The images shown on food packaging play an important role in the processes of identification, categorisation and the generation of expectations, since the consumer uses the images to infer information about the product. However, a given image may convey different meanings (e.g. in a food package, “fire” may mean barbecued or spicy), so it is very important for producers and designers to understand the factors responsible for consumers inferring a specific meaning. This paper addresses this problem and shows experimentally that the consumer tends to infer the meaning from the image which is most congruent with the product it is displayed with. 65 participants carried out two speeded classification tasks which results show an interaction between the product (congruent vs. incongruent) and the image (with fire vs. without fire): products congruent with a meaning of fire were categorised more quickly when shown with fire than without it, while products incongruent with a meaning of fire were categorised more slowly when shown with fire than without it. In addition, the results show that stimuli were categorised more quickly when the interpretation of fire was literal (e.g. barbecue) than in those that were metaphorical (e.g. spiciness), indicating that the rhetorical style of the image (literal or metaphorical) influences the cognitive effort required to process it. These contributions improve our understanding of the effect of the images shown on packaging in the communication between packaging and consumers.