Elamur Research⌕
☰
home / papers / 2026 / On the Cultural Anachronism and Temporal Reasoning in Vision Language Models

On the Cultural Anachronism and Temporal Reasoning in Vision Language Models

Paper · 2026 · Mukul Ranjan, Prince Jha, Khushboo Kumari et al.

Summary

Vision-language models are increasingly used in museums, digital archives and educational tools. The authors point to a failure they call cultural anachronism: a model explains a historical object through concepts, materials, techniques or cultural frames from the wrong era. To measure it, they built TAB-VLM, a benchmark on Indian cultural heritage covering artifacts from prehistoric to modern times. They gathered roughly 220,000 candidate artifacts from online sources, sorted them into eight historical periods by hand, and used expert validation and balancing to reduce them to 1,600 items. These items support 600 multiple-choice questions built from task-specific prompt templates. Ten proprietary and open models were tested on images alone. All of them performed poorly. The best, GPT-5.2, answered just under three-fifths of the questions correctly, and ordering artifacts in time was especially weak. Larger models and different architectures did not close the gap. The authors therefore argue that the problem comes from how models are trained and cannot be fixed by scaling alone. This matters for any heritage application, and especially for non-Western cultures that are poorly represented in training data.

Key points

  • TAB-VLM has 600 questions split evenly across six tasks (100 each):
    • Chronological Sequencing: order 4 artifacts out of 24 possible permutations.
    • The Odd-One-Out Period task
    • The Material Availability task
    • The Manufacturing Technique task
    • The Period Grouping task
    • The Style Period Attribution task
  • The dataset pipeline went from about 220,000 collected artifacts to 1,600 curated ones, classified into 8 periods with expert checking and duplicate removal.
  • Ten models were evaluated, including GPT-5.2, GPT-4o, GPT-4o-mini and the Qwen2-VL, Qwen2.5-VL and InternVL3 series. The top overall score was 58.7% (GPT-5.2), and no model exceeded about 38% on chronological sequencing.
  • Suggested remedies:
    • training data with time annotations from diverse cultures
    • contrastive objectives that separate neighbouring periods
    • checking for anachronism before deployment
  • Stated limitations:
    • only Indian artifacts are covered
    • no textual metadata is given to the models
    • period labels are discrete, so transitional pieces fit poorly
    • only a subset of available models was tested

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Mohamed bin Zayed University of Artificial IntelligenceAbu Dhabi · United Arab Emirates124 works · 5,676 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 3Research subjects and languages the work is about.Language models & pretraining · 3,320 worksReasoning · 1,426 worksVision & language · 657 works

Cite

Cite as Mukul Ranjan, Prince Jha, Khushboo Kumari and Zhiqiang Shen. 2026. On the Cultural Anachronism and Temporal Reasoning in Vision Language Models. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1514

@inproceedings{ranjan2026cultural,
  title = {{On the Cultural Anachronism and Temporal Reasoning in Vision Language Models}},
  author = {Mukul Ranjan, Prince Jha, Khushboo Kumari and Zhiqiang Shen},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1514},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1514}
}

References

43 references, 16 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Yi Fung, Ruining Zhao, Jae Doo, Chenkai Sun, and Heng Ji. 2024. Massively multi-cultural knowledge acquisition & lm benchmarking. arXiv preprint arXiv:2402.09369.
  2. Muhammad Farid Adilazuarda, Sagnik Mukherjee, Pradhyumna Lavania, Siddhant Singh, Alham Fikri Aji, Jacki O’Neill, Ashutosh Modi, and Monojit Choudhury. 2024. Towards measuring and modeling" culture" in llms: A survey. arXiv preprint arXiv:2403.15412.
  3. Isabel O Gallegos, Ryan A Rossi, Joe Barrow, Md Mehrab Tanjim, Sungchul Kim, Franck Dernoncourt, Tong Yu, Ruiyi Zhang, and Nesreen K Ahmed. 2024. Bias and fairness in large language models: A survey. Computational Linguistics, 50(3):1097– 1179.
  4. Sara Ghaboura, Ketan More, Ritesh Thawkar, Wafa Alghallabi, Omkar Thawakar, Fahad Shahbaz Khan, Hisham Cholakkal, Salman Khan, and Rao Muhammad Anwer. 2025. Time travel: A comprehensive benchmark to evaluate lmms on historical and cultural artifacts. arXiv preprint arXiv:2502.14865. find
  5. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  6. EunJeong Hwang, Bodhisattwa Prasad Majumder, and Niket Tandon. 2023. Aligning language models to user opinions. arXiv preprint arXiv:2305.14929.
  7. Hyerin Hwang, Chan-Woo Park, Hee-Kwon Kim, and Jae-Ho Lee. 2025. Cats: cultural-heritage classification using llms and distribute model. npj Heritage Science, 13(1):76. find
  8. Mohamed Fazli Imam, Chenyang Lyu, and Alham Fikri Aji. 2025. Can multimodal llms do visual temporal understanding and reasoning? the answer is no! arXiv preprint arXiv:2501.10674. — [arXiv:2501.10674](https://arxiv.org/abs/2501.10674) find
  9. Raghav Jain, Daivik Sojitra, Arkadeep Acharya, Sriparna Saha, Adam Jatowt, and Sandipan Dandapat. 2023. Do language models have a common sense regarding time? revisiting temporal commonsense reasoning in the era of large language models. In Proceedings of the 2023 Conference on Empirical Methods in… find
  10. Nithish Kannen, Arif Ahmad, Marco Andreetto, Vinodkumar Prabhakaran, Utsav Prabhu, Adji Bousso Dieng, Pushpak Bhattacharyya, and Shachi Dave. 2024. Beyond aesthetics: Cultural competence in text-to-image models. arXiv preprint arXiv:2407.06863. — [arXiv:2407.06863](https://arxiv.org/abs/2407.06863) find
  11. Julia Kharchenko, Tanya Roosta, Aman Chadha, and Chirag Shah. 2024. How well do llms represent values across cultures? empirical analysis of llm responses based on hofstede cultural dimensions. arXiv preprint arXiv:2406.14805. — [arXiv:2406.14805](https://arxiv.org/abs/2406.14805) find
  12. Dohwan Ko, Ji Soo Lee, Wooyoung Kang, Byungseok Roh, and Hyunwoo J Kim. 2023. Large language models are temporal and causal reasoners for video question answering. arXiv preprint arXiv:2310.15747. find
  13. Cheng Li, Damien Teney, Linyi Yang, Qingsong Wen, Xing Xie, and Jindong Wang. 2024b. Culturepark: Boosting cross-cultural understanding in large language models. arXiv preprint arXiv:2405.15145. find
  14. Zongxia Li, Xiyang Wu, Hongyang Du, Fuxiao Liu, Huy Nghiem, and Guangyao Shi. 2025. A survey of state of the art large vision language models: Alignment, benchmark, evaluations and challenges. arXiv preprint arXiv:2501.02189. — [arXiv:2501.02189](https://arxiv.org/abs/2501.02189) find
  15. Percy Liang, Rishi Bommasani, Tony Lee, Dimitris Tsipras, Dilara Soylu, Michihiro Yasunaga, Yian Zhang, Deepak Narayanan, Yuhuai Wu, Ananya Kumar, and 1 others. 2022. Holistic evaluation of language models. arXiv preprint arXiv:2211.09110. find
  16. Chen Cecilia Liu, Fajri Koto, Timothy Baldwin, and Iryna Gurevych. 2023a. Are multilingual llms culturally-diverse reasoners? an investigation into multicultural proverbs and sayings. arXiv preprint arXiv:2309.08591. find
  17. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024. Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306.
  18. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023b. Visual instruction tuning. Advances in neural information processing systems, 36:34892– 34916.
  19. Shudong Liu, Yiqiao Jin, Cheng Li, Derek F Wong, Qingsong Wen, Lichao Sun, Haipeng Chen, Xing Xie, and Jindong Wang. 2025. Culturevlm: Characterizing and improving cultural understanding of vision-language models for over 100 countries. arXiv preprint arXiv:2501.01282. — [arXiv:2501.01282](https://arxiv.org/abs/2501.01282) find
  20. John McLeod. 2015. The History of India. Bloomsbury Publishing USA. find
  21. Shravan Nayak, Kanishk Jain, Rabiul Awal, Siva Reddy, Sjoerd van Steenkiste, Lisa Anne Hendricks, Karolina Stańczak, and Aishwarya Agrawal. 2024. Benchmarking vision language models for cultural understanding. arXiv preprint arXiv:2407.10920.
  22. OpenAI. 2025a. Gpt-5 system card. Published Aug 13, 2025; system card for the GPT-5 model family. find
  23. J Richard Landis and Gary G Koch. 1977. The measurement of observer agreement for categorical data. biometrics, pages 159–174. find
  24. OpenAI. 2025b. Update to gpt-5 system card: Gpt-5.2. Published Dec 11, 2025; update describing GPT-5.2 model family. find
  25. Cheng Li, Mengzhuo Chen, Jindong Wang, Sunayana Sitaram, and Xing Xie. 2024a. Culturellm: Incorporating cultural differences into large language models. Advances in Neural Information Processing Systems, 37:84799–84838.
  26. Piotr Padlewski, Max Bain, Matthew Henderson, Zhongkai Zhu, Nishant Relan, Hai Pham, Donovan Ong, and 1 others. 2024. Vibe-eval: A hard evaluation suite for measuring progress of multimodal language models. arXiv preprint arXiv:2405.02287.
  27. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, and 1 others. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, pages 8748–8…
  28. Abhinav Rao, Akhila Yerukola, Vishwa Shah, Katharina Reinecke, and Maarten Sap. 2024. Normad: A benchmark for measuring the cultural adaptability of large language models. arXiv preprint arXiv:2404.12464. — [arXiv:2404.12464](https://arxiv.org/abs/2404.12464) find
  29. Sahapedia. 2025. Museums of india. https://www.museumsofindia.org/. Accessed: 2025-08-02. ↗
  30. Nithya Sambasivan, Erin Arnesen, Ben Hutchinson, Tulsee Doshi, and Vinodkumar Prabhakaran. 2021. Re-imagining algorithmic fairness in india and beyond. In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency, pages 315–328. find
  31. Iryna Siliutina, Olena Tytar, Marina Barbash, Nataliia Petrenko, and Larysa Yepyk. 2024. Cultural preservation and digital heritage: challenges and opportunities. Amazonia Investiga, 13(75):262–273. find
  32. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clément Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, and Joe Brew. 2019. Huggingface’s transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771.
  33. Jiahui Yu, Zirui Wang, Vijay Vasudevan, Legg Yeung, Mojtaba Seyedhosseini, and Yonghui Wu. 2022. Coca: Contrastive captioners are image-text foundation models. arXiv preprint arXiv:2205.01917.
  34. Chenhan Yuan, Qianqian Xie, Jimin Huang, and Sophia Ananiadou. 2024. Back to the future: Towards explainable temporal reasoning with large language models. In Proceedings of the ACM on Web Conference 2024, pages 1963–1974. find
  35. Jinguo Zhu, Weiyun Wang, Zhe Chen, Zhaoyang Liu, Shenglong Ye, Lixin Gu, Hao Tian, and 1 others. 2025. Internvl3: Exploring advanced training and test-time recipes for open-source multimodal models. arXiv preprint arXiv:2504.10479.
  36. Zhaochen Su, Jun Zhang, Tong Zhu, Xiaoye Qu, Juntao Li, Min Zhang, and Yu Cheng. 2024. Timo: Towards better temporal reasoning for language models. arXiv preprint arXiv:2406.14192.
  37. Nanna Bonde Thylstrup. 2022. The ethics and politics of data sets in the age of machine learning: Deleting traces and encountering remains. Media, Culture & Society, 44(4):655–671. find
  38. Georgios Trichopoulos. 2023. Large language models for cultural heritage. In Proceedings of the 2nd International Conference of the ACM Greek SIGCHI Chapter, pages 1–5. find
  39. Ted Underwood, Laura K Nelson, and Matthew Wilkens. 2025. Can language models represent the past without anachronism? arXiv preprint arXiv:2505.00030. — [arXiv:2505.00030](https://arxiv.org/abs/2505.00030) find
  40. Ujjwal Upadhyay, Mukul Ranjan, Zhiqiang Shen, and Mohamed Elhoseiny. 2025. Time blindness: Why video-language models can’t see what humans can? arXiv preprint arXiv:2505.24867. — [arXiv:2505.24867](https://arxiv.org/abs/2505.24867) find
  41. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, and 1 others. 2024. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint arXiv:2409.12191.
  42. Yuqing Wang and Yun Zhao. 2023. Tram: Benchmarking temporal reasoning for large language models. arXiv preprint arXiv:2310.00835. find
  43. Zeqing Wang, Shiyuan Zhang, Chengpei Tang, and Keze Wang. 2025. Timecausality: Evaluating the causal ability in time dimension for vision language models. arXiv preprint arXiv:2505.15435. — [arXiv:2505.15435](https://arxiv.org/abs/2505.15435) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ranjan-2026-cultural-anachronism-temporal-reasoning-vision/