Group Behavior as a Complex System
It is natural for people to focus on the behavior of single individuals because our own introspection provides us with motivation and perspective at this level. However, interacting groups of people also create emergent organizations at a higher level than the individual. Interacting ants create colony architectures that no ant intends. Populations of neurons create structured thought, permanent memories, and adaptive responses that no neuron can comprehend by itself. Similarly, people create group-level behaviors that are not intended by any person. Social phenomena such as rumors, the emergence of a standard currency, transportation systems, the World Wide Web, resource harvesting, crowding, and scientific establishments arise because of individuals’ beliefs and goals, but the eventual form that these phenomena take is rarely the goal of any individual.
One goal of our research is to conduct experiments on patterns of human group organization. We have developed an internet-based experimental platform that allows groups of 20-200 people to interact with each other in real time on networked computers. The experiments implement virtual environments where participants can see the moment-to-moment actions of their peers and immediately respond to their environment. Our second goal is to develop computational, agent-based models of the results of the experiments. Several models of group behavior exist, but rarely are these models tested against detailed data sets obtained from controlled laboratory settings. Often times, there is a disconcerting mismatch between the simplicity of formal models and the complexities of the real-world situations. For example, although simple cellular automata models can account for the diffusion of information in a spatially distributed population, these basic models need to be supplemented with many post-hoc assumptions to account for actual information diffusion situations, such as the spread of pesticide application techniques in a farming community. To be sure, the formal models can be elaborated to include relevant real-world factors. However, these elaborations all too often improve the model-to-data fit at the cost of creating models that are opaque, unconstrained, inelegant, and poor explanatory tools because the models’ behavior is almost as hard to fathom as the social phenomenon being explained.
Our strategy for bridging the gap between computational models and group behavior phenomena is to create relatively simple laboratory situations involving groups of people interacting in idealized environments according to easily stated “game rules.” The assumptions underlying our psychological experiments correspond almost exactly to the assumptions of the computational models, and so the models can be aptly applied without sacrificing their concise explanatory value and genuine predictiveness. It might be thought that by employing only simple experimental environments, we will blind ourselves to many of the complexities of group social behavior. However, we have been frequently surprised by results showing that even highly idealized experimental environments and tasks give rise to emergent group behavior such as spontaneous synchronization of choices, population oscillations, and the creation of discrete sub-populations. As many others have noted, complex behavior can emerge from agents acting in accord with simple rules. The complexity is based on the interactions of the agents with each other, and these interactions are a key element of all of our experiments.
Foraging for Resources
A problem faced by all mobile organisms is how to search their environment for resources. Animals forage their environment for food, web-users surf the internet for desired data, and businesses mine the land for valuable minerals. When an organism forages in an environment that consists, in part, of other organisms that are also foraging, then unique complexities arise. The resources available to an organism are affected not just by the foraging behavior the organism itself, but also by the simultaneous foraging behavior of all of the other organisms. The optimal resource foraging strategy for an organism is no longer a simple function of the distribution of resources and movement costs, but it is also a function of the strategies adopted by other organisms.
In this line of experiments, we develop and explore a novel experimental technique for studying human foraging behavior (Goldstone & Ashpole, 2004; Goldstone, Ashpole, & Roberts, in press). We have developed an experimental platform that allows many human participants to interact in real-time within a common virtual environment. A large volume of time-varying data can be collected from participants as they vie for resources. resource pools can be created within this environment, and we record the moment-by-moment exploitation of these resources by each human participant.
Group Path Formation
This line of research explores spatial group choice behavior. When people make choices within a group, they are frequently influenced by the choices made by others. In particular, there is a striking similarity of choices and behaviors by people in a group, as is revealed by any casual observation of highschools, bars, or scientific meetings. This similarity may have several causes. People in a group may simply be similar to one another, which may be why they form a group in the first place. Social psychologists have conducted a considerable amount of research pursuing a second possibility, that people are motivated to conform to the group in order to curry the group’s favor. However, we want to focus on a third potential cause: early choices by group members change the environment and hence attractiveness of choices for subsequent group members.
In particular, later choosers can take advantage of the efforts of earlier choosers. When VHS became more popular than Beta, then smart people chose VHS because the popularity of VHS led to more movie titles being released on VHS. Similarly, Microsoft may be evil, but its popularity makes exchanging files and obtaining software relatively easy for its users, insuring its further popularity. A similar popularity advantage explains why the QWERTYIOP keyboard continues to be chosen by most people despite its demonstrated inferiority to other keyboard arrangements.
Another aspect of taking advantage of predecessors’ choices is by standing on shoulders of giants. That is, many times, initial pioneers reduce the costs for followers who pursue the same path. For example, recent cognitive neuroscience researchers can perform fMRI experiments rapidly and efficiently because of prior researchers’ developments of brain imaging. This is an extremely powerful social force. In fact, everything we know as culture is built up in this fashion – by following and extending the innovations of predecessors. Moore’s law is a particularly striking case of this, where for the last 40 years there has been a constant rate exponential increase in the number of transistors per integrated circuit, all because technological advances pursue paths of previous innovations and extend them. The specific concrete instantiation of this that we explore is the evolution of spatial path systems. Early trail blazers through a jungle use machetes to make slow progress in building paths – progress that is capitalized on and extended by later trekkers, who may then widen the trail, then later put stones down, then gravel, and then asphalt.
Our specific question is: what kind of trail systems do people spontaneously produce when they are motivated to take advantage of the trails left by others, and in the process of so doing, further reinforce these trails. This interest in emergent trail systems was shared by Dwight Eisenhower, who, when he was president of Columbia University was asked how the university should arrange the sidewalks to best interconnect the campus buildings. He responded that they should first plant grass seed, let the grass grow, see where the grass became worn by people’s footsteps, and install the sidewalks in the most worn patches.
To study group path formation, participants are shown a configuration of cities, and are told to move to a randomly selected city to earn points. Points are earned for each destination reached, but are deducted for each traveled step. Critically, the steps do not have equal costs. The travel cost for each spot is inversely related to the number of times the spot was previously visited by all participant.
The Dissemination of Innovations in Networked Groups
Humans are uniquely adept at adopting one others’ innovations. Cultural identity is largely due to the dissemination of concepts, beliefs, and artifacts across people. Imitation is commonly thought to be the last resort for dull and dim-witted individuals. However, cases of true imitation are rare among non-human animals, requiring complex cognitive processes of perception, analogical reasoning, and action preparation. When combined with variation and adaptation, imitation is one of the most powerful methods for quick and effective learning.
In social psychology, there has been a long and robust literature on conformity in groups. To some degree, conformity is found because people desire to obtain social approval from others. For example, sometimes when people give their answers privately, they are less likely to conform to the group’s opinion than when responding publicly. However, at other times, the conformity runs deeper than this, and people continue to conform to the group’s opinion even privately. In this series of experiments and modeling, we are interested in the use of information provided by others even when social approval motivations are minimized because the group members never meet one another.
The adoption of others’ ideas has been a major field of research not only in social psychology, but also in economics, political science and sociology. It is common in models of collective action to make an individual’s decision to participate based upon their expectations for how many other people will participate. A common outcome of a collective “I’ll do it if you do it” mentality, is for “tipping points” to arise in which adding a couple more participants to an action leads to a positive feedback cycle in which still more participants sign on, leading to an exponential increase in participation for a time. This behavior is a sensible policy both because the likelihood of success of an innovation depends upon its public adoption rate and because other people may have privileged information unavailable to the individual making a choice. The potential cost of this bandwagon behavior is wasted time, money, and effort in adopting new innovations. At the group level, another problem with bandwagons is that the full range of possible solutions is not well explored if the population is overly homogeneous.
This set of studies explores the diffusion of innovative ideas among a group of participants, each of whom is trying to individually find the best solution that they can to a search problem. The work fills an important gap in research. There are several excellent computational models for how agents in a population exchange information. There is also excellent work in social psychology on how individuals conform or use information provided by others. Field work also explores actual small groups of people engaged in cooperative problem solving. However, there is very little work with laboratory-controlled conditions that explores the dynamics of a group of participants solving problems as they exchange information. Our internet platform provides an excellent method for providing and measuring controlled interactions between participants.
Our Selected Relevant Papers to Group Behavior as a Complex System
Visit our paper repository for abstracts. Clicking on the paper below will download a PDF version of the paper, but the repository has additional formats.
Other Related Research on Group Behavior as a Complex System
Bibb Latane and Andrej Nowak have modeled the spread of opinions in groups by cellular automata models. Their compuational model, dynamical social impact theory, is one of the few computational models in social psychology for group behavior.
One of Latane’s students, James Kennedy has developed the Particle Swarm algorithm with Russ Ebehart, and together they have written the book Swarm Inteligence. This book integrates social psychology research on group behavior with computational approaches to communal search.
My wife Katy Borner continues to do research on diffusion of ideas in scholarly networks of people.
The Complex Systems and Networks is a complex network of people at Indiana University interested in complex networks
I am associatedwith Marco Janssen’s project to study the emergence of rules in the Tragedy of the Commons dilemmas
Leigh Tesfatsion has an excellent web site on agent-based computational economics
The Journal of Artificial Societies and Social Simulation regularly covers research on agent-based models of social phenomena
Robert Axelrodstudies the evolution of cooperation, and has developed models of cultural transmission of ideas