According to current theory, memories are initially encoded through hippocampal (HC) and neocortical (NC) interactions, and subsequently consolidated during slow-wave-sleep (SWS) into hippocampus-independent neocortical schemas. In this sleep phase, specific neural oscillations, Sharp-Wave-Ripples (SWR) and K-complexes (KC), dominate, respectively, HC and NC.Complementary Learning Systems Theory (CLST)1 proposes that new memories can gradually be integrated into the NC, without catastrophically interfering2 with existing knowledge. However, it assumes a massive replay of previous knowledge interleaved with the replay of new memory.In Artificial Neural Networks, similarity-weighted interleaved learning facilitates the efficient acquisition of new information by employing a reduced set of older items, selected based on their representational similarity to the new information. This approach minimizes interference and augments data efficiency3..We propose a biological embodiment of this model, which assumes that cortical KCs accompanied by hippocampal SWR denote the reactivation of novel information whereas KCs without accompanying SWR denote the reactivation of preexisting memories in NC. This model provides insight into memory consolidation processes in the brain. We introduce a novel memory task in virtual reality and an experimental framework designed to evaluate the retention of recent and remote memories, enabling a direct test of this hypothesis.References: 1. McClelland et al., 1995; 2. McClosky and Cohen, 1989; 3. Saxena et al., 2022;