跳转至

情感目标导向理论的计算实现

文章背景与核心概要

长期以来，情感的计算建模一直面临着一种二分困境：一方面是描述性的、基于“快照”的评估模型；另一方面是缺乏坚实心理学基础、粒度粗糙的信号驱动架构。本文引入了情感目标导向理论（Goal-Directed Theory, GDT）的首个高保真计算实现，旨在弥合这一鸿沟。在此框架下，情感不再被视作事后的标签，而是作为智能体内部处理循环中差异检测与行动选择之间持续交互所产生的涌现功能副产品。

通过系统性的模拟实验（如骰子和走廊任务），作者证明了复杂的情感特征——例如预期的“提升”或失败的“崩溃”——能够自然地从目标差异与行动选择预期之间的简单交互中涌现出来，完全无需依赖专门的评估模块。通过将每个计算元素直接映射到心理学理论，这项工作为深入理解并嵌入在核心智能体行为中的情感建立了一个透明、可测试的框架，推动该领域从“黑盒”启发式方法向精细的、机械化的情感理解迈进。

Metadata (元数据)

  • arXiv ID: 2609.06654
  • Primary Subject: Artificial Intelligence (cs.AI)
  • Submission Date: September 6, 2026
  • Authors: Bernhard Hilpert, Tamás Szűcs, Joost Broekens, Agnes Moors

Executive Summary (执行摘要)

Computational modeling of emotion has long struggled with a dichotomy: descriptive, "snapshot-based" appraisal models on one hand, and granular, signal-driven architectures lacking solid psychological grounding on the other.

情感的计算建模长期以来一直苦于一种二分法：一方面是描述性的、基于“快照”的评估模型，另一方面是缺乏坚实心理学基础的细粒度、信号驱动架构。

This paper introduces the first high-fidelity computational implementation of the Goal-Directed Theory (GDT) of affect. In this framework, affect is modeled not as a post-hoc label, but as an emergent functional byproduct arising from the continuous interplay between discrepancy detection and action selection within an agent's internal processing cycles.

本文介绍了情感目标导向理论（Goal-Directed Theory, GDT）的首个高保真计算实现。在此框架中，情感不被建模为事后标签，而是作为智能体内部处理循环中差异检测与行动选择之间持续交互所产生的涌现功能副产品。

Through principled simulations (such as Dice and Corridor tasks), the authors demonstrate that complex affective profiles—such as an anticipatory "lift" or a failure "crash"—emerge naturally from straightforward interactions between goal-discrepancy and action-selection expectancies, entirely bypassing the need for dedicated appraisal modules. By mapping every computational element directly to psychological theory, this work bridges the gap toward a mechanistic, transparent understanding of affect embedded within core agent behavior.

通过原则性的模拟（如骰子和走廊任务），作者证明了复杂的情感特征（例如预期的“提升”或失败的“崩溃”）可以从目标差异和行动选择预期之间的直接交互中自然涌现，完全无需专门的评估模块。通过将每个计算元素直接映射到心理学理论，这项工作弥合了差距，朝着对嵌入在核心智能体行为中的情感进行机械化、透明化的理解迈出了重要一步。

Abstract (摘要)

Computational modeling of emotion has long faced a tension between descriptive, "snapshot-based" appraisal models and granular, signal-driven architectures that often lack appropriate psychological grounding. This paper addresses this gap by presenting the first high-fidelity computational implementation of the Goal-Directed Theory (GDT) of affect. In this framework, affect is not a post-hoc label but a functional byproduct emerging from the continuous interplay between discrepancy detection and action selection within an agent's internal processing cycles. We evaluate the model through a series of principled simulations (Dice/Corridor tasks) designed to isolate affective signatures and dynamics during multi-step goal pursuit. Results demonstrate that complex affective profiles, like an anticipatory "lift" and a failure "crash", emerge naturally from simple interactions between goal-discrepancy and action-selection expectancies without requiring additional dedicated modules. By ensuring every computational component maps directly to components of the psychological theory, this work establishes a transparent, testable framework that enables a continuous "simulation-empiry" research loop. Our work contributes to moving the field beyond "black-box" heuristics toward a granular, mechanistic understanding of affect, integrated into the core of agent behavior.

情感的计算建模长期以来面临着一种张力：描述性的、基于“快照”的评估模型与通常缺乏适当心理学基础的细粒度、信号驱动架构之间的张力。本文通过提出情感目标导向理论（GDT）的首个高保真计算实现来弥补这一空白。在此框架下，情感不是事后标签，而是智能体内部处理循环中差异检测与行动选择之间持续交互所产生的功能副产品。我们通过一系列旨在隔离多步目标追求过程中的情感特征与动态的原则性模拟（骰子/走廊任务）来评估该模型。结果表明，复杂的情感特征（如预期的“提升”和失败的“崩溃”）自然地从目标差异与行动选择预期之间的简单交互中涌现，而无需附加的专用模块。通过确保每个计算组件直接映射到心理学理论的组件，这项工作建立了一个透明、可测试的框架，从而能够实现持续的“模拟-经验”研究循环。我们的工作有助于推动该领域超越“黑盒”启发式方法，走向对情感的细粒度、机械化理解，并将其整合到智能体行为的核心中。