Correlator beware: Correlation has limited selectivity for fMRI data analysis

Groups of time-courses created from fMRI data by the frequently used correl ation analysis are often highly heterogeneous. This heterogeneity is due to the limited selectivity of correlation when trying to match brain time-cou rses to an externally imposed activation paradigm. Thus, this process unnec essarily generates many type I errors (false positives). Furthermore, as a consequence of the heterogeneity, time-courses identified and grouped by co rrelation may in fact describe different activations. After demonstrating t his inadequacy, we give one particular approach to partition such a heterog eneous group into internally more homogeneous subgroups, using Kendall's co efficient of concordance W, and show its applicability and application to b oth simulated and in vivo data. Such group partition and "purification" wil l help subsequent inferential methods to deal more efficiently with false p ositives.