The rapid advancement of generative models has made synthetic images increasingly realistic, challenging reliable detection. Existing methods are often limited to end-to-end classification or monolithic reasoning, and thus fail to model structured forensic reasoning and heterogeneous visual evidence. We revisit synthetic image detection from a cognitive perspective and propose a Heuristic-to-Reasoning cognitive skill learning framework for evidence-based forensic analysis. Given an input image, our framework first extracts heuristic perceptual clues, selects the optimal forensic skill, and then performs skill-conditioned reasoning for evidence extraction and decision making. To support this paradigm, we introduce ClueAegis-Bench, which decomposes synthetic image detection into explicitly annotated forensic cognitive skills for structured evaluation beyond binary classification. Based on this benchmark, we propose ClueAegis (Cognitive-skill Learning for Unified Evidence-based Synthetic Image Detection), a two-stage agentic framework that conducts heuristic skill selection followed by evidence-guided reasoning through skill-conditioned toolchains. This design reformulates synthetic image detection as a configurable multi-skill reasoning process that bridges perception, skill selection, and forensic reasoning. Extensive experiments show that ClueAegis achieves state-of-the-art performance while improving cross-domain generalization and robustness. It also provides transparent reasoning trajectories and structured forensic evidence, offering a more explainable alternative to conventional end-to-end detectors.

Aspect, a linguistic category describing how actions and events unfold over time, is traditionally characterized by three semantic properties: stativity, durativity and telicity. In this study, we investigate whether and to what extent these properties are encoded in the verb token embeddings of the contextualized spaces of two English language models – BERT and GPT-2. First, we propose an experiment using semantic projections to examine whether the values of the vector dimensions of annotated verbs for stativity, durativity and telicity reflect human linguistic distinctions. Second, we use distributional similarity to replicate the notorious Imperfective Paradox described by Dowty (1977), and assess whether the embedding models are sensitive to capture contextual nuances of the verb telicity. Our results show that both models encode the semantic distinctions for the aspect properties of stativity and telicity in most of their layers, while durativity is the most challenging feature. As for the Imperfective Paradox, only the embedding similarities computed with the vectors from the early layers of the BERT model align with the expected pattern.