Pith. sign in

REVIEW 15 cited by

CausalVQA: A Physically Grounded Causal Reasoning Benchmark for Video Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.09943 v1 pith:IS4B7SZS submitted 2025-06-11 cs.CV cs.AI

CausalVQA: A Physically Grounded Causal Reasoning Benchmark for Video Models

classification cs.CV cs.AI
keywords modelsunderstandingbenchmarkcausalvqaphysicalquestionsreal-worldreasoning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We introduce CausalVQA, a benchmark dataset for video question answering (VQA) composed of question-answer pairs that probe models' understanding of causality in the physical world. Existing VQA benchmarks either tend to focus on surface perceptual understanding of real-world videos, or on narrow physical reasoning questions created using simulation environments. CausalVQA fills an important gap by presenting challenging questions that are grounded in real-world scenarios, while focusing on models' ability to predict the likely outcomes of different actions and events through five question types: counterfactual, hypothetical, anticipation, planning and descriptive. We designed quality control mechanisms that prevent models from exploiting trivial shortcuts, requiring models to base their answers on deep visual understanding instead of linguistic cues. We find that current frontier multimodal models fall substantially below human performance on the benchmark, especially on anticipation and hypothetical questions. This highlights a challenge for current systems to leverage spatial-temporal reasoning, understanding of physical principles, and comprehension of possible alternatives to make accurate predictions in real-world settings.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 15 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. RetroHolmes: When Semantic Plausibility Fails Retrospective Physical Process Reasoning

    cs.MM 2026-07 conditional novelty 7.0

    Current VLMs fail retrospective physical reachability and causal reconstruction on RetroHolmes; a simple analysis-by-synthesis loop with video simulation reduces bias and belief-conflict sensitivity.

  2. PhysEditWorld: A Large-Scale Dataset Toward Physics-Editable World Models

    cs.CV 2026-06 unverdicted novelty 7.0

    PhysEditWorld is a new dataset of over 60 million frames from 12 UE5 cinematic scenes with synchronized multimodal signals and explicit gravity labels, built via replay to support physics-editable world models.

  3. YoCausal: How Far is Video Generation from World Model? A Causality Perspective

    cs.CV 2026-05 unverdicted novelty 7.0

    YoCausal benchmark shows video diffusion models detect the arrow of time but lack genuine causal understanding relative to humans.

  4. What-If World: A Causal Benchmark for General World Models in Embodied Scenarios

    cs.CV 2026-05 unverdicted novelty 7.0

    What-If World is a new paired-prompt benchmark showing that nine state-of-the-art video generation models achieve at most 52% on causal intervention tests and cluster near 28% for open-source systems.

  5. CRONOS: Benchmarking Counterfactual Physical Consistency in Video Models

    cs.CV 2026-05 unverdicted novelty 7.0

    CRONOS benchmark shows recent open-source video generators fail to preserve physical consistency under controlled changes to viewpoint, scene, object category, and appearance.

  6. CaST-Bench: Benchmarking Causal Chain-Grounded Spatio-Temporal Reasoning for Video Question Answering

    cs.CV 2026-05 unverdicted novelty 7.0

    CaST-Bench creates a benchmark with causal-chain annotations and novel metrics showing that current VLMs struggle to construct precise grounded causal chains in video QA.

  7. CaST-Bench: Benchmarking Causal Chain-Grounded Spatio-Temporal Reasoning for Video Question Answering

    cs.CV 2026-05 unverdicted novelty 7.0

    Introduces CaST-Bench, a dataset of 2,066 causal questions on 1,015 videos with annotated causal chains and metrics to evaluate VLMs on spatio-temporal causal reasoning.

  8. Act2See: Emergent Active Visual Perception for Video Reasoning

    cs.CV 2026-05 unverdicted novelty 7.0

    Act2See trains VLMs via supervised fine-tuning on verified reasoning traces to interleave active frame calls within text CoTs, yielding SOTA results on video reasoning benchmarks.

  9. SCP: Spatial Causal Prediction in Video

    cs.CV 2026-03 unverdicted novelty 7.0

    SCP defines a new benchmark task for predicting spatial causal outcomes beyond direct observation and shows that 23 leading models lag far behind humans on it.

  10. Causal Scaffolding for Physical Reasoning: A Benchmark for Causally-Informed Physical World Understanding in VLMs

    cs.DB 2026-06 unverdicted novelty 6.0

    Introduces CausalPhys benchmark with causal graphs and CRFT fine-tuning to improve VLMs' causal physical reasoning accuracy and interpretability.

  11. Cosmos 3: Omnimodal World Models for Physical AI

    cs.CV 2026-06 unverdicted novelty 6.0

    Cosmos 3 presents a unified omnimodal world model family based on mixture-of-transformers that processes language, vision, audio, and action for Physical AI applications.

  12. PhysEditWorld: A Large-Scale Dataset Toward Physics-Editable World Models

    cs.CV 2026-06 unverdicted novelty 5.0

    PhysEditWorld supplies 12 UE5 scenes, 60+ million frames, and explicit gravity labels via a replay paradigm to support gravity-faithful and physically editable world models.

  13. How Should World Models Be Evaluated for Embodied Decision-Making? A Decision-Making-Centric Position

    cs.LG 2026-06 unverdicted novelty 5.0

    The paper proposes an L0-L7 evidential ladder for evaluating world models in embodied decision-making, prioritizing interventional action fidelity and policy optimization utility over visual plausibility.

  14. OpenWorldLib: A Unified Codebase and Definition of Advanced World Models

    cs.CV 2026-04 conditional novelty 4.0

    OpenWorldLib defines world models as perception-centered systems with interaction and long-term memory, and provides a modular inference codebase unifying interactive video, 3D, reasoning, and VLA tasks.

  15. OpenWorldLib: A Unified Codebase and Definition of Advanced World Models

    cs.CV 2026-04 unverdicted novelty 4.0

    OpenWorldLib offers a standardized codebase and definition for world models that combine perception, interaction, and memory to understand and predict the world.