44
Welcome!

Collective Behavior CARA Presentation

Embed Size (px)

Citation preview

Welcome!

Uninformed Individuals Promote Democratic Consensus

in Animal Groups

By Couzin et al. (Princeton) 2011

Presented by Dmitry & Marion

Collective Behavior

- aggregation of individuals & coordination of behavior

- different types of groups (social structure, period…)

- costs of aggregation:* easily detectable by predators* selfishness* cost of cooperation

- relatedness: important factor in decision-making processes

- lot of benefits

Democratic consensus?

Consensus: general agreementDemocracy: majority opinion

Democratic consensus: decision-making process in which the opinion of every agent has the same importance

Advantages

- unwillingness/splitting is costly- will of the majority

Opinion strength role

- intransigence > number

despotism

save time and energy

cost of a bad decision

- strong opinion ≠ optimal opinion cost for the group

back to the paper...

The main question is...

Whether and, if so, under which conditions, a self-interested and strongly opinionated minority can exert its influence on group movement decisions?

And the answer is...

Uninformed individuals who lack a preference or are uninformed about the features on which the collective decision is being made, play a central role in achieving consensus.

To answer the question, 3 models were used:

1. Spatial models (vectors in space)

2. Adaptive network model

3. Convention game

In all of them...

- N1 - strongly opinionated majority (e.g. 6 specimen)

- N2 - strongly opinionated minority (e.g. 5 specimen)

- N3 - uninformed individuals (from 0 to 100)

All models have shown:

- When the strength of minority preference is stronger than that of majority, minority dictates the outcome.

- Adding uninformed individuals to the mix at first returns control towards the majority, but as more of the uninformed are added, they neither lend support to majority, nor to minority.

More about each model...

(briefly)

1. Spatial Models - social component of individuals’ position (they want to be close to each other, but not too close = attraction and repulsion)

g - goal or preference vector

d - desired direction of travel

Spatial model results

In other words...

Uninformed individuals damp any other group’s preferences and make it so that the outcome is determined purely by numbers.

2. Adaptive-network model

- Individuals are the nodes, their interactions are the edges >> the network is created

- Each individual may have a preference (left/right), uninformed may also spontaneously switch their preference

- The more interactions an uninformed individual has with the ones who have strong preferences, the more likely they are to be influenced.

Local Contagion

Adaptive network model results

3. Convention game model

Describes self-enforcing normative behavior in social and economic systems.

The basic assumption is that people prefer to follow the major preference of their immediate surrounding, because doing otherwise may be costly.

3. Convention game model

All the models have shown that...

- Strongly opinionated minority may dictate the outcome for the whole group.

- When the number of those who just don’t care is low, the majority will dictate the outcome even if their preference is lower than that of minority.

- However, as the number of those who don’t care increases, the number of outcomes will be more or less evenly split between the majority and the minority targets.

Uninformed individuals- Inhibit the influence of a strongly opinionatedminority

- Return control to the numerical majority (if their number is slightly higher than the sum of minority and majority)

- However, as uninformed individuals increase in number, they help efface the influence of preferences and strong opinions and instead make sheer numbers become more important than ideology.

Now...

The experiment...

Validation: schooling fishNotemigonus crysoleucas

3 groups:- N1: majority/weaker preference

- N2: minority/strong preference - N3: uninformed (untrained)

3 repetitions7 days of training

Experiment

N1= 6 N2= 5

3 conditions: N3=0, 5, 10

- 6 trials/day, 2 blocks of 3 conditions

- fish were tested only 1/day

- 108 trials

- 1st fish arrived at 2 body length = decision

Consistency with spatial modelSpatial model

Perspectives

Similar research has been done in social network studies of infectious disease contagion and rumor propagation...

Local Contagion

House, T, Keeling M, “Household structure and infectious disease propagation”, 2009

Zoom Out...

Media reaction“It could be a scientific fact that apolitical individuals, when pressed for a decision, will shun the minority view”

Tags: government-and-politics, animal-behaviour

“As Congress proves itself increasingly dysfunctional and captive to extremists. A team of researchers concluded that without all our know-nothing fellow citizens, things might be even worse.”

“Uninformed people catch a lot of flak in society [...] New research, though, suggests these know-nothings may be more vital to democracy than anyone has given them credit for.”

Tags: Democracy, Democrats, Public Opinion, Republicans, Voting

“The findings challenge the commonly held idea that an outspoken minority can manipulate uncommitted voters.”

“a recent article in Science [...] demonstrates how uninformed voters [...] can serve as ballast in a democracy.”

“the golden shiner [...] can tell us a lot about how American democracy works.”

“use fish as a model electorate”

“golden shiners, which naturally associate the color yellow with food” Have you even read the article guys?!Tags: civic education, civic engagement, Princeton fish study

Critics (Ours and Yours)- uninformed individual?

* advantage of a group* environmental cues* uninformed ≠ no preference

- fish model: * uninformed? bias to yellow?* small sample size: statistics* majority/minority* only 3 conditions

- process and results depend on the stakes

Our conclusions

+ interesting study promoting debate

+ well-designed models and convincing results

- lack of conclusion & perspectives

- lack of delimitation of the topic => will to create debate?

- “due to the time-consuming nature of training and constraints related to obtaining enough fish for replication”

Thank you for your attention

All the videos you’ve seen in one playlist:

www.bit.ly/CARAVideos

This presentation:

slideshare