Action schema: A computational representation for modeling social agenc

The research reported in this dissertation is motivated by an apparent paradox in organizational behavior. At times, organizational behavior looks rational. At other times it appears "anarchic". Rational behavior is explained by the model of consequential rationality which views organizational structure as a governance mechanism and actors as consequential rational choosers (March and Simon, 1993). This model fails to account for the behavior of actors and organizations under conditions where goals and technology are ambiguous. Critics of consequential rationality propose that organizational action is shaped by institutional norms and that the meaning of action emerges from sense-making following action (Weick, 1979). Consequential rationality and non-rational action are discussed in organization theory as two modes of action that are salient in different types of organizations (Meyer and Scott, 1983) without an underlying unifying model. In this dissertation we propose the model of practical rationality as a synthesis of the two models. Most computational models of organizations are based on the model of consequential rationality (Masuch and LaPotin, 1989; Carley et al, 1992; Levitt et al, 1994; Jin and Levitt, 1993). We propose a construct for practical rationality called an action proposition and a computational representation for it called the action schema. A design for the action schema has been implemented as part of a simulation engine to support discrete-event modeling and simulation of organizations based on the model of practical rationality. Our approach to modeling organizations presents a view of organizations as "open systems" and focuses on practice as the unit of analysis.;To demonstrate the value of the model of practical rationality in resolving the paradox of organizational behavior a hypothetical restaurant organization was simulated. By varying slack and formalization, the salience of conflict among the organization's goals is varied, leading to different degrees of ambiguity faced by the actors. The simulation experiments revealed that at high levels of slack the organization appears tightly-coupled at all levels of formalization, however at low slack and low and medium formalization the organization appears much more loosely-coupled. Thus the simulation model produces behavior that under some conditions appears purposive and in others 'anarchic'.