Skip to main content
SpringerLink
Log in

Can Machines and Humans Use Negation When Describing Images?

  • Conference paper
  • First Online:
Human and Artificial Rationalities (HAR 2023)

Part of the book series: Lecture Notes in Computer Science ((LNCS,volume 14522))

Included in the following conference series:

  • 62 Accesses

Abstract

Can negation be depicted? It has been claimed in various areas, including philosophy, cognitive science, and AI, that depicting negation through visual expressions such images and pictures is challenging. Recent empirical findings have shown that humans can indeed understand certain images as expressing negation, whereas this ability is not exhibited by machine learning models trained on image data. To elucidate the computational ability underlying the understanding of negation in images, this study first focuses on the image captioning task, specifically the performance of models pre-trained on large linguistic and image datasets for generating text from images. Our experiment demonstrates that a state-of-the-art model achieves some success in generating consistent captions from images, particularly in photographs rather than illustrations. However, when it comes to generating captions containing negation from images, the model is not as proficient as humans. To further investigate the performance of machine learning models in a more controlled setting, we conducted an additional analysis using a Visual Question Answering (VQA) task. This task enables us to specify where in the image the model should focus its attention when answering a question. As a result of this setting, the model’s performance was improved. These results will shed light on the disparities in the attentional focus between humans and machine learning models.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution
Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 59.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 79.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    https://github.com/pharmapsychotic/clip-interrogator.

  2. 2.

    https://huggingface.co/spaces/pharma/CLIP-Interrogator.

  3. 3.

    https://github.com/dandelin/vilt.

  4. 4.

    https://huggingface.co/dandelin/vilt-b32-finetuned-vqa.

References

  1. Ahn, M., et al.: Do as I can, not as I say: grounding language in robotic affordances. In: CoRL 2023, PMLR, vol. 205, pp. 287–318 (2023)

    Google Scholar 

  2. Altman, S.: The structure of primate social communication. In: Altman, S. (ed.) Social Communication Among Primates, pp. 325–362. University of Chicago Press, Chicago (1967)

    Google Scholar 

  3. Anderson, P., et al.: Vision-and-language navigation: interpreting visually-grounded navigation instructions in real environments. In: CVPR 2018, pp. 3674–3683. IEEE (2018)

    Google Scholar 

  4. Bender, E. M., Koller, A.: Climbing towards NLU: on meaning, form, and understanding in the age of data. In: ACL 2020, pp. 5185–5198 (2020)

    Google Scholar 

  5. Bernardi, R., Pezzelle, S.: Linguistic issues behind visual question answering. Lang. Linguist. Compass 15(6), elnc3.12417 (2021). https://doi.org/10.1111/lnc3.12417

  6. Cohn, N.: The Visual Language of Comics: Introduction to the Structure and Cognition of Sequential Images. Bloomsbury Academic, London (2013)

    Google Scholar 

  7. Gokhale, T., Banerjee, P., Baral, C., Yang, Y.: VQA-LOL: visual question answering under the lens of logic. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J.-M. (eds.) ECCV 2020. LNCS, vol. 12366, pp. 379–396. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-58589-1_23

    Chapter  Google Scholar 

  8. Horn, L.R.: A Natural History of Negation. University of Chicago Press, Chicago (1989)

    Google Scholar 

  9. Kim, W., Son, B., Kim, I.: ViLT: vision-and-language transformer without convolution or region supervision. In: ICML 2021. PMLR, vol. 139, pp. 5583–5594 (2021)

    Google Scholar 

  10. Li, J., Li, D., Xiong, C., Hoi, S.: BLIP: bootstrapping language-image pre-training for unified vision-language understanding and generation. In: ICML 2022. PMLR vol. 162, pp. 12888–12900 (2022)

    Google Scholar 

  11. Lin, T.-Y., et al.: Microsoft COCO: common objects in context. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) ECCV 2014. LNCS, vol. 8693, pp. 740–755. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-10602-1_48

    Chapter  Google Scholar 

  12. Manmadhan, S., Kovoor, B.C.: Visual question answering: a state-of-the-art review. Artif. Intell. Rev. 53, 5705–5745 (2020). https://doi.org/10.1007/s10462-020-09832-7

    Article  Google Scholar 

  13. Matsui, Y., et al.: Sketch-based manga retrieval using manga109 dataset. Multimed. Tools Appl. 76(20), 21811–21838 (2017). https://doi.org/10.1007/s11042-016-4020-z

    Article  Google Scholar 

  14. van Miltenburg, E., Morante, R., Elliott, D.: Pragmatic factors in image description: the case of negations. In: VL 2016, pp. 54–59. ACL (2016)

    Google Scholar 

  15. Park, D. H., Darrell, T., Rohrbach, A.: Robust change captioning. In: ICCV 2019, pp. 4624–4633, IEEE (2019)

    Google Scholar 

  16. Qiu, Y., Satoh, Y., Suzuki, R., Iwata, K., Kataoka, H.: Indoor scene change captioning based on multimodality data. Sensors 20(17), 4761 (2020). https://doi.org/10.3390/s20174761

    Article  Google Scholar 

  17. Radford, A., et al.: Learning transferable visual models from natural language supervision. In: ICML 2021. PMLR vol. 139, pp. 8748–8763 (2021)

    Google Scholar 

  18. Russell, B.: The Problems of Philosophy. Oxford University Press, Oxford (1912)

    Google Scholar 

  19. Sato, Y., Mineshima, K.: How diagrams can support syllogistic reasoning: an experimental study. J. Log. Lang. Inf. 24, 409–455 (2015). https://doi.org/10.1007/s10849-015-9225-4

    Article  MathSciNet  Google Scholar 

  20. Sato, Y., Mineshima, K.: Visually analyzing universal quantifiers in photograph captions. In: Giardino, V., Linker, S., Burns, R., Bellucci, F., Boucheix, J.M., Viana, P. (eds.) Diagrams 2022. LNCS, vol. 13462, pp. 373–377. Springer, Cham (2022). https://doi.org/10.1007/978-3-031-15146-0_34

    Chapter  Google Scholar 

  21. Sato, Y., Mineshima, K., Ueda, K.: Can negation be depicted? Comparing human and machine understanding of visual representations. Cogn. Sci. 47(3), e13258 (2023). https://doi.org/10.1111/cogs.13258

    Article  Google Scholar 

  22. Søgaard, A.: Grounding the vector space of an octopus: word meaning from raw text. Minds Mach. 33(1), 33–54 (2023). https://doi.org/10.1007/s11023-023-09622-4

    Article  Google Scholar 

  23. Yoshikawa, Y., Shigeto, Y., Takeuchi, A.: STAIR captions: constructing a large-scale Japanese image caption dataset. In: ACL 2017, pp. 417–421 (2017)

    Google Scholar 

  24. Wittgenstein, L.: Notebooks 1914-1916. In: Anscombe, G.E.M., von Wright, G.H. (eds.) University of Chicago Press, Chicago (1984). (Original Work Published 1914)

    Google Scholar 

Download references

Acknowledgements

All comic images in this paper are from the Manga-109 dataset and are licensed for use. “HighschoolKimengumi vol. 20” p. 139 \(\copyright \) Motoei Niizawa/Shueisha for illust 1 of Fig. 1, “MoeruOnisan vol19” p. 58 \(\copyright \) Tadashi Sato/Shueisha for illust 2 of Fig. 1, “OL Lunch” p. 9 \(\copyright \) Yoko Sanri/Shogakukan. Regarding photographs, they are retrieved from MS-COCO. The COCO image id is #449681 for Photo 7, #163084 for Photo 2, #51587 for Photo 3, #65737 for Photo 1.

This study was supported by Grant-in-Aid for JSPS KAKENHI Grant Number JP20K12782 and JP21K00016 as well as JST CREST Grant Number JPMJCR2114.

Author information

Authors and Affiliations

  1. Ochanomizu University, Tokyo, Japan

    Yuri Sato

  2. Keio University, Tokyo, Japan

    Koji Mineshima

Authors
  1. Yuri Sato
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Koji Mineshima
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Yuri Sato .

Editor information

Editors and Affiliations

  1. Université Paris 8, Saint Denis, France

    Jean Baratgin

  2. Université Paris 8, Saint-Denis, France

    Baptiste Jacquet

  3. Osaka City University, Osaka, Japan

    Hiroshi Yama

Appendix

Appendix

https://www.dropbox.com/scl/fi/pv5kyujtsz1a3l76kkn9t/app_HAR23.pdf?rlkey=tpgclqok7w49pqmg7cf1g9423 &dl=0.

Rights and permissions

Reprints and permissions

Copyright information

© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Check for updates. Verify currency and authenticity via CrossMark

Cite this paper

Sato, Y., Mineshima, K. (2024). Can Machines and Humans Use Negation When Describing Images?. In: Baratgin, J., Jacquet, B., Yama, H. (eds) Human and Artificial Rationalities. HAR 2023. Lecture Notes in Computer Science, vol 14522. Springer, Cham. https://doi.org/10.1007/978-3-031-55245-8_3

Download citation

  • DOI: https://doi.org/10.1007/978-3-031-55245-8_3

  • Published:

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-031-55244-1

  • Online ISBN: 978-3-031-55245-8

  • eBook Packages: Computer ScienceComputer Science (R0)

Publish with us

Policies and ethics

Search

Navigation