Cortical thickness and the shape of psychiatric risk: a cautionary note on causality