← davidlitman.com

AI / Neuro

Salk Institute — computational neuroscience

Current · Terry Sejnowski lab

Courtyard and reflecting pool at the Salk Institute, La Jolla
Salk Institute, La Jolla

I work in the computational neuroscience lab of Francis Crick Chair Professor Terry Sejnowski at the Salk Institute.

I am currently extremely interested in mechanistic interpretability research. Right now I am investigating the internal ontologies of transformer-based chess engines. I developed an open source toolkit and visualizer library for mechanistic interpretability of these models: https://github.com/chessformer-lens/chessformer_lens

Screenshots of the interactive chess engine visualizer
Screenshots of the interactive chess engine visualizer
Residual stream view
Residual stream view
Logit lens microscope view
Logit lens microscope view
Attention heads and GAB templates view
Attention heads and GAB templates view

Here is a more accessible blog post of some older results: Fork Around and Find Out (Part 2): One Head Does the Summing.

ABCD dataset work

Big Data & Machine Learning

I spent a year leveraging unsupervised learning algorithms on massive corpuses of developmental data to investigate the behavioral, neural, and genetic (GWAS) correlates of a patient subset identified by clustering on ABCD study datasets. Publication upcoming as soon as final data set is released for research from NIH.

UCSB LEAP lab — emotional temporal coding

fMRI · MVPA & RSA

Contributed to a graduate student’s dissertation on emotional temporal coding: multivoxel pattern analysis (MVPA) and representational similarity analysis (RSA), with region-of-interest masks in frontal pole and related areas (FSLeyes).

Recent books

Relevant to neuroscience

Annotated reading list

Favorite papers with notes; a rough sketch of deepest interests: