| V. |
Human-Machine
Systems: Analysis at the whole-system level. These methods consider how the entire system, consisting of all the machines and all the humans, is supposed to work as a whole in order to accomplish the overall system goal. More traditional human factors approaches are primarily focused on determining what role in the system individual human operators will play (system-oriented methods). Cognitively-oriented methods, such as
Cognitive Work Analysis, focus on the fundamental characteristics of the work domain and the cognitive demands that are imposed on humans operating in those domains. These methods complement the
CTA and Knowledge Elicitation methods in Section I by mapping out the structure and purpose of the domain, allowing analysts to identify which cognitive strategies arise from actual domain demands and which are workarounds due to poorly designed systems. |
| |
|
|
| |
A. |
Cognitively
Oriented Methods
|
| |
|
1. |
Cognitive Work Analysis (CWA):
An integrated approach to human-centered system design developed primarily by Rasmussen and described in Vicente's Cognitive Work Analysis book. The approach is divided into five stages: Work Domain Analysis, Control Task Analysis, Strategies Analysis, Social Organization and Cooperation Analysis, and Worker Competencies Analysis. While this approach is broad in the scope of its analysis activities, its primary focus is on the work domain. The work domain imposes fundamental constraints on behavior that any agent, be they human or machine, must satisfy for success. The complexity of behavior that arises in performing work tasks is not a function of the complexity of the agent, but rather a function of the complexity of the world. The work domain is modeled as an Abstraction Hierarchy (AH), which shows goal-means relationships on different levels of the hierarchy, including functional purpose, abstract function, generalized function, physical function, and physical form. The AH shows how higher level functions (such as mission goals) may be supported by lower level functions and physical systems (such as sensor systems). It can show how the same physical system can support multiple higher level functions. The Work Domain Analysis Workbench (WDAW) is a software tool currently under development to support system development using CWA.
References
|
| |
|
2. |
Applied Cognitive Work Analysis (ACWA):
A streamlined and pragmatic version of Cognitive Work Analysis that attempts to bridge the gap between Cognitive Task Analysis and software development, namely decision support development. The approach also aims to represent the results of knowledge elicitation using a goal-means decomposition, which is a modified version of Rasmussen's Functional Abstraction Hierarchy. The goal-means decomposition focuses explicitly on the goals to be accomplished in the work domain, the relationships between goals, and the means to achieve goals, including the decisions required, and the information required to make those decisions. A software tool called CACSE (Computer-Aided Cognitive Systems Engineering) is under development to help in using ACWA to support decision support system design.
References
|
| |
|
3. |
Cognitive Function Analysis (CFA):
An integrated approach to human-centered system design developed by Boy. Boy defines cognitive functions as a mapping from a task to an activity. Tasks are things that the user of a system is required to do. Activities are the actual actions workers perform to complete a task. The phases of this approach include design of a set of primitive cognitive functions through the use of participatory design and domain analysis, definition of evaluation criteria to guide the distribution of cognitive functions among agents, and incremental design and assessment of cognitive functions by designers, users, and usability specialists that are used to build active design documents during the lifecycle of an artifact. Active design documents provide interactive and dynamic explanations about the way a machine should be or actually is used, as well as a trace of the design rationale as a function of usability criteria. Active design documents include interaction descriptions, interface objects, and contextual links. The approach has been used successfully for applications in the commercial aviation domain.
References
|
| |
|
4. |
COADE Framework (COgnitive Analysis
Design and Evaluation): A systematic approach to the development of cognitively-centered systems comprising a set of activities for cognitive analysis, design, and evaluation. In the ANALYSE phase, behavioral Task Analysis, Cognitive Task Analysis, and cognitive performance analysis and behavioral task performance analysis are undertaken to generate cognitive requirements. Two representations of human activities in the system are generated at this stage: a Behavior Model representing the goal structure of the task, and a Cognitive Model representing the knowledge and mental processes applied in tasks. In the DESIGN phase, the cognitive requirements are translated into system design specifications. The EVALUATE phase takes place throughout the entire process, and aims to control the quality of the intermediate and final products in the development process. The COADE approach is concerned specifically with the application area of decision support for military command and control.
References
|
| |
|
5. |
The Perceptual Control Theory (PCT)
Approach: was developed by William T. Powers in the 1970's and claims that all human behavior occurs as a result of a perceptually driven, goal-referenced feedback system. A PCT-based approach to human-centered systems design has been proposed. This approach deals with the goals and knowledge of humans in the system, but is equally concerned with the sensory, perceptual, and psychomotor requirements of the human machine interface. The approach produces a hierarchical goal analysis based on PCT. The Integrated Performance Modeling Environment (IPME) software (see Section II, Part B) incorporates the PCT modeling framework and allows for the understanding of operator information processing limitations and their impact on perceptions of workload, error production, and performance.
References
|
| |
|
|
|
| |
B. |
System
Oriented Methods
|
| |
|
1. |
Information Flow Analysis: A
flow chart of the information and decisions required to carry out the
functions of a complex system is constructed. Aptima and Micro Analysis
and Design are currently working to develop a model-driven software tool
to represent and analyze information flow in complex systems called
IFAST.
References
|
| |
|
2. |
Functional Flow Analysis:
The system is decomposed into the functions it must support. Function-flow diagrams are constructed to show the sequential or information-flow relationships between system functions. Petri nets may be used as a modeling formalism to implement function-flow diagrams.
References
|
| |
|
3. |
Function Allocation:
A set of informal techniques for determining which system functions should be performed by people and which should be performed by physical components. The system functions identified in a function flow analysis may be allocated to specific physical systems. A classical human factors technique to conduct function allocation involves the use of a Fitts list, which describes tasks that humans perform well and tasks that machines perform well.
References
|
| |
|
4. |
Mission and Scenario Analysis:
An approach to designing a system based on what the system has to do (the mission), especially using concrete examples, or scenarios. In the HCI domain, these scenarios are typically called use cases. In the
Cognitive Task Analysis arena, scenario events that draw on cognitive processes (such as monitoring, planning, or decision making) may be identified.
References
|
| |
|
5. |
Signal Flow Graph Analysis:
This technique identifies the important variables within a system and enables their relationship to be detailed. An output variable from the system is selected and all the variables that can influence this are identified. The variables that affect these are then identified until all the output and input variables have been found and linked together. Variables are shown as nodes connected by lines to show their causal dependencies.
References
|
|