Introduction & Aims
Animals in nature sense their surroundings by actively engaging with them and processing the resulting signals according to their utility. Traditionally, the neuronal circuitry of sensory signalling and behaviour has been primarily explored using head-fixed rodents [1]. However, head-fixation induces long-term stress [2], severely limits natural movements [3], and disrupts sensory perception and exploration strategies crucial for sensory-based tasks [4].
To address these limitations, we developed an experimental architecture for investigating these capacities in freely moving rodents. We aimed for the setup to allow for the exploration of context-dependent decision-making, flexible planning, and the abstraction of sequential rules. The setup enables study of foraging behaviour in freely moving mice without restricting their movement or access to nutrition. The maze includes automated animal tracking, stimulus presentation and reward delivery.
Methods
We built the modular labyrinth with ‘Makerbeam XL’ posts and PVC/acrylic panels, opaque in visible light but transparent under infrared, encouraging mouse exploration while enabling machine vision tracking [5]. The panels slot into the posts and can be replaced with devices such as 3D-printed food pellet dispensers and rotating tactile gratings, controlled by servo motors and activated by microcontrollers. Mouse entry into specific regions of interest (ROIs) is tracked using OpenCV and a Python state machine.
Results
We developed an automated sensory maze that combines allocentric and egocentric navigation. We found that mice readily habituate to this modular maze, are motivated to explore it with no need for water or food restriction, and perform object-in-place recognition as described in the literature.
Automation in the maze provides high levels of experimental control while enabling simple and flexible behavioural task design. Device (stimulus or reward) motion is triggered when the mouse enters specific user-configured ROIs. The system supports flexible maze reconfiguration and scalability. An animal can encounter multiple stimuli as it moves from the labyrinth’s origin to any endpoint, permitting the experimenter to set up complex rules or conditions governing whether the mouse will be rewarded, involving chains or sequences of stimuli, such as tactile or auditory. The entrance connects to a home cage, permitting free movement between the maze and the cage.
The maze is being used with two different types of stimuli. (1) Sequences of auditory stimuli, delivered according to animal presence in an ROI, are used to investigate how sensory predictability, as evaluated through the learning and processing of sequential structure, influences the emergence of intrinsic reward and subsequent decision making. (2) Sequences of oriented tactile cues are controlled on a trial-by-trial basis and predict the location of rewards in different maze chambers.
Conclusion
Our maze offers a practical and cost-effective approach for studying a wide range of cognitive behaviours in laboratory settings. Through this automated maze we model foraging behaviour, allowing mice to perform abstract sensory-guided sequential tasks, demonstrating context-dependent decision-making and probabilistic rule learning. The platform is designed to be easy to modify and share, through its use of inexpensive and readily available components.