Cognition vs time as constraints in the structuring of human social - - PowerPoint PPT Presentation

Cognition vs time as constraints in the structuring

• f human social networks

Robin Dunbar

British Academy Centenary Project

Institute of Cognitive & Evolutionary Anthropology University of Oxford

robin.dunbar@anthro.ox.ac.uk

Convergence of Three Projects Convergence of Three Projects

• British Academy’s “Lucy Project”

http://www.liv.ac.uk/lucy2003/

–

Liverpool (Archaeology + Psychology), Kent (Social Psychology)

–

how social bonds work

–

cognition and brain evolution (Social Brain Hypothesis)

EPSRC/ESRC DTESS Project

http://www.informatics.man.ac.uk/research/groups/isd/projects/dtess

–

Manchester Business School + Sheffield Hallam

–

Integrating Small-Groups-as-Dynamic-Systems Theory with Social Brain Hypothesis

EU-FP7 SOCIALNETS Project

http://www.social-nets.eu/

–

Computer Sciences at Cambridge and Cardiff; + EU partners

– How to design better networking technology

The Social Brain Hypothesis The Social Brain Hypothesis

Predicted group size for

humans is ~150

“Dunbar’s Number”

Primates have big brains because they live in a complex social world

Human Human Social Networks Social Networks

These all have mean sizes of 100-200

Neolithic villages 6500 BC 150-200 military units (company) (N=10) 180 * Hutterite communities (N=51] 107 Nebraska Amish parishes (N=8) 113 business organisation <200 ideal church congregations <200 Doomsday Book villages 150 C18th English villages 160 * GoreTex Inc’s structure 150 Research sub-disciplines (N=13) 100-200

Small world experiments (N=2) 134 Hunter-Gatherer communities 148 Xmas card networks 154

Maximum Network Size

• 3

• 3

• 3

• 2

• 2

• 2

• 2

• 1

• 1

• 1

• 1

• 9

• 7

• 4

• 2

Number of Cases

“Reverse” Small World Experiments

Killworth et al (1984)

1 10 100 1000 10000 10 20 30

Hunter-Gatherer Societies

Dunbar (1993)

Xmas Card Networks

Hill & Dunbar (2003)

Individual Tribes

What Makes it Work? What Makes it Work?

Personalised relationships Trust Expectations of reciprocity In traditional societies:

– kinship – a shared history

The Atapuerca “family”

[Homo heidelbergensis]

Hidden Structure of Social Hidden Structure of Social Networks Networks

Stable points in

group size at: 5-7 12-15 ~35 ~80? ~150

Residual Contact Frequency

7.4 7.3 5.0 3.4 1.4 .1

• .7
• 1.7
• 2.7
• 3.6
• 3.9
• 4.1
• 4.4
• 4.6
• 4.7
• 7.6

Cumulative Network Size

160 140 120 100 80 60 40 20 Maximum Contacted

Hill & Dunbar (2003)

?

The Fractal Periodicity of The Fractal Periodicity of Human Group Sizes Human Group Sizes

Peak at ω=5.4 Peak at ω=5.2

Xmas Card Database

Social Groupings Database [N=60]

Scaling ratio = exp(2π/ω) = 3.2 and 3.3

Zhou, Sornette, Hill & Dunbar (2005)

Horton Order Analysis of Hunter-Gatherer Group Sizes

Hamilton et al (2007)

Intimacy, Frequency and Trust Intimacy, Frequency and Trust

Relationship between

frequency of contact and intimacy

Trust and obligation

seem to be important

Emotional Closeness

10 9 8 7 6 5 4 3 2 1

Mean Time Since Last Contact (Months)

8 6 4 2

Hill & Dunbar (2003)

The Circles of Acquaintanceship The Circles of Acquaintanceship

• A hierarchically inclusive

series of levels of acquaintanceship

• Levels reflect

familiarity and emotional closeness

• There are at least

TWO more layers at ~500 and ~1500

[is this where weak “work” ties lie?]

5 15 50 150

Intensity

500 1500

Friends Friends ≠ ≠ Kin Kin

Friends and Kin are not

the same thing

Friendship requires

emotional closeness

We have no choice

about Kin

Hence: Friendships are

fragile…. ….Kinship is robust

[We put up with them even though we don’t particularly like them]

Total network size

Mean emotional closeness to unrelated alters

Total network size

Mean emotional closeness to related alters

Unrelated Alters Related Alters

Small Medium Large

Network Size

Structure of Networks Structure of Networks

For relationships indexed on a

1-10 scale:

Among UNRELATEDs:

– medium strength links predominate – large networks exhibit more

STRONG links

Among RELATEDs:

– Weak and Medium links

predominate

– large networks exhibit more WEAK

links

Total network size

Median percentage of related network

Total network size

Median percentage of unrelated network

Unrelated Alters Related Alters

Medium Weak Strong Strong Weak Medium <55 55-82 >82 <55 55-82 >82

Total Network Size

Blood Blood is is Thicker Thicker than water than water

Kin are given

priority over Friends

Kinship may

reduce the cognitive load?

Related network size

Unrelated network size

10 20 30 40 50 60

Total Kin

20 40 60 80

Total Non-Kin

250 complete

networks

80 close networks Total Kin

N P Maximum Network Size 6 0.011 150.0 8 0.002 146.1 10 0.001 144.5 12 0.004 145.3 14 0.004 141.8 16 0.001 136.3

Estimating the Limit on Network Size

Related network size

Unrelated network size

Two Unresolved Questions Two Unresolved Questions

Is the limit at:

• higher level, with the

internal structure a consequence of fragmentation [top

down]?

• lower level, with

higher levels simply being small-world emergent properties

[bottom-up]?

Are human groupings limited by:

⇒ frequency of interaction ⇒ capacity for emotional closeness [i.e. cognition]

A Role for the Social Brain A Role for the Social Brain

Intentionality as a reflexively hierarchical sequence of belief states

The Levels of Intentionality

…that may be very costly in computational terms

1000 2000 3000 4000 5000 6000

Frontal Lobe Volume (cc)

1 2 3 4 5

Achievable Intentionality Level

Humans Apes Monkeys

The Limits to Intentionality... The Limits to Intentionality...

A natural limit at 5th order intentionality:

“I intend that you believe that Fred understands that we want him to be willing to [do something]…” [level 5]

20 40 60 80 100 120 2 3 4 5 6 7 ToM Physical

% Correct Intentionality Level

Kinderman, Dunbar & Bentall (1998).

The Story The Story-

• Teller

Teller’ ’s Art s Art

BUT…

Shakespeare had to do SIX

Iago Othello Desdemona

Othello - An Everyday Story of Deception

• The audience

has to do FIVE

• rders of

intentionality Stories (especially “origins” stories) are an integral part of community-bonding

4 2 3 1 6 5

Cassio

Is Is Mentalising Mentalising Costly? Costly?

Two Experiments Two Experiments

Reaction Time Experiment N = 8 Mentalising vs Memory (controlling for order) accuracy: p = 0.919 RT: p < 0.05

Functional Imaging Experiment fMRI [BOLD] 5 stories with 20 mentalising and memory questions @ levels 2, 3 and 4 N=17

The Cognitive Demands of The Cognitive Demands of Mentalising Mentalising? ?

Areas with significant parametric effects on the contrast [intentionality > memory] at p=0.001 uncorrected After FWE correction [p=0.05]: right TPJ, bilateral TP, right inferior FG, cerebellum

Lewis, Birch & Dunbar (in prep)

Significant effects for parametric contrast [ToM>memory] masked by nonparametric contrast [ToM>memory]

(p<0.005 uncorrected)

fMRI N=17

Temporal- Parietal junction

Cognitive Limits to Sociality? Cognitive Limits to Sociality?

Achievable intentionality level

indexed from stories

5th order seems to be the limit

Level of intensionality

9th 8th 7th 6th 5th 4th 3rd 2nd 1st

Frequency of failure 20 10

Level of intensionality 10 8 6 4 2 Clique size 30 20 10

[Stiller & Dunbar 2006]

Intentionality correlates

with clique size

We now have two neuroimaging

studies to support this

A Volumetric Perspective A Volumetric Perspective

Optimised Optimised VBM VBM with modulation with modulation

[N=29 subjects, aged 18 [N=29 subjects, aged 18-

• 50]

50]

Orbitofrontal

Grey matter volume

correlates of network size for

ToM > memory contrast

[corrected p<0.005]: Middle frontal gyrus Orbitofrontal area Dorsolateral PFC ACC Hippocampus Amygdalla

Lewis, Browne & Dunbar (in prep)

among others, most bilaterally Masked analysis for both ToM and network size

Social Bonding Social Bonding Primate Primate-

• Style

Style

Primate social bonds

seem to involve two distinct components:

An emotionally intense

component [=grooming]

A cognitive component

[=brain size + cognition]

Why Does Grooming Work? Why Does Grooming Work?

endorphins are relaxing They create a psycho-

pharamological environment for building trust?

Group Size

160 140 120 100 80 60 40 20

Social Time (%)

50 40 30 20 10 Predicted for Humans

2 4 6 8 10 12 Number of Grooming Partners Sal Naltrex Sal Morph

[Keverne et al 1979]

An experimental study with monkeys Opiates block social drive; Opiate-blockers enhance social drive

How Much Time Should How Much Time Should Humans Spend Grooming? Humans Spend Grooming?

If humans

bonded their groups as primates do….

Grooming time

would be about ~45% of total day time

Group Size

160 140 120 100 80 60 40 20

Social Time (%)

50 40 30 20 10 Predicted for Humans

The bonding gap

Physical Interaction Physical Interaction may be Critical may be Critical… …. .

A touch is worth a

thousand words….

We underestimate the importance of physical contact Touch may be critical in establishing “honesty”

Three Ways Three Ways to Bridge to Bridge the Gap? the Gap?

Millions Years BP

3.5 3.0 2.5 2.0 1.5 1.0 .5 0.0

• .5

Predicted Grooming Time (%)

50 40 30 20 10

Laughter a cross-cultural trait shared with chimpanzees Music and dance Religion and its rituals

Australopiths Modern humans

• H. erectus

Archaic humans

An Opium for the Masses?

Religious practices are

• ften well suited to

stimulate endorphins

Endorphins:

⇒ make you relaxed ⇒ may trigger the release of

• xytocins (creating sense
• f “euphoric love”)

⇒ enhance sense of communality ⇒ positively influence immune system

Medieval flagellants Whirling dervishes [an Islamic Sufi sect] Bernini’s Ecstacy of St Theresa of Avila

Laughter The Best Medicine?

0.5 1 1.5 2 2.5 3 3.5 4 Strangers Acquaintances Cooperation ( in GBP) Neutral Comedy

In a Public Goods Game (Prisoner’s Dilemma) Ss were more generous to strangers (but not friends) after watching a comedy video

van Vugt et al (submitted)

A human universal

Lessons for Networking Lessons for Networking Technology? Technology?

Constraint may be

internal rather than technical

Why do people want to

contact each other?

• Are all contacts

really equal?

• Can technology

ever replace face-to-face?

Texting:

averaging 120 texts per day to just 2 people

Technology:

may slow relationship decay rate, but be poor for creating new ones

Conclusions Conclusions

There are cognitive constraints on sociality Human social groupings are structured in

discrete layers

Does Cognition or Time (or both) limit network

size and structure?

So…. – Will cognition limit electronic networks? – Can technology help us to overcome this?