In DA I.2–5, Aristotle offers a series of critical discussions of earlier Greek definitions of the soul. The status of these discussions and the role they play in the justification of Aristotle’s theory of soul in DA II–III is controversial. In contrast to a common view, I argue that these discussions are not dialectical but philosophical. I also contend that Aristotle does not consider earlier philosophical definitions of soul to be endoxa, but rather contradoxa – beliefs about which the many and the wise disagree among themselves. Through an analysis of Plato’s and Empedocles’s definitions of soul, I show that these definitions are nevertheless treated by Aristotle as potential scientific principles for explaining two of the soul’s per se attributes: causing motion and cognition in animate bodies. The main role of the critical discussions in DA I.2–5 is to show that all such earlier definitions of soul fail this explanatory task. Nevertheless, I show that these chapters are not wholly aporetic. Aristotle makes progress by solving two scientific puzzles within them: whether the soul has spatial parts, and whether ‘soul’ refers to a uniform entity across biological species.