Controlling uncertainty: A review of human behavior in complex dynamic environments.

Abstract

Complex dynamic control tasks (CDC tasks) are a type of problem-solving environment used for examining many cognitive activities (e.g., attention, control, decision making, hypothesis testing, implicit learning, memory, monitoring, planning, and problem solving). Because of their popularity, there have been many findings from diverse domains of research (Economics, Engineering, Ergonomics, Human Computer Interaction (HCI), Management, Psychology), which remain largely disconnected from each other. The objective of this article is to review theoretical developments and empirical work on CDC tasks, and to introduce a novel framework (Monitoring and Control framework) as a tool for integrating theory and findings. The main thesis of the Monitoring and Control framework is that CDC tasks are characteristically uncertain environments, and subjective judgments of uncertainty guide the way in which monitoring and control behaviors attempt to reduce it. The article concludes by discussing new insights into continuing debates and future directions for research on CDC tasks. Controlling Uncertainty 3Key words: Monitoring, Control, Uncertainty, Dynamic, Complex tasks, Causality, Agency Controlling Uncertainty 4Controlling Uncertainty: A Review of Human Behavior in Complex Dynamic Environments Problem solving is regarded by many (Anderson & Lebiere, 1998;Laird, Newell, & Rosenbloom, 1987;Newell & Simon, 1972;Sweller, 2003) as the most important cognitive activity in everyday and professional contexts, especially problem solving of the kind needed to control complex dynamic environments. As the complexity of the systems that we interact with in our daily lives grows (e.g., phones, computers, automated driving systems) so too does the importance of research on how we learn to control dynamic environments. Before detailing the issues surrounding the various research domains that investigate control behaviors, the illustrations below provide a general flavor of the kind of real world and laboratory tasks in which control behaviors have been studied. Illustration 1: Ecosystem Control System. In a laboratory-based task, people are presented with a simulated environment of an African landscape with different flora and fauna. The goal is to manage the environment in order to improve the living conditions of the native population.This involves controlling the system over several weeks by managing many interconnected variables that have a complex feedback process, both negative and positive, and with time delays.Illustration 2: Automated Pilot System. Trained pilots are required to take part in flight scenarios (including take-off and landing) in a flight simulator, in which they control the system while responding to a variety problems, including delays between their commands and their actual execution in the system, failings in the system, and miscued problems from the control panel. Controlling Uncertainty 5Illustration 3: Solid Incineration Plant. A municipal solid waste incineration plant is a largescale industrial syste...

Hierarchical Bayesian models as formal models of causal reasoning

“…In general, performance in causal induction increases substantially, when people can observe temporal relations or can intervene (Lagnado and Sloman 2004). Research on the control of complex systems shows that participants often gain very limited knowledge about the underlying causal structure despite extensive learning experience (Osman 2010). HBMs can explain this finding as the available evidence in these studies was often compatible with numerous causal hypotheses, and participants normally lacked domain specific knowledge (see, Hagmayer, Meder, Osman, Mangold, and Lagnado 2010, for a more detailed discussion).…”

A cognitive prosthesis for complex decision-making

“…It has been demonstrated on numerous occasions both in field studies and laboratory experiments that human decision-makers confronted with complex situations fail to perform satisfactorily despite their well-intended efforts (Frensch and Funke, 1995;Gonzalez, Vanyukov, and Martin, 2005;Osman, 2010a;Quesada, Kintsch, and Gomez, 2005;Yasarcan, 2009). Using computer-simulations of complex situations, Dörner (1996) noted particular examples of behaviors leading to successful performance (e.g., active learning and hypothesis testing) as well as numerous instances of poor behaviors leading to failure, which were generally linked to cognitive limitations and poor understanding and/or decision-making strategies (e.g., thinking in terms of isolated cause-and-effect relationships).…”

Dealing with the uncertainties of climate engineering: Warnings from a psychological complex problem solving perspective

“…From a psychological perspective, we can conceive of this decision situation as a complex problem because very generally speaking, complex problems require dealing with (psychological) uncertainty [2]. Therefore, in the present paper, we wish to stress the psychological problem solving perspective: in order to do so we will first give an introduction to the concept of complex problems and how it applies to the CE situation.…”