Feature selection leads to divergent neurobiological interpretations of brain-based machine learning biomarkers
Brendan D. Adkinson, Matthew Rosenblatt, Huili Sun, Javid Dadashkarimi, Link Tejavibulya, Corey Horien, Margaret L. Westwater, Raimundo X. Rodriguez, Stephanie Noble, Dustin Scheinost · 2026
Read Feature selection leads to divergent neurobiological interpretations of brain-based machine learning biomarkers by Brendan D. Adkinson, Matthew Rosenblatt, Huili Sun, Javid Dadashkarimi, Link Tejavibulya, Corey Horien, Margaret L. Westwater, Raimundo X. Rodriguez, Stephanie Noble, Dustin Scheinost (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13388108.txt.
Description
Brendan Adkinson and nine co-authors published a 2026 neuroimaging study warning that a routine modeling step can quietly rewrite a study's biological story. Brain based machine learning models predict behavior and symptoms from connectivity data. Because the data are huge, researchers first select a subset of features, then describe the surviving network as the biology behind the trait.
Across four large datasets covering more than twelve thousand people and thirteen outcomes, the team tested what gets thrown away. Connections discarded by selection could still predict the outcomes well on their own. In other words, the kept network is not the unique signature it is often presented as. Feature selection, they argue, trades modeling convenience against faithfulness to the brain, and interpretation should own that trade.
Catalog record
| CREATOR | Brendan D. Adkinson, Matthew Rosenblatt, Huili Sun, Javid Dadashkarimi, Link Tejavibulya, Corey Horien, Margaret L. Westwater, Raimundo X. Rodriguez, Stephanie Noble, Dustin Scheinost |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |