Psychology and Brain Sciences | Topical Seminar: Multiagent Modeling of Social Behavior
P657 | 15357 | Smith, E.


Topic: Multiagent Modeling of Social Behavior
Many significant forms of human social behavior arise not because of
one individual’s desires or decisions, but as an outcome of the
interactions of multiple individuals, each autonomously responding
to their own goals and understandings of the situation.   Examples
of such “emergent” behaviors include escalation of conflict,
convergence of groups to premature consensus or “groupthink,”
interactions of buyers and sellers in markets, patterns of mate
choice, emergence of fads and fashions, and the spread of
individuals’ reputations through gossip.  Multiagent modeling is an
innovative technique that aims at understanding such behaviors, by
investigating the consequences of different assumptions about the
knowledge, goals, and behavioral tendencies of individual agents,
the nature of interactions among agents, and the surrounding
environment as a source of resources or dangers.  In this way
multiagent modeling crosses conceptual levels, linking properties of
individual agents and agent-to-agent interaction with emergent
outcomes in entire groups or populations.

The goal of this course is to enable students to understand the
behavior of multiagent systems by analyzing multiple examples in
diverse areas. Students will learn the Netlogo language, and will
use it to produce a meaningful multiagent model over the course of
the semester. We will particularly emphasize multiagent thinking as
an approach for building theories about human social behavior,
drawing on the ability of this approach to simultaneously model
interacting processes at the multiple levels (individual cognition
and behavior, social interaction, and emergent group behavior) that
are relevant for human social behavior.

The course would be appropriate  for graduate students in many areas
of psychology, cognitive science, or related fields who are
interested in multiagent modeling, as well as for those interested
in deepening their understanding of topics within social psychology.
No prior background in computational modeling will be assumed.