In an appeal to establish transformative learning's (TL) boundaries, Kegan (2000) implored researchers to identify "what from transforms?" As a social psychologist, I am particularly drawn to theorists who identify the self as the form that transforms; however, I argue that a model of the self that specifies how it is represented in memory offers advantages to the prevailing models of the self-advanced by transformative learning theorists. Since a multiple self-framework specifies how the self is represented in memory, it is more specific than previous models and it is amenable to research methods beyond retrospective reports. Additionally, it offers several explanations of how transformation may occur and provides an alternative explanation for incomplete transformations. In doing so, it adds to theories attempting to identify the form that transforms.