Please enable JavaScript.
Coggle requires JavaScript to display documents.
From Social Learning to Spread of Trends (Spread of Trends (Network…
From Social Learning to Spread of Trends
Social Learning
Evolution
van Leeuwen, Cronin + Haun (2014)
Didn't think apes imitate but they do on occasion
However mostly just a human trait
Haun et al (2014)
Children, chimpanzees, and orangutans
Watch 3 conspecifics use same DIFFERENT coloured box at near by location
Trained to put balls in one coloured box to gain a reward
Test: given new balls
Children switched to demonstrators' box; chimps/orangutans did not
Behaviour change in children not others
Social learning/conforming to majority in children
Why?
Same but manipulate is test is public or private
Children switched more in public > suggests partly conformity based not informational learning
However some private switching suggests learning is still present
Imitate for learning and fear of exclusion
Why did social learning evolve?
Boyd et al (2011)
Beyond best invention of a single person
Accumulation of adaptations
Combination of several models
Easily adapted to circumstances
Incapable of inventing something completely new/big > built upon something that is already there
Necessary
What do we socially learn?
Not just goal oriented
Apes imitate outcome-oriented (relevant/rational) actions, children and adults also imitate outcome-irrelevant aspects of actions
Overimitation
Interpret as causal
Lyons et al (2011)
New/unknown so copy everything as they think it's relevant
Dont understand function
Even under time pressure in competitive scenarios s still did irrelevant actions
BUT increases with age (wold expect opposite)
Box opening task
Affiliative rather than instrumental goal
Children with autism over imitate less that typically developing children even though they imitate the instrumental actions to the same degree (
Marsh et al, 2013
)
Autistic = less socially oriented so less likely to over-imitate suggesting social aspect
Increases when primed with social ostracism (for conventional actions ) (
Waston-Jones, Whitehouse + Clegg, 2014
)
Blue dots moving around screen to prime ostracism or affiliation
Social interpretation
Increases with convention cues (
Legare, Wen, Hermann & Whitehouse, 2015
)
Traditions not just goals
This is how we do it, no obvious goal
Need to learn conventions: Many beneficial practices survived ritualistically without understanding of mechanism
Spread of Trends
What can spread?
Things can spread through population
Cause of similarity/clustering
Induction (behave same way)
Homophily (flock together)
Confounding (other factors)
Behaviours + norms
Socially closer a person is, the greater the influence
Christakis + Fowler (2007)
Even obesity
Socially closer a person is to an obese person, the more likely they are to become obese as well
Mutual friend > ego-perceived friend > alter-percieved friend
Less likely to be due to similar circumstances otherwise neighbour would have effect
Potential mechanisms
Tolerance of obesity
Copying lifestyle
Physiological imitation
Geographical distance does not play the same role
Social distance = important
Fowler + Christakis (2008)
Also other things such as emotions (clustered)
The more happy people surrounding a person, and the closer they are, the more likely the person is to be happy
Dynamics of Diffusion
Network Dynamics
Properties of scale free networks
No. of connections follows a power law: there are few people that are connected to many but many people who are connected to few
Short path between every two individuals
Centrality
No. of links
Who are neighbours?
How often is individual on shortest path between 2 other individuals?
Network size
There is an inverse relationship between sample size and the weight given to any source
More people = less weight therefore less influence
Monster + Lev-Ari (2018)
Adoption of hashtags on twitter
No. of users someone follows
The fewer users someone follows, the more likely they are to adopt very hashtag they encounter
People with smaller networks are important for spreading trends
Network homophily vs weak ties
Strong ties tend to be homophiles, while weak ties tend to bring new information and serve as bridges (
Granovetter, 1973
)
Weak ties
Weinmann (1983)
Spread of info and influence in a Kibbutz (closed community)
Within groups: central > marginals (all friends interact)
Between groups: marginals > centrals (others friends therefore can bring in new information)
% of info via each tie type
Intra = own group gossip from central
Inter = other group gossip from acquaintances
When it comes to influence, central in-group members are important
Homophily
Links are not random, people associate with similar others
Friendships in US high school (black, white, hispanics)
Currarini, Jackson + Pin (2009)
8% of US adults reported discussing important matters with someone of another race (
Marsden, 1987
)
10% of men and 32% of women name other sex members as close friends (
Verbrugge, 1977
)
Exists for both +ive and -ive ties
Causes
Cognitive biases
Structural biases
Baseline opportunities (majority/minority)
Institues (schools, work place), geography, family ties
Misperceptions of homophile (esp. attitudes)
Goel, Mason + Watts (2010)
FB questionnaire in Jan 2008, political issues relevant for presidential campaign, for self and friends
Homophily more than would be expected by base-rate
People overestimate how similar their friends; opinions are to thiers
Over-estimation is stronger for weak ties (>10 mutual friends + don't discuss politics)
Boutyline + Willer (2015)
Twitter study
How many of users' ties followed the same political figures
Hyp = conservatives and people with extreme opinions are less tolerant of ambiguity and change so will seek more homophilous networks
Results
Overall homophily - 11%, 1 out of 9 ties follows the same political figure
Greater homophile among conservatives
Greater homophile among those with extreme values (voting history of followed congress members)
Huckfeldt, Mendez + Osborn (2004)
Name people with whom one discusses politics
Networks are much more homophilous than chance but more have heterogeneity
Number of x supporters in the network increases number of reasons to like x and to dislike y
More homogeneous networks are associated with more polarised attitudes towards candidate (positive-negative)
Mutz (2002)
Heterogeneous networks reduce voting > ambiguity
Even after controlling for demographic variables and political knowledge
True for different elections with different telephone survey samples
Further analysis shows this is mostly due to
Confrontation avoidance (bigger effect for individual high on trial)
More ambivalent attitudes towards candidates
Centola (2011)
Ps from fitness forum
6 neighbours - similar in age, gender, BMI or random
See health behaviour of neighbours
Measure = start only diet diary
Result = similar means more likely
Multiple runs all of which showed greater adoption of diet diary in homophilous network
Increased for obese and non obese individuals, but % wise = greater on obese
Network homophile increased exposure to new health behaviour similar for obese/non but increased the odds of adoptions of behaviour more for obese people