The general goal of this thesis is to elucidate children’s early understandings of pictorial meanings, and how one can know anything about them. My central aim is to explore how picture comprehension develops during children’s first 3 years of life, through semiotic-theory-derived analyses of meaning relations. In so doing, I hope to contribute to the study of both semiotic theory’s psychological basis and the role of semiotic processes in cognitive development: specifically, in children’s experiences of pictorial meanings. In an experimental object retrieval test, including pictures, I show the importance of studying concrete instances of children’s experiences. Among its key results is that, for a group of children who are close to the threshold of being able to use the picture to solve the retrieval task, indexical cuing assists their understanding. One central claims is that the picture sign reflects a dual semiotic process: on the one hand, picture understanding relies on recognition of perceptual similarities; on the other, it draws on communicative processes that are intrinsic to all sign constructions. This duality is particularly interesting when it comes to looking at children’s development of picture understanding. Through similarity relations, children perceive accurate – but initially private and incomplete – understanding of pictures. At the same time though, children are alert to communicative meanings from the start.