Collective cognition and behavior

Collective behavior is result of individuals cognition and behavior and collectives influence individuals

A murmuration, done by birds Might be like a ripple effect, not a phenomenon with a leader Flocking emerges from the simple rules governing how a bird responds to its closest neighbors

It is model through Boids model of flocking

It has three simple rules: Alignment, Cohesion, Separation A boid only applies the rules to boids in its neighborhood. (Its neighborhood is not 360 degrees) No single bird is the leader- flocking is self-organized

Steer towards average heading of neighbors

Move closer to nearby neighbors

Move away from neighbors that are too close

A minimal model of how people decide to locate in a city that will produce segregation

It is a grid, representing a city There are X’s that represent members of one group and O’s that represent members of another group Agents are placed randomly in the grid The assumption is that an agent is satisfied if it is surrounded by a certain percentage of agents like itself Calculate the agents to know whether they are satisfied or dissatisfied. The agents that are dissatisfied are relocated (can be done randomly) and recalculate satisfaction. So, you repeat the process again and again.

Nothing, segregation is locked in, hysteresis Only escape this if agents start preferring diversity, not just tolerating it

The Schelling model is incomplete, very simple Doesn’t take into account agent variability in preferences and income, housing options vary in cost, zoning, loan discrimination, and housing discrimination You can add many of these to the model straightforwardly

When tendencies of individuals change (ex.become more tolerant), they can still be locked into old behaviors/patterns (segregation) Collective tendencies often don’t simply reflect individual tendencies

The fact that we use all of the information we have. Result of a rational individual’s behavior Ex. A first person said red, since they received a red marble. The second person said red, since they received a red marble and heard the person before say red. The third person says red, even though they have a purple marble, since they heard two people say red. The cascade continues with people saying red even though in reality there are more purple marbles then red marbles.

Boids model is a 2d plane Schelling model is a 2d grid Networks

Political views Rumors Fashion Language Change Health behavior Culture

Procedure The behavior done by subjects is signing up for an online health forum Subjects were randomly assigned to one social network Subjects get an email when a neighbor signs up, like a bit of a nudge to sign up Subjects can’t communicate and are anonymous to each other Findings: A clustered-lattice network is faster to spread sign-ups through emails than a random network. Subjects who get more emails return to the forum more.

Simple contagion (disease) can spread after a single contact thus random networks with many long-range connections are efficient But most behavior change is a complex contagion, which requires multiple contacts. Clustered networks can be more efficient Intuitions about how network structure influences behavior can be unreliable

High degree of clustering No long-range connections

Low degree of clustering Only long-range connections

Emergence Self-organization Social network structure matters Robustness to perturbation

Simple local rules of cognition and behavior in individual interactions lead to the emergence of a complex global phenomenon on the group level Group does not reduce to unit (agent/individual)

Order emerges even when control is de-centralized (no leader)

But how it matters depends on the phenomenon Disease spread (random network) vs behavior spread (clustered network)

Remove a unit, or disrupt interaction between a few units, similar behavior/cognition emerges Being able to adjust