Custom AutoNoMouse
AutoNoMouseBase is a base class for running tasks automatically without a real animal. It is useful for testing and debugging your task code — you can simulate nose pokes, update a virtual animal position, and inject mock trial data, all without needing a physical subject in the box.
Each virtual animal is a class that inherits from AutoNoMouseBase, written in any
file of your project’s code folder. Its TASK_NAME says which task it is for (the
task’s class name), so a project can have one for each task. A class without
TASK_NAME is used for any task that has no virtual animal of its own.
It is started and stopped from the VIRTUAL MOUSE tab of the MONITOR screen while the
task is running, where its parameters (PARAMS) can also be changed. The same tab can
inject simulated trials directly into the session data, without running the task’s
state machine, to fill a session quickly and test the plots or the training protocol.
Read the base class in village/custom_classes/auto_no_mouse_base.py to understand the full interface before subclassing it.
Creating a Custom AutoNoMouse Class
from village.custom_classes.auto_no_mouse_base import AutoNoMouseBase, AutonomouseParam
class MyVirtualAnimal(AutoNoMouseBase):
"""Automated task execution without a real animal."""
TASK_NAME = "Bpod1Habituation" # the task it is for
# Parameters shown (and editable) in the VIRTUAL MOUSE tab:
# name, type, default, label, min, max, tooltip
PARAMS = [AutonomouseParam("pause", float, 1.0, "pause (s)", 0.0, 10.0)]
def run_trial(self) -> None:
"""Sequence of actions for one trial."""
self.poke(port=1, duration=0.1)
self.wait(self.pause)
A complete example, for the delay task of the demo project: auto_follow_the_light.py.
Available Attributes and Methods
Define these in your subclass:
TASK_NAME: The class name of the task it is for. Leave it empty for any task.PARAMS: A list ofAutonomouseParam(name, type, default, label, minimum, maximum, tooltip). Each one becomes an attribute (self.<name>) and a control in theVIRTUAL MOUSEtab.run_trial(): Called over and over while the virtual animal runs, once per trial. Define the sequence of actions the virtual animal performs — pokes, position updates, waits.self.taskis the running task, so it can read what the task decided for the trial (e.g. the correct side).inject_trial(**params): Appends one simulated trial row directly to the session data (self.task.session_df), bypassing the state machine. It receives the current values ofPARAMS.
And call these:
inject_trials(n, interval=1.0, **params): Callsinject_trialn times, one everyintervalseconds.poke(port, duration=0.1): Simulates a nose-poke in and out on the given Bpod port (only with Bpod). It sends Out, In 10 ms later, and Out again afterduration. The first Out does nothing when the port is already out, but a port with nothing connected reads as in, so without it the In would not be registered. With nothing connected, the first poke of each trial therefore also records an extra Out, 10 ms before the In.set_position(x, y): Updates the virtual animal’s position and trace, as seen by the camera tracking system.wait(seconds): Pauses execution for the given duration, stopping early if the virtual animal is stopped.