MLLM能否解碼創造性跳躍？——引入C4用於跨概念理解

摘要

多模態大語言模型（MLLMs）的創造能力在設計、溝通、教育以及人機協作中至關重要；然而，與以準確度為導向的任務相比，這類能力因明確目標與獎勵訊號稀缺而仍難以評估。跨概念理解是支撐接受性創造力的核心認知能力，它使感知者能夠從不明顯但具有意義的概念關係中還原出所欲傳達的含義。我們將題目建構操作化為跨概念編碼，並將模型推論操作化為跨概念解碼。我們提出 C4——一個受認知啟發、用於評估基於成語之跨概念創造力（Chengyu-based Cross-Concept Creativity）的框架。其編碼組件沿著橋接路徑，將目標槽位映射至可意象的替代概念；這些路徑位於人工標註並經第三方審查的跨概念網絡中，因而能實現批量生成具有明確結構、以橋接數量與深度作為難度指標、且帶有精確答案的題目。利用此框架，我們建構了 C4 評估集（C4-Eval），包含 184 個合成題目，以及 37 個從線上來源收集、由人類創作的跨概念成語修辭。我們為所收集的修辭表達人工建構並審查其跨概念關係、橋接路徑與推理過程。每個 C4-Eval 題目在五種任務設定下實例化，共產生 884 個主要的答案還原案例。在十個受評測的 MLLM 中，最強的封閉模型達到 50.7% 與 48.0% 的主要準確率，而開源模型的表現則明顯較低。候選項約束能大幅提升準確率，但橋接提示與解釋要求僅帶來溫和的增益。這些結果揭示了當前 MLLM 在透過跨概念關係解碼創造性編碼意義方面存在顯著差距。程式碼位於補充材料中。

English

Creative capabilities of MLLMs matter in design, communication, education, and human--AI collaboration, yet remain difficult to evaluate because explicit targets and reward signals are scarce compared with accuracy-oriented tasks. Cross-concept understanding is a core cognitive capacity underlying receptive creativity. It enables a perceiver to recover intended meaning from non-obvious but meaningful conceptual relations. We operationalize item construction as cross-concept encoding and model inference as cross-concept decoding. We introduce C4, a cognition-inspired evaluation framework for Chengyu (Chinese idiom)-based Cross-Concept Creativity. Its encoding component maps target slots to imageable substitute concepts along bridge paths in a manually annotated and third-party-reviewed cross-concept network, enabling batch generation with explicit structure, difficulty indexed by bridge count and depth, and exact answers. Using this framework, we instantiate the C4 Evaluation Set (C4-Eval), comprising 184 synthetic items and 37 human-created cross-concept chengyu figures collected from online sources. We manually construct and review cross-concept relations, bridge paths, and reasoning processes for the collected figures. Each C4-Eval item is instantiated in five task settings, yielding 884 primary answer-recovery cases. Across ten evaluated MLLMs, the strongest closed models reach 50.7% and 48.0% primary accuracy, while open-source models remain substantially lower. Candidate constraints improve accuracy sharply, but bridge hints and explanation requests provide only modest gains. These results expose a substantial gap in how current MLLMs decode creatively encoded meaning through cross-concept relations. The code is in the supplementary material.