Experimental models of crowd behaviors

Theories pertaining to group behavior are quite broad in scope. Most of the traditional psychological theories involve describing how individuals perceive themselves as belonging to a particular group (group cohesion), assimilating the ideals of that group (identification), and assuming the associated social roles (e.g. Ashe, Milgram, and Zimbardo). Since this early social-psychological research, other theories (e.g. contagion theory, convergence theory, and emergent-norm theory) have attempted to identify the processes by which individual behavior becomes assimilated into a cohesive group action. These theories attempt to explain how a group may change their "law-abiding" behavior to one of civil disruption or actual destruction. For instance, Turner's Emergent Norm Theory proposes the social rule structure changes amongst the crowd once a significant number of participants partake in the non-normative behavior (Aguirre, Wenger, & Vigo, 1998). The resultant behavior (e.g. burning cars) is assimilated as a "normal" and acceptable behavior for that time and place. Generally, these theories differ in the role of communication, previous experience/culture, and social expectations have upon the initiation of unlawful behavior. Moreover, as suggested by Thompson and Fine in their review of Socially Shared Cognition (Thompson & Fine, 1999), the historical view of social psychology has been from the perspective of the individual's cognitions within the context/environment of a group of people. The idea that individuals are more reactive to outside influences and have less of an individualistic, cognitively-driven motivations and behavior is a rather recent occurrence.

The main issue here is that the fields of social psychology and sociology have focused most of their efforts on the initiation process. They are still trying to understand how the groups form and assimilate certain ideals and attitudes. These are theoretical arguments that will likely continue for the foreseeable future.

Very few scientific disciplines have examined how groups respond to antagonistic outside influences (i.e. police intervention or military force). In fact, most research in the area of coherent group processes are in the areas of organizational structure, and, therefore, focus on how organizations can maintain their motivation to perform and maximize their cohesiveness [e.g. (Osterloh & Frey, 2000)]. Still, there is little scientific research on how groups, crowds, and riots respond to external interventions (e.g. non-lethal interventions such as loud sounds, rubber bullets, and clubbing). The lack of experimental data for how groups will respond to such interventions has left the military and law enforcement to use non-systematic retrospective analyses of case reports to try to piece-together the social interplay that occurs.

We have created 4 theoretical group-types that will exemplify the rudimentary behaviors seen in certain crowds. Using these modeled scenarios, we will be able to assess changes in the group dynamic as deterrent stimuli are presented. This will provide one of the first experimental treatments testing the efficacy of a hypothesized group/crowd-deterrent device in freely-behaving individuals under semi-controlled conditions. Still, it should be recognized that the extent to which we can manipulate the experimental situation to "create" these conditions is limited. We cannot create motivational states that are conducive to violence and destruction, but, we can create a motivational state that is based upon an incentive. Therefore, using positive incentives as a replacement for the malicious motivational state, we can still measure changes in motivated behavior in responses to the deterrence stimuli (e.g. sound).

If the continuum is simply bisected (for the sake of clarity), we leave ourselves two levels: peaceful and malicious. A 2 x 2 matrix forms. Our approach here does not specifically focus on the processes of social-cognition that may lead to more or less group cohesion as much as the behavior itself. Obviously, the motivation behind those behaviors can be varied. Our interest is to examine 4 categories of crowd behavior based on 2 variables: group coherence and intent. Group coherence represents the implicit or explicit affiliation to the group as an identified entity. A coherent group infers that the congregation of individuals has developed a sense of group-identity. As reviewed by Vider, the sense of coherence can emerge through proximity, but an additional contributing factor is the emergence of a leader who the individuals in the group come to identify and follow (Vider, 2004). An incoherent group is a mass of individuals who are only seen from the outside as one unit because of physical proximity, but the duration of their position in proximity to each other or the physical spacing has not elicited a sense of group identification where each individual has a particular affiliation to those surrounding him/her (e.g. people walking on a city block versus people waiting 20 minutes in line for a bus).

Our second characteristic, intent, refers to the general motivation of the individuals in the group. Obviously this would fall on a continuum from peaceful to malicious. Malicious intent can involve behaviors that cause disruption of others' behaviors such that attaining a future goal becomes more difficult (e.g. disrupting traffic flow during a march) to the actual destruction of person or property. Obviously, we cannot ethically elicit malicious behaviors in subjects. We can, however, use a scenario that could, behaviorally, mimic possible malicious behavior. In this case, the behavior of stealing items is being modeled. We could have chosen other behaviors (such as throwing balls at a target to simulate throwing rocks at people or property), but we wanted a situation where the location for the sound to be directed toward would be static. We are additionally adding a goal to provide a motivational factor to the behavior (albeit not malicious). The motivational factor does provide incentive for producing the behavior, which could be seen as similar in at least directing the behavior toward a specific goal of removing items from a location (just without the malicious intent).

Within the matrix, we can see how 4 distinct group behaviors evolve. People walking in a common area or sitting in a city-square would constitute general meandering; they have no distinct group identity and are being generally peaceful. On the other hand, a coherent group of behaving individuals, acting in a peaceful manner, would reflect a march or a sit-in. Malicious intent can be categorized as either pillaging or looting. The distinction is whether there is clear organized group intent towards damaging persons or property (pillaging) or it is individualistic mass destruction (everyone-on-their-own, looting for self-satisfaction and self-gain). Still, these group designations are transient; they can change based on individual and environmental factors. A peaceful march can be infiltrated by malicious individuals that direct it to become more destructive in behavior. Similarly, when a professional sport team wins a championship, people who would be considered meandering in our categorizations, take to the streets and begin to burn cars in "celebration". Not only is it important to recognize that groups can shift in their categorizations, it is also essential to first document how each type responds to deterrence.

Other factors can contribute to the formation of a group and the likelihood of certain actions of that group. Personality factors and the occurrence of misattributions to local individuals that the person has self-identified (i.e. the false consensus effect) have been studied in individuals who self-report violent trends in their social behavior at sporting events (Russell & Arms, 1995). Not surprisingly, those that were found to be more aggressive were more prone to misattribute similar aggressive states in others. Others have examined the possible contributory role of environmental factors to group aggression events. Temperature, light/darkness and noise level have been implicated as possible modulating factors (Russell, 2004). Further, still others have tried to mathematically represent how the movement of crowds through urban areas can be modeled (Yamori, 1998). Although many of these internal and external factors could contribute to the motivation state of an individual, as would cultural factors and group density, they are beyond the scope of this initial step in experimentation of intervention-responsiveness.

Therefore, our first-generation experimentation of our group dynamics model focuses on the basic characterizations of cohesion and directive motivation. As we collect additional data from our experiments, we will revise this theory, as needed, and add the additional factors mentioned.