Presence, as the feeling of actually being in a virtual place, is an important metric for VR learning applications due to its positive impact on learning. In the literature, the level of immersion a system can provide is assumed to be the main factor influencing presence. This paper investigates this assumption for virtual environments in abstract learning contexts that are not based on reality. This is done by building a physical replica of a VR learning game and comparing both in terms of learning outcomes and presence. The study shows that subjects do not have a consistent view of what can be described as real in abstract contexts, suggesting that the distinction between reality-based virtual environments and those in abstract contexts is important in terms of presence.