Convergent Representations of Computer Programs in Human and Artificial Neural Networks

What aspects of computer programs are represented by the human brain during comprehension? We investigate this question by analyzing brain recordings derived from functional magnetic resonance imaging (fMRI) studies of programmers comprehending Python code. We first evaluate a selection of static and dynamic code properties, such as abstract syntax tree (AST)-related and runtime-related metrics and study how they relate to neural brain signals. Then, to learn whether brain representations encode fine-grained information about computer programs, we train a probe to align brain recordings with representations learned by a suite of ML models trained on code. We find that both the Multiple Demand and Language systems–brain systems which are responsible for very different cognitive tasks, encode specific code properties and uniquely align with machine learned representations of code. These findings suggest at least two distinct neural mechanisms mediating computer program comprehension and evaluation, prompting the design of code model objectives that go beyond static language modeling. We make all the corresponding code, data, and analysis publicly available at https://github.com/ALFA-group/code-representations-ml-brain