This paper develops a theory of Intelligent Action in which behavior is not modeled as reward maximization, but as the outcome of interacting epistemic forces evolving over a structured cognitive manifold. Seven fundamental forces govern cognition and produce a Behavior Triad that evaluates, learns, and decides when to act, while a unified energy framework integrates contradiction, emotional alignment, contextual uncertainty, and risk. Emotional state functions as a control mechanism for behavior, and a contextual depth variable enables detection of instability and unknown factors. The framework includes mechanisms for operating safely near the limits of knowledge and establishes a theoretical connection to the boundary of what computable reasoning systems can achieve, positioning it as the second part of the trilogy.