According to the Source Monitoring Framework, the origin of a memory is remembered through the retrieval of specific features (e.g. perceptive, sensitive, affective signals). In two source discrimination tasks, we studied the ability of cuttlefish to remember the modality in which an item had been presented several hours ago. In Experiment 1, cuttlefish were able to retrieve the modality of presentation of a crab (visual vs olfactory) sensed before 1 h and 3 hrs delays. In Experiment 2, cuttlefish were trained to retrieve the modality of the presentation of fish, shrimp, and crabs. After training, cuttlefish performed the task with another item never encountered before (e.g. mussel). The cuttlefish successfully passed transfer tests with and without a delay of 3 hrs. This study is the first to show the ability to discriminate between two sensory modalities (i.e. see vs smell) in an animal. Taken together, these results suggest that cuttlefish can retrieve perceptual features of a previous event, namely whether they had seen or smelled an item. Can you tell whether you truly enjoyed your last holiday? According to the Source-Monitoring Framework (SMF), answering such a question requires you to revisit your personal past and retrieve specific features belonging to your memories (e.g. affective, perceptual and contextual features 1). For instance, I can remember that I went to my parents' home town (contextual features), and that we spent evenings talking or playing music (perceptual features) in a joyful atmosphere (affective feature). To remember these specific details and moments, I travelled mentally back through my personal past and engaged in episodic cognition processes, projecting myself in space and time to relive and re-experience the content of those personal memories, integrating the contextual, perceptual and affective features. Travelling mentally back into one's personal past is referred to as episodic memory, while retrieving specific features belonging to these episodic memories is a cognitive capacity involving source-memory processes. Source-memory is embedded into the episodic memory, and triggers semantic processes aiming at retrieving the origin of a memory and enabling to distinguish between two or more episodic memories. In humans, source-memory is mostly studied using item versus source-memory discrimination task. As the memory of the source relies on the recall of specific characteristics of a prior situation, participants are asked to retrieve the features of the context in which items were previously encountered. In such studies, participants have to recall the items they encountered earlier in opposition to new items (item memory), and then retrieve the context in which they were presented (e.g. whether the target items were read or mentally imagined 2 ; their spatial location 3 ; the list to which they belonged 4 ; the colour of the item 5 , etc; source-memory). Only few studies have focussed on source-memory in non-human animals. One single experiment mimicked the item versus source procedure in monkeys 6. Rhesus monkeys learnt to respond differently to two images (i.e. the first needed to be simply touched and the second one should be classified as bird, fish, flower, or human). At test, four images were presented (the two previously seen images and two distractors) and half of the monkeys needed to retrieve the image previously simply touched, and half of the monkeys needed to retrieve the image previously classified. Monkeys showed their ability to discriminate between the two sources when test was presented after a short delay, but they made source-memory mistakes when tested after a long delay, while still avoiding distractors (item memory preserved). Crystal and colleagues 7 studied rat's ability to discriminate between self-or externally-generated information. This study focused on another type of source-memory, called reality monitoring (i.e. did I learn this information myself, or did I learn it from someone else? 1). Apart from monkeys and rats, source-memory has not been