Facing Uncertainty: The Role of Norms and Formal Institutions As Shared Mental Models
The paper opens with the following background concepts: Bounded rationality (goal-oriented behavior with limited cognition and adaptive learning processes); uncertainty (agents do not know underlying probability distributions); two broad categories of cognitive processes (low-cost intuitive (S1) or costly deductive (S2)); heuristics (easily accessible, largely S1 processes that produce impressionistic judgments); mental models (cognitive frameworks that incorporate conceptions of categories with conceptions of cause-effect relationships and that combine S1 and S2 processes to generate judgments); institutions (commonly understood social prescriptions or “rules of the game in a society,” both informal, as in social norms, or formal, as in laws); punctuated equilibrium processes (long periods of incremental change interrupted by bursts of rapid change).
The core argument of the paper involves four related assertions. First, the impressionistic judgments of heuristics respond to shared narratives in a fashion that at times leads to conformity effects. Conformity effects imply positive-feedback loops (with respect to cognition and/or behavior) that can, in turn, can generate dominant narratives. Narratives, whether shared at some level or dominant, affect beliefs and expectations and, consequently, actions. Dominant narratives can generate cognitive monocultures across groups that shape beliefs, expectations, and behavior.
Second (and closely related), mental models follow the dynamics of punctuated equilibrium processes. Heuristics and narratives both influence and become incorporated into mental models. Hence conformity dynamics apply to the adoption and sharing of mental models and, on a larger scale, to the creation of institutions (see next assertion) and to processes of institutional change. Once established, mental models facilitate two basic types of learning: relatively low-cost hypothesis testing within the framework of an extant (or reigning) mental model; or high-cost reevaluative learning that involves reformulating founding precepts of a mental model. Accordingly, mental models exhibit the dynamic properties of punctuated equilibria—stasis, rapid change, path-dependence.
Third, both informal and formal institutions are types of shared mental models. Institutions incorporate and reflect shared conceptions of fundamental social categories as well as shared conceptions of cause-effect relationships (a type of narrative). Institutions offer prescriptions with respect to certain types of behavior for certain categories of individuals who act within specific social contexts (e.g., a social norm may specify that when an elderly person enters a bus, young people should offer a seat). A functional norm reflects a shared understanding of its behavioral prescription that then communicates (or signals) a type of causal relationship.
Fourth, Institutions coordinate cognition, expectations, and actions across groups of individuals who would otherwise face paralyzing uncertainty. Equivalently, in complex social environments where the potential range of interactions would generate unmanageable uncertainty, institutions can limit the required cognitive steps needed to formulate workable expectations. In terms of Brian Arthur’s (1992) concept of a problem-complexity boundary (degree of complexity beyond which individuals cannot reasonably estimate outcomes), institutions, by limiting expected actions, vastly expand the domain of circumstances for which individuals (in given social contexts) may operate within the boundary. Institutions thus underlie the ability of individuals to face uncertainty with respect to their complex social environments.
For each of these assertions, the paper uses variations of game-theoretic reasoning to illustrate and substantiate. Classical game theory (founded a concept of substantive rationality whereby individual agents possess sufficient cognitive ability to estimate best responses to strategic encounters) can illustrate conformity effects implied by the first assertion. Multi-player games of assurance offer relatively simple illustrations. Social network analysis can illustrate such processes more precisely.
Concerning the second assertion, evolutionary game-theoretic reasoning—in which agents “inherit” certain behavioral or cognitive practices (from prior acculturation) and then, over time, adapt to encountered circumstances—can represent learning properties that accompany the emergence of heuristics and mental models and their punctuated equilibrium proprieties. The underlying premise here draws from Ronald Heiner’s (1983) reliability condition. This condition specifies when certain behavioral or cognitive responses will, on average, generate relatively favorable outcomes for agents under conditions of uncertainty, where key underlying probabilities are unknown.
Concerning the third and fourth assertions, epistemic game theory—a game-theoretic variant that facilitates modeling shared understandings—can illustrate the sharing conceptions, narratives, and specifically institutions (as shared mental models). Epistemic games can illustrate how institutions coordinate cognition (and responses) across groups of individuals via “signals” that agents use to coordinate understandings and activities.
The paper concludes by linking the four assertions together in a discussion of relationships between shared narratives, imitation cascades, and dominant narratives—which may sometimes emerge from heuristics—that, in turn, shape the contents of shared mental models. These dynamics interact and contrast with the ability of institutions to (sometimes) foster manageable understanding in complex social environments—enhancing predictability. The conclusion proceeds to link these dynamics to the generation and (often partial) resolution of collective-action problems on both behavioral and cognitive levels.
 The arguments in this abstract rely on many sources, only three of which are explicitly mentioned. The partial list of sources the end lists some of the others.
 Here substantive rationality is a useful simplifying assumption (clearly inconsistent with uncertainty as defined here) that can illustrate conformity effects. Classical game theory can also illustrate the definition of an institution (informal or formal) by representing “the rules of the game” in myriad social contexts.