts.
• Lack of Integration: There is a gap in understanding how non-social cognitive traits (e.g., impulsivity, reinforcement learning) relate to social behaviors (e.g., dominance, chasing) in large, semi-naturalistic groups over long periods.
• Manual Intervention: Traditional assays often require human handling, which can introduce confounds and distort natural behavior.

2. Methodology

The authors developed and utilized a novel platform called the NoSeMaze (Non-invasive Sensor-rich Maze) to address these gaps.

• Experimental Setup:

  • Platform: A semi-naturalistic, open-source, modular system featuring an open-field arena, a housing arena with nesting material, and two RFID-equipped tubes connecting them.
  • Subjects: 79 adult male C57BL/6J mice (two cohorts: younger 16–30 weeks and older 55–97 weeks). This included a secondary manipulation where 26 mice had conditional deletion of Oxytocin Receptors in the Anterior Olfactory Nucleus ($OXTR^{\Delta AON}$).
  • Design: An "accelerated longitudinal" design. Mice lived in groups of 9–10 for ~3 weeks (a "round"), after which they were returned to home cages, reshuffled, and placed into new groups with different members for subsequent rounds. This allowed for testing stability across changing social contexts.
  • Data Collection: Automated, continuous tracking over >4,000 mouse-days without human intervention.

• Behavioral Assays Integrated:

  1. Social Rank (Incidental Competition): Mice must cross tubes to access food/water. When two mice enter a tube from opposite ends simultaneously, a "tube competition" occurs. The dominant mouse pushes the other back. RFID detections automatically record wins/losses to calculate David's Scores and Elo ratings.
  2. Chasing: Voluntary, active pursuit of one mouse by another through the tubes, quantified as the fraction of initiated chases vs. being chased.
  3. Reinforcement Learning: A self-paced go/no-go olfactory task at a water lickport. Mice learn to lick for water upon a rewarded odor (CS+) and withhold licking for an unrewarded odor (CS-). Contingencies reverse every 3 days. Metrics include pre-CS licking (impulsivity), hit/rejection rates, and switch latencies.

• Statistical Analysis:

  • Used Spearman correlations, Intraclass Correlation Coefficients (ICCs) for stability, and Linear Mixed-Effects Models (LMEs) to account for repeated measures.
  • K-means clustering to identify distinct learning strategies.
  • Principal Component Analysis (PCA) to test orthogonality between social and cognitive domains.

3. Key Contributions

• The NoSeMaze Platform: A high-throughput, automated system enabling long-term, longitudinal phenotyping of social and cognitive behaviors in group-housed mice under semi-naturalistic conditions.
• Disentangling Stability vs. Emergence: By systematically reshuffling group compositions, the study provides robust evidence that social rank is a stable, internalized trait rather than just a product of immediate group dynamics.
• Multidimensional Sociality: The study decouples "social rank" (derived from incidental competition) from "chasing" (proactive aggression), showing they are distinct but related dimensions of social position.
• Orthogonality of Domains: Demonstrates that social behaviors (rank/chasing) and non-social cognitive styles (impulsivity/flexibility) are largely independent, challenging the notion of unified "behavioral types."

4. Key Results

A. Stability of Social Rank and Chasing

• Internalized Rank: Social rank (David's scores) derived from incidental tube competitions was highly stable across different group compositions (Spearman's $\rho \approx 0.57$ between rounds 1 and 2; ICCs 0.55–0.89). This suggests individuals carry a stable "social identity."
• Chasing Stability: The propensity to initiate chases was also highly stable across rounds ($\rho \approx 0.75$), indicating it is a consistent individual trait.
• Hierarchy Structure: Hierarchies were non-despotic, transitive, and stable. Higher body weight correlated with fewer losses but not more wins, suggesting weight acts as a passive buffer against subordination rather than a driver of dominance.

B. Reinforcement Learning Strategies

• Three Clusters: K-means clustering identified three distinct learning strategies:
  1. Impulsive Go-Learners: High pre-CS licking, fast CS+ adaptation, poor CS- inhibition.
  2. Cautious No-Go Learners: Low impulsivity, slow CS+ adaptation, high CS- inhibition.
  3. Flexible Learners: Balanced performance, high adaptability to both CS+ and CS- reversals.
• Stability: These cognitive styles showed moderate-to-high stability across rounds, particularly for "Flexible" and "Impulsive" learners.

C. Relationship Between Social and Cognitive Domains

• Rank vs. Chasing: High-ranking mice initiated more chases. However, the strength of this correlation depended on hierarchy clarity. In groups with low transitivity (ambiguous hierarchies), chasing was strongly linked to rank (used to negotiate status). In highly transitive groups, chasing was less linked to rank, suggesting stable hierarchies reduce the need for proactive signaling.
• Chasing Dynamics: Chasing was concentrated among high-ranking individuals and often occurred reciprocally within this "elite" group, suggesting it functions as a mechanism for negotiating relative standing among top-tier mice rather than just enforcing dominance over subordinates.
• Social-Cognitive Orthogonality:
  • No significant correlation was found between social rank/chasing and specific cognitive metrics (impulsivity, reward sensitivity, flexibility) after correction for multiple comparisons.
  • PCA confirmed that social and cognitive features loaded onto distinct principal components, supporting the view that they are independent dimensions of individuality.
  • Exception: "Cautious" learners initiated fewer chases than "Flexible" learners, but overall, the domains remained largely dissociable.

D. Genotype Effects ($OXTR^{\Delta AON}$)

• Mice with OXTR deletion in the AON showed a transient reduction in social rank during the first week of a new group, which normalized by week 3.
• No effects were observed on chasing behavior or reinforcement learning strategies.
• This suggests OXT-dependent olfactory learning is critical for initial social recognition and clique formation but not for the maintenance of stable, internalized dominance hierarchies.

5. Significance

• Paradigm Shift: The study moves beyond forced, short-term assays to demonstrate that social position in mice is a multifaceted, stable trait that persists across changing environments.
• Resilience Modeling: By establishing that social rank and cognitive styles are stable individual differences, the NoSeMaze provides a powerful framework for studying resilience factors in psychiatric and neurological disorders.
• Mechanistic Insight: The findings clarify that "dominance" is not a monolithic behavior; it comprises stable competitive outcomes (rank) and dynamic, context-sensitive proactive signaling (chasing).
• Translational Potential: The platform's compatibility with longitudinal neural interventions (e.g., chemogenetics, fMRI) and its open-source nature make it a versatile tool for linking neural circuit dynamics to complex social behaviors over the lifespan.

In conclusion, the paper establishes that individual differences in behavior drive social hierarchies in mice, with social rank and proactive chasing representing stable, internalized traits that are largely independent of non-social cognitive styles.