This research was carried out in the context of the I SEA project, focused on the development of a non-obtrusive method for evaluating science communication initiatives. It was intended to explore the social representations about the main concepts of the project: deep sea (message content), virtual reality (medium used), and science communication (field of the study). Through a snowball sampling strategy, we recruited 294 adult participants (87 males and 205 females) to answer an online questionnaire that included association questions about the stimuli related to the project’s main concepts. Participants were also asked to rank their answers according to their perceived importance. Prototypical and similitude analyses were conducted with the software IRAMUTEQ (version 0.7 alpha 2).  Preliminary results suggest that deep sea is represented as a dark, blue, cold, deep and immense unknown region, populated with fishes, sharks, corals and submarines, inspiring fear, tranquility and curiosity.  Pollution is referred by few participants. Virtual reality is objectified in specific equipment such as headsets and anchored to computers and games, which appear associated with immersion and interactivity. Notwithstanding the associated risk, it appears to be a promising symbol for the technology of the future. Other drawbacks appear in the form of illusion, alienation, falsity and addiction, provoking a tension between promises and perils. Science communication evoked very diversified ideas, suggesting that its social representation is poorly structured. For the participants, it is presently an important, relevant and essential activity, consisting of sharing, communicating scientific knowledge, though research papers, conferences and media. The audience and its role are not very visible in the representation. Results are significant as they allow us to understand that the topics are paradoxically next to and far away from the audience, feeding the reflection on how to communicate and to evaluate cutting-edge knowledge in non-formal spaces.