top of page
  • Writer's pictureThe Rivers School

Joshua Do ’24: The Tye Lab at the Salk Institute at UCSD

A student at a desk with a laptop and monitor screen.
Josh's work-from-home setup

Ever since I entered high school, I began noticing social dynamics and hierarchical patterns within my school, classrooms, and friend groups. Observing dominant and subordinate roles in my own social circles, I wondered whether there was a natural source for these types of patterns and behaviors. This summer, I have worked towards potential neuroscientific solutions to my observations by working with Dr. Cazares at the prestigious Tye Lab of the Salk Institute with the University of California San Diego (UCSD).

Salk by the Numbers

Throughout the animal kingdom, social species naturally establish social hierarchies and use placement within these hierarchies to guide the appropriate expression of dominant behaviors that are critical for survival and social stability. Social recognition utilizes the orbitofrontal cortex (OFC), a region best known for encoding inferential information relevant to ongoing task decisions, including social ranking information. With an already conducted experiment with mice, the lab's approach was to consider the ethologically relevant phenomena in which 1) dominant mice have priority access to resources in social contexts and 2) mice display predatory behavior in their pursuit and capture of live crickets for consumption.

The first two weeks of the summer were dedicated to setting up a personal account linked to a VPN that could directly access the confidential data stored in the Tye Lab in San Diego. With three 2.5-hour Zoom meetings, I finally completed the detailed security training and could start my work.

Screen to access lab data through Salk VPN

My main role was to meticulously sort through video experiments of two mice in pursuit of a cricket and use a program called “SLEAP” (an open-source deep-learning framework) to ultimately be able to predict which mouse will get the cricket and prove ethological social competition. Each folder of experiments contained 15 videos with different socially ranked mice. In my dataset, the mouse with two tail marks was the most dominant, the mouse with one tail mark was the second, and the one with three tail marks was the least dominant. With this in mind, I created a skeleton that pinpointed various parts of the mouse. I was directed to do 10 frames per video with both mice labeled and showing their social rank by color: the orange skeleton displayed the subordinate mouse and the blue displayed the dominant.

When labeling, it is important to be as accurate as possible. Since the labeled data will be entered into SLEAP to train the program to use the skeleton in order to determine mice behavior, every point must be at the tip of each body part and the labeling must be as random as possible to get the best data. As each video contained thousands of frames, labeling took several hours to complete.

Example of labeling a single frame on a video

Example of a labeled mouse

After finalizing 150 labeled videos, I was immediately tasked to begin labeling for the cricket. However, this time, Dr. Cazares had me search for, organize, and then create the skeleton edge for the cricket myself. As the cricket was a smaller entity with fewer skeleton points, I was able to complete the labeling process rather quickly. I ended with a total of over 300 labeled videos for the summer that will be used to train the SLEAP program.

Example of a labeled cricket

A successful outcome of these experiments will 1) define the relationship between competitive predatory capability and social rank and 2) characterize the ethological and behaviorally-relevant role for OFC encoding of social ranking information. I am hopeful that my contribution will help push for this outcome.

Intern hard at work

Throughout this summer internship, I learned patience, which is a simple yet difficult skill in a work setting. I had to acknowledge that my work will not show immediate solutions, nor will my work answer the questions to my observations on social groups in society; however, I took away the gift of being a part of something that has yet to be done. I learned how to manage time on my own to finish a large task with consistent, everyday progress. I learned communication skills as time differences, online meetings, and relying on email were inevitable challenges during my time. Lastly, I learned the ability to prove ethological social behaviors through lab work.

As I finish off my final tasks as an intern, I am excited about what I will do with my newfound skills and learnings. As this neuroscience experience was a space for me to funnel my scattered thoughts on human behavior into an academic field, I appreciate the deeper insight I have gained from being a part of such a monumental movement. As my work and data will be used in a discovery that has never been done before, I am honored that my interest was utilized for findings that I will definitely see in the future.

Beautiful view of the Salk Institute in San Diego


bottom of page