Do Humans and Deep Convolutional Neural Networks Use Visual Information Similarly for the Categorization of Natural Scenes?
Convolutional neural networks
Spatial frequencies
Visual categorization
Journal
Cognitive science
ISSN: 1551-6709
Titre abrégé: Cogn Sci
Pays: United States
ID NLM: 7708195
Informations de publication
Date de publication:
06 2021
06 2021
Historique:
revised:
19
05
2021
received:
24
01
2020
accepted:
31
05
2021
entrez:
25
6
2021
pubmed:
26
6
2021
medline:
30
9
2021
Statut:
ppublish
Résumé
The investigation of visual categorization has recently been aided by the introduction of deep convolutional neural networks (CNNs), which achieve unprecedented accuracy in picture classification after extensive training. Even if the architecture of CNNs is inspired by the organization of the visual brain, the similarity between CNN and human visual processing remains unclear. Here, we investigated this issue by engaging humans and CNNs in a two-class visual categorization task. To this end, pictures containing animals or vehicles were modified to contain only low/high spatial frequency (HSF) information, or were scrambled in the phase of the spatial frequency spectrum. For all types of degradation, accuracy increased as degradation was reduced for both humans and CNNs; however, the thresholds for accurate categorization varied between humans and CNNs. More remarkable differences were observed for HSF information compared to the other two types of degradation, both in terms of overall accuracy and image-level agreement between humans and CNNs. The difficulty with which the CNNs were shown to categorize high-passed natural scenes was reduced by picture whitening, a procedure which is inspired by how visual systems process natural images. The results are discussed concerning the adaptation to regularities in the visual environment (scene statistics); if the visual characteristics of the environment are not learned by CNNs, their visual categorization may depend only on a subset of the visual information on which humans rely, for example, on low spatial frequency information.
Identifiants
pubmed: 34170027
doi: 10.1111/cogs.13009
pmc: PMC8365760
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
e13009Informations de copyright
© 2021 The Authors. Cognitive Science published by Wiley Periodicals LLC on behalf of Cognitive Science Society (CSS).
Références
PLoS Comput Biol. 2014 Dec 18;10(12):e1003963
pubmed: 25521294
Nat Neurosci. 2016 Mar;19(3):356-65
pubmed: 26906502
J Neurosci. 2015 Jul 8;35(27):10005-14
pubmed: 26157000
Biol Psychol. 2006 Apr;72(1):35-45
pubmed: 16157440
Network. 2003 Aug;14(3):391-412
pubmed: 12938764
J Vis. 2009 Apr 30;9(4):29.1-15
pubmed: 19757938
J Neurosci. 2011 Nov 23;31(47):17052-7
pubmed: 22114274
Nat Neurosci. 2003 Jun;6(6):624-31
pubmed: 12740580
PLoS One. 2017 Dec 22;12(12):e0189828
pubmed: 29272283
J Cogn Neurosci. 2019 Jan;31(1):109-125
pubmed: 30188778
Nat Rev Neurosci. 2010 Nov;11(11):773-83
pubmed: 20959860
J Neurosci. 2018 Aug 15;38(33):7255-7269
pubmed: 30006365
F1000Res. 2020 Jun 11;9:
pubmed: 32566136
J Vis. 2009 Jan 08;9(1):2.1-16
pubmed: 19271872
Elife. 2018 Mar 07;7:
pubmed: 29513219
IEEE Trans Neural Netw Learn Syst. 2019 Nov;30(11):3212-3232
pubmed: 30703038
Network. 2011;22(1-4):4-44
pubmed: 22149669
PLoS Comput Biol. 2014 Nov 06;10(11):e1003915
pubmed: 25375136
J Cogn Neurosci. 2001 May 15;13(4):454-61
pubmed: 11388919
Percept Psychophys. 2001 Nov;63(8):1314-29
pubmed: 11800459
Philos Trans R Soc Lond B Biol Sci. 2017 Feb 19;372(1714):
pubmed: 28044013
Front Psychol. 2017 Feb 02;8:142
pubmed: 28210237
Vision Res. 2006 Sep;46(18):2901-13
pubmed: 16782164
Proc Natl Acad Sci U S A. 2014 Jun 10;111(23):8619-24
pubmed: 24812127
Vision Res. 1997 Dec;37(23):3367-83
pubmed: 9425550
Neuroimage. 2006 Aug 15;32(2):583-91
pubmed: 16750397
Vision Res. 1995 Jan;35(1):7-24
pubmed: 7839612
Psychon Bull Rev. 2005 Feb;12(1):43-65
pubmed: 15948283
Cognition. 1999 Jan 1;69(3):243-65
pubmed: 10193048
Dev Sci. 2005 Mar;8(2):151-61
pubmed: 15720373
Perception. 1987;16(1):89-101
pubmed: 3671045
Ophthalmic Physiol Opt. 1992 Apr;12(2):229-32
pubmed: 1408179
J Vis. 2015;15(8):14
pubmed: 26114677
Annu Rev Neurosci. 2001;24:1193-216
pubmed: 11520932
PLoS One. 2013 May 31;8(5):e65103
pubmed: 23741468
J Exp Psychol Hum Percept Perform. 1993 Aug;19(4):878-98
pubmed: 8409864
PLoS Comput Biol. 2016 Apr 28;12(4):e1004896
pubmed: 27124699
Network. 2001 Aug;12(3):241-53
pubmed: 11563528
PLoS Comput Biol. 2018 Dec 7;14(12):e1006613
pubmed: 30532273
Nature. 1996 Jun 6;381(6582):520-2
pubmed: 8632824
Psychon Bull Rev. 2011 Oct;18(5):840-7
pubmed: 21779945
Neurosci Biobehav Rev. 2017 Mar;74(Pt A):44-57
pubmed: 28089884
Annu Rev Neurosci. 1999;22:29-47
pubmed: 10202531
Science. 2006 Jul 28;313(5786):504-7
pubmed: 16873662
Appl Opt. 1987 Jan 1;26(1):157-70
pubmed: 20454092
Psychophysiology. 2020 Oct 14;:e13704
pubmed: 33090526
Wiley Interdiscip Rev Dev Biol. 2014 Jan-Feb;3(1):1-40
pubmed: 24902832
Nat Hum Behav. 2017 Sep;1(9):657-664
pubmed: 31024135
J Cogn Neurosci. 2007 Aug;19(8):1241-58
pubmed: 17651000
Nat Neurosci. 2005 Dec;8(12):1643-6
pubmed: 16306891
Psychon Bull Rev. 2001 Sep;8(3):454-69
pubmed: 11700896
J Physiol. 1969 Jul;203(1):237-60
pubmed: 5821879
J Opt Soc Am A. 1987 Dec;4(12):2379-94
pubmed: 3430225
J Exp Psychol Gen. 2015 Jun;144(3):551-69
pubmed: 25938178
Rev Neurosci. 2013;24(1):89-104
pubmed: 23183741
Cognition. 1998 Jul;67(1-2):147-79
pubmed: 9735539
IEEE Trans Pattern Anal Mach Intell. 2016 Oct;38(10):1943-55
pubmed: 26599615
Sci Rep. 2016 Jun 10;6:27755
pubmed: 27282108
Neuroimage. 2000 Jun;11(6 Pt 1):634-43
pubmed: 10860792
Front Psychol. 2011 Dec 07;2:365
pubmed: 22162973
J Neurosci. 2009 Sep 9;29(36):11409-16
pubmed: 19741147
Behav Res Methods. 2012 Jun;44(2):314-24
pubmed: 22083660
Front Psychol. 2017 Sep 12;8:1551
pubmed: 28955272
Cogn Psychol. 2009 Mar;58(2):137-76
pubmed: 18762289
J Vis. 2011 Sep 20;11(10):14
pubmed: 21933932
Psychol Bull. 1992 Jul;112(1):24-38
pubmed: 1529037
Curr Biol. 2002 Jul 23;12(14):R476-7; author reply R478
pubmed: 12176342