# Counterfactual Tests for Measuring Chain-of-Thought Faithfulness in Visual Language Models

arXiv: https://arxiv.org/abs/2609.06704 · HTML: https://mlsanity.com/paper/2609.06704/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 18 of its 48 references that are in the map.
- First posted on arXiv 6 Sep 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 18 of 18)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Neural Machine Translation by Jointly Learning to Align and Translate](https://mlsanity.com/paper/1409.0473/) (2014)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [Qwen-Image Technical Report](https://mlsanity.com/paper/2508.02324/) (2025)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Reasoning Models Don't Always Say What They Think](https://mlsanity.com/paper/2505.05410/) (2025)
- [Kimi-VL Technical Report](https://mlsanity.com/paper/2504.07491/) (2025)
- [Visual Entailment: A Novel Task for Fine-Grained Image Understanding](https://mlsanity.com/paper/1901.06706/) (2019)
- [Ovis2.5 Technical Report](https://mlsanity.com/paper/2508.11737/) (2025)
- [Are DeepSeek R1 And Other Reasoning Models More Faithful?](https://mlsanity.com/paper/2501.08156/) (2025)
- [R-4B: Incentivizing General-Purpose Auto-Thinking Capability in MLLMs via Bi-Mode Annealing and Reinforce Learning](https://mlsanity.com/paper/2508.21113/) (2025)
- [On the Faithfulness of Visual Thinking: Measurement and Enhancement](https://mlsanity.com/paper/2510.23482/) (2025)
- [Explanation-Driven Counterfactual Testing for Faithfulness in Vision-Language Model Explanations](https://mlsanity.com/paper/2510.00047/) (2025)
- [Verbosity Tradeoffs and the Impact of Scale on the Faithfulness of LLM Self-Explanations](https://mlsanity.com/paper/2503.13445/) (2025)
- [VFaith: Do Large Multimodal Models Really Reason on Seen Images Rather than Previous Memories?](https://mlsanity.com/paper/2506.11571/) (2025)

## Abstract (arXiv)

Chain-of-thought (CoT) may often look plausible, yet it may not faithfully reflect the model's decision-making process. While methods for measuring the faithfulness of CoTs for textual inputs have been increasingly introduced, using these methods for visual inputs is not straightforward. In this work, we adapt the family of counterfactual methods for measuring CoT faithfulness, namely the Counterfactual Test (CT) and Correlational Counterfactual Test (CCT), to visual inputs, and call them vCT and vCCT, respectively. Using vCT and vCCT, we benchmark eight recent open-source Vision Language Models (VLMs) on two datasets. Our analysis shows that CoTs do not reliably track visual evidence that influences model predictions: they may omit the removed object even when its removal causes a large prediction shift, yet mention it when the shift is small. We further find that Predict-then-Explain explanations align more strongly with perturbation-induced probability shifts than pre-answer CoTs, while binary vCT scores are often nearly saturated. We also include a reconstruction control, in which images pass through the same editing pipeline without object removal, and find that the main object-removal intervention induces larger shifts than reconstruction alone. We construct and release Counter-SNLI-VE and Counter-A-OKVQA, two datasets of image pairs that differ by a single object.