{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,27]],"date-time":"2026-02-27T06:17:28Z","timestamp":1772173048978,"version":"3.50.1"},"update-to":[{"DOI":"10.1371\/journal.pcbi.1009572","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2022,5,25]],"date-time":"2022-05-25T00:00:00Z","timestamp":1653436800000}}],"reference-count":68,"publisher":"Public Library of Science (PLoS)","issue":"5","license":[{"start":{"date-parts":[[2022,5,13]],"date-time":"2022-05-13T00:00:00Z","timestamp":1652400000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100000781","name":"European Research Council","doi-asserted-by":"publisher","award":["741134"],"award-info":[{"award-number":["741134"]}],"id":[{"id":"10.13039\/501100000781","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>Humans rely heavily on the shape of objects to recognise them. Recently, it has been argued that Convolutional Neural Networks (CNNs) can also show a shape-bias, provided their learning environment contains this bias. This has led to the proposal that CNNs provide good mechanistic models of shape-bias and, more generally, human visual processing. However, it is also possible that humans and CNNs show a shape-bias for very different reasons, namely, shape-bias in humans may be a consequence of architectural and cognitive constraints whereas CNNs show a shape-bias as a consequence of learning the statistics of the environment. We investigated this question by exploring shape-bias in humans and CNNs when they learn in a novel environment. We observed that, in this new environment, humans (i) focused on shape and overlooked many non-shape features, even when non-shape features were more diagnostic, (ii) learned based on only one out of multiple predictive features, and (iii) failed to learn when global features, such as shape, were absent. This behaviour contrasted with the predictions of a statistical inference model with no priors, showing the strong role that shape-bias plays in human feature selection. It also contrasted with CNNs that (i) preferred to categorise objects based on non-shape features, and (ii) increased reliance on these non-shape features as they became more predictive. This was the case even when the CNN was pre-trained to have a shape-bias and the convolutional backbone was frozen. These results suggest that shape-bias has a different source in humans and CNNs: while learning in CNNs is driven by the statistical properties of the environment, humans are highly constrained by their previous biases, which suggests that cognitive constraints play a key role in how humans learn to recognise novel objects.<\/jats:p>","DOI":"10.1371\/journal.pcbi.1009572","type":"journal-article","created":{"date-parts":[[2022,5,13]],"date-time":"2022-05-13T13:40:40Z","timestamp":1652449240000},"page":"e1009572","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":18,"title":["Feature blindness: A challenge for understanding and modelling visual object recognition"],"prefix":"10.1371","volume":"18","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6868-9655","authenticated-orcid":true,"given":"Gaurav","family":"Malhotra","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1523-227X","authenticated-orcid":true,"given":"Marin","family":"Dujmovi\u0107","sequence":"additional","affiliation":[]},{"given":"Jeffrey S.","family":"Bowers","sequence":"additional","affiliation":[]}],"member":"340","published-online":{"date-parts":[[2022,5,13]]},"reference":[{"issue":"3","key":"pcbi.1009572.ref001","doi-asserted-by":"crossref","first-page":"415","DOI":"10.1016\/j.neuron.2012.01.010","article-title":"How does the brain solve visual object recognition?","volume":"73","author":"JJ DiCarlo","year":"2012","journal-title":"Neuron"},{"issue":"8","key":"pcbi.1009572.ref002","doi-asserted-by":"crossref","first-page":"333","DOI":"10.1016\/j.tics.2007.06.010","article-title":"Untangling invariant object recognition","volume":"11","author":"JJ DiCarlo","year":"2007","journal-title":"Trends in cognitive sciences"},{"key":"pcbi.1009572.ref003","doi-asserted-by":"crossref","first-page":"403","DOI":"10.1146\/annurev-vision-091517-034103","article-title":"Invariant recognition shapes neural representations of visual input","volume":"4","author":"A Tacchetti","year":"2018","journal-title":"Annual review of vision science"},{"key":"pcbi.1009572.ref004","doi-asserted-by":"crossref","first-page":"271","DOI":"10.1146\/annurev.psych.55.090902.142005","article-title":"Object perception as Bayesian inference","volume":"55","author":"D Kersten","year":"2004","journal-title":"Annu Rev Psychol"},{"issue":"7","key":"pcbi.1009572.ref005","doi-asserted-by":"crossref","first-page":"771","DOI":"10.1016\/j.visres.2010.09.027","article-title":"Contributions of ideal observer theory to vision research","volume":"51","author":"WS Geisler","year":"2011","journal-title":"Vision research"},{"issue":"7553","key":"pcbi.1009572.ref006","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"Y LeCun","year":"2015","journal-title":"nature"},{"issue":"1","key":"pcbi.1009572.ref007","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/srep32672","article-title":"Deep networks can resemble human feed-forward vision in invariant object recognition","volume":"6","author":"SR Kheradpisheh","year":"2016","journal-title":"Scientific reports"},{"key":"pcbi.1009572.ref008","first-page":"646","article-title":"Measuring invariances in deep networks","volume":"22","author":"I Goodfellow","year":"2009","journal-title":"Advances in neural information processing systems"},{"issue":"4","key":"pcbi.1009572.ref009","doi-asserted-by":"crossref","first-page":"e1004896","DOI":"10.1371\/journal.pcbi.1004896","article-title":"Deep neural networks as a computational model for human shape sensitivity","volume":"12","author":"J Kubilius","year":"2016","journal-title":"PLoS computational biology"},{"key":"pcbi.1009572.ref010","unstructured":"Ritter S, Barrett DG, Santoro A, Botvinick MM. Cognitive psychology for deep neural networks: A shape bias case study. In: International conference on machine learning. PMLR; 2017. p. 2940\u20132949."},{"key":"pcbi.1009572.ref011","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1146\/annurev-vision-082114-035447","article-title":"Deep neural networks: a new framework for modeling biological vision and brain information processing","volume":"1","author":"N Kriegeskorte","year":"2015","journal-title":"Annual review of vision science"},{"key":"pcbi.1009572.ref012","doi-asserted-by":"crossref","first-page":"142","DOI":"10.3389\/fpsyg.2017.00142","article-title":"Perception science in the age of deep neural networks","volume":"8","author":"R VanRullen","year":"2017","journal-title":"Frontiers in psychology"},{"issue":"4","key":"pcbi.1009572.ref013","doi-asserted-by":"crossref","first-page":"305","DOI":"10.1016\/j.tics.2019.01.009","article-title":"Deep neural networks as scientific models","volume":"23","author":"RM Cichy","year":"2019","journal-title":"Trends in cognitive sciences"},{"key":"pcbi.1009572.ref014","doi-asserted-by":"crossref","first-page":"104365","DOI":"10.1016\/j.cognition.2020.104365","article-title":"Deep learning and cognitive science","volume":"203","author":"P Perconti","year":"2020","journal-title":"Cognition"},{"issue":"12","key":"pcbi.1009572.ref015","doi-asserted-by":"crossref","first-page":"e1003963","DOI":"10.1371\/journal.pcbi.1003963","article-title":"Deep neural networks rival the representation of primate IT cortex for core visual object recognition","volume":"10","author":"CF Cadieu","year":"2014","journal-title":"PLoS Comput Biol"},{"issue":"3","key":"pcbi.1009572.ref016","doi-asserted-by":"crossref","first-page":"356","DOI":"10.1038\/nn.4244","article-title":"Using goal-driven deep learning models to understand sensory cortex","volume":"19","author":"DL Yamins","year":"2016","journal-title":"Nature neuroscience"},{"key":"pcbi.1009572.ref017","first-page":"133504","article-title":"Deep neural networks in computational neuroscience","author":"TC Kietzmann","year":"2018","journal-title":"BioRxiv"},{"issue":"11","key":"pcbi.1009572.ref018","doi-asserted-by":"crossref","first-page":"1761","DOI":"10.1038\/s41593-019-0520-2","article-title":"A deep learning framework for neuroscience","volume":"22","author":"BA Richards","year":"2019","journal-title":"Nature neuroscience"},{"issue":"8","key":"pcbi.1009572.ref019","doi-asserted-by":"crossref","DOI":"10.1073\/pnas.2011417118","article-title":"An ecologically motivated image dataset for deep learning yields better models of human vision","volume":"118","author":"J Mehrer","year":"2021","journal-title":"Proceedings of the National Academy of Sciences"},{"issue":"8","key":"pcbi.1009572.ref020","doi-asserted-by":"crossref","first-page":"1798","DOI":"10.1109\/TPAMI.2013.50","article-title":"Representation learning: A review and new perspectives","volume":"35","author":"Y Bengio","year":"2013","journal-title":"IEEE transactions on pattern analysis and machine intelligence"},{"issue":"4","key":"pcbi.1009572.ref021","doi-asserted-by":"crossref","first-page":"541","DOI":"10.1162\/neco.1989.1.4.541","article-title":"Backpropagation applied to handwritten zip code recognition","volume":"1","author":"Y LeCun","year":"1989","journal-title":"Neural computation"},{"issue":"23","key":"pcbi.1009572.ref022","doi-asserted-by":"crossref","first-page":"8619","DOI":"10.1073\/pnas.1403112111","article-title":"Performance-optimized hierarchical models predict neural responses in higher visual cortex","volume":"111","author":"DL Yamins","year":"2014","journal-title":"Proceedings of the national academy of sciences"},{"issue":"5294","key":"pcbi.1009572.ref023","doi-asserted-by":"crossref","first-page":"1926","DOI":"10.1126\/science.274.5294.1926","article-title":"Statistical learning by 8-month-old infants","volume":"274","author":"JR Saffran","year":"1996","journal-title":"Science"},{"issue":"6","key":"pcbi.1009572.ref024","doi-asserted-by":"crossref","first-page":"499","DOI":"10.1111\/1467-9280.00392","article-title":"Unsupervised statistical learning of higher-order spatial structures from visual scenes","volume":"12","author":"J Fiser","year":"2001","journal-title":"Psychological science"},{"issue":"6971","key":"pcbi.1009572.ref025","doi-asserted-by":"crossref","first-page":"244","DOI":"10.1038\/nature02169","article-title":"Bayesian integration in sensorimotor learning","volume":"427","author":"KP K\u00f6rding","year":"2004","journal-title":"Nature"},{"key":"pcbi.1009572.ref026","unstructured":"Goodfellow IJ, Shlens J, Szegedy C. Explaining and harnessing adversarial examples. arXiv preprint arXiv:14126572. 2014."},{"key":"pcbi.1009572.ref027","doi-asserted-by":"crossref","unstructured":"Nguyen A, Yosinski J, Clune J. Deep neural networks are easily fooled: High confidence predictions for unrecognizable images. In: Proceedings of the IEEE conference on computer vision and pattern recognition; 2015. p. 427\u2013436.","DOI":"10.1109\/CVPR.2015.7298640"},{"key":"pcbi.1009572.ref028","doi-asserted-by":"crossref","first-page":"e55978","DOI":"10.7554\/eLife.55978","article-title":"What do adversarial images tell us about human vision?","volume":"9","author":"M Dujmovi\u0107","year":"2020","journal-title":"Elife"},{"key":"pcbi.1009572.ref029","doi-asserted-by":"crossref","unstructured":"Torralba A, Efros AA. Unbiased look at dataset bias. In: CVPR 2011. IEEE; 2011. p. 1521\u20131528.","DOI":"10.1109\/CVPR.2011.5995347"},{"key":"pcbi.1009572.ref030","doi-asserted-by":"crossref","first-page":"57","DOI":"10.1016\/j.visres.2020.04.013","article-title":"Hiding a plane with a pixel: examining shape-bias in CNNs and the benefit of building in biological constraints","volume":"174","author":"G Malhotra","year":"2020","journal-title":"Vision Research"},{"issue":"11","key":"pcbi.1009572.ref031","doi-asserted-by":"crossref","first-page":"665","DOI":"10.1038\/s42256-020-00257-z","article-title":"Shortcut learning in deep neural networks","volume":"2","author":"R Geirhos","year":"2020","journal-title":"Nature Machine Intelligence"},{"issue":"3","key":"pcbi.1009572.ref032","doi-asserted-by":"crossref","first-page":"353","DOI":"10.1016\/0010-0285(77)90012-3","article-title":"Forest before trees: The precedence of global features in visual perception","volume":"9","author":"D Navon","year":"1977","journal-title":"Cognitive psychology"},{"issue":"1","key":"pcbi.1009572.ref033","doi-asserted-by":"crossref","first-page":"38","DOI":"10.1016\/0010-0285(88)90024-2","article-title":"Surface versus edge-based determinants of visual recognition","volume":"20","author":"I Biederman","year":"1988","journal-title":"Cognitive psychology"},{"issue":"3","key":"pcbi.1009572.ref034","doi-asserted-by":"crossref","first-page":"299","DOI":"10.1016\/0885-2014(88)90014-7","article-title":"The importance of shape in early lexical learning","volume":"3","author":"B Landau","year":"1988","journal-title":"Cognitive development"},{"issue":"1","key":"pcbi.1009572.ref035","doi-asserted-by":"crossref","first-page":"13","DOI":"10.1111\/1467-9280.00403","article-title":"Object name learning provides on-the-job training for attention","volume":"13","author":"LB Smith","year":"2002","journal-title":"Psychological science"},{"key":"pcbi.1009572.ref036","doi-asserted-by":"crossref","unstructured":"Long B, Konkle T. The role of textural statistics vs. outer contours in deep CNN and neural responses to objects. In: Conference on Computational Cognitive Neuroscience; 2018. p. 4.","DOI":"10.32470\/CCN.2018.1118-0"},{"key":"pcbi.1009572.ref037","unstructured":"Geirhos R, Rubisch P, Michaelis C, Bethge M, Wichmann FA, Brendel W. ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness. arXiv preprint arXiv:181112231. 2018."},{"key":"pcbi.1009572.ref038","article-title":"The Origins and Prevalence of Texture Bias in Convolutional Neural Networks","volume":"33","author":"K Hermann","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"pcbi.1009572.ref039","unstructured":"Feinman R, Lake BM. Learning inductive biases with simple neural networks. arXiv preprint arXiv:180202745. 2018."},{"key":"pcbi.1009572.ref040","article-title":"What shapes feature representations? Exploring datasets, architectures, and training","volume":"33","author":"K Hermann","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"1","key":"pcbi.1009572.ref041","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1037\/xge0000387","article-title":"Relational discovery in category learning","volume":"147","author":"MB Goldwater","year":"2018","journal-title":"Journal of Experimental Psychology: General"},{"key":"pcbi.1009572.ref042","volume":"vol. 2","author":"S Haykin","year":"1999","journal-title":"Neural Networks: A comprehensive foundation"},{"key":"pcbi.1009572.ref043","doi-asserted-by":"crossref","first-page":"495","DOI":"10.1016\/j.neuropsychologia.2016.07.006","article-title":"Early differential sensitivity of evoked-potentials to local and global shape during the perception of three-dimensional objects","volume":"89","author":"EC Leek","year":"2016","journal-title":"Neuropsychologia"},{"issue":"1","key":"pcbi.1009572.ref044","doi-asserted-by":"crossref","first-page":"331","DOI":"10.1016\/j.neuron.2015.03.015","article-title":"Medial prefrontal cortex predicts internally driven strategy shifts","volume":"86","author":"NW Schuck","year":"2015","journal-title":"Neuron"},{"issue":"4","key":"pcbi.1009572.ref045","first-page":"1162","article-title":"When does fading enhance perceptual category learning?","volume":"39","author":"H Pashler","year":"2013","journal-title":"Journal of Experimental Psychology: Learning, Memory, and Cognition"},{"issue":"6","key":"pcbi.1009572.ref046","doi-asserted-by":"crossref","first-page":"1016","DOI":"10.1037\/0012-1649.38.6.1016","article-title":"Statistical regularities in vocabulary guide language acquisition in connectionist models and 15-20-month-olds","volume":"38","author":"LK Samuelson","year":"2002","journal-title":"Developmental psychology"},{"issue":"2","key":"pcbi.1009572.ref047","doi-asserted-by":"crossref","first-page":"195","DOI":"10.1111\/j.1467-7687.2007.00665.x","article-title":"Knowledge embedded in process: the self-organization of skilled noun learning","volume":"11","author":"E Colunga","year":"2008","journal-title":"Developmental Science"},{"key":"pcbi.1009572.ref048","unstructured":"Shah H, Tamuly K, Raghunathan A, Jain P, Netrapalli P. The pitfalls of simplicity bias in neural networks. arXiv preprint arXiv:200607710. 2020."},{"key":"pcbi.1009572.ref049","article-title":"Generalisation in humans and deep neural networks","volume":"31","author":"R Geirhos","year":"2018","journal-title":"Advances in neural information processing systems"},{"key":"pcbi.1009572.ref050","doi-asserted-by":"crossref","first-page":"99","DOI":"10.1146\/annurev-neuro-072116-031526","article-title":"Toward a rational and mechanistic account of mental effort","volume":"40","author":"A Shenhav","year":"2017","journal-title":"Annual review of neuroscience"},{"issue":"12","key":"pcbi.1009572.ref051","doi-asserted-by":"crossref","first-page":"899","DOI":"10.1038\/s41562-018-0401-9","article-title":"Mental labour","volume":"2","author":"W Kool","year":"2018","journal-title":"Nature human behaviour"},{"key":"pcbi.1009572.ref052","unstructured":"Zhang C, Bengio S, Hardt M, Recht B, Vinyals O. Understanding deep learning requires rethinking generalization. arXiv preprint arXiv:161103530. 2016."},{"key":"pcbi.1009572.ref053","unstructured":"Tsvetkov C, Malhotra G, Evans BD, Bowers JS. Adding biological constraints to deep neural networks reduces their capacity to learn unstructured data. In: Proceedings of the 42nd Annual Conference of the Cognitive Science Society; 2020."},{"issue":"4","key":"pcbi.1009572.ref054","doi-asserted-by":"crossref","first-page":"R136","DOI":"10.1016\/j.cub.2009.12.014","article-title":"Working memory","volume":"20","author":"A Baddeley","year":"2010","journal-title":"Current biology"},{"issue":"4","key":"pcbi.1009572.ref055","doi-asserted-by":"crossref","first-page":"160","DOI":"10.1016\/j.tics.2011.02.005","article-title":"Visual crowding: A fundamental limit on conscious perception and object recognition","volume":"15","author":"D Whitney","year":"2011","journal-title":"Trends in cognitive sciences"},{"issue":"10","key":"pcbi.1009572.ref056","doi-asserted-by":"crossref","first-page":"13","DOI":"10.1167\/12.10.13","article-title":"Grouping, pooling, and when bigger is better in visual crowding","volume":"12","author":"M Manassi","year":"2012","journal-title":"Journal of Vision"},{"issue":"5","key":"pcbi.1009572.ref057","doi-asserted-by":"crossref","first-page":"180","DOI":"10.1111\/1467-8721.01256","article-title":"Inattentional blindness: Looking without seeing","volume":"12","author":"A Mack","year":"2003","journal-title":"Current Directions in Psychological Science"},{"key":"pcbi.1009572.ref058","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1016\/j.conb.2019.06.001","article-title":"A new framework for understanding vision from the perspective of the primary visual cortex","volume":"58","author":"L Zhaoping","year":"2019","journal-title":"Current opinion in neurobiology"},{"key":"pcbi.1009572.ref059","doi-asserted-by":"crossref","first-page":"399","DOI":"10.1146\/annurev-vision-091718-014951","article-title":"Deep learning: the good, the bad, and the ugly","volume":"5","author":"T Serre","year":"2019","journal-title":"Annual review of vision science"},{"issue":"8","key":"pcbi.1009572.ref060","doi-asserted-by":"crossref","first-page":"19","DOI":"10.1167\/jov.21.8.19","article-title":"Controlling the spatial dimensions of visual stimuli in online experiments","volume":"21","author":"JW Brascamp","year":"2021","journal-title":"Journal of Vision"},{"key":"pcbi.1009572.ref061","doi-asserted-by":"crossref","unstructured":"He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition; 2016. p. 770\u2013778.","DOI":"10.1109\/CVPR.2016.90"},{"key":"pcbi.1009572.ref062","unstructured":"Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:14091556. 2014."},{"key":"pcbi.1009572.ref063","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","volume":"25","author":"A Krizhevsky","year":"2012","journal-title":"Advances in neural information processing systems"},{"key":"pcbi.1009572.ref064","unstructured":"Yosinski J, Clune J, Bengio Y, Lipson H. How transferable are features in deep neural networks? arXiv preprint arXiv:14111792. 2014."},{"key":"pcbi.1009572.ref065","unstructured":"Kingma DP, Ba J. Adam: A method for stochastic optimization. arXiv preprint arXiv:14126980. 2014."},{"key":"pcbi.1009572.ref066","unstructured":"Paszke A, Gross S, Chintala S, Chanan G, Yang E, DeVito Z, et al. Automatic differentiation in PyTorch. 31st Conference on Neural Information Processing Systems. 2017."},{"key":"pcbi.1009572.ref067","volume-title":"Pattern recognition and machine learning","author":"CM Bishop","year":"2006"},{"key":"pcbi.1009572.ref068","volume-title":"Machine learning: a probabilistic perspective","author":"KP Murphy","year":"2012"}],"updated-by":[{"DOI":"10.1371\/journal.pcbi.1009572","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2022,5,25]],"date-time":"2022-05-25T00:00:00Z","timestamp":1653436800000}}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009572","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,5,25]],"date-time":"2022-05-25T13:47:19Z","timestamp":1653486439000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009572"}},"subtitle":[],"editor":[{"given":"Blake A","family":"Richards","sequence":"first","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2022,5,13]]},"references-count":68,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2022,5,13]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1009572","relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/2021.10.20.465074","asserted-by":"object"}]},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,5,13]]}}}