reach is a small Python library for creating forelimb reaching tasks for mice or other rodents used for neuroscientific research. It has been developed for my own implementation of a Raspberry Pi-based visually-guided reaching task using one or two reaching targets and visual stimuli -- both head-fixed and freely-moving -- but has been designed to work with arbitrary 'backends' that may implement other stimulus or reach target configurations. reach is concerned with controlling hardware, responding to events during the task, and organising the resulting data in a format that makes analysis easy.

The code and data representation follows a hierarchical object-oriented structure:

Data is saved per animal in a JSON file. This data can be loaded as follows:

from reach import Mouse

data_dir = '/path/to/your/data/'
mouse = Mouse.init_from_file(
    mouse_id='Mouse1',
    data_dir=data_dir,
)

A training session can be run, appending its results to the data stored in Mouse. You would then want to save the data back to JSON to keep the new data:

mouse.train(backend)
mouse.save_data_to_file(data_dir)

backend here is what interfaces with the hardware controlling the task. Details on creating custom backends is here.