Latent Dynamics · 2026-10-14
AffectWorld: Benchmarking Emotion-Driven Future World Prediction
Changhao Song, Yacong Chen, Hui Gao, Yazhou Zhang, Peng Zhang
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
AffectWorld is a benchmark for emotion-driven future world prediction, evaluating the chain Emotion → Action → World State. It includes 1,498 everyday video samples plus a 233-sample counterfactual subset where only an agent’s emotion is intervened. The authors report evaluation of five state-of-the-art world models using eight metrics (three tasks plus counterfactual), revealing gaps in emotion-driven action consequences and emotional intervention sensitivity.
Why it matters
The authors argue existing world model benchmarks mainly assess physical plausibility and perceptual quality, while emotion-driven world evolution remains under-tested. AffectWorld targets whether models can treat emotion as a causal variable that shapes actions and resulting social/physical future states, providing a testbed for emotion-conditioned future prediction.
Method
- Construct AffectWorld from 1,498 real-world everyday videos with annotations for event context, target/plausible emotions, plausible action tendencies, and atomic world-state constraints; a 233-sample counterfactual subset holds scene/event fixed while intervening on the agent’s emotion.
- Evaluate with three progressive tasks: Task 1 World-State Consistency (identity, spatial structure, temporal coherence), Task 2 Emotion-Action Consistency (event-emotion plausibility and emotion-action consistency), Task 3 Causal Consistency (Emotion → Action → World State constraints).
- Use a hybrid evaluation pipeline: automatic perception metrics for Task 1 and human-calibrated VLM/LLM semantic judgments for Tasks 2–3 and counterfactual emotion sensitivity.
Limitation
The authors state that the task can suffer from sparse must_not constraints: “The uniform S_not = 1.0 indicates that prohibited changes are rarely violated, but this largely reflects the scarcity of must_noto constraints in the dataset.”
Abstract (from arXiv)
The world encompasses both physical environments and human actors. Predicting its evolution requires understanding both physical laws and how emotions shape human actions. However, existing world model benchmarks primarily assess physical plausibility and perceptual quality, leaving emotion-driven world evolution largely untested.We introduce AffectWorld, a benchmark that evaluates the chain Emotion-Action-World State through three tasks: World-State Consistency, Emotion-Action Consistency, and Causal Consistency. It comprises 1,498 everyday video samples annotated with plausible emotions, action tendencies, and atomic world-state constraints. A counterfactual subset of 233 samples tests whether changing an agent's emotion under fixed scene and event conditions yields corresponding changes in actions and consequences. We define eight metrics across the three tasks and an additional counterfactual metric, combining automatic visual assessment with human-calibrated VLM/LLM judgments. Evaluations of five SoTA world models (e.g., Cosmos3) required world-state change scores of 0.100-0.280 and counterfactual emotion sensitivity scores of 0.027-0.070 on a [0, 1] scale. These results reveal substantial gaps in generating the consequences of emotion-driven actions and adapting future predictions to emotional interventions. AffectWorld provides a testbed for advancing world models that capture how emotions shape future physical and social states.
Related papers
- Reading, Writing, Sharing, and Interacting with the World through Video
- WorldOracle: Benchmarking Action-Conditioned World Models against Real-World Executions
- PrismWAM: Decoupling State Representations in World Action Models
- Foresight at the Event Boundary: Evaluating Physical Prediction in Video World Models
- MindWorldBench: Evaluating Mental-State-to-Behavior Reasoning in Image-to-Video Generation
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-13 · Indexed by Pub-AI: 2026-10-14 UTC
AffectWorld: Benchmarking Emotion-Driven Future World Prediction
Preprint posted: 2026-10-13 · Original record ↗
- arxiv
- 2610.18305
Explicit source versions (1)
- v1 · 2026-10-13
- Preprint available
Source date: 2026-10-13 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
The world encompasses both physical environments and human actors. Predicting its evolution requires understanding both physical laws and how emotions shape human actions. However, existing world model benchmarks primarily assess physical plausibility and perceptual quality, leaving emotion-driven world evolution largely untested.We introduce AffectWorld, a benchmark that evaluates the chain Emotion-Action-World State through three tasks: World-State Consistency, Emotion-Action Consistency, and Causal Consistency. It comprises 1,498 everyday video samples annotated with plausible emotions, action tendencies, and atomic world-state constraints. A counterfactual subset of 233 samples tests whether changing an agent's emotion under fixed scene and event conditions yields corresponding changes in actions and consequences. We define eight metrics across the three tasks and an additional counterfactual metric, combining automatic visual assessment with human-calibrated VLM/LLM judgments. Evaluations of five SoTA world models (e.g., Cosmos3) required world-state change scores of 0.100-0.280 and counterfactual emotion sensitivity scores of 0.027-0.070 on a [0, 1] scale. These results reveal substantial gaps in generating the consequences of emotion-driven actions and adapting future predictions to emotional interventions. AffectWorld provides a testbed for advancing world models that capture how emotions shape future physical and social states.