The Evolution of Prestige-Biased Transmission
Karthik PanchanathanDepartment of Anthropology, UCLACenter for Behavior, Evolution, and Culture
[email protected]://buddha.bol.ucla.edu/
PrestigeDominance
Henrich and Gil-White 2001Two Routes to Status
A prestige-bias may improve social learning
Henrich and Gil-White 2001
Social learning
Culture
Henrich and Gil-White’s Argument
1. Rank models and imitate the prestigious (“selective” imitation)
Learn from a distance
Learn in close proximity
2. Trade deference for proximity
(“deferential” imitation)
Wright’s ‘island model’
Each sub-population has a uniquely adaptive behavior
emigrantsimmigrants
residents
Life History
1. Juveniles socially learn (random, selective, or deferential imitation)
2. Juveniles attempt to innovate (if necessary)
3. Some individuals migrate to a new social group
4. Reproduction proportionate to fitness
Social Learning Rules
1. Random Imitators- Choose a model at random- Imitate from a distance
2. Selective Imitators- Choose a prestigious model- Imitate from a distance
3. Deferential Imitators- Choose a prestigious model- Trade deference for intimate learning
Costs and Benefits
1. Adults who possess the adaptive behavior and don’t migrate receive a fitness benefit b
2. Fixed cost for selective imitation
3. Fixed cost for deferential imitation
4. Market-based deference cost
Hypothesis 1:Does a prestige-bias increase adaptability?
Cultural Adaptation
Adaptation through individual learning and social learning
i
si s
s ss s i
sss i sss
Social learning biasesincrease adaptation
s s
i
s si
s s ss s si
s s sss s ssi
time
Random Imitation
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00
Freq
uenc
y of
Ada
ptive
Beh
avio
r (q)
e = 0.001
I = 0.1
I = 0.01
I = 0.001
0.001 0.004 0.016 0.064 0.256
Migration Rate (m)
e = 0.01
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00e = 0.1
Freq
uenc
y of
A
dapt
ive
Beha
vior
(q)
0
1
.001 .256.016
Migration Rate (m)
.001 .01 .1
Social Learning Errors (e)
.1
.01
.001
Innovation Rates
Constraints: - innovation- migration- social learning error
q̂ ≈ i
i + m + e
Selective imitation
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00
Freq
uenc
y of
Ada
ptive
Beh
avio
r (q)
Selective Imitators (s=1, d=0)
e = 0.01e = 0.1
e = 0.2
0.001 0.004 0.016 0.064 0.256
Migration Rate (m)
Deferential Imitators (s=0, d=1)
I = 0.001
I = 0.01
I = 0.1
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00Selective Deferential Imitators (s=1, d=1)
Freq
uenc
y of
A
dapt
ive
Beha
vior
(q)
0
1
.001 .256.016
Migration Rate (m)
Social Learning Error Rates
.01
.1
.2
Constraints: - migration- social learning error
q̂ ≈ 1−m− e
Deferential imitation
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00
Freq
uenc
y of
Ada
ptive
Beh
avio
r (q)
Selective Imitators (s=1, d=0)
e = 0.01e = 0.1
e = 0.2
0.001 0.004 0.016 0.064 0.256
Migration Rate (m)
Deferential Imitators (s=0, d=1)
I = 0.001
I = 0.01
I = 0.1
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00Selective Deferential Imitators (s=1, d=1)
Freq
uenc
y of
A
dapt
ive
Beha
vior
(q)
0
1
.001 .256.016
Migration Rate (m)
q̂ ≈ 1−mConstraints: - migration
Conclusion 1: Selective and deferential imitation
increase population-wide adaptation
Hypothesis 2: Does natural selection favor a prestige-bias?
Hypothesis 2a: When is selective imitation
favored over random imitation?
Selective imitators are always present at equilibrium, though rarely going to fixation
5
benefit / selective imitation cost ratio
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00
Deg
ree
of s
elec
tive
imita
tion
(s)
e = 0.001
0.001 0.004 0.016 0.064 0.256
Migration Rate (m)
e = 0.01
b cs= 5
10
20
0.001 0.004 0.016 0.064 0.2560.00
0.25
0.50
0.75
1.00e = 0.1
.001 .256.016
Migration Rate (m)
.001 .01 .1
Social Learning Errors (e)
Equi
libri
um F
requ
ency
of
Sel
ectiv
e Im
itato
rs
0
1
10 20
Hypothesis 2b: When is deferential imitation
favored over selective imitation?
0.001 0.004 0.016 0.064 0.256
Migration Rate (m)
0
5
10
15
20
25
30
Bene
fit−C
ost R
atio(b
c d)
e = 0.2
e = 0.1
e = 0.05
Selection favors either all or no deferential imitators
.05social learning
error rate
.001 .256.016
Migration Rate (m)
.1
.2
Bene
fit–F
ixed
D
efer
ence
Cos
t R
atio
0
30
15
b
cd≈ 1
e
Conclusion 2:Selection likely favors
selective and deferential imitation
But you left out all the interesting complexities of the real world?
Acknowledgments
Rob Boyd Sarah Mathew
The Evolution of Prestige-Biased Transmission
Karthik PanchanathanDepartment of Anthropology, UCLACenter for Behavior, Evolution, and Culture
[email protected]://buddha.bol.ucla.edu/