The Interplay Between Random Field Theory and Permutation Inference Methods

• Activity Number: 323
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Thomas Nichols*+
• Companies: University of Warwick
• Keywords: Random Field Theory ; Permutation inference ; multiple testing ; brain imaging ; fMRI

Abstract

As a student, I swore at Random Field Theory (RFT) papers as I found them completely impenetrable. As a junior faculty member, I critiqued RFT methods, using simulations and permutation to show how conservative RFT can be in low-n settings (an issue later addressed with Keith Worsley's discrete local maxima work). But before long, though, I had written a paper using Random Field methods (with Keith, on nonstationary cluster size inference) and soon wrote another (without Keith, on cluster mass inference). In this talk I will use the cluster mass statistic as a case study to illustrate the interplay between RFT and resampling methods development. In particular, I'll show how permutation tests can be used to evaluate the accuracy of RFT methods on real data, as well as motivate novel RFT research.