Building an Artificial Agent Based on the Unified Dynamic Model of Mind (UDMM)

Abstract This paper proposes a novel approach to building artificial agents inspired by the Unified Dynamic Model of Mind (UDMM). The UDMM frames cognition as a dynamic, predictive interaction between internal generative models and the external world. We explore how this model can be implemented computationally, offering a new architecture for intelligent systems capable of adaptive behavior, emotional simulation, intentionality, and learning. We outline the core architecture, including the perception–prediction loop, the simulation of "virtual attractors," and the representation of time. Emotions are functionally emulated as cognitive responses to prediction errors, and intentionality is framed as a global system constraint. A foundational implementation is available in a public code repository, enabling further development and collaboration.