Linguistic concepts are self-generating choice architectures.

While the idea of a 'concept' has been defined in diverse ways, researchers in the cognitive science of language have largely agreed that linguistic concepts are objects, whether mental or physical, that bits of language stand for. This O-axis view (where O = object), focusing on sign-object relations, sees linguistic concepts as ideas that stand in a static relation to signs, with the function of mediating relations between agents and their environments. But this is only half the story. Because every linguistic concept is moored to a bit of language, and bits of language are mostly learned and encountered in sequences of social interaction, then we must look not only at what signs stand for (their ), but at the , or rational responses, that they elicit. By focusing on sign-interpretant relations, and thus taking an I-axis view (where I = interpretant), we not only acknowledge the direct link between concepts and social interaction, we also discover causal mechanisms that explain how linguistic concepts are distributed in relatively stable form in populations. We find that while concepts are indeed mental objects, they function as choice architectures in the dynamic flow of situated language usage. This article is part of the theme issue 'Concepts in interaction: social engagement and inner experiences'.