Affordable, open-source operant conditioning chambers designed for use in natural habitats across many animal species
Open-source apparatuses that you can build yourself to enable comparative cognition
research in an animal’s natural habitat
First Goal of Outdoor Operant Conditioning Chamber
Build your own operant conditioning chamber:
Testing whether squirrels can imagine hidden information and make decisions based on what might be there
Non-invasive research conducted in the animals' natural environment with simple, accessible equipment
Methods designed for researchers, students, and wildlife enthusiasts to participate in cognitive studies
Studying logical reasoning and problem-solving abilities in wild squirrels using our operant
conditioning chambers
A simple yet powerful test of animal cognition, adapted from rat studies to understand
whether squirrels can imagine hidden information
Squirrels learn through reinforcement: press the lever when the right cues appear (light + sound together), and receive food rewards.
By covering the light during testing, we reveal whether squirrels make decisions based on visible information alone, or if they imagine what might be hidden.
This experiment tests whether squirrels form mental representations—imagining that a hidden light might still be on, even when they can't see it.
The experiment uses simple, affordable equipment (Arduino/Raspberry Pi, LED, speaker) making it accessible for researchers, students, and enthusiasts.
If you are a squirrel-loving person and want to contribute to on-going scientific research you can do in a backyard please click here to learn more.
Our research explores a fascinating question: when information is hidden from view, do squirrels behave as if they’re imagining what might be there? This “covered cue” experiment tests whether squirrels possess representational reasoning—the ability to form mental images of hidden objects and events.
Using a simple operant conditioning setup with a lever, light, sound, and food rewards, we teach squirrels that a light + sound combination predicts food. Then, we test what happens when the light is covered: Do they act as if the hidden light might still be on?
This experiment, adapted from similar studies with rats, uses accessible equipment suitable for citizen scientists and field researchers. Our goal is to understand the cognitive capabilities of wild squirrels while making this type of research accessible to a broader community.
We are working with squirrels in the wild to replicate a classical logical fallacy first identified in humans and then later recently replicated with rats in a lab.
We are working with A conjunction effect or Linda problem is a bias or mistake in reasoning where adding extra details (an “and” statement or logical conjunction; mathematical shorthand: ∧ ) to a sentence makes it appear more likely.[1] Logically, this is not possible, because adding more claims can make a true statement false, but cannot make false statements true: If A is true, then A ∧ B might be false (if B is false). However, if A is false, then A ∧ B will always be false, regardless of what B is. Therefore, A ∧ B cannot be more likely than A.
Instead of language as was used in the human studies, simple light and sound cues, a tiny occluder, and careful observation are the ways we work with the squirrels.
This is a work-in-progress field study run by Oomvelt. We are adapting this previous UCLA lab experiment that revealed a human-like logical fallacy in rats to using it with squirrels in natural settings.
The goal: a clear, repeatable way to compare logical reasoning across species using low-cost, open hardware.
Creating food and water sources for animals can have unintended effects and consequences even with the most wonderful of intentions. Please refer to your state and federal naturalist societies for best practices and codes of ethics around creating human-provided food and water sources for wildlife observation and studies.
Watch squirrels interact with our operant conditioning chambers and solve cognitive challenges
in real-time
This video demonstrates a squirrel interacting with our testing apparatus. Watch as the subject learns the association between cues and rewards, demonstrating the cognitive abilities that make the covered cue experiment possible.
Day 1: Each time the squirrel presses the lever, immediately deliver one food pellet. Continue until the squirrel learns to press consistently.
Day 2: Switch to a partial reward schedule — for example, give food only about once every four presses. This keeps the squirrel motivated to press the lever even when food isn’t guaranteed.
Now the squirrel will learn when pressing the lever earns food and when it doesn’t. You’ll use one sound and one light. Each daily session should last around 45–60 minutes, with about 30 trials spaced by 1–2 minutes of rest.
Now you’ll test whether the squirrel behaves differently when the light is covered versus visible but off.
Each squirrel will complete two test sessions on separate days:
• Day 1: The light is visible (uncovered) but stays off the entire session.
• Day 2: The light is completely covered so that the squirrel can’t see it, even if it might be on. During each test session:
• Present 8 sound trials (sound plays for 30 seconds, no rewards given).
• Separate each trial by about 2 minutes.
• Record how often the squirrel presses the lever during and before each sound.
Researcher, writer, and artist
PhD in physics, software engineer and artist. He builds mechatronic systems for art and for science.
Oliver has worked in AI for the past seven years, first at Meta and currently at FuriosaAI.
A UCLA Psychology Professor and Chair of Behavioral Neuroscience.
How to do a study with squirrels in your area using our open-source operant conditioning chamber
you can build yourself!
Our plans and goals for future research
This study explores how squirrels make decisions when some information is hidden — for example, when a light that usually signals food is covered up. We want to find out whether squirrels behave as if they are imagining that the hidden light might still be on, even when they can’t see it. This experiment adapts a similar study originally done with rats, using simpler equipment suitable for citizen scientists.
Watch Andrew McGregor’s TedX talk How Bionic Rats and Humans are Partnering
We presented a talk titled “Porting Open-Source Operant Conditioning Technologies to Natural Settings” and gave a demo of our open-source operant conditioning technologies at the 2025 Biennial Meeting for the International Society for Comparative Cognition in Bogotá
Sign Up now by providing your email and stay updated with our research content.
Interested in research collaboration, hosting equipment, or learning more? Let’s talk.
