The Process-Dissociation Procedure is used to derive estimates of controlled and automatic memory from performance measured under inclusion and exclusion instructions. A controversial assumption of the model is that the two aspects of memory are independent. Recently, Hirshman (2004, Psychological Review, 111(2), 553-560) has proposed a weaker version of this model that assumes only an ordinal relationship between each dependent variable and the underlying parameters of the model and derived a set of conclusions that can be drawn from ordinal differences in inclusion and exclusion performance. I show how the original independence model, Hirshman's generalization, and other models (e.g. redundancy, exclusivity) may be interpreted as parameterizations of a space consisting of performance under inclusion and exclusion instructions. This representation yields a simple geometric interpretation of Hirshman's inferences and offers a basis for constructing relatively straightforward tests of these and other related models.