{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,27]],"date-time":"2026-02-27T21:04:52Z","timestamp":1772226292785,"version":"3.50.1"},"reference-count":27,"publisher":"MIT Press","issue":"3","content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,2,27]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Object detection and recognition are fundamental functions that play a significant role in the success of species. Because the appearance of an object exhibits large variability, the brain has to group these different stimuli under the same object identity, a process of generalization. Does the process of generalization follow some general principles, or is it an ad hoc bag of tricks? The universal law of generalization (ULoG) provides evidence that generalization follows similar properties across a variety of species and tasks. Here, we tested the hypothesis derived from ULoG that the internal representations underlying generalization reflect the natural properties of object detection and recognition in our environment rather than the specifics of the system solving these problems. Neural networks with universal-approximation capability have been successful in many object detection and recognition tasks; however, how these networks reach their decisions remains opaque. To provide a strong test for ecological validity, we used natural camouflage, which is nature's test bed for object detection and recognition. We trained a deep neural network with natural images of \u201cclear\u201d and \u201ccamouflaged\u201d animals and examined the emerging internal representations. We extended ULoG to a realistic learning regime, with multiple consequential stimuli, and developed two methods to determine category prototypes. Our results show that with a proper choice of category prototypes, the generalization functions are monotone decreasing, similar to the generalization functions of biological systems. Critically, we show that camouflaged inputs are not represented randomly but rather systematically appear at the tail of the monotone decreasing functions. Our results support the hypothesis that the internal representations underlying generalization in object detection and recognition are shaped mainly by the properties of the ecological environment, even though different biological and artificial systems may generate these internal representations through drastically different learning and adaptation processes. Furthermore, the extended version of ULoG provides a tool to analyze how the system organizes its internal representations during learning as well as how it makes its decisions.<\/jats:p>","DOI":"10.1162\/neco.a.1483","type":"journal-article","created":{"date-parts":[[2026,2,4]],"date-time":"2026-02-04T21:30:59Z","timestamp":1770240659000},"page":"328-372","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":0,"title":["Object Detection, Recognition, Deep Learning, and the Universal Law of Generalization"],"prefix":"10.1162","volume":"38","author":[{"given":"Faris B.","family":"Rustom","sequence":"first","affiliation":[{"name":"Neuroscience Program, Computational Neuroscience and Vision Lab, Center for Systems Neuroscience, Boston University, Boston, MA 02215, USA frustom@bu.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Rohan","family":"Sharma","sequence":"additional","affiliation":[{"name":"Neuroscience Program, Computational Neuroscience and Vision Lab, Boston University, Boston, MA 02215, USA roohan@bu.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Haluk","family":"\u00d6\u011fmen","sequence":"additional","affiliation":[{"name":"Laboratory of Perceptual and Cognitive Dynamics, Department of Electrical and Computer Engineering, University of Denver, CO 80210, USA ogmen@du.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Arash","family":"Yazdanbakhsh","sequence":"additional","affiliation":[{"name":"Department of Psychological and Brain Sciences, Computational Neuroscience and Vision Lab, Center for Systems Neuroscience, Boston University, Boston, MA 02215, USA yazdan@bu.edu"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"281","published-online":{"date-parts":[[2026,2,27]]},"reference":[{"key":"2026022715162902500_B1","author":"Battleday","year":"2017","journal-title":"Modeling human categorization of natural images using deep feature representations"},{"key":"2026022715162902500_B2","doi-asserted-by":"publisher","first-page":"77","DOI":"10.1109\/2.33","article-title":"The ART of adaptive pattern recognition by a self-organizing neural network","volume":"21","author":"Carpenter","year":"1988","journal-title":"Computer"},{"key":"2026022715162902500_B3","doi-asserted-by":"publisher","first-page":"346","DOI":"10.1016\/S0022-2496(03)00013-0","article-title":"The generalized universal law of generalization","volume":"47","author":"Chater","year":"2003","journal-title":"Journal of Mathematical Psychology"},{"key":"2026022715162902500_B4","doi-asserted-by":"publisher","first-page":"8522","DOI":"10.1038\/ncomms9522","article-title":"A higher order visual neuron tuned to the spatial amplitude spectra of natural scenes","volume":"6","author":"Dyakova","year":"2015","journal-title":"Nature Communications"},{"issue":"12","key":"2026022715162902500_B5","doi-asserted-by":"publisher","first-page":"2379","DOI":"10.1364\/JOSAA.4.002379","article-title":"Relations between the statistics of natural images and the response properties of cortical cells","volume":"4","author":"Field","year":"1987","journal-title":"Journal of the Optical Society of America"},{"key":"2026022715162902500_B6","doi-asserted-by":"publisher","first-page":"9803","DOI":"10.1073\/pnas.1809787115","article-title":"Measurement invariance explains the universal law of generalization for psychological perception","volume":"115","author":"Frank","year":"2018","journal-title":"Proceedings of the National Academy of Sciences, National Academy of Sciences"},{"key":"2026022715162902500_B7","doi-asserted-by":"publisher","first-page":"193","DOI":"10.1007\/BF00344251","article-title":"Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position","volume":"36","author":"Fukushima","year":"1980","journal-title":"Biological Cybernetics"},{"key":"2026022715162902500_B8","doi-asserted-by":"publisher","first-page":"183","DOI":"10.1016\/0893-6080(89)90003-8","article-title":"On the approximate realization of continuous mappings by neural networks","volume":"2","author":"Funahashi","year":"1989","journal-title":"Neural Networks"},{"issue":"11","key":"2026022715162902500_B9","doi-asserted-by":"publisher","first-page":"1173","DOI":"10.1038\/s41562-020-00951-3","article-title":"Revealing the multidimensional mental representations of natural objects underlying human similarity judgments","volume":"4","author":"Hebart","year":"2020","journal-title":"Nature Human Behavior"},{"issue":"10","key":"2026022715162902500_B10","doi-asserted-by":"publisher","first-page":"e0223792","DOI":"10.1371\/journal.pone.0223792","article-title":"THINGS: A database of 1,854 object concepts and more than 26,000 naturalistic object images","volume":"14","author":"Hebart","year":"2019","journal-title":"PLOS One"},{"key":"2026022715162902500_B11","doi-asserted-by":"publisher","first-page":"359","DOI":"10.1016\/0893-6080(89)90020-8","article-title":"Multilayer feedforward networks are universal approximators","volume":"2","author":"Hornik","year":"1989","journal-title":"Neural Networks"},{"key":"2026022715162902500_B12","doi-asserted-by":"publisher","first-page":"551","DOI":"10.1016\/0893-6080(90)90005-6","article-title":"Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks","volume":"3","author":"Hornik","year":"1990","journal-title":"Neural Networks"},{"key":"2026022715162902500_B13","volume-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2009"},{"key":"2026022715162902500_B14","doi-asserted-by":"publisher","first-page":"84","DOI":"10.1145\/3065386","article-title":"ImageNet classification with deep convolutional neural networks","volume":"60","author":"Krizhevsky","year":"2017","journal-title":"Communications of the ACM"},{"key":"2026022715162902500_B15","doi-asserted-by":"publisher","first-page":"45","DOI":"10.1016\/j.cviu.2019.04.006","article-title":"Anabranch network for camouflaged object segmentation","volume":"184","author":"Le","year":"2019","journal-title":"Computer Vision and Image Understanding"},{"key":"2026022715162902500_B16","volume-title":"Camouflaged object segmentation","author":"Le"},{"key":"2026022715162902500_B17","doi-asserted-by":"publisher","first-page":"109","DOI":"10.1016\/S0079-7421(08)60536-8","article-title":"Catastrophic interference in connectionist networks: The sequential learning problem","volume":"24","author":"McCloskey","year":"1989","journal-title":"Psychology of Learning and Motivation"},{"key":"2026022715162902500_B18","doi-asserted-by":"crossref","first-page":"3","DOI":"10.7551\/mitpress\/1477.003.0005","article-title":"Learning and connectionist representations","volume-title":"Attention and performance 14: Synergies in experimental psychology, artificial intelligence, and cognitive neuroscience","author":"Rumelhart","year":"1993"},{"issue":"1","key":"2026022715162902500_B19","doi-asserted-by":"publisher","DOI":"10.1093\/biomethods\/bpae080","article-title":"Deep learning and transfer learning for brain tumor detection and classification","volume":"9","author":"Rustom","year":"2024","journal-title":"Biology Methods and Protocols"},{"key":"2026022715162902500_B20","doi-asserted-by":"publisher","first-page":"229","DOI":"10.1007\/s42113-020-00073-z","article-title":"Training deep networks to construct a psychological feature space for a natural-object category domain","volume":"3","author":"Sanders","year":"2020","journal-title":"Computational Brain and Behavior"},{"issue":"6","key":"2026022715162902500_B21","doi-asserted-by":"publisher","first-page":"509","DOI":"10.1037\/h0042354","article-title":"Stimulus and response generalization: tests of a model relating generalization to distance in psychological space","volume":"55","author":"Shepard","year":"1958","journal-title":"Journal of Experimental Psychology"},{"issue":"237","key":"2026022715162902500_B22","doi-asserted-by":"publisher","first-page":"1317","DOI":"10.1126\/science.3629243","article-title":"Toward a universal law of generalization for psychological science","volume":"80","author":"Shepard","year":"1987","journal-title":"Science"},{"issue":"360","key":"2026022715162902500_B23","doi-asserted-by":"publisher","first-page":"652","DOI":"10.1126\/science.aaq1118","article-title":"Efficient coding explains the universal law of generalization in human perception","volume":"80","author":"Sims","year":"2018","journal-title":"Science"},{"key":"2026022715162902500_B24","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1017\/CBO9780511852053.001","article-title":"Animal camouflage","volume-title":"Animal camouflage","author":"Stevens","year":"2011"},{"key":"2026022715162902500_B25","first-page":"1","article-title":"Going deeper with convolutions","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Szegedy","year":"2015"},{"key":"2026022715162902500_B26","article-title":"MATLAB version 9.7 (R2019b)","year":"2019"},{"key":"2026022715162902500_B27","doi-asserted-by":"publisher","first-page":"43290","DOI":"10.1109\/ACCESS.2021.3064443","article-title":"MirrorNet: Bio-inspired camouflaged object segmentation","volume":"9","author":"Yan","year":"2021","journal-title":"IEEE Access"}],"container-title":["Neural Computation"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/neco\/article-pdf\/38\/3\/328\/2579376\/neco.a.1483.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/neco\/article-pdf\/38\/3\/328\/2579376\/neco.a.1483.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,27]],"date-time":"2026-02-27T20:16:42Z","timestamp":1772223402000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/neco\/article\/38\/3\/328\/135156\/Object-Detection-Recognition-Deep-Learning-and-the"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,27]]},"references-count":27,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2026,2,27]]},"published-print":{"date-parts":[[2026,2,27]]}},"URL":"https:\/\/doi.org\/10.1162\/neco.a.1483","relation":{},"ISSN":["0899-7667","1530-888X"],"issn-type":[{"value":"0899-7667","type":"print"},{"value":"1530-888X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,3]]},"published":{"date-parts":[[2026,2,27]]}}}