创造性测量研究综述

A Review of Creativity Assessment Research

摘要: 创造性作为推动人类文明进步的核心心理能力，其科学测量一直是心理学、认知科学及教育研究的关键课题。本文系统回顾了创造力测量的主要范式演进。首先阐述了从发散思维、远距离联想的单一认知成分理论，到生成–评估双过程动态系统模型的认知理论基础变迁。继而详细剖析了传统测量范式，包括应用广泛但受限于主观评分和知识混淆的发散思维测验、生态效度高但实施成本昂贵的一致性评估技术，以及测量范围狭窄的远距离联想测验。随着认知神经科学的发展，研究开始探索创造力的神经关联，发现其依赖于默认模式网络、执行控制网络与突显网络的动态耦合，然而神经指标目前仍难以作为个体层面的实用评估工具。新兴的计算方法，如基于语义距离的客观评分和语义网络拓扑分析，虽在客观性与可扩展性上优势显著，但仍面临文化依赖性、适当性量化困难等挑战。本文最后展望未来，提出多模态数据融合、开发文化公平任务、结合计算建模与理论发展等方向，旨在构建更为客观、全面且生态效度更高的创造力评估体系。

Abstract: Creativity, as a core psychological capacity driving the advancement of human civilization, has long been a central subject of inquiry in psychology, cognitive science, and educational research. This paper provides a systematic review of the evolution of major paradigms in creativity assessment. It begins by delineating the shift in cognitive theoretical foundations, from early models focusing on singular components like divergent thinking and remote association, to contemporary dynamic system models such as the dual-process (generation-evaluation) framework. The paper then provides a detailed analysis of traditional assessment paradigms, including the widely used yet methodologically limited Divergent Thinking Tests (constrained by subjective scoring and knowledge confounds), the ecologically valid but resource-intensive Consensual Assessment Technique, and the narrowly focused Remote Associates Test. With the advent of cognitive neuroscience, research has begun to explore the neural correlates of creativity, revealing its reliance on the dynamic coupling of the Default Mode Network, the Executive Control Network, and the Salience Network; however, neural metrics currently remain impractical for individual-level assessment. Emerging computational approaches, such as objective scoring based on semantic distance and semantic network topology analysis, offer significant advantages in objectivity and scalability, yet they still confront challenges, including cultural bias and the difficulty in quantifying appropriateness. Finally, the paper outlines future directions, proposing the integration of multimodal data, the development of culturally fair tasks, and the synergy between computational modeling and theoretical advancement, aiming to construct a more objective, comprehensive, and ecologically valid creativity assessment system.