Contrast sensitivity functions in autoencoders

Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal images. Here we reassess the relevance of low-level vision tasks in the explanation of the contrast sensitivity functions (CSFs) in light of 1)...

Full description

Saved in:
Bibliographic Details
Published in:Journal of vision (Charlottesville, Va.) Vol. 22; no. 6; p. 8
Main Authors: Li, Qiang, Gomez-Villa, Alex, Bertalmío, Marcelo, Malo, Jesús
Format: Journal Article
Language:English
Published: United States The Association for Research in Vision and Ophthalmology 19.05.2022
ISSN:1534-7362, 1534-7362
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal images. Here we reassess the relevance of low-level vision tasks in the explanation of the contrast sensitivity functions (CSFs) in light of 1) the current trend of using artificial neural networks for studying vision, and 2) the current knowledge of retinal image representations. As a first contribution, we show that a very popular type of convolutional neural networks (CNNs), called autoencoders, may develop human-like CSFs in the spatiotemporal and chromatic dimensions when trained to perform some basic low-level vision tasks (like retinal noise and optical blur removal), but not others (like chromatic) adaptation or pure reconstruction after simple bottlenecks). As an illustrative example, the best CNN (in the considered set of simple architectures for enhancement of the retinal signal) reproduces the CSFs with a root mean square error of 11% of the maximum sensitivity. As a second contribution, we provide experimental evidence of the fact that, for some functional goals (at low abstraction level), deeper CNNs that are better in reaching the quantitative goal are actually worse in replicating human-like phenomena (such as the CSFs). This low-level result (for the explored networks) is not necessarily in contradiction with other works that report advantages of deeper nets in modeling higher level vision goals. However, in line with a growing body of literature, our results suggests another word of caution about CNNs in vision science because the use of simplified units or unrealistic architectures in goal optimization may be a limitation for the modeling and understanding of human vision.
AbstractList Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal images. Here we reassess the relevance of low-level vision tasks in the explanation of the contrast sensitivity functions (CSFs) in light of 1) the current trend of using artificial neural networks for studying vision, and 2) the current knowledge of retinal image representations. As a first contribution, we show that a very popular type of convolutional neural networks (CNNs), called autoencoders, may develop human-like CSFs in the spatiotemporal and chromatic dimensions when trained to perform some basic low-level vision tasks (like retinal noise and optical blur removal), but not others (like chromatic) adaptation or pure reconstruction after simple bottlenecks). As an illustrative example, the best CNN (in the considered set of simple architectures for enhancement of the retinal signal) reproduces the CSFs with a root mean square error of 11% of the maximum sensitivity. As a second contribution, we provide experimental evidence of the fact that, for some functional goals (at low abstraction level), deeper CNNs that are better in reaching the quantitative goal are actually worse in replicating human-like phenomena (such as the CSFs). This low-level result (for the explored networks) is not necessarily in contradiction with other works that report advantages of deeper nets in modeling higher level vision goals. However, in line with a growing body of literature, our results suggests another word of caution about CNNs in vision science because the use of simplified units or unrealistic architectures in goal optimization may be a limitation for the modeling and understanding of human vision.
Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal images. Here we reassess the relevance of low-level vision tasks in the explanation of the contrast sensitivity functions (CSFs) in light of 1) the current trend of using artificial neural networks for studying vision, and 2) the current knowledge of retinal image representations. As a first contribution, we show that a very popular type of convolutional neural networks (CNNs), called autoencoders, may develop human-like CSFs in the spatiotemporal and chromatic dimensions when trained to perform some basic low-level vision tasks (like retinal noise and optical blur removal), but not others (like chromatic) adaptation or pure reconstruction after simple bottlenecks). As an illustrative example, the best CNN (in the considered set of simple architectures for enhancement of the retinal signal) reproduces the CSFs with a root mean square error of 11% of the maximum sensitivity. As a second contribution, we provide experimental evidence of the fact that, for some functional goals (at low abstraction level), deeper CNNs that are better in reaching the quantitative goal are actually worse in replicating human-like phenomena (such as the CSFs). This low-level result (for the explored networks) is not necessarily in contradiction with other works that report advantages of deeper nets in modeling higher level vision goals. However, in line with a growing body of literature, our results suggests another word of caution about CNNs in vision science because the use of simplified units or unrealistic architectures in goal optimization may be a limitation for the modeling and understanding of human vision.Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal images. Here we reassess the relevance of low-level vision tasks in the explanation of the contrast sensitivity functions (CSFs) in light of 1) the current trend of using artificial neural networks for studying vision, and 2) the current knowledge of retinal image representations. As a first contribution, we show that a very popular type of convolutional neural networks (CNNs), called autoencoders, may develop human-like CSFs in the spatiotemporal and chromatic dimensions when trained to perform some basic low-level vision tasks (like retinal noise and optical blur removal), but not others (like chromatic) adaptation or pure reconstruction after simple bottlenecks). As an illustrative example, the best CNN (in the considered set of simple architectures for enhancement of the retinal signal) reproduces the CSFs with a root mean square error of 11% of the maximum sensitivity. As a second contribution, we provide experimental evidence of the fact that, for some functional goals (at low abstraction level), deeper CNNs that are better in reaching the quantitative goal are actually worse in replicating human-like phenomena (such as the CSFs). This low-level result (for the explored networks) is not necessarily in contradiction with other works that report advantages of deeper nets in modeling higher level vision goals. However, in line with a growing body of literature, our results suggests another word of caution about CNNs in vision science because the use of simplified units or unrealistic architectures in goal optimization may be a limitation for the modeling and understanding of human vision.
Author Malo, Jesús
Li, Qiang
Bertalmío, Marcelo
Gomez-Villa, Alex
Author_xml – sequence: 1
  givenname: Qiang
  surname: Li
  fullname: Li, Qiang
  organization: Image Processing Lab, Parc Cientific, Universitat de Valéncia, Spain, qiang.li@uv.es
– sequence: 2
  givenname: Alex
  surname: Gomez-Villa
  fullname: Gomez-Villa, Alex
  organization: Computer Vision Center, Universitat Autónoma de Barcelona, Spain, agomezvi@cvc.uab.cat
– sequence: 3
  givenname: Marcelo
  surname: Bertalmío
  fullname: Bertalmío, Marcelo
  organization: Instituto de Óptica, Spanish National Research Council (CSIC), Spain, marcelo.bertalmio@csic.es
– sequence: 4
  givenname: Jesús
  surname: Malo
  fullname: Malo, Jesús
  organization: Image Processing Lab, Parc Cientific, Universitat de Valéncia, Spain, jesus.malo@uv.eshttp://isp.uv.es
BackLink https://www.ncbi.nlm.nih.gov/pubmed/35587354$$D View this record in MEDLINE/PubMed
BookMark eNptkV1LwzAUhoNM3Ife-AOklyK05rNJbwQZfsHAG70ObZJqRpfMJh3s39uyTaZ4dQ6c97zP4T1TMHLeGQAuEcwQyvnt0m8yjLM8EydgghihKSc5Hh31YzANYQkhhgyiMzAmjAlOGJ2AbO5dbMsQk2BcsNFubNwmdedUtN6FxLqk7KI3Tnlt2nAOTuuyCeZiX2fg_fHhbf6cLl6fXub3i1QRLmKqtOYK4poXAmolKqEqRWuWFwoXFVZI1BRSUxWVZoQpUZuC1ZqVjHJRUU0JmYG7ne-6q1ZGKzMc2ch1a1dlu5W-tPL3xNlP-eE3skCUISJ6g-u9Qeu_OhOiXNmgTNOUzvguSJzneQEFZ6yXXh2zfiCHkHoB3AlU60NoTS2VjeWQT4-2jURQDn-Q_R8kxjKXA_7mz8rB9R_xN3JOi1o
CitedBy_id crossref_primary_10_3389_fnins_2023_1208882
crossref_primary_10_3389_fcomp_2023_1275026
crossref_primary_10_1016_j_neunet_2023_04_032
crossref_primary_10_3389_fpsyg_2024_1415958
crossref_primary_10_3758_s13423_023_02281_7
crossref_primary_10_3390_e24101442
crossref_primary_10_1007_s11571_024_10184_z
crossref_primary_10_1016_j_neunet_2025_107189
crossref_primary_10_1016_j_neucom_2023_127143
Cites_doi 10.1152/jn.1997.78.2.1045
10.2352/ISSN.2470-1173.2017.14.HVEI-113
10.1167/5.9.6
10.1371/journal.pcbi.1006897
10.1016/S0031-3203(02)00325-4
10.1167/jov.20.4.23
10.1016/S0042-6989(97)00183-1
10.1007/s11263-015-0816-y
10.1080/09500340.2011.606374
10.1113/jphysiol.1985.sp015591
10.1113/jphysiol.1966.sp008107
10.1016/0042-6989(81)90092-4
10.1523/JNEUROSCI.5023-14.2015
10.1109/ICIP.2002.1038898
10.1146/neuro.2001.24.issue-1
10.1167/jov.20.7.17
10.1126/science.171.3972.694
10.1109/30.125072
10.1371/journal.pcbi.1004927
10.1038/nn.4244
10.3109/0954898X.2011.638888
10.1364/JOSAA.3.000300
10.1364/JOSAA.27.000852
10.1016/0042-6989(83)90161-X
10.5594/j18266XY
10.1109/MSP.2008.930649
10.1016/j.neuron.2013.12.014
10.1162/neco.1992.4.4.559
10.1016/j.visres.2020.07.010
10.1109/TIP.2005.860345
10.1073/pnas.1905334117
10.1371/journal.pone.0086481
10.1049/ip-f-1.1981.0061
10.1038/s41583-020-0277-3
10.1167/jov.21.9.1996
10.1038/nrn3136
10.1037/h0041403
10.1007/BF00288786
10.1109/TIP.83
10.1109/PROC.1975.9801
10.1007/978-1-84882-491-1
10.1364/JOSA.70.001458
10.1038/s41598-020-73113-0
10.1016/S0042-6989(00)00021-3
10.1371/journal.pcbi.1009028
10.1162/neco.1992.4.2.196
10.1523/JNEUROSCI.22-14-06158.2002
10.1364/JOSA.69.001340
10.1038/356716a0
10.1109/TIT.1974.1055250
10.1146/vision.2015.1.issue-1
10.1080/09500349708232904
10.1016/j.neunet.2004.03.008
10.1371/journal.pcbi.1003963
10.1167/19.4.8
10.1113/jphysiol.1968.sp008574
10.1073/pnas.1403112111
10.1038/srep27755
10.1364/JOSAA.15.002036
10.1016/j.visres.2006.01.019
ContentType Journal Article
Copyright Copyright 2022 The Authors 2022
Copyright_xml – notice: Copyright 2022 The Authors 2022
DBID AAYXX
CITATION
NPM
7X8
5PM
DOI 10.1167/jov.22.6.8
DatabaseName CrossRef
PubMed
MEDLINE - Academic
PubMed Central (Full Participant titles)
DatabaseTitle CrossRef
PubMed
MEDLINE - Academic
DatabaseTitleList
PubMed
MEDLINE - Academic
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod fulltext_linktorsrc
Discipline Medicine
EISSN 1534-7362
ExternalDocumentID PMC9145138
35587354
10_1167_jov_22_6_8
Genre Journal Article
GroupedDBID ---
29L
2WC
53G
5GY
5VS
AAFWJ
AAYXX
ABIVO
ACGFO
ADBBV
AENEX
AFPKN
ALMA_UNASSIGNED_HOLDINGS
BAWUL
BCNDV
CITATION
CS3
DIK
DU5
E3Z
EBS
EJD
F5P
FRP
GROUPED_DOAJ
GX1
KQ8
M~E
OK1
OVT
P2P
RNS
RPM
TR2
TRV
W2D
W8F
XSB
NPM
7X8
5PM
ID FETCH-LOGICAL-c378t-cdd7c02f7980dc8b8cbc4f569c29b2c18f404eb9bd535c8fe95fd5a5478b4d433
ISICitedReferencesCount 11
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001001008200007&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1534-7362
IngestDate Thu Aug 21 17:21:38 EDT 2025
Thu Jul 10 22:49:28 EDT 2025
Mon Jul 21 06:00:23 EDT 2025
Tue Nov 18 20:52:14 EST 2025
Sat Nov 29 03:46:22 EST 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 6
Language English
License http://creativecommons.org/licenses/by-nc-nd/4.0
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c378t-cdd7c02f7980dc8b8cbc4f569c29b2c18f404eb9bd535c8fe95fd5a5478b4d433
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
OpenAccessLink http://dx.doi.org/10.1167/jov.22.6.8
PMID 35587354
PQID 2666908755
PQPubID 23479
ParticipantIDs pubmedcentral_primary_oai_pubmedcentral_nih_gov_9145138
proquest_miscellaneous_2666908755
pubmed_primary_35587354
crossref_citationtrail_10_1167_jov_22_6_8
crossref_primary_10_1167_jov_22_6_8
PublicationCentury 2000
PublicationDate 20220519
PublicationDateYYYYMMDD 2022-05-19
PublicationDate_xml – month: 5
  year: 2022
  text: 20220519
  day: 19
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle Journal of vision (Charlottesville, Va.)
PublicationTitleAlternate J Vis
PublicationYear 2022
Publisher The Association for Research in Vision and Ophthalmology
Publisher_xml – name: The Association for Research in Vision and Ophthalmology
References Kubilius (bib49) 2019
Wallace (bib85) 1992; 38
Ingling (bib44) 1983; 23
Simoncelli (bib79) 2001; 24
Cottaris (bib21) 2019; 19
Simoncelli (bib77) 1998; 38
Cadieu (bib15) 2014; 10
Firestone (bib28) 2020; 117
Antolík (bib3) 2016; 12
Gutmann (bib38) 2014; 9
Cottaris (bib22) 2020; 20
Taubman (bib83) 2001
Kelly (bib46) 1979; 69
Atick (bib4) 1992; 4
Hurvich (bib42) 1957; 64
Gomez-Villa (bib34) 2020; 176
Prenger (bib72) 2004; 17
Gutiérrez (bib37) 2006; 15
van den Oord (bib70) 2014; 15
Watson (bib88) 2005; 5
Geirhos (bib32) 2019
Malo (bib57) 2022
Malo (bib58) 2002
Hyvärinen (bib43) 2009
Welles (bib92) 1946
Güçlü (bib36) 2015; 35
Legge (bib51) 1981; 21
Wuerger (bib95) 2020; 20
Martinez-Uriegas (bib66) 1994
Soh (bib80) 2021
Watson (bib89) 2016
Akbarinia (bib2) 2021; 21
Martinez-Otero (bib65) 2014; 81
Carandini (bib18) 2012; 13
Hepburn (bib39) 2020
Bertalmío (bib12) 2020; 10
Cadena (bib14) 2019; 15
Wilson (bib94) 1973; 13
Martinez-Uriegas (bib67) 1997
Geirhos (bib31) 2020
Atick (bib5) 1992; 4
Cai (bib16) 1997; 78
Abadi (bib1) 2016
Berardino (bib11) 2017
Mullen (bib69) 1985; 359
Mannos (bib60) 1974; 20
Díez-Ajenjo (bib24) 2011; 58
Wichmann (bib93) 2017; 10
LeRoy (bib53) 1959
Reid (bib74) 2002; 22
Batty (bib9) 2017
Goodfellow (bib35) 2016
Marr (bib61) 1982
Campbell (bib17) 1968; 197
Reid (bib73) 1992; 356
Tao (bib82) 2018
Kriegeskorte (bib48) 2015; 1
Zhang (bib98) 2017; 26
Lillicrap (bib54) 2020; 21
Karklin (bib45) 2011
Malo (bib56) 2020; 10
Gomez-Villa (bib33) 2019
Enroth-Cugell (bib25) 1966; 187
Cichy (bib19) 2016; 6
Burg (bib13) 2021; 17
Poggio (bib71) 2021
Russakovsky (bib76) 2015; 115
Atick (bib6) 2011; 22
Valois (bib84) 1971; 171
Laparra (bib50) 2010; 27
Legge (bib52) 1980; 70
Freleng (bib30) 1963
Malo (bib59) 1997; 44
Martinez (bib63) 2019
Ballé (bib7) 2017
Wang (bib86) 2009; 26
Marr (bib62) 1976
Epifanio (bib26) 2003; 36
Morgan (bib68) 2006; 46
Stockman (bib81) 2000; 40
Yamins (bib96) 2016; 19
Watson (bib90) 1986; 3
Baydin (bib10) 2018; 18
Hunt (bib41) 1975; 63
Lindsey (bib55) 2019
Clarke (bib20) 1981; 128
Flachot (bib29) 2020
Ruderman (bib75) 1998; 15
Watson (bib87) 2013; 122
Martinez (bib64) 2018; 13
Kingma (bib47) 2017
Barlow (bib8) 1961; 1
Donen (bib23) 1963
Watson (bib91) 2002
Esteve (bib27) 2020
Yamins (bib97) 2014; 111
Hepburn (bib40) 2022
References_xml – volume: 78
  start-page: 1045
  issue: 2
  year: 1997
  ident: bib16
  article-title: Spatiotemporal receptive field organization in the lateral geniculate nucleus of cats and kittens
  publication-title: Journal of Neurophysiology,
  doi: 10.1152/jn.1997.78.2.1045
– volume: 10
  start-page: 36
  year: 2017
  ident: bib93
  article-title: Methods and measurements to compare men against machines
  publication-title: Electronic Imaging,
  doi: 10.2352/ISSN.2470-1173.2017.14.HVEI-113
– start-page: 12309
  volume-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 19)
  year: 2019
  ident: bib33
  article-title: Convolutional neural networks can be deceived by visual illusions
– volume: 15
  start-page: 2061
  issue: 60
  year: 2014
  ident: bib70
  article-title: The student-t mixture as a natural image patch prior with application to image compression
  publication-title: Journal of Machine Learning Research,
– start-page: 1
  volume-title: ArXiV: Computer Vision and Pattern Recognition,
  year: 2022
  ident: bib57
  article-title: Paraphrasing Magritte's observation
– year: 1963
  ident: bib23
– volume: 5
  start-page: 717
  issue: 9
  year: 2005
  ident: bib88
  article-title: A standard model for foveal detection of spatial contrast
  publication-title: Journal of Vision,
  doi: 10.1167/5.9.6
– volume: 15
  start-page: e1006897
  issue: 4
  year: 2019
  ident: bib14
  article-title: Deep convolutional models improve predictions of macaque V1 responses to natural images
  publication-title: PLoS Computational Biology,
  doi: 10.1371/journal.pcbi.1006897
– volume: 36
  start-page: 1799
  issue: 8
  year: 2003
  ident: bib26
  article-title: Linear transform for simultaneous diagonalization of covariance and perceptual metric matrix in image coding
  publication-title: Pattern Recognition,
  doi: 10.1016/S0031-3203(02)00325-4
– volume: 20
  start-page: 23
  year: 2020
  ident: bib95
  article-title: Spatio-chromatic contrast sensitivity under mesopic and photopic light levels
  publication-title: Journal of Vision,
  doi: 10.1167/jov.20.4.23
– volume: 38
  start-page: 743
  issue: 5
  year: 1998
  ident: bib77
  article-title: A model of neuronal reponses in visual area MT
  publication-title: Vision Research,
  doi: 10.1016/S0042-6989(97)00183-1
– volume: 115
  start-page: 211
  year: 2015
  ident: bib76
  article-title: Imagenet large scale visual recognition challenge
  publication-title: International Journal of Computer Vision,
  doi: 10.1007/s11263-015-0816-y
– volume: 58
  start-page: 1
  year: 2011
  ident: bib24
  article-title: Red-green vs. blue-yellow spatio-temporal contrast sensitivity across the visual field
  publication-title: Journal of Modern Optics,
  doi: 10.1080/09500340.2011.606374
– volume: 359
  start-page: 381
  year: 1985
  ident: bib69
  article-title: The CSF of human colour vision to red–green and yellow–blue chromatic gratings
  publication-title: Journal of Physiology,
  doi: 10.1113/jphysiol.1985.sp015591
– volume: 187
  start-page: 516
  year: 1966
  ident: bib25
  article-title: The contrast sensitivity of retinal ganglion cells on the cat
  publication-title: Journal of Physiology (London),
  doi: 10.1113/jphysiol.1966.sp008107
– volume-title: International Conference on Learning Representations, ICLR
  year: 2019
  ident: bib55
  article-title: The effects of neural resource constraints on early visual representations
– start-page: 1
  volume-title: Arxiv: Quantitative Biology,
  year: 2020
  ident: bib27
  article-title: Psychophysical estimation of early and late noise
– volume: 21
  start-page: 457
  issue: 4
  year: 1981
  ident: bib51
  article-title: A power law for contrast discrimination
  publication-title: Vision Research,
  doi: 10.1016/0042-6989(81)90092-4
– volume: 35
  start-page: 10005
  issue: 27
  year: 2015
  ident: bib36
  article-title: Deep neural networks reveal a gradient in the complexity of neural representations across the ventral stream
  publication-title: Journal of Neuroscience,
  doi: 10.1523/JNEUROSCI.5023-14.2015
– start-page: 41
  volume-title: Proceedings of the IEEE International Conference on Image Processing
  year: 2002
  ident: bib91
  article-title: Video quality measures based on the standard spatial observer
  doi: 10.1109/ICIP.2002.1038898
– volume: 24
  start-page: 1193
  issue: 1
  year: 2001
  ident: bib79
  article-title: Natural image statistics and neural representation
  publication-title: Annual Review of Neuroscience,
  doi: 10.1146/neuro.2001.24.issue-1
– volume: 1
  start-page: 217
  year: 1961
  ident: bib8
  article-title: Possible principles underlying the transformation of sensory messages
  publication-title: Sensory Communication,
– start-page: 121
  volume-title: Proceedings of the IEEE International Conference on Image Processing (ICIP)
  year: 2020
  ident: bib39
  article-title: Perceptnet: A human visual system inspired neural network for estimating perceptual distance
– volume-title: Proceedings of the Advances in Neural Information Processing Systems,
  year: 2011
  ident: bib45
  article-title: Efficient coding of natural images with a population of noisy linear-nonlinear neurons
– volume-title: Servei de Publicacions de la Universitat de Valencia
  year: 2002
  ident: bib58
  article-title: ColorLab: A Matlab Toolbox for color science and calibrated color image processing
– volume-title: Jpeg 2000: Image compression fundamentals, standards and practice
  year: 2001
  ident: bib83
– volume: 20
  start-page: 17
  year: 2020
  ident: bib22
  article-title: A computational observer model of spatial contrast sensitivity: Effects of photocurrent encoding, fixational eye movements, and inference engine
  publication-title: Journal of Vision,
  doi: 10.1167/jov.20.7.17
– volume: 171
  start-page: 694
  year: 1971
  ident: bib84
  article-title: Contours and contrast: Responses of monkey lateral geniculate nucleus cells to luminance and color figures
  publication-title: Science,
  doi: 10.1126/science.171.3972.694
– volume-title: The Pink Panther Show
  year: 1963
  ident: bib30
– volume: 38
  start-page: xviii
  issue: 1
  year: 1992
  ident: bib85
  article-title: The JPEG still picture compression standard
  publication-title: IEEE Transactions on Consumer Electronics,
  doi: 10.1109/30.125072
– volume: 12
  start-page: e1004927
  issue: 6
  year: 2016
  ident: bib3
  article-title: Model constrained by visual hierarchy improves prediction of neural responses to natural scenes
  publication-title: PLoS Computational Biology,
  doi: 10.1371/journal.pcbi.1004927
– volume: 19
  start-page: 356
  year: 2016
  ident: bib96
  article-title: Using goal-driven deep learning models to understand sensory cortex
  publication-title: Nature Neuroscience,
  doi: 10.1038/nn.4244
– volume: 22
  start-page: 4
  year: 2011
  ident: bib6
  article-title: Could information theory provide an ecological theory of sensory processing?
  publication-title: Network: Computation in Neural Systems,
  doi: 10.3109/0954898X.2011.638888
– volume-title: The FBI story
  year: 1959
  ident: bib53
– volume: 3
  start-page: 300
  year: 1986
  ident: bib90
  article-title: Window of visibility: A psychophysical theory of fidelity in time-sampled visual motion displays
  publication-title: Journal of The Optical Society of America A-optics Image Science and Vision,
  doi: 10.1364/JOSAA.3.000300
– year: 2017
  ident: bib47
  article-title: Adam: A method for stochastic optimization
– volume: 27
  start-page: 852
  issue: 4
  year: 2010
  ident: bib50
  article-title: Divisive normalization image quality metric revisited
  publication-title: Journal of the Optical Society of America A,
  doi: 10.1364/JOSAA.27.000852
– volume: 23
  start-page: 1495
  year: 1983
  ident: bib44
  article-title: The relationship between spectral sensitivity and spatial sensitivity for the primate r-g x-channel
  publication-title: Vision Research,
  doi: 10.1016/0042-6989(83)90161-X
– volume: 122
  start-page: 18
  issue: 2
  year: 2013
  ident: bib87
  article-title: High frame rates and human vision: A view through the window of visibility
  publication-title: SMPTE Motion Imaging Journal,
  doi: 10.5594/j18266XY
– year: 2020
  ident: bib31
  article-title: Beyond accuracy: Quantifying trial-by-trial behaviour of CNNs and humans by measuring error consistency
– volume: 26
  start-page: 98
  issue: 1
  year: 2009
  ident: bib86
  article-title: Mean squared error: Love it or leave it? A new look at signal fidelity measures
  publication-title: IEEE Signal Processing Magazine,
  doi: 10.1109/MSP.2008.930649
– volume: 81
  start-page: 943
  issue: 4
  year: 2014
  ident: bib65
  article-title: Statistical wiring of thalamic receptive fields optimizes spatial sampling of the retinal image
  publication-title: Neuron,
  doi: 10.1016/j.neuron.2013.12.014
– volume: 4
  start-page: 559
  issue: 4
  year: 1992
  ident: bib4
  article-title: Understanding retinal color coding from first principles
  publication-title: Neural Computation,
  doi: 10.1162/neco.1992.4.4.559
– volume: 176
  start-page: 156
  year: 2020
  ident: bib34
  article-title: Color illusions also deceive CNNs for low-level vision tasks: Analysis and implications
  publication-title: Vision Research,
  doi: 10.1016/j.visres.2020.07.010
– start-page: 3533
  volume-title: Proceedings of the Neural Information Processing Systems, 30
  year: 2017
  ident: bib11
  article-title: Eigen-distortions of hierarchical representations
– volume: 15
  start-page: 189
  issue: 1
  year: 2006
  ident: bib37
  article-title: Regularization operators for natural images based on nonlinear perception models
  publication-title: IEEE Transactions on Image Processing,
  doi: 10.1109/TIP.2005.860345
– volume: 117
  start-page: 26562
  issue: 43
  year: 2020
  ident: bib28
  article-title: Performance vs. competence in human–machine comparisons
  publication-title: Proceedings of the National Academy of Sciences of the United States of America,
  doi: 10.1073/pnas.1905334117
– volume-title: Deep Learning
  year: 2016
  ident: bib35
– volume: 9
  start-page: e86481
  issue: 2
  year: 2014
  ident: bib38
  article-title: Spatio-chromatic adaptation via higher-order canonical correlation analysis of natural images
  publication-title: PloS One,
  doi: 10.1371/journal.pone.0086481
– volume-title: Vision: A computational approach
  year: 1982
  ident: bib61
– volume: 128
  start-page: 359
  year: 1981
  ident: bib20
  article-title: Relation between the Karhunen Loève and cosine transforms
  publication-title: IEE Proceedings F Communications, Radar and Signal Processing,
  doi: 10.1049/ip-f-1.1981.0061
– volume: 21
  start-page: 335
  issue: 6
  year: 2020
  ident: bib54
  article-title: Backpropagation and the brain
  publication-title: Nature Reviews Neuroscience,
  doi: 10.1038/s41583-020-0277-3
– volume-title: Proceedings Advances in Neural Information Processing Systems,
  year: 2019
  ident: bib49
  article-title: Brain-like object recognition with high-performing shallow recurrent ANNs
– start-page: 117
  year: 1994
  ident: bib66
  article-title: Chromatic-achromatic multiplexing in human color vision
– volume-title: 5th International Conference on Learning Representations (ICLR)
  year: 2017
  ident: bib9
  article-title: Multilayer recurrent network models of primate retinal ganglion cell responses
– volume-title: arxiv Comp. Sci
  year: 2016
  ident: bib1
  article-title: TensorFlow: Large-scale machine learning on heterogeneous distributed systems
– volume: 21
  start-page: 1996
  issue: 9
  year: 2021
  ident: bib2
  article-title: Contrast sensitivity is formed by visual experience and task demands
  publication-title: Journal of Vision
  doi: 10.1167/jov.21.9.1996
– volume: 13
  start-page: 51
  issue: 1
  year: 2012
  ident: bib18
  article-title: Normalization as a canonical neural computation
  publication-title: Nature Reviews Neuroscience,
  doi: 10.1038/nrn3136
– volume: 64
  start-page: 384
  issue: 6
  year: 1957
  ident: bib42
  article-title: An opponent-process theory of color vision
  publication-title: Psychological Review,
  doi: 10.1037/h0041403
– volume: 13
  start-page: 55
  issue: 2
  year: 1973
  ident: bib94
  article-title: A mathematical theory of the functional dynamics of cortical and thalamic nervous tissue
  publication-title: Kybernetik,
  doi: 10.1007/BF00288786
– volume: 26
  start-page: 3142
  issue: 7
  year: 2017
  ident: bib98
  article-title: Beyond a gaussian denoiser: Residual learning of deep CNN for image denoising
  publication-title: IEEE Transactions on Image Processing,
  doi: 10.1109/TIP.83
– volume: 63
  start-page: 693
  year: 1975
  ident: bib41
  article-title: Digital image processing
  publication-title: Proceedings of the IEEE,
  doi: 10.1109/PROC.1975.9801
– year: 2009
  ident: bib43
  publication-title: Natural image statistics: A probabilistic approach to early computational vision
  doi: 10.1007/978-1-84882-491-1
– volume: 70
  start-page: 1458
  issue: 12
  year: 1980
  ident: bib52
  article-title: Contrast masking in human vision
  publication-title: Journal of the Optical Society of America,
  doi: 10.1364/JOSA.70.001458
– year: 2019
  ident: bib63
  article-title: In praise of artifice reloaded: Caution with natural image databases in modeling vision
  publication-title: Frontiers in Neuroscience
– volume: 10
  start-page: 16277
  year: 2020
  ident: bib12
  article-title: Evidence for the intrinsically nonlinear nature of receptive fields in vision
  publication-title: Scientific Reports,
  doi: 10.1038/s41598-020-73113-0
– volume: 40
  start-page: 1711
  issue: 13
  year: 2000
  ident: bib81
  article-title: The spectral sensitivities of the middle- and long-wavelength-sensitive cones derived from measurements in observers of known genotype
  publication-title: Vision Research,
  doi: 10.1016/S0042-6989(00)00021-3
– volume: 17
  start-page: e1009028
  issue: 6
  year: 2021
  ident: bib13
  article-title: Learning divisive normalization in primary visual cortex
  publication-title: PLoS Computational Biology,
  doi: 10.1371/journal.pcbi.1009028
– volume: 4
  start-page: 196
  issue: 2
  year: 1992
  ident: bib5
  article-title: What does the retina know about natural scenes?
  publication-title: Neural Computation,
  doi: 10.1162/neco.1992.4.2.196
– volume-title: International Conference on Learning Representations, ICLR
  year: 2022
  ident: bib40
  article-title: On the relation between statistical learning and perceptual distances
– volume: 22
  start-page: 6158
  year: 2002
  ident: bib74
  article-title: Space and time maps of cone photoreceptor signals in macaque lateral geniculate nucleus
  publication-title: Journal of Neuroscience,
  doi: 10.1523/JNEUROSCI.22-14-06158.2002
– volume: 69
  start-page: 1340
  issue: 10
  year: 1979
  ident: bib46
  article-title: Motion and vision. ii. stabilized spatio-temporal threshold surface
  publication-title: Journal of the Optical Society of America,
  doi: 10.1364/JOSA.69.001340
– volume: 356
  start-page: 716
  year: 1992
  ident: bib73
  article-title: Spatial structure of cone inputs to receptive fields in primate lateral geniculate nucleus
  publication-title: Nature,
  doi: 10.1038/356716a0
– start-page: 747
  volume-title: Proceedings International Conference on Pattern Recognition (ICPR)
  year: 2021
  ident: bib80
  article-title: Deep universal blind image denoising
– volume: 10
  issue: 18
  year: 2020
  ident: bib56
  article-title: Spatio-chromatic information available from different neural layers via gaussianization
  publication-title: Journal of Mathematical Neuroscience,
– volume: 20
  start-page: 525
  year: 1974
  ident: bib60
  article-title: The effects of a visual fidelity criterion of the encoding of images
  publication-title: IEEE Transactions on Information Theory,
  doi: 10.1109/TIT.1974.1055250
– volume: 1
  start-page: 417
  issue: 1
  year: 2015
  ident: bib48
  article-title: Deep neural networks: A new framework for modeling biological vision and brain information processing
  publication-title: Annual Review of Vision Science,
  doi: 10.1146/vision.2015.1.issue-1
– start-page: 8174
  volume-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
  year: 2018
  ident: bib82
  article-title: Scale-recurrent network for deep image deblurring
– volume-title: International Conference on Learning Representations (ICLR)
  year: 2019
  ident: bib32
  article-title: Imagenet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
– volume: 44
  start-page: 127
  issue: 1
  year: 1997
  ident: bib59
  article-title: Characterization of the human visual system threshold performance by a weighting function in the gabor domain
  publication-title: Journal of Modern Optics,
  doi: 10.1080/09500349708232904
– volume: 13
  issue: 10
  year: 2018
  ident: bib64
  article-title: Derivatives and inverse of cascaded linear+nonlinear neural models
  publication-title: PLoS One,
– year: 2020
  ident: bib29
  article-title: Deep neural models for color discrimination and color constancy
– volume: 17
  start-page: 663
  issue: 5-6
  year: 2004
  ident: bib72
  article-title: Nonlinear v1 responses to natural scenes revealed by neural network analysis
  publication-title: Neural Networks,
  doi: 10.1016/j.neunet.2004.03.008
– volume-title: From Marr's Vision to the problem of human intelligence (CBMM Memo No. 118)
  year: 2021
  ident: bib71
– volume-title: International Conference on Learning Representations (ICLR
  year: 2017
  ident: bib7
  article-title: End-to-end optimized image compression
– volume-title: The stranger
  year: 1946
  ident: bib92
– volume: 10
  start-page: e1003963
  issue: 12
  year: 2014
  ident: bib15
  article-title: Deep neural networks rival the representation of primate it cortex for core visual object recognition
  publication-title: PLoS Computational Biology,
  doi: 10.1371/journal.pcbi.1003963
– volume-title: From understanding computation to understanding neural circuitry (AI Memo No. AIM-357)
  year: 1976
  ident: bib62
– start-page: 81
  year: 1997
  ident: bib67
  article-title: Color detection and color contrast discrimination thresholds
  publication-title: In Proceedings of the OSA Annual Meeting ILS-XIII
– volume: 18
  start-page: 1
  issue: 153
  year: 2018
  ident: bib10
  article-title: Automatic differentiation in machine learning: A survey
  publication-title: Journal of Machine Learning Research,
– volume: 19
  start-page: 8
  issue: 4
  year: 2019
  ident: bib21
  article-title: A computational-observer model of spatial contrast sensitivity: Effects of wave-front-based optics, cone-mosaic structure, and inference engine
  publication-title: Journal of Vision,
  doi: 10.1167/19.4.8
– volume: 197
  start-page: 551
  year: 1968
  ident: bib17
  article-title: Application of Fourier analysis to the visibility of gratings
  publication-title: Journal of Physiology,
  doi: 10.1113/jphysiol.1968.sp008574
– volume: 111
  start-page: 8619
  year: 2014
  ident: bib97
  article-title: Performance-optimized hierarchical models predict neural responses in higher visual cortex
  publication-title: Proceedings of the National Academy of Sciences of the United State of America,
  doi: 10.1073/pnas.1403112111
– start-page: 1
  volume-title: Proceedings Human Vision and Electronic Imaging HVEI16,
  year: 2016
  ident: bib89
  article-title: The pyramid of visibility
– volume: 6
  start-page: 27755
  year: 2016
  ident: bib19
  article-title: Comparison of deep neural networks to spatio-temporal cortical dynamics of human visual object recognition reveals hierarchical correspondence
  publication-title: Scientific Reports,
  doi: 10.1038/srep27755
– volume: 15
  start-page: 2036
  issue: 8
  year: 1998
  ident: bib75
  article-title: Statistics of cone responses to natural images: Implications for visual coding
  publication-title: Journal of the Optical Society of America A,
  doi: 10.1364/JOSAA.15.002036
– volume: 46
  start-page: 2412
  issue: 15
  year: 2006
  ident: bib68
  article-title: Predicting the motion after-effect from sensitivity loss
  publication-title: Vision Research,
  doi: 10.1016/j.visres.2006.01.019
SSID ssj0020501
Score 2.4716845
Snippet Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal...
Three decades ago, Atick et al. suggested that human frequency sensitivity may emerge from the enhancement required for a more efficient analysis of retinal...
SourceID pubmedcentral
proquest
pubmed
crossref
SourceType Open Access Repository
Aggregation Database
Index Database
Enrichment Source
StartPage 8
Title Contrast sensitivity functions in autoencoders
URI https://www.ncbi.nlm.nih.gov/pubmed/35587354
https://www.proquest.com/docview/2666908755
https://pubmed.ncbi.nlm.nih.gov/PMC9145138
Volume 22
WOSCitedRecordID wos001001008200007&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAON
  databaseName: DOAJ Directory of Open Access Journals
  customDbUrl:
  eissn: 1534-7362
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0020501
  issn: 1534-7362
  databaseCode: DOA
  dateStart: 20010101
  isFulltext: true
  titleUrlDefault: https://www.doaj.org/
  providerName: Directory of Open Access Journals
– providerCode: PRVHPJ
  databaseName: ROAD: Directory of Open Access Scholarly Resources
  customDbUrl:
  eissn: 1534-7362
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0020501
  issn: 1534-7362
  databaseCode: M~E
  dateStart: 20010101
  isFulltext: true
  titleUrlDefault: https://road.issn.org
  providerName: ISSN International Centre
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1Lj9MwELbYBSEuiDflUQWBkBBKyDpxbB8X6IpDt4DURb1FieNAUTcpbVqtOPDbmXGcNIEelgOXJHKtuPKMJjOeme8j5IUCr5hrLl2dJQiqzaibZMqHmCfMAxlAxKJNo_CYTyZiNpOfbFnR2tAJ8KIQFxdy-V9FDWMgbGyd_Qdxty-FAXgGocMVxA7XSwke8aZWybp6vcbadEsOgZ-vtmo82VQl4ldmtvp9j3Nat5ybFC_m48sKPNIttg2autjE6xwgjE09wGdQs69tMU95rn-6X5DQqGmiacN-TP4vzk1-_n1pu4WUXpS7o3GbDAL7BZPeHvcOJiCmRUxT2bOlocsDa2z1njFrgCntKFrXmor9Nj7CLPP3cutR6kWe2H3Jmuz95GN8cjYex9PRbPpy-cNFjjHMxVvClQNylXIm0Qae_hq14bnPfIuuW_9FC2YLy73ZLdZ3X_6KSf4sre34KtNb5KaVo3NcK8dtckUXd8j1U1tGcZd4jY44HR1xWh1x5oXT1ZF75OxkNH33wbXEGa4KuKhclWVc-TTnUviZEqlQqQpzFklFZUrVkchDP9SpTDMWMCVyLVmesQSh3dIwC4PgPjksykI_JE5EkyzSgvOU6vAIblpxcClzhb66YGpAXjX7ESuLKo_kJovYRJcRj2HvYkrjKBYD8rydu6yxVPbOetZsawymDvNXSaHLzToGXzKSyMDABuRBvc3te5AlgAcsHBDeE0A7AWHU-78U828GTl0iWXUgHl1i3cfkxk7Tn5DDarXRT8k1ta3m69WQHPCZGJpTnaFRrd-EiJZs
linkProvider ISSN International Centre
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Contrast+sensitivity+functions+in+autoencoders&rft.jtitle=Journal+of+vision+%28Charlottesville%2C+Va.%29&rft.au=Li%2C+Qiang&rft.au=Gomez-Villa%2C+Alex&rft.au=Bertalm%C3%ADo%2C+Marcelo&rft.au=Malo%2C+Jes%C3%BAs&rft.date=2022-05-19&rft.issn=1534-7362&rft.eissn=1534-7362&rft.volume=22&rft.issue=6&rft.spage=8&rft_id=info:doi/10.1167%2Fjov.22.6.8&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1534-7362&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1534-7362&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1534-7362&client=summon