Discussion of imalsogreg's comment: The claim that non primary cortex is the first to show pitch correlates is dependent on the assumption that nonsignificant activation differences in subcortical/primary cortical areas imply that those areas don't process pitch. But how do you judge that pitch isn't just encoded there in a way not detectable by fMRI? For example temporal coding hasn't been shown to have an fMRI correlate, but it definitely is a form of place coding in hippocampus. Could pitch be rate coded in primary cortex? To be fair, the author acknowledges this issue. But it's not clear how to get around it.