Cognitive conflicts in dynamic systems

INCOSE International Symposium, 2006

With the addition of advanced technologies and automation within systems, the tasks that operators are asked to complete have grown more complex. This has led to greater opportunities for malfunctions, critical incidents, and catastrophes. A major contributor to these problems is the human operator and the technology not being explicitly designed as a single Joint Cognitive System (JCS). The operator is typically only blithely considered during system design and is typically expected to adapt to inputs from the other subsystems. Thinking of the operator and the automation/technology from a holistic perspective, as a single JCS, instead of separate subsystems, allows for more effective, agile, and robust decision‐making. The operator must have a complete understanding of the automated processes and the work domain, leading to more informed decision‐making. Thinking of the entities as co‐agents, rather than two isolated components that must interface, allows for more effective problem...

Communications of the …, 2006

NYU Working Paper No. IS-86-080, 1986

Quite simply, the fact is that we do not understand very much about designing complex, computer based information systems. I mean that we don't know what system design consists of, we don't know how it is done, and we don't know how to teach it. Furthermore, ...

1999

In the paper is a concern for the decreasing capability of humans to appreciate that which they have an increasing ability to manipulate. Articulated in a 1970s U of Penn systems sciences dissertation, this was ascribed to a seriously limited capacity for context appreciation. This was proposed as a consequence of an historic emphasis on reductionistic ideas and analytical thoughts, not holistic ideals and systemic thinking. This was similar to the notion that humans become adept at achieving short-term results, but not at managing their longer-term consequences. The systems science platform of the 1970s, especially the planks developed by the Ackoff/Ozbekhan/ Trist group at U of Penn, argued that appreciation of connections was a better route into context-sensitive systems than was traditional analysis of parts. While there was ambiguity as to what this meant it was sufficiently attractive to encourage much fruitful research and several innovative dissertations. Concern for context...

IEEE Intelligent Systems, 2002

Since then, this field has taught us many important things, including two major lessons. First, the road to user-hostile systems is paved with designers' user-centered intentions. Even smart, clever, well-intentioned people can build fragile, hostile devices that force the human to adapt and build local kludges and workarounds. Worse still, even if you are aware of this trap, you will still fall into it. Second, technology developers must strive to build truly human-centered systems. Machines should adapt to people, not the other way around. Machines should empower people. The process of designing machines should leverage what we know about human cognitive, perceptual, and collaborative skills. Time to rethink We're in a new ballgame, in which the modern "sociotechnical" workplace is characterized by changing collaborative mixes of humans and machines. Advances in technology have opened new horizons that are changing the nature of work and education, including distance learning, distance collaboration, training support, and performance support. 1-3 Consider, for example, the notion from human factors engineering of task analysis: you can decompose jobs into invariant linear or tree-like sequences of actions (and some cognitions). This notion has a long history, dating to the applied psychological research in Europe dubbed psychotechnics in the late 1800s. (This notion of task differs from the AI notion of "generic tasks." 4) Studies of the modern workplace suggest that significant problems can arise when you design systems based on a decomposition of tasks into invariant sequences of prescribed steps. Sometimes, people might appear to be conducting linear sequences of actions, when they are actually engaging in context-sensitive, knowledge-driven choices among alternative actions. 5,6 Would loss of the uplink to the weather radar keep a forecaster from crafting a forecast? No, the forecaster can work around it because knowledge permits the creation of alternative paths to the goal. When you are forced to adapt, kicking and screaming, to a new software upgrade and are frustrated by changes in functionality, are you totally paralyzed? No, you can craft a workaround. The point is not that something is inherently wrong about the notion of a task as an expression of a particular goal, but that task analysis as it has been applied can sometimes be limiting. When regularly occurring sequences are regarded as invariant and therefore predefined, systems designed on this basis can run a substantial risk of being flawed. Specifically, you can expect them to lead to fragilities, hostilities, and automation surprises. 3,7 In short, they might not be human-centered. Over the past decade, research activities have converged on new notions of "cognitive field research" and new frameworks that point toward methodologies for crafting human-centered systems. 8-13 Understanding Research over the past decade has led to T he widespread introduction of the personal computer, beginning about 1970, helped spawn the field of inquiry called cognitive engineering, which concerns itself with such things as interface design and user friendliness.