They Are Not the Same: Direct Causes Are Not Grounded Emotion Explanations

2026-05-24

AI Summary

This study addresses a critical limitation in current Emotion-Cause Pair Extraction (ECPE) research, which oversimplifies emotion explanation as binary pair prediction and conflates direct triggers with contextually grounded explanations, thereby neglecting essential contextual support. To rectify this, the authors explicitly distinguish between these two notions, introduce the “emo-context” concept, and construct the IEMO-MECP dataset with fine-grained annotations. Through dialogue-based emotion analysis and multi-model benchmarking, they reveal a fundamental misalignment between prevailing ECPE paradigms, evaluation metrics, and the actual task of explanatory reasoning. Experiments show that while 90.9% of original positive samples remain valid emotion-cause pairs, models excel at identifying direct triggers yet perform significantly worse on structurally complex but well-supported explanations, indicating that high ECPE scores do not reflect genuine explanatory capability.

Abstract

Emotion-Cause Pair Extraction (ECPE) was introduced to explain why an emotion occurs, but this goal is now often reduced to binary pair/non-pair prediction. This proxy is useful for direct-cause extraction, yet easy to over-read as evidence grounded emotion explanation. We show that this interpretation is only partially valid. In IEMO-MECP, 90.9% of original positives remain emo-cause and 95.0% of original negatives remain non-pair, confirming that the binary ECPE task is largely preserved. The problem is that direct triggers alone do not constitute a grounded explanation. Emo-context, an utterance that helps interpret a target emotion without directly causing it, appears on both sides of the original boundary and is enriched near binary uncertainty, showing that the binary boundary has no stable place for such discourse evidence. Across evaluated ECPE models, direct triggers are recovered more reliably than contextual support. Under shortcut pressure, this imbalance becomes consequential. Binary-trained models assign higher pair scores to nearby lexically similar non-pair candidates than to evidence supported but structurally harder emo-cause and emo-context pairs. Thus, pair scores can reward convenient attributions over grounded explanations. High binary ECPE performance indicates that a model can identify direct triggers; it does not indicate that the model has explained the emotion. Code is publicly available at https://github.com/panzhzh/ECPExsame.

Problem

• Emotion-Cause Pair Extraction
• grounded explanation
• direct cause
• emo-context
• binary prediction

Innovation

• Emotion-Cause Pair Extraction
• grounded explanation
• direct cause
• emo-context
• shortcut learning