Global-matching models of memory argue that knowledge emerges from the interaction between presented cues and traces of past experiences. But these models generally rely on the use of independent episodic traces, unable to account for global interactions between learned situations (see Versace et al., 2009). Enactivism (Varela, 1993) could theoretically take advantage of an inter-dependent processing of traces to account for abstraction processing using only sensorimotor covariances (Hutto & Myin, 2012), but no mathematical formalization of an enactivist memory has yet been proposed. In this paper, we propose the ATHENA model as an enactivist mathematical formalization of Act-In theories (Versace et al., 2014) within MINERVA2 (Hintzman, 1986) non-specific traces: ATHENA is a fractal model which keeps track of former processes that led to the emergence of knowledge, and is therefore able to process contextual processes (abstraction manipulation). We present three simulations designed to test ATHENA’s ability to construct, learn, and manipulate emergent abstractions.