top of page
認知行動神経機構部門
Laboratory of Cognitive and Behavioral Neuroscience
Our tools
At Takahashi Laboratory, we develop experimental and analytical tools to study animal behavior and neural dynamics across laboratory and natural settings. Our platforms are designed to increase experimental flexibility, improve data quality, and enable measurements that are difficult to achieve with conventional approaches.
Current examples include the Reconfigurable Maze, Neurologger, and vmTracking.
tools showcase
Reconfigurable Maze: Modular Behavioral Testing Platform
The Reconfigurable Maze is a modular behavioral testing system that allows researchers to build and rapidly reconfigure a wide range of maze layouts within a single experimental framework. Unlike conventional fixed mazes, it enables precise, repeatable changes in spatial structure while maintaining a stable testing environment. This makes it particularly well suited for studies of spatial learning, memory, and cognitive flexibility (Hoshino, et. al, iScience, 2020, Sawatani, et al., JoVE, 2022).
A key strength of the system is that flexibility does not come at the expense of experimental rigor. The modular components are designed for geometric precision, reproducibility, and ease of use, allowing investigators to move efficiently between different configurations while preserving comparability across trials. The platform has also been shown to support reliable behavioral and neural measurements, making it useful for linking maze geometry to hippocampal coding and adaptive behavior.
By replacing multiple fixed apparatuses with a single scalable system, the Reconfigurable Maze offers a practical and versatile approach for behavioral neuroscience.
Neurologger: Wireless Neural Activity Logger
The Neurologger is a lightweight wireless recording device designed to capture neural activity in animals moving freely through natural or experimental environments. By storing electrophysiological signals locally on the device rather than transmitting them through cables, it overcomes a major limitation of conventional neural recording systems: physical tethering. This enables neural measurements during behaviors that are otherwise difficult or impossible to study, including flight, swimming, and large-scale exploration (Ide and Takahashi, Micromachines, 2023).
Because the system operates without real-time wired connections, it preserves natural behavior while maintaining high-fidelity data acquisition. This makes the Neurologger especially valuable for neuroethology, where the goal is to understand brain activity under ecologically relevant conditions (Takahashi et al., Science advances, 2022). The platform has supported studies of navigation and spatial cognition in freely behaving animals, including work examining neural representations during movement in complex outdoor environments
More broadly, the Neurologger extends neuroscience beyond conventional laboratory constraints, enabling direct investigation of the neural basis of natural behavior (Takahashi et al., Animal Biotelemetry 2023).
vmTracking: Virtual Marker Multi-Animal Tracking
vmTracking (Virtual Marker Tracking) is a computational framework for high-accuracy pose tracking of multiple animals in video data, particularly in crowded or occlusion-rich settings. A major challenge in multi-animal tracking is that identities are often lost when individuals overlap, cross, or temporarily disappear from view. vmTracking addresses this problem by introducing virtual markers—persistent digital identifiers that help maintain the identity of each subject across time (Azechi & Takahashi, PLoS Biology, 2025).
The approach builds on existing markerless pose-estimation frameworks such as DeepLabCut and SLEAP, combining their flexibility with improved identity preservation (Azechi & Takahashi, JoVE, 2025).. Rather than relying on physical tags attached to animals, vmTracking assigns software-based labels that remain linked to each individual throughout the tracking process. This substantially reduces identity switches, lost tracks, and the need for manual correction.
vmTracking is particularly useful for studies of social interaction, collective behavior, and group movement, where accurate individual-level tracking is essential but technically challenging. Its utility extends beyond animal research to other complex multi-agent settings, offering a general strategy for reliable pose-based tracking under conditions of frequent occlusion (Yin et al., Sports Engineering, 2026).
bottom of page