Studying Cognitive Systems in Context: Preface to the Special Section

Abstract

MULTIPLE PURPOSES, MULTIPLE HISTORIES, MULTIPLE THREADS Intellectually and historically, many lines of development have intertwined and contributed to the current interest in field research on cognitive and collaborative activity across multiple human and machine agents. For instance, in Europe a line of work began to examine social, psychological, and cognitive activities in the workplace in response to larger political and social changes in the late 1960s - a tradition that came to be called work analysis (see De Keyser, 1992, for a brief overview). Another example is knowledge acquisition, which became a flourishing activity in the 1980s to fuel the demand for expert systems and other applications of artificial intelligence (AL), especially in North America (see Gaines & Boose, 1988, for the results of the first workshop on this topic). One fairly widely used label for studies of cognition and collaboration in the workplace is cognitive task analysis (see Schraagen, Chip-man, & Shalin, in press). Around 1980 cognitive task analysis (CTA) and similar labels emerged independently and in parallel in several lines of work as a natural consequence of the evolution of work, the increasing computerization of work, and, in the United States, the recognitivization of psychology (see Hoffman, Woods, & Annett, 2000). Among the threads that began to use or draw a link to the term CTA were 1. the study of expertise for the development of different forms of computer-aided training (e.g., Chi, Feltovich, & Glaser, 1981; Glaser et al., 1985; McKeithen, Reitman, Reuter, & Hirtle, 1981) -- what might be labeled a cognitive learning thread; 2. the need to aid human cognition and performance in complex, high-consequence settings (e.g., Hollnagel & Woods, 1983; Rasmussen, 1986; Rasmussen & Rouse, 1981) -- what might be labeled a functionalist cognitive engineering thread; 3. the need to understand work cultures as a result of the consequences of technology change (e.g., Hutchins, 1990, 1995a, 1995b; Jordan & Henderson, 1995; Suchman 1987) - what might be called an ethnography of work thread; 4. the need to understand work that is conducted using computers, and to support the design of human-computer interfaces (e.g., Card, Moran, & Newell, 1983; Kieras, 1988) - what might be called a cognitive simulation thread; and 5. the inability of formal or normative models of decision making to capture observations from field studies of experts in complex settings (e.g., Klein, 1989; Klein, Orasanu, Calderwood, & Zsambok, 1993) - what might be called a naturalistic decision-making thread. These and other strands arise from attempts to understand the interaction of cognition, collaboration, and complex artifacts by studying cognitive systems in context. The reference point for all these trends is the field, where teams of practitioners confront significant problems aided by technological and other types of artifacts. When interest in fieldwork flares, U.S. experimental psychology responds with attacks questioning the ability of fieldwork to add to the knowledge base (e.g., Banaji & Crowder, 1989). Those using natural history techniques of whatever type respond with general justifications for research in the wild (Hoffman & Deffenbacher, 1993; Woods, 1993). Others ignore the shouting and proceed with their studies in the field based on their local driving forces and their particular historical links. However, the different purposes, histories, and threads have resulted in a fragmented landscape, parochialism, imperialism, and a proliferation of labels. For example, terms such as cognitive task analysis (Hollnagel & Woods, 1983; Roth & Woods, 1989), knowledge elicitation (Hoffman, 1987; Hoffman, Shadbolt, Burton, & Klein, 1995), ethnomethodology (Suchman, 1987), cognitive work analysis (Vicente, 1999), concept mapping (Means & Gott, 1988), naturalistic decision making (Klein, 1989) and others abound, and a single label may be used to refer to many very different methods. …

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 75 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.