Computational models of information processing

The necessity and desire to understand the nature of information permeates virtually all aspects of scientific endeavors in both natural and social systems. This is particularly true in research that seeks to understand how various forms of organizations arise and function. This dissertation is dedicated towards understanding one important concern in the study of information in organizations: that of information integrity. There are two aspects of information integrity that are interesting to examine in management contexts: First, because the definition and quantification of information is often nebulous, the onus to preserve information integrity, from ensuring a successful information transfer to minimizing distortion, to a large extent falls on the information recipient. Second, unlike in physical systems, in which there tends to be a clear demarcation between the information processing system and the environment, the boundary between the two in management contexts is usually more diffuse, which can give rise to complex interactions. The structure of the environment can thus have a significant influence on an information recipient\u27s ability to process information and to preserve its integrity. I present two computational models to develop theories about these aspects: In the first, I look at how an organization\u27s strategic effort to acquire (and therefore receive) information co-evolves with its absorptive capacity in different types of environment. Here, loss of information integrity is defined by acquisition failure. The model suggests that an exploitative information acquisition strategy could better preserve information integrity and eventually generate a more diverse knowledge stock than an explorative strategy could, thereby challenging common assumptions. The model also highlights several environmental and cognitive parameters that modulate the relationship between information acquisition strategy and its outcome. In the second model, I look at information processing in the context of event forecasting. The model is built on the idea that events have structural signatures that are given by the web of causal relationships from which those events arise. As forecasters receive information about an event, their failure to preserve the integrity of the event\u27s structural information hurts forecast performance, but interestingly, some events have structural characteristics that buffer against this effect