Page 420 -

17 Social Constraint 423

Contribution
Already Axelrod’s classical model provides a causal explanation for norm spread-
ing. This includes a designation of mechanisms of norm transmission and normative
transformation. An investigation of the functional effect of norms on the society is
left aside. This orientation remained predominant in this line of research. Typically,
models with a game theoretic background concentrate on the question of norm
dynamics. They ask how a behaviour regularity emerges in an agent population.
This is the problem of the rational choice tradition in sociological theory, namely,
the perspective of norms as an aggregated product of a sum of individual actions.

Transformation
It is striking that, except for the model by Sen and Airiau, the transformation of
individual behaviour in all models is driven by some kind of sanctions. However,
also in the Sen and Airiau model, agents react to losses of utility values. This is
the causal mechanism of norm spreading. The great advantage of this account is to
shed light on the process of norm change. As it has become apparent in discussing
Mr. Smith, this process can also be observed in human societies. However, norm
change is only barely captured by the functional account of role theory. On the other
hand, the models of this tradition only include a restricted functional perspective:
on an individual level, the agents’ choice of action is guided by the functional
consideration of calculating the expected utility. However, a corresponding analysis
on the social macro-level can be found only in Savarimuthu et al.’s model.

Transmission

With regard to the transmission of norms, it is striking that social learning is
implemented in many game theoretic models by a replicator dynamics. Typically,
this is interpreted as social learning by imitation. If applied in a context where no
real natural selection, rather than some kind of learning is at work, then using a
replicator dynamics amounts to saying: somehow the individuals learn in a way
that—measured by the relative overall success of their type of behaviour—more
successful types of behaviour become more frequent. As an effect, this may be true.
However, no mechanism is indicated. In this dimension, the models struggle with the
same kind of problem as functional analysis which the individualistic programme
tries to resolve, namely, the lack of a causal explanation.

Implementation
From the perspective of the role theory of action, a weakness of this approach
becomes apparent, one immediately related to the game theoretic problem descrip-
tion. Agents are faced with a strategic (binary) decision situation. Thus, they have
a fixed set of behaviour (Moss 2001). For this reason, behaviour change can be
implemented by dynamical propensities (i.e. the propensity to defect is dynamically