Hi Stan
we have found positive correlations between performance (in this case CVLT) and thickness, so it is certainly possible. And negative correlations aren't necssarily false - you could certainly imagine that successful pruning for example could help performance. Have you visually inspected the surfaces for accuracy?
cheers Bruce
On Tue, 5 Feb 2013, Berg, S.F. van den wrote:
Dear Freesurfer experts,
We investigated the relation between cortical thickness and performance on several cognitive tasks within a large group of Parkinson?s disease patients, but are slightly puzzled by the results. We obtained several, both in the vertex-wise analysis in qdec and in the SPSS analysis in the a-priori parcelated areas, negative correlations (i.e. better performance relates to a thinner cortical area) for all our neuropsychological tasks. These negative correlations were unexpected and we are having a hard time interpreting them. After correcting for multiple comparisons, all effects failed to reach the statistical threshold, rendering no results at all. When we compared our patient group on a structural level with healthy controls, we did find expected results that made sense.
We also ran the exact same analyses (same group, same data) in VBM and there found several positive task-related correlations in expected areas.
We noticed that in previous literature almost no studies investigated the relation between cortical thickness and cognitive task-performance. This made us wonder whether Freesurfer is suited for these correlation-based analyses, or is it better to be used in between group analyses? And do you have any idea on how to explain our negative correlations? Is it possible we might have done something wrong?
Many thanks,
Stan