Proceedings of the Annual Meeting of the Cognitive Science Society

Human categorical speech perception (CP) is traditionally modelled as deterministic to maximise individual efficiency, yet empirical evidence suggests it is intrinsically probabilistic or fuzzy. This study investigates this apparent cognitive inefficiency using an agent-based model of emergent vowel systems. We simulated populations playing Imitation Games under varying noise levels to compare the robustness of rigid (strong CP) versus fuzzy (weak CP) perception mechanisms. Contrary to the assumption that precision yields optimality, results demonstrate that the perceptual uncertainty of weak CP acts as a crucial buffer against environmental noise. Weak CP populations not only achieved higher communicative success but also maintained categories with significantly longer lifespans and lower cognitive maintenance costs. We argue that individual-level perceptual inefficiency is an adaptive feature essential for robust social coordination, functioning as a mechanism of loose coupling that prevents systemic collapse.