Hello,
I am using the BIDs app provided here:
I am interested in comparing the ability of different regions to classify two experimental conditions, speech vs. tone (non-speech). The classification accuracy of one region, the superior temporal gyrus (STG) was quite high, which is what I expected (as the STG is specialized for language processing). However, I also expected that the classification accuracy of the whole brain should have been higher than the STG by itself. Instead, it was lower than the STG by itself.
Likewise, the classification of the STG and another region (the lateral occipital cortex, LOC) added together was lower than the STG by itself; instead, it fell between the classification accuracy of the STG by itself and the LOC by itself. It seems that at least with the settings I am using for the app now, the classification accuracy of a region is more similar to an average of how good each of its subparts it at classification, rather than a sum of the classification accuracies of each of its subparts.
Does anyone know if this is how the app always works, or is there a way to change its behavior to match the expectation that the classification accuracy of the whole brain should be better than the STG by itself?
Thanks much,
Stefan Bartell
https://sites.udel.edu/q-lab/