Personal Science Week - 250703 Brain Reaction Time
Resurrecting 10-Year-Old Brain Reaction Time Testing with Modern AI
Ten years ago last month, I presented findings at the Quantified Self 15 conference in San Francisco that seemed almost too good to be true: fish oil supplements significantly improved my cognitive performance. The data was solid—99.9% statistical confidence—but the tools required to replicate the study were complex R scripts that few people could use.
This week, I'll revisit that 2015 experiment and show how modern AI tools let us resurrect and democratize research that was previously locked away in academic code.
The Original Fish Oil Experiment
After decades of research in psychology, the late UC Berkeley professor Seth Roberts had developed a daily “Brain Reaction Time” (BRT) test—a 4-minute cognitive vigilance assessment that he claimed could uncover subtle environmental effects that influence our mental performance. His test, developed in the arcane and messy statistical programming language R, relied on precise measurements of a person's ability to stay focused and react quickly to visual stimuli. Using his test over several years, Seth discovered dozens of insights that let him optimize his sleep, mood, productivity, and much more.
Unfortunately, Seth's program required a complicated software installation that few could master, so after his untimely death in 2014, his research languished in the backwaters of disciples like Alex Chernavsky and a few other diehards like me, leading to my presentation at that Quantified Self event.
Seth had freely released his source code, which somehow I managed get running for myself. In a four-month experiment, I tracked my BRT scores alongside dozens of variables: sleep, exercise, supplements, and diet. Of all the factors I measured, fish oil showed the clearest cognitive benefit.
Here's what the data looked like:
What surprised me even more: BRT scores seemed to have no correlation with sleep quality, alcohol or other factors you'd think would directly affect performance. Seth once told me, in complicated neuroscience language that was beyond me at the time, that this was expected. The test specifically cancelled out factors except those from the direct environment.
The Problem: Technical Barriers
Seth Roberts passed away unexpectedly in 2014, leaving behind his sophisticated R analysis scripts that few could use. His BRT methodology required:
Complex statistical analysis (t-tests, correlation matrices)
Data visualization in a powerful, but difficult to use software package
Complicated installation on a laptop or desktop (no mobile)
Programming expertise to modify or extend
Seth’s scientific approach was sound, but his tools were inaccessible. Over the years, several of us had considered rebuilding Seth’s creation in an easier-to-use form, but one by one each of us moved on to other projects.
Modern AI: Resurrection in Hours
Fast-forward to last weekend. I had been playing with the Neurable brain computer interface (as mentioned in PSWeek250625), when Alex Chernavsky suggested I try comparing it Seth’s old BRT app.
Meanwhile, a new AI-powered platform, Bolt was running a free promotion to demonstrate the seamless ability of their tool to create full-fledged professional software projects—no programming experience required.
I was on the way out the door when I read Alex’ suggestion, but I thought maybe Bolt would be worth a try. My initial prompt was, literally, “Look at this Github repo and turn it into an interactive web app”. It worked on the first shot! I can’t describe how easy this was!
The result, now live at brt.personalscience.com is a complete cognitive tracking platform that:
Implements the original BRT vigilance test, using Seth’s precise algorithms.
Creates interactive visualizations
Works on both desktop and mobile
Requires zero programming knowledge
Stores data and lets you track progress over time
It’s exactly Seth’s original algorithm with a modern web app interface. To make it even better, in less than an hour, using simple English language prompts, I added a login feature and a database to track user history, plus miscellaneous user interface tweaks. Believe it or not, the hardest part was getting it posted to the personalscience.com URL. (ChatGPT o3 managed to crack that one, thankfully).
Try it yourself! Hold your mouse (or finger, if on mobile) just above the screen and hit the red circle as soon as it appears. Do a few warmup exercises first, then repeat 20 times for your final score. Takes less than five minutes.
Create an account (free) and it will track all your tests, letting you generate charts like this.
Remember, lower is better, so that slight rise in the middle represents a situation that was less than ideal for my brain health (in this case, it was after a long afternoon of no eating).
Personal Science Implications
Seth Roberts pioneered the idea that individuals could conduct rigorous self-experiments outside traditional academic structures. His work influenced the entire Quantified Self movement, but remained technically inaccessible.
Tools like Bolt.new don't just make development faster—they democratize scientific methodology itself. Personal scientists can now:
Revive abandoned research projects
Rapidly prototype experimental designs
Share replicable tools, not just results
Lower barriers for n-of-1 studies
What This Means for You
If you're interested in cognitive enhancement, try the BRT test yourself with your own experiments:
Supplements (fish oil, vitamin D, nootropics)
Diet changes
Exercise timing
Sleep interventions
Stress management techniques
But more importantly, consider what research you've always wanted to do but couldn't because of technical barriers. Those barriers are dissolving rapidly.
About Personal Science
We're entering an era where the tools of science are becoming accessible to everyone. The same rigorous methodology that drives professional research—hypothesis formation, controlled experimentation, statistical analysis—can now be deployed by individuals for personal optimization.
Seth Roberts understood this vision decades ahead of his time. Now we have the tools to make it reality.
What experiment will you resurrect or create?
If you have other topics you'd like to discuss, or if you try the BRT test yourself, please let us know.
Thanks for another brilliant post, Richard—fantastic on so many levels. Great memories of Seth.
Nice work, Richard! (And thanks for the shout-out). I created an account and ran through the test once. Just curious about a difference between your test vs. Seth's test that he implemented in R code. Seth's script presented the user with a random digit 2 through 8, and the user had to hit the corresponding key on the keyboard. Your test just has a single type of stimulus (the red circle). Was Seth's protocol too complex for Bolt to implement? Or did you have other reasons to simplify the app?