Rethinking Inner Speech through Linguistic Active Inference

This paper introduces the Linguistic Active Inference Theory (LAIT), which proposes that inner speech enhances the brain’s ability to navigate uncertainty by augmenting predictive processes for perception and action. By leveraging language’s unique properties - its efficiency in representing sensorimotor information, ability to extend across time and space, and generativity in constructing novel predictions - inner speech enables predictive processing to transcend immediate experience, encoding complex sensory experiences into linguistic forms for perceptual inference, while decoding abstract goals through situated action simulations for active control. By unifying previously disparate aspects of inner speech research, LAIT provides a comprehensive framework explaining how its diverse functions, varied phenomenology, and theoretical models emerge from its implementation of perceptual inference and active control through dynamic linguistic predictions, adapting to computational demands and fluctuating uncertainties across contexts. This synthesis generates novel testable hypotheses, underscores the need for methodological innovations in studying inner speech dynamics, and opens new perspectives on both related mental phenomena and the broader role of symbolic systems in cognition.