Active Inference and the Primacy of the ‘I Can’