WJ Clancey — Visualizing Practical Knowledge
Shouldn’t he be observing the setting more broadly, in a theory-neutral way? On the
other hand, his time is short; he has just a few days to get information, and his
information must be useful. The templates of an article provide a way of relating data
usefully—the ideas of comparison, distribution, formation, and typicality are central to
biology. These are disciplined ways of thinking, orientations that make the biologist an
efficient, insightful observer.
Nevertheless, the biologist’s highly disciplined fieldwork was not preordained, but itself
developed during the first few days of the expedition. Originally, the group leader
envisaged studying lemmings when he saw the oases the year before. In early excursions
after his arrival in the third phase of HMP-98, the biologist saw no lemmings, but decided
to place some traps anyway. When no lemmings were trapped, the item of interest
became the oases themselves, leading to activities of excavating, mapping, and
measuring. Revisiting the same locations with a larger team, new discoveries were made
(bones, more feathers) and additional theories were articulated about oases’ formation.
Returning home, the biologist investigated previous related work, and eventually wrote a
paper that placed his study in the context of other biological formations in the Canadian
Arctic.
Although the idea of learning on the job may sound dubious to a manager of office work
and most businesses, for a scientist learning on the job is a necessity. If the scientist is not
forming new ideas and new theories, he is not working. Just as learning occurs during a
typical work day, we can track the conceptual change that occurs during a week’s
expedition.
Apollo’s two-day stays on the moon didn’t provide much opportunity for forming new
ideas and plans. Time was strictly controlled, and two days is not much time for shifting
objectives and plans. But in a week at Haughton, scientists experience broad shifts in
their objectives and plans, as observations and interpretations feed back to shape the next
day’s activities, producing new objectives, new observations, and eventually a different
understanding of what one is doing at Haughton.
Brahms: Multiagent simulation of situated action
Brahms is a tool for modeling (describing) and simulating practical knowledge (Clancey,
et al., 1998). We developed Brahms in order to make visible, to visualize, practical
14