Chris Thornton

University of Sussex, University of Calgary

Papers

Total Citations

325

H-Index

About

Chris Thornton’s research sits at the intersection of machine learning, cognitive science, and evolutionary computation, with a particular focus on the fundamental limits of learning and representation. His most influential work, *Trading Spaces: Computation, Representation, and the Limits of Uninformed Learning* (1997, 297 citations), is a landmark contribution that explores how certain regularities in data are only statistically visible after systematic recoding—a process that confronts the infinite space of Turing machines. This work challenges assumptions about uninformed learning and has shaped debates on representation in AI. Thornton also made notable contributions to the critique of genetic algorithms, coining the “building block fallacy” (1997, 13 citations), which questions the theoretical foundations of schema analysis in evolutionary search. His later work on *In Situ Motion Capture of Speed Skating* (2012, 13 citations) demonstrates a practical turn, using Kinect depth imaging to escape the constraints of treadmill-based analysis. Though his citation counts vary, Thornton’s theoretical contributions—especially on the limits of learning and representation—remain essential reading for students and researchers probing the foundational challenges of artificial intelligence and cognitive science.