Course Overview#
The goal ⚽️ of neuroscience is to understand how the nervous system controls behavior, not only in the simplified environments of the lab 🔬, but also in the natural 🌳🌷 environments for which nervous systems evolved.
In pursuing this goal ⚽️, neuroscience research is supported by an ever-larger toolbox 🛠, ranging from optogenetics to connectomics. However, often these tools are coupled with reductionist approaches for linking nervous systems and behavior. This course will introduce advanced techniques for measuring 📐 and analysing 🔍 behavior, as well as three fundamental principles as necessary to understanding biological behavior:
1️⃣ morphology and environment;
2️⃣ action-perception closed loops and purpose; and
3️⃣ individuality and historical contingencies [GMG19].
What will you learn 👀?#
This course will emphasize✨ the philosophical and observational skills required to understand behavior, while also providing training in motion capture technologies and computer vision methods that can assist in the collection and analysis of video recorded 📹 behavior datasets.
Focusing 🔍 on the tool DeepLabCut, students will analyse either their own original video dataset or datasets of general interest and have the opportunity to practice tracking, pose estimation, action segmentation, kinematic analysis and modeling of behavior.
By the end 🔚 of the course, you will:
1️⃣ be familiar with modern and historical frameworks for studying the behavior of living biological systems
2️⃣ practice methods for carefully and precisely observing and defining behaviors
3️⃣ understand the limits and capabilities of computer vision
4️⃣ develop an intuition for how to build experimental setups that can take advantage of tools such as DeepLabCut
This course shares and promotes open-source software✨! We encourage students to try 🆕 ideas, share insights, and connect with the open-source community.
Structure of the course 🚧#
Each day, from Monday to Thursday, will take off 🚀 with lectures presenting innovative, related work. As the course is spread around the world 🌎, you can watch those pre-recorded lectures at your own time. We will release them day by day 🗓!
We will also do 4h of 🔥 practical work daily. The ✨TAs will help you better understand the course material and apply it practically. We have an excellent teaching assistant-to-student ratio of around 5:1! You will interact with your teaching assistants locally in different satellites 🛰 (Buenos Aires, Nairobi, London, Warsaw, Munich, Colorado Springs, and Okinawa) or online.
You can choose between using your data to create a DeepLabCut model and then analyze kinematics, and cluster behaviors or perform actions segmentation and working with one of our example datasets.
Please do not hesitate to reach out to us if you have any questions (⁇)
Content#
Day 1 – What is animal behavior 🤔 ?#
Historical and current theoretical frameworks for the study of behavior in living biological systems.
Practical exercises for training 🏋️♀️ skills in observing and defining behaviors.
Day 2 – Pose Tracking I#
Fundamentals of machine learning, computer vision, and deep learning
Introduction to DeepLabCut
Creating a tailored DeepLabCut model for the data you are working with.
Day 3 – Pose Tracking II#
Evaluating, utilizing and optimizing your DeepLabCut model from day 2
Multi-animal tracking
Live tracking
Day 4 – Analysis by eye 👁 and by computer 💻#
Movement kinematics in living biological systems
Action segmentation and clustering – when does a behavior start and end?
Analyse your original video dataset of behavior
Day 5 – Working on your data and discussion#
Discussion of advanced behavioral analysis topics and potential pitfalls
Keep analyzing your data and share your work (student presentations 📽)
References
Alex Gomez-Marin and Asif A. Ghazanfar. The life of behavior. Neuron, 104(1):25–36, October 2019. URL: https://doi.org/10.1016/j.neuron.2019.09.017, doi:10.1016/j.neuron.2019.09.017.