Paul E. Griffiths argues that most research on the emotions has been as misguided as Aristotelian efforts to study "superlunary objects" - objects outside the moon's orbit. Such subjects exist, of course, but studying them as a group produces no useful results because they share no traits other than an arbitrarily defined location. Similarly, Griffiths show that "emotion," as currently defined, groups together psychological states of very different, and thus not comparable, kinds. According to Griffiths, theoretical research on emotions took a wrong turn by not fully exploring the relevant empirical evidence. Griffiths provides a detailed overview of this material, drawing on ethology, evolutionary biology, neuroscience, cognitive psychology, and anthropology of the emotions. He identifies and assesses the relative merits of three main theoretical approaches - affect program theory, evolutionary psychology, and social constructionism.