Strongly partitive accounts deviate radically from the common view of a single, unified knower or self within each 'person', proposing instead an account of multiple knowers. This view is justified by consideration of mental conflict, and objections, including the view that conflict does not require strong partitioning, that there exists a tension between 'persons' and 'parts', and the problem of homunculi, are found not to hold. However, the problems of proposing partitioning ad hoc and ad libitum are genuine concerns that any account of mental plurality must address. The realist account of cognition, proposing that cognition is a relation between subject and object terms existing independently of the cognitive relation, provides a conceptual basis for evaluating strongly partitive accounts. On this view, any account of knowers, either singular or plural, must be capable of specifying their intrinsic qualities. Pears's account is found to be problematic here since it fails to meet this logical requirement. Maze's account is found to satisfy this requirement, providing inprinciple means of characterizing the multiple knowers.