The alleged vagueness of visual images and the lack of a univocal coding scheme make it difficult to be sure about the propositions to which image-makers are committed. This is particularly problematic for the analysis of multimodal discourse from an argumentation studies perspective, because it makes it hard for the analyst to establish the argumentative nature and relevance of visuals. The paper explores how insights from Relevance Theory can be applied in order to determine the commitments of image-makers. In particular, it has recourse to the inferential processes involved in the recovery of explicit and implicit content in order to analyse a series of covers from The Economist, where visuals in combination with verbal text cue allusions to films and paintings. It argues that these multimodal allusions are not simply used to attract the audience's attention but also help the analyst to reconstruct the argument of the cover.