The Training Village: an open platform for continuous testing of rodents in cognitive tasks

The paper introduces the Training Village, an open-source, automated, and affordable home-cage system that enables continuous, personalized cognitive training and monitoring of rodents to overcome the labor and expertise barriers of traditional behavioral research.

The Gist

Imagine you are trying to teach a group of squirrels to solve a complex puzzle to get a nut. In a traditional lab, a scientist acts like a strict drill sergeant: they wake up, grab one squirrel at a time, carry it to a small room, force it to sit at the puzzle for an hour, then carry it back to its cage. They do this every day, often starving the squirrels beforehand to make them hungry enough to work. It's stressful for the squirrels, exhausting for the scientist, and the data is limited to just that one hour of "forced" work.

Now, imagine a different approach: The Training Village.

This paper introduces a new, open-source system that turns the lab into a smart, self-running "village" where the animals live, sleep, and play together, but can also choose to go to a "gym" (the puzzle room) whenever they feel like it.

Here is a breakdown of how it works, using simple analogies:

1. The Setup: A Smart Neighborhood

Think of the Training Village (TV) as a high-tech apartment complex for mice and rats.

• The Home Cages: These are the living rooms. The animals live in groups here, just like they would in the wild, with toys, tunnels, and plenty of space. They aren't isolated; they can socialize and sleep naturally.
• The Corridor: This is the hallway connecting the living rooms to the gym. It has smart doors and cameras that act like a bouncer.
• The Gym (Operant Box): This is where the cognitive tasks happen. It's a small room with screens, lights, and a water dispenser that gives sweet water as a reward.

2. How the Animals "Work"

In the old way, the scientist decides when the animal works. In the Training Village, the animal decides.

• The Motivation: Instead of starving the animals, they have access to slightly sour water in their home cages. The "gym" offers sweet, sugary water. The animals learn that if they want the tasty treat, they have to go to the gym and solve the puzzle.
• The Bouncer System: The corridor is the key innovation. It uses RFID tags (like tiny ID chips under the skin) and cameras.
  • If a mouse tries to enter the gym, the system checks: "Is the gym empty? Has this mouse worked recently?"
  • If yes, the door opens.
  • If the gym is busy or the mouse just finished a session, the door stays closed.
  • This prevents a "stampede" where all the mice try to enter at once, and ensures everyone gets a fair turn.

3. The "Brain" of the Village

The whole system is run by a Raspberry Pi (a tiny, cheap computer, about the size of a credit card) that acts as the village manager.

• 24/7 Operation: The village never sleeps. Animals can work at 3:00 AM or 3:00 PM, whenever they feel motivated.
• Remote Control: The scientists don't need to be in the lab. They can check the system from their phones or laptops. If a door gets stuck or a mouse gets sick, the system sends an alert (like a text message) to the scientist.
• Auto-Grading: The computer watches how well the animal is doing. If a mouse gets the puzzle right 90% of the time, the system automatically makes the puzzle harder for the next session. If it's struggling, it makes it easier. No human needs to adjust the settings.

4. Why This is a Game-Changer

The researchers tested this with mice and rats on several difficult mental tasks (like remembering where a light was, or choosing the best path for a reward). Here is what they found:

• Fair Play: Even though there is only one "gym" for a group of 10–12 animals, the animals naturally take turns. They don't let the "bossy" mice dominate; everyone gets to work roughly the same amount. It's like a well-organized coffee shop where everyone waits their turn without a manager telling them to.
• More Data, Less Stress: Because the animals can work whenever they want, they do more trials than in traditional labs, but they are less stressed because they aren't being dragged around or starved.
• Real Life Context: The system tracks what the animals do between tasks. Do they sleep more after a hard puzzle? Do they hang out with friends? This gives scientists a complete picture of the animal's life, not just a snapshot of one hour of forced work.
• Cost-Effective: Traditional high-tech labs cost tens of thousands of dollars. The Training Village is built with 3D-printed parts and open-source code, costing about $6,000. It's like building a custom smart home instead of buying a luxury mansion.

The Big Picture

The Training Village is like upgrading from a manual typewriter to a self-driving car. It removes the human error and exhaustion from the equation. It allows scientists to study how the brain learns and changes over months or years, with animals living happy, healthy lives.

This isn't just about making science easier; it's about making it better. By letting animals work at their own pace in a natural environment, the data scientists get is more accurate, more reliable, and more relevant to how brains actually function in the real world. It's a win for the animals, a win for the scientists, and a win for the future of medical research.

Technical Summary

1. Problem Statement

The study of neural mechanisms underlying complex cognitive functions in rodents faces a significant bottleneck: the adaptation of human-like cognitive tasks requires prolonged, labor-intensive training periods. Current solutions have distinct limitations:

• Manual Training: Requires significant researcher time, introduces subjective biases, disrupts animal circadian rhythms, and often necessitates food/water deprivation, causing stress.
• Multi-box Systems: While they scale up testing, they still require manual transfer of animals between home cages and operant boxes, occupy large physical spaces, and are costly to maintain.
• Existing Home-Cage Systems: Many are limited to simple ethological tasks, rely on single-housed animals (ignoring social dynamics), or lack modularity to integrate diverse cognitive paradigms and commercial hardware.

There is a critical need for an affordable, open-source, fully automated system that allows group-housed rodents to train autonomously in complex cognitive tasks 24/7 while maintaining high welfare standards and enabling longitudinal monitoring.

2. Methodology: The Training Village (TV)

The authors developed the Training Village (TV), a modular, open-source system designed to bridge the gap between home-cage living and complex operant conditioning.

Hardware Architecture

• Core Components: The system consists of three main elements:
  1. Home Cages: Enriched environments (mimicking Eco-HAB designs) where groups of animals live socially.
  2. Operant Box: A custom chamber where cognitive tasks are performed. The design is modular, supporting both a standard 3-port box and a touchscreen T-maze.
  3. Sorting Corridor: A transparent tube connecting the home cages to the operant box, equipped with motorized doors, RFID readers, and video cameras.
• Control Unit: A Raspberry Pi 5 serves as the central controller, managing sensors, actuators, and cameras. It runs custom open-source Python software.
• Identification & Tracking: Animals are implanted with RFID microtransponders. The corridor uses a dual-detection system (RFID + real-time video analysis) to ensure only one animal enters the operant box at a time.
• Sensors: Includes load cells for weight monitoring, temperature/humidity sensors, and infrared cameras for 24/7 surveillance.
• Cost: Approximately €5,500 per unit, significantly cheaper than commercial automated systems.

Software & Automation

• State Machine Logic: The system operates via a state machine that manages access. When an animal enters the corridor, the system checks its ID, verifies the inter-session interval (refractory period), and opens the door to the operant box.
• Adaptive Training: The system automatically adjusts task parameters (difficulty, stage) based on real-time performance data stored in a local database.
• Remote Supervision: A Graphical User Interface (GUI) allows remote monitoring via VNC/AnyDesk. It features live video feeds, real-time behavioral plots, and a Telegram alarm system that notifies researchers of errors, low weight, or system failures.
• Motivation Strategy: Instead of strict water deprivation, animals have free access to slightly sour water (citric acid) in home cages and are rewarded with sweetened water in the operant box, maintaining health while driving task engagement.

3. Key Contributions

• First Modular, Open-Source Platform: The TV is the first system to simultaneously combine continuous autonomous training, group housing, complex cognitive task compatibility, and remote supervision in an affordable, open-source package.
• Self-Regulated Training: It enables animals to self-initiate training sessions at their own pace, eliminating experimenter-imposed schedules and deprivation stress.
• Integration of Home Cage & Task Data: By integrating with Eco-HAB, the TV links spontaneous home cage behaviors (sleeping, eating, socializing) with cognitive task performance.
• Scalability: The system supports both mice and rats (with scaled hardware) and can be adapted to various behavioral paradigms without changing the core infrastructure.

4. Key Results

The system was validated across 12 groups of rodents (11 mouse groups, 1 rat group; 121 total subjects) performing three distinct cognitive tasks:

1. Visuospatial Delayed-Response (3AFC): Testing short-term memory.
2. Visual/Auditory Discrimination (2AFC): Testing perceptual discrimination.
3. Two-Armed Bandit (2AB): Testing reinforcement learning and decision-making.

Performance Metrics:

• Efficient Resource Sharing: Animals self-organized to share a single operant box with ~80% occupancy during training periods. The system achieved an average of ~134 hours of data collection per week.
• Balanced Access: Despite social hierarchies, box usage was highly egalitarian (Gini coefficient < 0.15), indicating no single animal dominated access.
• High Engagement: Animals spent ~64% of their time inside the box engaged in the task. Engagement levels increased over months of training, and self-trained animals showed equal or slightly higher accuracy compared to manually trained controls.
• Longitudinal Stability: The system generated thousands of trials per animal over months without significant drop-off in performance or motivation.
• Circadian & Welfare: Animals maintained natural circadian rhythms. Task performance did not disrupt sleep cycles significantly. Weight monitoring confirmed animals remained healthy without strict deprivation.
• Behavioral Linkage: The system successfully correlated home cage behaviors (e.g., time spent in nesting vs. eating cages) with task engagement, revealing that animals often transitioned from sleeping to task-seeking immediately after a session ended.

5. Significance and Impact

• Accelerating Brain Research: The TV removes the "human bottleneck" in rodent training, allowing for high-throughput, longitudinal studies of learning, memory, and cognitive decline that were previously unfeasible.
• Improved Translational Validity: By enabling continuous, stress-free testing that mimics natural fluctuations in motivation and arousal, the data generated is likely more robust and clinically relevant than data from short, forced daily sessions.
• Animal Welfare (3Rs): The system adheres to the principles of Replacement, Reduction, and Refinement by:
  • Refinement: Eliminating transport stress and strict deprivation.
  • Reduction: Generating massive datasets from fewer animals due to the efficiency of continuous testing.
  • Social Housing: Allowing animals to live in enriched groups rather than isolation.
• Open Science: As an open-source platform with available hardware designs and software, it democratizes access to advanced cognitive testing, fostering reproducibility and inter-laboratory collaboration.

In conclusion, the Training Village represents a paradigm shift in behavioral neuroscience, transforming rodent cognitive training from a manual, episodic process into a fully automated, continuous, and ethologically rich longitudinal study platform.