A Decision-Making Model with Interoceptive Expectations

This article addresses the influence of intelligent interbody sensory systems on decision-making in high-risk tasks. It proposes a decision-making method grounded in interoceptive anticipatory reasoning, incorporating a desire value function for three reference points: the bottom line, goal, and current situation of the three-reference point model. Through this approach, agents can not only forecast future environmental changes but also navigate a multi-task environment by constraining potentially harmful tasks. In instances where the agent's interoceptive system conflicts with external environmental goals, the desire value for the external goal is diminished. However, with the emergence of a new external goal, a superimposed effect takes place, allowing for the execution of high-risk tasks through interoceptive constraints. Addressing this issue, the article investigates the changing rules of interoception in psychology across diverse situations. Additionally, it utilizes the NARS reasoning platform to exemplify the impact of interoceptive tasks on decision-making in high-risk urban fire situations. Experimental results reveal that the desire for external fire-fighting goals is influenced by the fire-fighting robot's internal sense of goal. Simultaneously, the robot's desire value to protect its own goal decreases when confronted with special situations. These research findings present novel insights into rapid decision-making in high-risk tasks and establish a research foundation for emotional construction.