What do 20th-century Indian paintings, Formula 1 virtual races, protein simulations, standardized test anxieties, and AI biases all have in common?
As much as they may sound like a confusing Jeopardy episode, they’re Princeton junior papers and independent work topics. Speaking to members of the Class of 2026 across disciplines, these papers often follow the same winding journey from curiosity to completion.
For juniors, many of whom are embarking on their first research projects in their chosen fields, the process starts with a daunting task: Ask a question no one has answered. Juniors described long processes of trying to come up with this central question.
Robert Mohan ’26 from the Department of Art and Archeology discovered his question over months spent poring over exhibition catalogs and auction essays. His focus landed on Manjit Bawa, an understudied South Asian modernist painter whose work engages religious icons and at times critiques the dogmatic use of this imagery.
“I found something really compelling about how he’s abstracting the human form,” Mohan explained. “The topic that I settled on was the role of religious icons in his artistic production and what role a religious icon might play within some of his secular critiques.”
Mohan’s hybrid approach, combining biography and the theory of art, challenges the Eurocentric idea that modern art comes after the fall of religion in society. Through the Department of Art and Archeology, he received funding to visit Bawa’s work in person at the Peabody Essex Museum in Massachusetts.
The South Asian Modern and Contemporary Art collection at the Peabody Essex Museum.
Courtesy of Robert Mohan ’26
Over in the Department of Economics, an inquiry by Andy Zhang ’26 took shape closer to home, through observing his younger sister’s experience with the new digital SAT format, which uses computer-adaptive testing (CAT) to adjust question difficulty based on performance.
“I wondered how people psychologically react to harder or easier questions mid-test — does anxiety help or hurt performance?” Zhang said.
The Department of Economics gives juniors an entire semester to write their “prospectus,” fleshing out their central question into a literature review and research design.
Most juniors in the department work with existing datasets, and Zhang said the department highly discouraged experiments for junior papers due to the difficult nature of obtaining data. However, Zhang felt “incredibly stuck” trying to find a topic that aligned with conventional junior papers in the department.
Instead, he proposed a rough idea for an experiment involving the digitized SAT, which economics professor Swati Bhatt found interesting. According to Bhatt, the topic was an understudied subject and relevant to recent changes in standardized testing, and hence he encouraged Zhang to pursue it.
Zhang designed an experiment inspired by the new computer-adaptive SAT, priming a subgroup of the participants to believe they were taking an adaptive test when it was actually a standard fixed test. He then measured their stress levels and performance scores.
The flowchart of Zhang’s survey, testing to see if CAT priming influences performance and stress.
Courtesy of Andy Zhang ’26
Meanwhile, Tamara Tymczyszyn ’26 took root in a philosophical psychology framework about AI morality for her junior paper in psychology. She created an experiment based on the Investor Trust Game using two AI players, where one player invests in the other based on the other player’s name and the other player decides whether to return the investment or not. The names insinuate demographic information, such as gender and race, and Tymczyszyn’s system aimed to uncover whether the AI players reciprocate trust unequally based on these factors.
The code snippet used to create the Investment Trust Game prompt.
Courtesy of Tamara Tymczyszyn ’26
“We know AI mirrors human prejudices, but how?” she asked. “I built trust games where AI ‘players’ receive demographic cues — like names signaling race or income — to see if algorithms reciprocate trust unequally.”
After each junior has their question, only half the battle is over. Executing them is the other half.
For Arav Raval ’26 from the Department of Electrical and Computer Engineering (ECE), building an autonomous racing simulator meant navigating decision-making in unprecedented situations. Using GPUDrive — a new, open-source simulator designed for machine learning in autonomous driving — he reverse-engineered Formula 1 tracks from low-resolution decal images to generate racecourse simulations.
“The codebase for GPUDrive had limited documentation — we basically taught ourselves to ‘see’ racetracks through edge detection and contour mapping,” he explained.
His fall independent work final paper details how AI drivers perform risk-reward calculations in competitive environments. “Unlike street driving, racing means every opponent is actively trying to outmaneuver you,” he explained.
While Raval’s simulator races on, chemistry major Kelih Henyo ’26 has been working to predict how small molecules might stabilize faulty proteins — a potential therapeutic avenue for treating conditions like Alzheimer’s. His project, developed on the Mpipi model, explores the molecular dynamics of the proteins commonly involved in neurodegenerative diseases.
“I’m not touching a single test tube,” Henyo said. “On computers, we’re using a thing called molecular dynamics, which allows you to model different proteins, molecules, [and] assign them masses, charges, so they behave exactly like a [real-life] protein with the same hydrophobicity.”
Henyo’s molecular dynamics required relentless troubleshooting, given the many moving parts in the simulation.
“There’s many different places where this could go wrong,” Henyo said. “Maybe the Python code I’m using to analyze the data, or maybe the actual way I set up the simulation, could be wrong.”
He said that his advisor Jerelle Joseph, assistant professor of Chemical and Biological Engineering, was instrumental in helping him troubleshoot molecular dynamics simulations with an intuitive grasp of systems behavior.
“Troubleshooting simulations is really hard, but she’s been really helpful,” he said. “I think I lose my mind a lot when I make mistakes. But [Joseph is] very kind, very sweet, very caring.”
Henyo said he learned to approach developing problems methodically to get the simulations to work, such as figuring out storage problems after consulting Princeton Research Computing.
Behind many intellectual triumphs also lies a less glamorous reality: navigating the bureaucratic hurdles of academia.
Zhang’s experiment required approval from Princeton’s Institutional Review Board (IRB), which he describes as “a lot of academic and bureaucratic work.” The IRB is a committee that reviews research involving human subjects to ensure that the research is ethical and complies with regulations.
For the IRB approval process, Professor Bhatt who is also acting as Zhang’s principal investigator, had to approve every change, even for minor changes such as page formatting. According to Zhang, the IRB took a few days to a week to approve each individual change, totaling to a process of over two months over winter break.
Many compelling findings often emerge when research challenges initial assumptions — or reveals entirely unexpected patterns.
Zhang discovered that participants who believed that they were taking an adaptive test felt significantly more confident on average. However, among overall participants that felt good about their performance, the primed and confident group surprisingly performed much worse compared to their peers who took the exam without special instructions — a phenomenon he attributes to overconfidence or misaligned expectations about test difficulty.
“Turns out, thinking you’re crushing the adaptive test makes you gloss over subtle hints in harder questions,” Zhang said.
For many juniors, the projects they have worked on for their junior papers have become training grounds for their senior thesis.
Raval plans to expand his simulations into real-world autonomous kart tests through Princeton Autonomous Vehicle Engineering (PAVE) and continue with research on racing simulations for his senior independent work or thesis.
PAVE’s exhibition with the Council of Science & Technology (CST) at the 35th Anniversary on March 26th, which also works with autonomous vehicles.
Courtesy of Arav Raval ’26
Even though Raval was not required to complete independent work as a junior in the Electrical and Computer Engineering department, he said he would recommend it to all students as a practice round before more vigorous senior thesis research.
“Taking on independent work as a junior taught me to fail fast and pivot faster,” Raval says. This is a takeaway for many juniors: The junior paper is less about answers than learning to navigate the intricate, messy research process.
And learn to navigate they must, as around this time next year, these students will all be turning in senior theses.
The Department of Art and Archeology requires its students to work on two different topics for the junior paper and senior thesis, so Mohan said he will pivot from Bawa’s modernist critiques to 19th-century art, potentially from a different region, for his senior thesis. Zhang and Tymczyszyn will also pivot to new projects for their theses with the economics and psychology departments.
Henyo hopes to create his own tool for other chemists and biologists to use for molecular dynamics simulations, like his project’s Mpipi model, to create new experiments.
As the final deadline for junior papers approaches on April 29, these juniors will be wrapping up their work and readying themselves for submission.
Chloe Lau ’27 is an assistant Features editor and a staff Prospect writer for the ‘Prince.’
Please send any corrections to corrections[at]dailyprincetonian.com.
Read More
Fifteen Princeton faculty, arts fellows, and alumni awarded Guggenheim Fellowship
Victoria DaviesFifteen Princeton affiliates have received the Guggenheim Fellowship. This year marks the 100th anniversary of the award. Among those celebrated are photographers, composers, and historians.
Fifteen Princeton affiliates have received the Guggenheim Fellowship. This year marks the 100th anniversary of the award. Among those celebrated are photographers, composers, and historians.
Judge grants temporary restraining order in lawsuit by U. and others against the DOE
Christopher BaoThe United States District Court of Massachusetts granted Princeton and other plaintiffs a temporary restraining order on Wednesday stopping the cap on indirect cost rates by the Department of Energy.
The United States District Court of Massachusetts granted Princeton and other plaintiffs a temporary restraining order on Wednesday stopping the cap on indirect cost rates by the Department of Energy.
Play the Friday Puzzle, ‘Cut to the Bone’
Lily HutchesonBe careful and try not to put on a show while solving this crossword by staff constructor Lily Hutcheson!
Be careful and try not to put on a show while solving this crossword by staff constructor Lily Hutcheson!