Pittsburgh, PA · Class of 2027

Larry Zhong

Class of 2027 at Shady Side Academy, boarding from Beijing. I build things at the intersection of computational neuroscience, brain-computer interfaces, and history. I can't leave something broken or missing without building a fix.

Peilin (Larry) Zhong
I.

About

I work at the intersection of neuroscience, computing, and history. Three of my projects turn out to ask one question: when does a measurement confidently report something that isn't there? A hippocampal CA1 model where the spectral statistic found a rhythm that was an artifact of its own input. A brain-computer interface where the hard part was telling a decoder's failure apart from the signal's. A surrogate model that certifies firing rates a real neuron cannot produce, reporting zero error while the cell stays silent. The pattern showed up before any of them: the first open-source BCI I trusted owed its accuracy to data leakage, and the single-electrode headset I had bought could not physically reach motor cortex. I moved to public datasets and rebuilt the analysis on real EEG.

The question is not only a technical one. I spent a year on a paper arguing that McCarthyism damaged American universities less through congressional pressure than through administrative capitulation, institutions failing the value they claimed to exist for. In an essay on psychiatric diagnosis I argued that the threshold for clinically significant is a judgment, not a measurement, and that where it falls decides whose hard life becomes a condition. I build historical arguments in Civilization VI mods too, where a claim about what a leader optimized for has to be written as numbers and can then be proved wrong by play.

I also build things that simply have to work: a dining-hall ordering system for my school, and a rating engine that turned my dorm's foosball table into a social hub. What drives me is curiosity about the brain: how it computes, remembers, and might one day interface with machines, and I care about using what I build to help people, from a scholarship-funded housing project for rural families to reading with kids.

II.

Currently

Three computational-neuroscience preprints out this year: one recovering emergent gamma in a reduced CA1 microcircuit, one diagnosing why motor-imagery BCIs fail, one showing that smooth surrogate models certify firing patterns a real neuron cannot produce.

Building E-Dining, a full-stack meal-ordering platform piloted with students and dining staff at Shady Side Academy.

Founder and co-president of the Computing Club, president of the Neuroscience Club at Shady Side Academy.

III.

Research
Interests

Computational neuroscience: biologically constrained microcircuit models, model simplification, and neural oscillations (theta / gamma).

Brain-computer interfaces: decoding motor imagery from EEG, and understanding why current pipelines fail.

Memory and the hippocampus: the CA1 circuit, and how structure gives rise to function.

I'm also drawn to the history and philosophy of science: how ideas, institutions, and technologies rise, fall, and get rebuilt.

IV.

Writing & Research

Computational Neuroscience · preprint

Screens Reject, Only Mechanism Confirms: Recovering Emergent Gamma in a Reduced Hippocampal CA1 Microcircuit

Independent research. It shows the standard spectral test is blind to genuine gamma on a theta-driven CA1 network, builds a comb-notch statistic to recover it, and confirms emergent gamma with a pre-registered PING test, reversing the study's own earlier conclusion. Python, with full code. Zenodo preprint.

Brain-Computer Interfaces · preprint

Diagnosing Motor-Imagery BCI Failure: A Three-Step Diagnostic Framework

Independent research on why motor-imagery BCIs fail. A three-step diagnostic: classDis (Lotte & Jeunet, 2018) measures class separability before any decoder is trained, a transfer gate separates subjects who decode within a session but not across one, and filter-bank CSP, a Riemannian classifier and permutation tests confirm what remains. Pooled across three cohorts (75 subjects), classDis tracks cross-session decoding at r = .66. Python, with full code. Zenodo preprint.

Computational Neuroscience · preprint

Smooth Surrogate Models Approve Impossible Firing Patterns: A Hidden Flaw in Writing Brain Codes

Independent research on the write side of a neural interface. Optimising stimulation against a smooth, differentiable surrogate of a neuron's rate-current curve certifies target firing rates the real cell cannot produce: the surrogate reports zero error while the neuron stays silent. Whether that happens is set by the cell's onset bifurcation class, and a feasibility screen against the true curve removes it. Python, with full code. Zenodo preprint.

U.S. History · research paper

McCarthyism's Effects on Academic Freedom and Higher Education

A year-long research paper arguing that the Second Red Scare damaged the American university less through congressional pressure than through administrative capitulation: loyalty oaths and HUAC hearings did the visible work, but it was universities policing themselves that made the damage stick. Institutions failed the value they claimed to exist for.

History · John Locke 2025 (Shortlisted)

Beyond Shame and Pride: A Balanced Approach to National History

National shame and national pride are usually treated as opposites. I argue they fail the same way: both replace historical understanding with an emotional verdict, and shame in particular paralyses the civic engagement it is meant to produce. What works instead looks like a truth commission, aimed at accuracy and reconciliation rather than punishment or celebration. Shortlisted, John Locke Global Essay Competition.

History · John Locke 2026 (Shortlisted)

The Most Famous Stranger in the World: Does Che Deserve His Iconic T-Shirt?

The prompt asks whether Che "deserves" the T-shirt. I argue the question smuggles in a premise: that the shirt is a verdict on the man. It stopped pointing at him decades ago. The essay weighs the executions honestly, citing a contested range rather than one convenient number, and then turns the image around: what the shirt reveals is the wearer, and a culture that no longer asks what the face on its chest is supposed to mean.

Psychology · John Locke 2026 (Shortlisted)

The Thermometer That Changes the Weather: Is Mental Illness Over-Diagnosed or Better Recognized?

Over-diagnosis or better recognition? The evidence supports both, and I defend that mixture. But settling the proportions would not settle the dispute, because both answers share a hidden premise: that a diagnosis only describes. It also acts. The threshold for "clinically significant" is a judgment, not a measurement, and where it falls decides whose hard life becomes a condition. Shortlisted, John Locke Global Essay Competition.

V.

Selected Work

Software · 2024–present

Foosball Elo Rating & Ranking System

I couldn't watch casual games stay scoreless, so I built a living rank system: offense/defense ratings, Fibonacci rank protection, a hidden rank, and a live web scoreboard. It ranked 32 players across 1,230 recorded player-games.

Game Design · 2026

The Cruelest Years: WWII Strategy Board Game

Designed and led the build of a complex WWII wargame from scratch: 3D-printed units, a full rulebook and DLC, six playtests, and 8+ iterations to keep it deep but approachable.

Game Modding · 2025–2026

Civilization VI Mods

Taught myself Lua, SQL, and XML to build original civilizations: part algorithm, part history research, part storytelling. Published to the Steam Workshop.

Web · 2025–present

E-Dining: Dining-Hall Ordering System

A system for pre-ordering packed lunches with nutrition and allergen info and kitchen feedback, built so the kitchen gets a clearer view of demand before service. Piloted with students and dining staff.

Web · 2025–present

Smart Schedule

SSA's rotating Day 1–8 schedule is confusing, so I built a web app that folds it into one clean view.

EEG · Cz · sensorimotor mu ~10 Hz
VI.

Awards & Activities

  • 2025, 2026John Locke Global Essay Competition, Shortlisted (History twice, Psychology)History & Psychology
  • 2025Brain Bee, Regional Runner-Up (2nd)Neuroscience
  • 2024Caribou Senior Coding Contest, 1st PlaceComputing
  • 2025Shady Hacks, Computing Award WinnerHackathon
  • 2025Stanford Math Tournament, Calculus 5th & Algebra HMMath
  • 2024Berkeley Math Tournament, Distinguished Honorable MentionMath
  • 2024–26Math League, National Qualifier (twice)Math
  • 2025, 2026AP Scholar · College Board National RecognitionCollege Board
  • 2024–presentFounder & Co-President · Computing Club · President · Neuroscience ClubShady Side Academy
  • 2025–presentStanford Neuroscience Journal Club (Stanford Neurosurgery)Member · remote
  • 2023–24The Duke of Edinburgh's International AwardBronze
VII.

Contact