Response to Sloman's Review of Affective Computing