Anna Ivanova at MIT: code-reading recruits the multiple-demand network, not language

In 2014 researchers at University of Passout scanned professional programmers reading code and initially concluded code behaved like language because language-related areas (notably Braka's region) lit up. Later, Anna Ivanova at MIT ran a more controlled fMRI study directly comparing the language network (activated by sentences) with the multiple-demand network (used for math, logic puzzles, planning). Ivanova's team found that experienced programmers rely primarily on the multiple-demand network when reading code, implying code comprehension is 'like solving a puzzle' rather than parsing a sentence.

Two linked mechanisms explain how programming changes cognition. First, pattern compression: beginners process each token anew, while experts chunk code into recognisable units (a loop, a missing base case, an off-by-one error) — the same chunking described in decades-earlier studies of chess grandmasters. Second, transfer and practice: coding repeatedly trains decomposition — breaking complex problems into ordered, solvable steps — which studies show transfers narrowly to computational thinking and certain reasoning skills, not to broad rises in general intelligence.

There is also measurable neuroplasticity: experienced programmers differ in regions tied to working memory, pattern recognition, and sustained attention, consistent with repeated mental practice. Equally important is an emotional habit learned through debugging: after '40 minutes tracking down a bug' that turns out to be 'one misplaced character' or 'one wrong comparison', developers learn to separate frustration from the next corrective step. This accidental emotional discipline and the slow accrual of chunked patterns over 'thousands of hours' or 'a decade' are the core ways coding reshapes thinking — quietly, one function at a time — so you begin to apply decomposition and focused problem-solving across non-programming problems like planning projects, untangling arguments, or fixing recipes.