mlsanityData to 7 Oct 2026

Are Deep Neural Networks Adequate Behavioural Models of Human Visual Perception?

arXiv 2305.17023 ↗First posted 26 May 2023cs.CV cs.AI cs.LGcited by 10

Not on the map. Why?

Abstract (arXiv)

Deep neural networks (DNNs) are machine learning algorithms that have revolutionised computer vision due to their remarkable successes in tasks like object classification and segmentation. The success of DNNs as computer vision algorithms has led to the suggestion that DNNs may also be good models of human visual perception. We here review evidence regarding current DNNs as adequate behavioural models of human core object recognition. To this end, we argue that it is important to distinguish between statistical tools and computational models, and to understand model quality as a multidimensional concept where clarity about modelling goals is key. Reviewing a large number of psychophysical and computational explorations of core object recognition performance in humans and DNNs, we argue that DNNs are highly valuable scientific tools but that as of today DNNs should only be regarded as promising -- but not yet adequate -- computational models of human core object recognition behaviour. On the way we dispel a number of myths surrounding DNNs in vision science.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Builds on 49 of its 145 references that are in the map. First posted on arXiv 26 May 2023.

Builds on

49 of 145 references.

References · 49

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  2. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  3. 2013Intriguing properties of neural networkscited by 5,114
  4. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,590
  5. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  6. 2017Arbitrary Style Transfer in Real-time with Adaptive Instance Normalizationcited by 1,998
  7. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  8. 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,842
  9. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,507
  10. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  11. 2019Natural Adversarial Examplescited by 1,401
  12. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
37 more
  1. 2018Sanity Checks for Saliency Mapscited by 1,149
  2. 2014Deep Learning in Neural Networks: An Overviewcited by 765
  3. 2023Scaling Vision Transformers to 22 Billion Parameterscited by 690
  4. 2017Dynamic Routing Between Capsulescited by 689
  5. 2020RobustBench: a standardized adversarial robustness benchmarkcited by 659
  6. 2017The (Un)reliability of saliency methodscited by 331
  7. 2018Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problemcited by 323
  8. 2019Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNetcited by 260
  9. 2015Explaining NonLinear Classification Decisions with Deep Taylor Decompositioncited by 248
  10. 2018Generalisation in humans and deep neural networkscited by 237
  11. 2018Deep Learning: A Critical Appraisalcited by 214
  12. 2018Markerless tracking of user-defined features with deep learningcited by 203
  13. 2021Partial success in closing the gap between human and machine visioncited by 185
  14. 2018Strike (with) a Pose: Neural Networks Are Easily Fooled by Strange Poses of Familiar Objectscited by 143
  15. 2022Revisiting Weakly Supervised Pre-Training of Visual Perception Modelscited by 116
  16. 2019Brain-Like Object Recognition with High-Performing Shallow Recurrent ANNscited by 90
  17. 2019Stacked Capsule Autoencoderscited by 82
  18. 2020Beyond accuracy: quantifying trial-by-trial behaviour of CNNs and humans by measuring error consistencycited by 77
  19. 2021Real-time gravitational-wave science with neural posterior estimationcited by 76
  20. 2018Adversarial Examples that Fool both Computer Vision and Time-Limited Humanscited by 73
  21. 2019Recurrence is required to capture the representational dynamics of the human visual systemcited by 58
  22. 2021Neural population geometry: An approach for understanding biological and artificial neural networkscited by 52
  23. 2022ViewFool: Evaluating the Robustness of Visual Recognition to Adversarial Viewpointscited by 48
  24. 2022ImageNet-X: Understanding Model Mistakes with Factor of Variation Annotationscited by 44
  25. 2016Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortexcited by 42
  26. 2021Trivial or impossible -- dichotomous data difficulty masks model differences (on ImageNet and beyond)cited by 35
  27. 2018Humans can decipher adversarial imagescited by 31
  28. 2021How Well do Feature Visualizations Support Causal Understanding of CNN Activations?cited by 28
  29. 2017Eigen-Distortions of Hierarchical Representationscited by 25
  30. 2018Solving Bongard Problems with a Visual Language and Pragmatic Reasoningcited by 21
  31. 2021Capturing the objects of vision with neural networkscited by 21
  32. 2020Going in circles is the way forward: the role of recurrence in visual inferencecited by 19
  33. 2022The Robustness Limits of SoTA Vision Models to Natural Variationcited by 15
  34. 2022Progress and limitations of deep networks to recognize objects in unusual posescited by 13
  35. 2020A neural network walks into a lab: towards using deep nets as models for human behaviorcited by 10
  36. 2022The developmental trajectory of object recognition robustness: children are like small adults but unlike big deep neural networkscited by 9
  37. 2020Are there any 'object detectors' in the hidden layers of CNNs trained to identify objects or scenes?cited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 3; 2024: 2; 2025: 4; 2026: 1.

10 papers.

  1. 2023Neither hype nor gloom do DNNs justicecited by 40
  2. 2023Visual cognition in multimodal large language modelscited by 38
  3. 2024How Aligned are Different Alignment Metrics?cited by 6
  4. 2025Adopting a human developmental visual diet yields robust, shape-based AI visioncited by 5
  5. 2024RTify: Aligning Deep Neural Networks with Human Behavioral Decisionscited by 3
  6. 2026Perceptual misalignment of texture representations in convolutional neural networkscited by 2
  7. 2025Seeing Eye to AI? Applying Deep-Feature-Based Similarity Metrics to Information Visualizationcited by 1
  8. 2025A Rigorous Behavior Assessment of CNNs Using a Data-Domain Sampling Regimecited by 1
  9. 2023Fixing the problems of deep neural networks will require better training data and learning algorithmscited by 0
  10. 2025A Causal Framework for Aligning Image Quality Metrics and Deep Neural Network Robustnesscited by 0