Episodic memory is traditionally conceived of as the remembering of events from one’s personal past, experienced with one’s own senses. However, hallucinations and dreams, vicarious experiences with other people, or even narrative experiences with literature, may all lead to memories that exhibit qualities typically considered to be indicative of episodic memory, such as vividness, sequentiality and a first-person perspective. Our project aims to account for this tension by arguing for a reframing of the boundaries of episodic memory as a natural kind. It does so through a naturalistic approach that compares the neurobiological features of both perception-based and certain non- perception-based memories. Specifically, these features are: (F1) the experience-like brain state that grounds the memory; (F2) the mechanism that constructs a scenario of the remembered event and reflects the immersive, sensorimotor and emotional character of the original experience; and, (F3) the minimal memory trace that acts as an appropriate causal link between the experience-like brain state and the remembering event and warrants the reliability and specificity of the memory. The comparison of these features raises a number of challenges that will be addressed through five work packages. In WP1, we assess the grounds to extend the natural kind of episodic memories to include cases that are not based on personal, perceptual experience. We understand a natural kind as the maximal class whose members are likely to share their properties because of some uniform underlying causal mechanism. WP2 addresses the factivity challenge, which is raised because memories based on non-veridical experiences are not grounded in actual events. In response, we develop a presuppositional account of factivity and explain reference relations in episodic memories in terms of referential parasitism. WP3 will elucidate how both perception- and non-perception-based memories, despite their phenomenological differences, feed into minimal memory traces by similar mechanisms. In WP4, we will develop an answer to how, in non-perception-based memories, the notion of memory perspective needs to be re-evaluated, by dissociating self from first-person perspective. Finally, in WP5, we will develop an embodied account that casts remembering in terms of an immersive, sensorimotor, and emotional scenario construction informed by a minimal trace and supplemented with semantic information, in contrast to prevailing imagistic views that conceive of memory as principally having sensory content. The results of these investigations will inform debates on the relation between memory and imagination. If it turns out that the natural kind of episodic memories needs extending, then our findings will have a potential impact on scientific study and clinical intervention contexts that target episodic memory.