Mathematical modeling of perception with focus on expectation effect and cross-modality

Abstract

Presented at ANSYS/Optis conference on "Digital Twin" in the French embassy in Japan.

Full text

״੓͹਼ཀྵϠυϨϱή
غଶްՎʀέϫηϠʖξϩΝ஦ৼͶ
Mathematical modeling of perception
with focus on expectation effect and cross-modality
༆ᖔ ऴ٤
Hideyoshi Yanagisawa, PhD
Associate Professor of Design Engineering
Graduate School of Engineering, The University of Tokyo
෼ཀྵͳ״੓͹ؔͶʹ͹Γ͑͵๑ଉ੓͍͗Ζ͹͖ʃ
෼ཀྵ
3K\VLFDO
״੓
3HUFHSWXDO
Artifacts Human
䠛
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 2
$JHQGD
״੓અܯָ
.DQVHL GHVLJQHQJLQHHULQJ
غଶްՎཀྵ࿨ ʛέϫηϠʖξϩ͹๑ଉ੓
([SHFWDWLRQHIIHFWWKHRU\t$ODZRI&URVVPRGDOSHUFHSWLRQ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 3
״੓અܯ
.DQVHL GHVLJQ
Loudness
Fluctuation
strength
Sharpness
Roughness
Structure
Behavior
Physical
phenomena Perception Emotion
Meanings
䝣䜯䞁䞉䝰䞊䝍
䝝䜴䝆䞁䜾 㡢䞉ὶయ䞉᣺ື 㡢㡪▱ぬ≉ᚩ
ᚰᆅ䜘䛔
Function/performance
Kansei design
e.g. 䝗䝷䜲䝲䞊
Ϡό͹ؽ೵ʀ੓೵ͶՅ͓ͱɾਕؔ͹״ְΏ৚ಊͶ૎͓Ζ
ܨݩʤH[SHULHQFHʥΝଲেͳͤΖઅܯ
Experience
Engineering design
୙ᛌ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 4

״੓અܯָ
.
.DQVHL
GHVLJQHQJLQHHULQJ
״੓અܯ͹ͪΌ͹ཀྵ࿨ɾ๏๑࿨ɾघ๑
״੓඾࣯͹ࢨඬԿ (YDOXDWLRQFULWHULD
״੓͹਼ཀྵϠυϨϱή 0DWKHPDWLFDOPRGHOLQJ
״੓અܯ͹๏๑࿨ɾࢩԋٗढ़ɾࣰભԢ༽ 0HWKRGDSS
੣඾Խ͹չԽઅܯ ࣙಊӣ఺͹҈ৼ״
૤ࡠकର״
ોࡒʀτέηοϡ͹࣯״
HJ
6RXQGYLEUDWLRQ 0DWHULDOWH[WXUH %HKDYLRU
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 5
״੓͹ࢨඬԿͳ״੓અܯʤչԽઅܯʥ
3HUFHLYHGTXDOLW\RISURGXFWVRXQG
չԽ੓͹ࢨඬԿͳ໪ඬԽ͹અఈ
અܯմʁ
ుಊૻ෫ؽΝώϋͲ௽ΖߑଆͶΓΕ৾ಊઊԓ
ఽୣر༫ིմੵʁ ຌର৾ಊ͗ࢩഓదͶӪڻ
೧ౕೖຌؽփָճৈʤٗढ़ʥ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 6
ຌ࣯ద͵໲ୌ $QHVVHQWLDOLVVXH
״੓൝ʰ1HZWRQbVODZʱ͹෈ࡑ
ଲেͶғଚ͢͵͏ԍ៹Ն೵͵Ҳൢ๑ଉ
Expectation Effect Theory
Yanagisawa, H. (2016) J. Sens. Stud.
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 7
๑ଉ͹ࡠΕ๏ʁ ؾ೴ͳԍ៹
,VDDF1HZWRQ

-RKDQQHV.HSOHU

ʰίϕϧʖ͹๑ଉʱ
ϔϧʖΦ͹؏଎υʖνͶخͰ͚
ؾ೴ద๏๑ LQGXFWLRQ
˥
࿯੗͗ʹ͹Γ͑Ͷಊ͚͖
ʰຬ༙Ӏྙ͹๑ଉʱ
ΝՀఈʤޮཀྵʥͳͪ͢
ԍ៹ద๏๑ GHGXFWLRQ
˥
෼ରҲൢ͗ɾ͵ͧɾ
ͨ͹Γ͑Ͷಊ͚͖
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 8
7\FKR %UDKH


Size-weight illusion (SWI)
୉͗͘͠ҡ͵ΖಋҲ॑ྖ͹෼ରΝർֳͤΖͳɾ
ঘ͠͏๏͚͗॑״ͣΖɿ
॑͠ >J@
ʻ
॑͠״>"@
ʾ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 9
㼣㼑㼕㼓㼔㼠
ᣢ䛱ୖ䛢䜛๓
Prior expectation
ᐇ㝿䛾㔜䛥
=
ᣢ䛱ୖ䛢䛯ᚋ
Posterior perception
<
ண ㄗᕪ䜢㐣኱ホ౯
ண ㄗᕪ ண ㄗᕪ
<
6L]HZHLJKLOOXVLRQͺ͵ًͧ͞Ζ͖ʃ
ぢ䛯┠䛾㔜䛥
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 10
പϭ΢ϱ͗੼ϭ΢ϱͶ
Morrot, G. et al. 2001, Brain and Language.
(Université de Bordeaux)
$QH[DPSOHRIH[SHFWDWLRQHIIHFWFURVVPRGDO
䝷䝧䝹䛷࿡䛾ホ౯䛜ኚ䜟䜛
Plassmann, H. et al. 2008, PNAS.
(CalTech)
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 11
どぬ䛻䜘䜛ᮇᚅ ぢ䛺䛔䛷ᣢ䛱ୖ䛢䜛
䠄ᮇᚅ䛺䛧䠅
㔜䛥
䊻ぢ䛶䛛䜙ᣢ䛱ୖ䛢䜛
䠄ᮇᚅ䛒䜚䠅
ྠ໬
㼍㼟㼟㼕㼙㼕㼘㼍㼠㼕㼛㼚
ᑐẚ
㼏㼛㼚㼠㼞㼍㼟㼠
ண ㄗᕪ 㼑㼞㼞㼛㼞
ᮇᚅ䛻㏆䛵䛟
ᮇᚅ䛛䜙㞳䜜䜛
䠎✀㢮䛾ᮇᚅຠᯝ 䡚ᑐẚ䛸ྠ໬
ᑐẚ䠄contrast䠅䠖 ᮇᚅ䛸䛾ᕪ䜢㐣኱ホ౯
ྠ໬䠄assimilation䠅䠖 ᮇᚅ䛸䛾ᕪ䜢㐣ᑠホ౯
౛䠅 㔜䛥䛾▱ぬ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 12

ಋԿͳଲർʁ ࣎ؔ࣢ʀۯؔ࣢
ྠ໬
assimilation
ᑐẚ
contrast
᫬㛫ⓗᩥ⬦
Temp o ra l
✵㛫ⓗᩥ⬦
Spatial
Size ĺweight
Color ĺsmell
Ebbinghaus illusionColor assimilation
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 13
ਈܨෘߺԿ͹ݬཀྵ
1HXUDOFRGLQJSULQFLSOHV
T
┿䛾≀⌮㔞
䠄౛䠊㔜䛥䠅
1
r
2
r
3
r
N
r
䠖
Nಶ䛾⚄⤒⣽⬊
䛾Ⓨⅆ⋡䝧䜽䝖䝹
Firing rate
ˆ
T
R
≀⌮㔞䛾᥎ᐃ
䠄౛䠊ឤ䛨䜛㔜䛥䠅
Encoding
䠄➢ྕ䠅
Decoding
䠄᚟ྕ䠅
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 14
」ᩘ䛾⚄⤒⣽⬊(population)䛾
Ⓨⅆ⋡ศᕸ䛷≀⌮㔞䜢➢ྕ໬
Pouget, A., et al. (2000)
Nature Rev. Neurosci.
ˆ
T
Likelihood
Physical Perceptual
ϗ΢ζΠϱ෰ߺԿ Bayesian decoding
ᮇᚅ ឤぬ
Prior Likelihood
()p
T
▱ぬ
ˆ
T
Posterior
(| )pR
T
(|)pR
T
|(|) () ( )ppRpR
TTT
vu
Prior
䠄ᮇᚅ䠅
Likelihood
䠄ឤぬ䠅
Posterior
䠄▱ぬ䠅
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 15
ᮇᚅ䛸ឤぬ䛾᥃䛡⟬䛷▱ぬ䛜Ỵ䜎䜛䠊
෈ࣰ֮੓ͳ֐ཛྷ 8QFHUWDLQW\QRLVH
Prior Likelihood
Posterior
()p
T
(| )pR
T
(|)pR
T
p
P
l
P
2
p
V
p
p
୙☜ᐇᛶ
Uncertainty
እ஘
Noise
22
ˆl
p
pl
PP
V
TV
v
ศᩓ䛾㏫ᩘ䛻䜘䜛ᮇᚅ್䛾㔜䜏௜䛡⥺ᙧ࿴
䠄ṇつศᕸ䛻ᚑ䛖ሙྜ䠅
2
l
V
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 16
ᮇᚅ䛸ឤぬ䛾䛖䛱ಙ㢗ᛶ䛜㧗䛔᪉䜢▱ぬ䛩䜛䠊

෈ࣰ֮੓ ʾ ֐ཛྷ ͹৖߻
ᑬᗘ㛵ᩘ䛻ᐤ䜛
ண 䜘䜚⢭ᗘ䛾䜘䛔ឤぬ䛾᪉䛜㢗䜚䛻䛺䜛
ˆ
T
Prior
୙☜ᐇ䛺ᮇᚅ
Likelihood
Posterior
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 17
෈ࣰ֮੓ ʽ ֐ཛྷ ͹৖߻
஦๓ศᕸ䛻ᐤ䜛
ឤぬ䜘䜚⢭ᗘ䛾䜘䛔ண 䛾᪉䛜㢗䜚䛻䛺䜛
ˆ
T
Prior
☜ᐇ䛺ᮇᚅ Likelihood
Posterior
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 18
ᮇᚅ䛸ឤぬ䛾⥘ᘬ䛝
ްིదෘߺԿ Efficient encoding
஦๓ศᕸ䛜୍ᵝ䛸௬ᐃ䛾ሙྜ䛾
⚄⤒⣽⬊㞟ᅋ䛾Ⓨⅆ☜⋡ศᕸ
෼ཀྵۯؔͶ͕͜Ζਈܨࡋ๖܊͹൅Ւི֮෾ා
T
T

() ()Fpxdx
T
T
f
³
)(p
T
T
1
r
2
r
3
r
4
r
5
r
R
Prior
䠄ᮇᚅ䠅
Likelihood
ᮇᚅ್࿘㎶䛻ከ䛟䛾⚄⤒⣽⬊䜢๭䜚ᙜ䛶䜛䠊⋓ᚓ䛩䜛᝟ሗ㔞䛾᭱኱໬
஦๓ศᕸ䛾〈㔝䛿
ᑬᗘ㛵ᩘ䛜ᕥྑ㠀
ᑐ⛠䛻䜖䛜䜐
1
()F
TT


Wei & Stocker (2015) Nature Neurosci.
Barlow, H. (1961) Sensory Communication, MIT
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 19
໮਼ౕؖ͹ඉଲস੓ͶΓΖൕ൅ްՎ
5HSXOVLRQELDVDJDLQVWH[SHFWDWLRQFDXVHVFRQWUDVW
ӷ҄ ⁓⁥⁥⁛ ⁛⁞⁓󰝧⁛⁡⁠ ݣൔ ⁕⁡⁠󰝧⁤⁓⁥󰝧
ˆ
T
ˆ
T
ண ㄗᕪ 㼑㼞㼞㼛㼞 ண ㄗᕪ 㼑㼞㼞㼛㼞
ᮇᚅ್䛻ᘬ䛝ᐤ䛫䜙䜜䜛
䠄ㄏᘬస⏝ attraction bias䠅
ᮇᚅ್䛛䜙㞳䜜䜛
䠄཯Ⓨస⏝ repulsion bias䠅
prior likelihood
䠄ᑐ⛠䠅
prior likelihood
䠄㠀ᑐ⛠䠅
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 20

غଶͳ״ְ͹౹߻ͶΓΖஎְϠυϩ
㻼㼑㼞㼏㼑㼜㼠㼕㼛㼚㻌㼙㼛㼐㼑㼘㻌㼎㼍㼟㼑㼐㻌㼛㼚㻌㼚㼑㼡㼞㼍㼘㻌㼏㼛㼐㼕㼚㼓㻌㼜㼞㼕㼚㼏㼕㼜㼘㼑㼟
㻼㼞㼕㼛㼞㻌㼗㼚㼛㼣㼘㼑㼐㼓㼑
㻼㼞㼕㼛㼞㻌㼕㼚㼒㼛㼞㼙㼍㼠㼕㼛㼚
㻸㼕㼗㼑㼘㼕㼔㼛㼛㼐
㻼㼛㼟㼠㼑㼞㼕㼛㼞
㻿㼠㼕㼙㼡㼘㼕
㻼㼞㼕㼛㼞
㻼㼑㼞㼏㼑㼕㼢㼑㼐
㼜㼔㼥㼟㼕㼏㼍㼘㻌㼢㼍㼘㼡㼑
(| )pR
T
()p
T
T
ˆ
T
(|)pR
T
㻼㼔㼥㼟㼕㼏㼍㼘㻌㼢㼍㼘㼡㼑
㻱㼤㼜㼑㼏㼠㼍㼠㼕㼛㼚
㻿㼑㼚㼟㼍㼠㼕㼛㼚
Körding, K., et al. (2004). Nature.
Bayesian decoding
㼜㼞㼕㼛㼞 㼘㼕㼗㼑㼘㼕㼔㼛㼛㼐
㼜㼛㼟㼠㼑㼞㼕㼛㼞
|(|) () ( )ppRpR
TTT
vu
Wei X et al. (2015) Nat. Neurosci.
T
ˆ
T
Efficient encoding
Yanagisawa, H. (2016) J. Sens. Stud.
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 21
غଶްՎ͹਼ؖϠυϩ
$IXQFWLRQDOPRGHORIH[SHFWDWLRQHIIHFW
Expectation Effect
Assimilation / contrast
Prediction error
Uncertainty
Noise
f
ᮇᚅ䛸ᐇ㝿䛾ᕪ
ᮇᚅ䛾୙☜ᐇᛶ
ឤぬ䛻ΰධ䛩䜛እ஘
ྠ໬䞉ᑐẚ䠈䛭䛾ᗘྜ䛔
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 22
㼜㼞㼕㼛㼞 㼘㼕㼗㼑㼘㼕㼔㼛㼛㼐
㼜㼛㼟㼠㼑㼞㼕㼛㼞
T
ˆ
T
ಋԿͳଲർ͹ড়݇ʤVLPXODWLRQʥ
Expectation effect䠄
䠄ᮇᚅຠᯝ䠅
020 40 60 80 100 120
-1
-0.5
0
0.5
1
1.5
2
Prediction error
䠄ண ㄗᕪ䠖ᮇᚅ䛸ᐇ㝿䛾ᕪ䠅
ᑐẚ 㼏㼛㼚㼠㼞㼍㼟㼠
ྠ໬ 㼍㼟㼟㼕㼙㼕㼘㼍㼠㼕㼛㼚
Yanagisawa, H. (2016) J. Sens. Stud.
ண ㄗᕪ䛾ቑ኱䛻క䛔䠈ྠ໬䛛䜙ᑐẚ䛻᥎⛣
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 23
ݗৄࡒྋ Size-weight illusion - SWI
॑͠ >J@
ʻ
॑͠״ʤଲർʥ
ʾ
Charpentier, A (1891). Arch. Physiol. Norm. Pathol.
:
Flanagan, J. R. et al. (2000). Nature Neurosci.
॑͠״ʤಋԿʥ
ʽ
Thesis 䠄ṇ䠅䠖
΍͢ɾಋԿ΍ਫ਼ͣΖ͵Δͻʀʀʀ
Antithesis 䠄཯䠅䠖
Synthesis 䠄ྜ䠅䠖
ᮇᚅຠᯝ⌮ㄽ Expectation effect theory
Yanagisawa, H. (2016) J. Sens. Stud.
1HZ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 24

6:,H[SHULPHQWDOUHVXOW
180 280 380 480 580 680 780 880 980 1080
ᑠ䝃䞁䝥䝹䛸ẚ㍑䛧䛯኱䝃䞁䝥䝹䛾㔜䛥ឤホ౯
䠄
Expectation effect
䠅
ண᝿䛸ᐇ㝿䛾㔜䛥䛾ᕪ䠄
Prediction error
䠅
(g)
small uncertainty & small noise
big uncertainty & small noise
small uncertainty & big noise
big uncertainty & big noise
㔜䛔
-2
䜔䜔㔜䛔
-1
ྠ⛬ᗘ
0
䜔䜔㍍䛔
1
㍍䛔
2
኱䛝䛔᪉䛜
Yanagisawa, H. (2016) J. Sens. Stud.
ᑐẚ
contrast
ྠ໬
assimilation
Anti-SWI
䠄኱䛝䛔᪉䛜㔜䛔䠅
SWI
䠄ᑠ䛥䛔᪉䛜㔜䛔䠅
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 25
எְ͠Ηͪ඾࣯
඾࣯ͶଲͤΖغଶ
඾࣯
ଲർ
༩଎ޣࠫ
ࣰࡏ͹඾࣯
״ಊ
ಋԿ
գঘ඲Ճ
غଶްՎ͹қັʁغଶҐ৏͹৖߻
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 26
҈સͶଲͤΖغଶ
҈સ੓
ଲർ
༩଎ޣࠫ
எְ͠Ηͪ҈સ
ࣰࡏ͹҈સ੓
ࣨ๮
ಋԿ
ڒ༲
غଶްՎ͹қັʁغଶҐԾ͹৖߻
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 27
غଶްՎ͹ΉͳΌ
ಋԿ
DVVLPLODWLRQ
ଲർ
FRQWUDVW
غଶҐ৏
([FHHGLQJH[SHFWDWLRQ
ʹ
գঘ඲Ճ
˗
״ಊ
غଶҐԾ
%HORZH[SHFWDWLRQ
˕
ڒ༲
ʹʹ
ࣨ๮
༩଎ޣࠫ HUURU VPDOO %LJ
෈ࣰ֮੓
XQFHUWDLQW\ 6PDOO 6PDOO
֐ཛྷ QRLVH %LJ 6PDOO
غଶްՎ
غଶͳ͹ࠫ
How?
What?
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 28

״੓ͺغଶͶଲ͢ͱ૮ଲదͲ͍Ζ
7KHRU\RIUHODWLYLW\LQSHUFHSWLRQ
෼ཀྵ
3K\VLFDO
״੓
3HUFHSWXDO
αϱτέηφ
&RQWH[WXDO
য়ڱɾ؂ڧ
઴ޛؖܐ
எࣟɾชԿ
࢙ܻ
VWLPXOL
غଶ
H[SHFWDWLRQ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 29
0HUFLEHDXFRXS͍Ε͗ͳ͑͟͡͏Ήͪ͢
Reality is merely an illusion, albeit a very
persistent one.
…
Physics and psychology are only different
attempts to link our experiences together
by way of systematic thought.
Albert Einstein
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 30
ึଏࣁྋ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 31
඾࣯͹எְͳ͢ͱ͹״੓඾࣯
౲Ε઴඾࣯
0XVWEH
੓೵඾࣯
3HUIRUPDQFH
ຳྙ඾࣯
$WWUDFWLYH
ैଏ
)XOILOOHG
෈ଏ
෈ຮଏ
ޒٮຮଏ
6DWLVILHG
Kano model
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 32

ॄ஄ෘߺԿ 3RSXODWLRQFRGLQJ
」
」ᩘ䛾⚄⤒⣽⬊
(
㞟ᅋ
)
䛾Ⓨⅆ⋡ศᕸ䛷≀⌮㔞䜢➢ྕ໬
⚄⤒⣽⬊䛾Ⓨⅆ⋡ r[Hz]
40ᗘ䛾⚄⤒⣽⬊⩌䛾Ⓨⅆ⋡䛸
ᑬᗘ㛵ᩘ
᪉ྥ䛻ᑐ䛩䜛⚄⤒⣽⬊⩌䛾Ⓨⅆศᕸ
Pouget, A., et al. (2000) Nature Rev. Neurosci.
ˆ
T
ᑬᗘ㛵ᩘ
(|)pR
T
౛䠅 ື䛝䛾᪉ྥ䛾coding @MT㔝䠄V5䠅
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 33
ߡࡱʁ ༩଎ޣࠫ
•ঘ͠͵༩଎ޣࠫͲͺغଶͶಋԿ
–գঘ඲Ճʀໃࢻ
–ݸΔΗͪ೟எࣁݱ͹ઇ༁
020 40 60 80 100 120
-1
-0.5
0
0.5
1
1.5
2
ண ㄗᕪ
ᑐẚ
ྠ໬
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo
•༩଎ޣࠫ͗୉͚͘͵Ζͳଲർ͗൅ਫ਼
–״ְ೘ྙ͹ΰ΢ϱΝ৏͝Ζ
–஭қͳ೟ࣟΝଇ਒
–న઀͵ߨಊʤյඈʀંۛʥ
–ָस͹ؽճ
34
ߡࡱʁ ෈ࣰ֮੓
෈ࣰ֮੓͗ঘ͠͏ΆʹغଶްՎ͹کౕ͗୉͘͏
ܨݩʀஎࣟ͗๝෍͵ਕɾ
਎ۛ੓ɾఱܗ੓͗߶͏ଲেΆʹ
ಋԿʀଲർ͹͏ͥΗ΍ک͚ਫ਼ͣΖɿ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 35
020 40 60 80 100 120
-1
-0.5
0
0.5
1
1.5
2
ண ㄗᕪ
ᑐẚ
ྠ໬
෈ࣰ֮੓͹қັ
•༩଎ɾغଶ͹৶೨͹ౕ߻͏ɿ༩଎੠ౕɿ
•෈ࣰ֮੓ͶӪڻͤΖҾࢢ
–ਇ͹ི֮෾ාʤྭɾγ΢αϫ YVೖ͹ड़͹๏ֱʥ
–ఱܗ੓ɾϕϫφν΢ϕʤ˕˕ͳ͏͓ͻ˕˕ʥ
–ંৰළౕʀ਎ۛ੓ʤྭɾஎ໌ౕɾϔϧϱχʥ
–ݺਕ͹எࣟʀܨݩʤྭɾϕϫʀΠϜɾ೧ྺʥ
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 36

༰ࣉͲͺࡪְ͕͗͘Ͷ͚͏
Fig. Discrimination accuracy as a function of real-size
difference and context for four age groups.
Doherty, M. J., Campbell, N. M., Tsuji, H., & Phillips, W. A.
(2010). The Ebbinghaus illusion deceives adults but not
young children. Developmental Science, 13(5), 714-721.
(c) 2018 Hideyoshi Yanagisawa, The University of Tokyo 37