Addressing the theory crisis in statistical learning research
Research into statistical learning, the ability to learn structured patterns in the environment, faces a theory crisis. Specifically, three challenges must be addressed: a lack of robust phenomena to constrain theories, issues with construct validity, and challenges with establishing causality. Here, we describe and discuss each issue in relation to several prominent statistical learning phenomena. We then offer recommendations to help address the theory crisis and move the field forward.
Authors
- Alice E. Milne (ORCID: https://orcid.org/0000-0002-2000-2842)
- Benjamin Wilson (ORCID: https://orcid.org/0000-0002-0471-8410)
- Christopher M. Conway (ORCID: https://orcid.org/0000-0003-2506-1304)
- Sonia Singh (ORCID: https://orcid.org/0009-0002-7679-5113)
- Holly Jenkins (ORCID: https://orcid.org/0000-0002-1055-642X)
Institutions
- Emory University (US)
- Grinnell College (US)
- University of Oxford (GB)
- Emory National Primate Research Center
- University College London (GB)
- Lancaster University (GB)
Publication Details
- Journal
- npj Science of Learning
- Published
- 2025-09-29
- DOI
- https://doi.org/10.1038/s41539-025-00359-6
- Primary Topic
- Gaussian Processes and Bayesian Inference
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Wellcome Trust
- Grinnell College
- National Institutes of Health