Can there be objective characterisations of action given G.E.M. Anscombe's idea that all action occurs "under a description"?