Journal on Policy and Complex Systems
tural IR theories . For instance , neorealism derives agents ’ behavior from the system in the assumption that all the units of the system would act similarly if posed is the same situation or under the same system ’ s constraints . Since the assumption of rationality implies that the units of the system share equal internal rules , equal environmental inputs result in equal and static behaviors . As Axelrod argues , rational choice and structural theories assume that behaviors are a given , and do “ not worry about where they come from ” ( Axelrod , 1997 , p . 95 ).
Conversely , in complex systems theory , rationality is not a given , but a construct . Rationality is an abstract concept that agents use to refer to a specific set of rules that they see as fit . However , this specific set of rules is not objective as in structural theories . Instead , it is subjective , or ( at best ) intersubjective among a defined group of agents that are part of the same breed or niche . Adaptive agents do not share the same set of rules of behavior , nor assign fitness to their internal rules equally . Indeed , depending on coevolution and adaptation , a single rule might have very different levels of fitness among the population ( Harrison , 2006a , p . 9 ). The fitness of rules even changes over time within an agent due to learning and coevolution . In sum , it is devious to talk about rationality , especially in a singular form , because there is no such a thing as a rational or irrational behavior . What we commonly define as rational behavior is a consequence of our actions and not a premise to them . The premise is adaptation : the higher order rule of CASs .
Adaptation allows us to go beyond behavioral assumptions that flatten the heterogeneity of the international system and assume states to be acting under unifying and static principles . Whereas structural theories generalize contextual and particular patterns of behavior to all the units of the system , complex systems theory goes one step deeper and generalizes how behavior emerges among the actors of the system . The difference between the two approaches is fundamental . In neorealism and structural liberalism , maximized power / security and utility are assumed as general behaviors ; in complexity theory , they are just some of the emergent behaviors that actors can happen to have . Just as for rationality , those behaviors are a consequence of actions and not a premise to them . They are partially incidental , and partially the outcome of a selection of fit rules made by agents .
In addition , by using the concept of adaptation , agents and systems can be theorized dynamic . Behaviors change over time in a process where evolutionary changes in one agent “ induce coevolutionary changes ” also in the other agents ( Gros , 2011 , p . 206 ). The system itself gains a role that goes beyond the one usually assigned to it by structural IR theories . Not only it acts as a constraint to the agents , as neorealism says , but it becomes a generator for new internal rules of behavior of the agents ( Hoffmann , 2006 , p . 98 ; Waltz , 1979 , p . 109 ).
Corpus : Information Processing and Self-Organization
Adaptive agents generate self-organization and entropy-defying behavior at
126