In this paper, Ostrom puts forth a theoretical framework that attempts to reconcile traditional rational choice theory of human behavior with recent evidence that people are actually able to solve some collective action problems without external rules. Traditional game theory shows that if people cooperate in collective action situations, the joint social outcome will be higher than if they merely act according to their natural short-term self-interest. Our author examines this assumption in order to determine: (i) which structural conditions lead to more cooperation, (ii) the implications for a behavior theory that allows for the impact of core relationships (reputation, trust, and reciprocity) on cooperation, and (iii) how these core relationships are impacted by the structural variables.

By examining existing literature, Ostrom identifies seven main structural variables that impact the likelihood of cooperation on a collective action problem. These are divided into factors that are repetition independent (number of participants, public vs. common-pool resources, heterogeneity, and face-to-face communication), and those that are not (information about past actions, how people are linked, and voluntary entry/exit). For each variable, Ostrom proscribes the direction of influence (positive or negative) along with a brief academic background. No comprehensive causal theory is offered, but each variable has been shown to have predictive power according to various models:

1) Number of participants: as group size increases, cooperation decreases. There are two reasons: with a larger group, it is easier to free-ride and transaction costs become higher.

2) Common-pool (subtractive) vs. public goods (fully shared): more cooperation occurs when a public good is in question, and less cooperation with common-pool goods.

3) Heterogeneity of participants: generally speaking, more diverse groups (in terms of payoffs, interests, and information) are less likely to cooperate.

4) Face-to-face communication: increased face-to-face communication correlates with more cooperative behavior.

5) Information about past actions: m ore information about past actions is likely to lead to more cooperation. This correlation diminishes with increasing group size.

6) How people are linked: Direct links (where parties interact directly) tend to lead to more cooperation than situations where actors draw resources from a generalized pool.

7) Entry and exit: Situations where parties can easily withdraw may lead to more cooperation. Because everyone has a veto, trust may develop over time.

Subsequently, Olstrom suggests that given the range of possible outcomes to collective action problems - and the impact of these variables on how people actually behave - a theory of complete self-interested rationality is incomplete. The alternate theory that Ostrom puts forth characterizes human behavior as boundedly rational and norm-based. Said another way, evidence suggests that people use a blend of norm-based behavior and rational behavior in a given situation. Given a lack of complete information about all possible outcomes, actions and strategies, people use heuristics (or rules-of-thumb). Norms (or learned internal valuations of certain actions) also guide their behavior. In particular, norms around retribution and reciprocity are important in collective action problems; group norms that promote trust can develop. Higher levels of trust can lead to higher levels of cooperation and, thus, higher net benefits for the group.

Using Ostrom's theory, both unsuccessful and successful collective action problems can be explained by the emergence of trust through reciprocity and reputation. It is also important to note that the relative influence of the structural variables depends on the degree to which they influence core relationships in collective action situations. The author concludes by suggesting avenues for further theoretical work. Specifically, she proposes the further development of scenarios of causal direction to tease out the relative influence/interaction of different structural variables and their interaction in fostering reputation, trust, and reciprocity.