In this presentation I propose a cognitive dynamics approach on meaning relevance problems. The problem of relevance -or how we decide what is adequate for our interpretation of the signs we encounter in the world- is a question that reappears continuously on semiotics and other disciplines concerned with meaning: Charles Peirce theorized about abduction, a form of reasoning that implies taking a bet on a hypothesis that needs to be verified, ¿but how do we decide such bet?; Structuralist semiotics debated extensively what makes an iconic sign similar to its object, and in that debate they found that is difficult to define what is the minimum quantity of elements the sign must have to be effectively iconic; Philosophy of Mind and Artificial Intelligence call ‘the Frame Problem’ the question about how we and machines can make efficient assumptions about what changes should be considered about an environment; Cognitive semantics have different theories about how mental interpretations develop, and every single one of those theories must face the question about what makes an interpretation appropriate in a particular context. In sum, all these problems require thinking about relevance. In this presentation I want to propose an approximation on relevance that conceives meaning as a trajectory across a cognitive landscape. Unlike conventional accounts of relevance, which presuppose mental processes as discrete rational operations on defined knowledge, my proposal suggests conceiving cognition as a fluid and emergent field of attractors or basins that get specified and modified when experiences appear, and conceiving meaning as a trajectory across the cognitive field; this means that relevance is not exclusively a rational process, but also a process of following tendencies to act. In this sense, relevant interpretations are defined by their trajectories in the mental landscape, and such landscape is defined by previous interpretations. If I do not have much information, ¿how do I decide if a vague sign is a bird or a plane? By strolling through the cognitive landscape that defines my semiotic tendencies. My proposal is mainly supported by two approaches: the categories of recurrences -proposed in other research projects by me- and the theories of dynamic neural populations of Walter Freeman III. The categories of recurrences explain what are the three general processes that ground an agent’s sense-making, or semiosis, and the theories of dynamic neural populations explain how such recurrences get indexed in the neural substrate and define the mental landscape, the predictive disposition of relevance. Additionally, I use Prototype Theory from Cognitive Semantics to explore a possible model of visualization for my proposal and to show some expected concrete applications of the approximation.