Both visual interfaces, textualized diagrams and image-native VLM input, beat text-only difficulty prediction on point estimates, but their difference is not statistically reliable, and image-native gains depend on broadening LoRA adaptation.
Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Predicting human item difficulty is central to educational assessment, where reliable estimates support fairness and effective test construction. Existing methods often depend on costly human calibration or item-level textual representations, providing limited evidence about the cognitive processes that make items difficult. We argue that difficulty should be viewed not only as a property of item text, but also as an observable consequence of the problem-solving burden an item induces. Large Reasoning Models (LRMs) offer scalable process evidence through reasoning traces, but such evidence must be structured to support interpretable modeling. To this end, we introduce Epi2Diff (Episode to Difficulty), a framework that maps LRM reasoning traces into cognitively grounded episode sequences. These episodes group trace segments into functional problem-solving states, enabling difficulty to be modeled through reasoning scale, effort allocation, and state transitions. Epi2Diff extracts compact episode-dynamic features and combines them with semantic item representations for human difficulty prediction. Experiments on four real-world human difficulty datasets show that Epi2Diff consistently outperforms strong baselines, including fine-tuned small language models, LLM in-context learning, and supervised LLM adaptation. On SAT-derived classification benchmarks, Epi2Diff achieves an 8.1% average relative gain over supervised LLM fine-tuning baselines. Further analyses show that harder items induce more effortful, iterative, and implementation-centered episode dynamics, rather than merely longer responses. These results demonstrate that cognitive episodes in LRM reasoning traces provide a predictive and interpretable process representation for human item difficulty, offering a new lens for educational measurement with reasoning models.
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Representing Visual Evidence for Item Difficulty Prediction: Visual Textualization and Image-Native Modeling
Both visual interfaces, textualized diagrams and image-native VLM input, beat text-only difficulty prediction on point estimates, but their difference is not statistically reliable, and image-native gains depend on broadening LoRA adaptation.