“[In] unguarded moments I do think that everything is concepts”, stated Murphy in his popular big book of concepts [1] emphasizing the pivotal role of concepts or “mental representations” in understanding the mind. Inspired by Murphy’s unguarded moments, we propose a framework asserting that in artificial neural networks (ANNs) everything is representations and mechanisms. Specifically, based on an interdisciplinary literature review, we propose a framework called the representations’ lifecycle. The framework consists of two main contributions: first, we propose a template that characterizes representational change in ANNs along compositional, hierarchical, and temporal dimensions. Second, the latter template allows for the characterization and demarcation of six representation-altering processes: abstract primitives’ integration, perceptual primitives’ integration, assembly, abstraction, differentiation and decay. Our framework provides the foundation for a more universal description of representational change in neural networks and thus, contributes to the broader efforts towards more transparent and explainable ANNs.