dingxingdi Skill: causal-and-counterfactual-video-reasoning
1. Capability Definition & Real Case
- Professional Definition: The capability to infer why an event happened and what would happen under a modified condition, by combining observed dynamics with causal structure rather than relying on surface co-occurrence.
- Dimension Hierarchy: Long-Horizon Reasoning->Inference and Planning->causal-and-counterfactual-video-reasoning
Real Case
[Case 1]
- Initial Environment: A football match video with repeated player interactions and a physically contested dribble. The question centers on one contact event between two players during active play.
- Real Question: Why did Player number 4 in white push down Player number 17 in purple during the match?
- Real Trajectory: Track the offensive possession before contact, identify the attempted dribble by Player 17, examine Player 4's defensive movement and timing, and infer the motive from the tactical context rather than the collision frame alone.
- Real Answer: Player number 4 in white was making a defensive play to prevent Player number 17 in purple from successfully dribbling past him.
- Why this demonstrates the capability: The answer is causal, not merely descriptive: the agent must explain the contact by reasoning about the prior offensive setup and the defender's goal. This is precisely the kind of higher-order inference that goes beyond naming visible actions.
Pipeline Execution Instructions
To synthesize data for this capability, you must strictly follow a 3-phase pipeline. Do not hallucinate steps. Read the corresponding reference file for each phase sequentially:
Phase 1: Environment Exploration Read the exploration guidelines to discover raw knowledge seeds:
references/EXPLORATION.mdPhase 2: Trajectory Selection Once Phase 1 is complete, read the selection criteria to evaluate the trajectory:
references/SELECTION.mdPhase 3: Data Synthesis Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data:
references/SYNTHESIS.md