Handbook of Psychology, Volume 5, Personality and Social Psychology

The Individual and Society 405

group level, with the interactants themselves continuing to
change their choices throughout the simulation.
In a different approach, Hegselman (1998) explored the
emergence of social support networks in a society. Individu-
als lived on a two-dimensional grid containing some unoc-
cupied sites and played a two-person “support game” with
all of their immediate neighbors. Each individual was char-
acterized by some probability of needing help. A needy indi-
vidual clearly benefited, of course, if he or she received help
from a neighbor, but providing help to a neighbor was
clearly costly. With this trade-off in mind, each individual’s
preferred neighborhood was one in which he or she could
obtain the degree of help needed while minimizing the help
he or she provided. Individuals were sometimes provided a
migration option that enabled them to move to a more desir-
able location within a certain radius. The results reveal how
support networks can evolve in a world of rational egoists
who are differentially needy, but similarly motivated to
choose partners in an opportunistic manner. Although social
support inevitably develops, the social networks that emerge
tend to be highly segregated. Individuals with a moderate
probability of becoming needy tend to form relationships
with one another, and also with individuals from somewhat
higher and lower risk classes. Interestingly, individuals at the
extremes of neediness—those with very high or very low
probabilities of needing help—tend to have the most diffi-
culty in establishing support relations. If they do manage to
form such relationships, their partners tend to be from the
same risk class.

Social Influence and the Emergence of Social Structure

The cellular automata model of social process that has been
analyzed most thoroughly concerns social influence (e.g.,
Lewenstein, Nowak, & Latané, 1993; Nowak, Lewenstein, &
Frejlak, 1996). The initial formulation of this model (Nowak
et al., 1990) focused on the emergence of public opinion in a
society characterized by a diversity of attitudes. The model
assumes that in the course of social interaction, individuals
are motivated to sample the degree of social support for their
position on a given topic. The model also assumes, in line
with social impact theory (Latané, 1981), that each individual
gives the greatest weight to the opinions of others who are
spatially closest to him or her and who have the greatest
strength (e.g., who are most influential or persuasive). An in-
dividual’s own opinion is also taken into consideration and is
weighted most heavily by virtue of spatial immediacy (i.e.,
distance is 0). After each round of interaction, the individual
compares the degree of support for each attitude position and
adopts the one with the strongest support in preparation for
the next round of interaction.

In the simulations, one individual is chosen (usually at

random), and influence is computed for each opinion in the

group. (The strength of influence of each opinion is ex-

pressed by the following formula.

Ii=





∑N

1

(

sj

d^2 ij

) 2 



1 / 2

where Ii denotes total influence, sj corresponds to the

strength of each individual, and dijcorresponds to the dis-

tance between individuals iandj.) If the resultant strength for

an opinion position is greater than the strength of the individ-

ual’s current position, his or her opinion changes to match the

prevailing position. This process is performed for each indi-

vidual. This procedure is repeated until there are no further

changes, which typically requires several rounds of simula-

tion, because a person who had previously changed his or her

position to match that of his or her neighbors may revert to

the original position if the neighbors change their opinions.

Figures 16.1 and 16.2 present representative results of the

computer simulations. Each box corresponds to an individ-

ual. The color of the box (light vs. dark gray) denotes the

individual’s position, and the height of the box corresponds

to the individual’s strength. In Figure 16.1, there is a majority

of 60% (light gray) and a minority of 40% (dark gray). The

majority and minority members are randomly distributed,

Figure 16.1 Initial distribution of opinions in the simulated group.

Figure 16.2 Final equilibrium of opinions in the simulated group.