Heran Hu, Yuhang Dong, Zhucun Xue, Kaiwei Chen, Yong Liu, Jiangning Zhang · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.28625 · cited by 0 works (OpenAlex) · open-access PDF
La compréhension des besoins humains et la sélection des réponses nécessitent des jugements distincts que l'exactitude agrégée peut obscurcir. Nous présentons PsycheMME, un référentiel vidéo informé par la psychologie, avec huit capacités couvrant les bases, les états, les besoins, les décisions d'interaction, la mémoire et le risque. En tirant parti du traitement des informations sociales et du cycle de perception-action, cette organisation sépare la compréhension humaine de la sélection des réponses. Nous construisons 1 000 questions de 630 épisodes à travers une réponse indépendante avec des références proposées retenues et une revue d'experts. Notre protocole Reasoning-Verified pèse le crédit de la réponse par le soutien et la coh
Machine translation (Meta M2M100); the English original follows.
Understanding human needs and selecting responses require distinct judgments that aggregate accuracy can obscure. We introduce PsycheMME, a video benchmark informed by psychology, with eight capabilities spanning grounding, states, needs, interaction decisions, memory, and risk. Drawing on Social Information Processing and the Perception-Action Cycle, this organization separates human understanding from response selection. We construct 1,000 questions from 630 episodes through independent answering with proposed references withheld and expert review. Our Reasoning-Verified protocol weights answer credit by the support and logical consistency of explicit rationales against annotated cues. Across 17 models, Timing has the lowest Verified scores and credit retention, with weaker support judgments for correct Timing answers. Models with similar Need scores still differ in answer completeness on open questions. Input controls and SFT/GRPO examine context dependence and adaptation. Evaluating answers alongside rationales locates where models fail across understanding, response selection, and justification.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Computer science, Benchmark (surveying), Consistency (knowledge bases), Construct (python library), Completeness (order theory), Context (archaeology)
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28625, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange