The ability to make analogical mapping - determining the correspondence between a pair of objects across different situations based on relational similarity - is thought to be important for different higher cognitive competences. Such higher cognitive competences must emerge out of sensorimotor competences through continuous evolution and development. Dynamic field theory (DFT) provides a theoretical language to explain a process of analogical mapping emerge out of neural networks needed for sensorimotor competences. The conference paper in CogSci 2024 proceedings demonstrates this point through a cognitive architecture autonomously generating a sequence of mental operations to describe the relational structure of one scene and to search for matching objects in another scene. The architecture is built based on other DFT architectures which accounted for scene representation and visually searching for an object described by nested phrases. The structure mapping between objects across scenes are represented by a set of mental maps defined over the index dimensions necessary for representation of relational structure. Additionally, the architecture can be influenced by both relational and featural similarity and qualitatively explain why younger children are more likely to selected featurally similar objects.