Your Brain on Code: What Brains and AI Models Both Learn
Shashank Srikant
NeurIPS · 2022Paper discussed
Using brain recordings to read what AI models of code have learned, and what people represent when they read programs.
Topic
Brain research, from fMRI studies of programmers reading code to the neuroscience behind mental illness and its treatment.
2 episodes
Neuroscience here ranges from imaging studies of everyday thinking to the research behind mental illness and its treatment. Two episodes of the podcast fall under it. In Episode 001, Thomas R. Insel, a psychiatrist and neuroscientist who directed the National Institute of Mental Health from 2002 to 2015, discusses how technology, data, and community-based care could change mental health care.
In Episode 005, Shashank Srikant turns to imaging. The episode reviews fMRI work on programmers reading code, in which two brain systems, the Multiple Demand system and the language system, each encode specific properties of code. His NeurIPS 2022 paper, written with Ben Lipkin, Anna Ivanova, Evelina Fedorenko, and Una-May O'Reilly, then maps those brain recordings onto the representations learned by machine learning models of code and reports at least two distinct neural mechanisms behind program comprehension. The episode shows how programs can be modified so that a model predicts stronger responses in particular brain regions. The first concerns care and the second concerns measuring brain activity. The episodes, with the book and paper behind them, are listed below.
Shashank Srikant
NeurIPS · 2022Paper discussed
Using brain recordings to read what AI models of code have learned, and what people represent when they read programs.
Dr. Thomas R. Insel
Penguin Press · 2022
Why mental health care needs signals people can understand, and where data and digital biomarkers fit.