Methods for handling uncertainty should be evaluated in terms of their cognitive compatibility with real-world decision makers. Bayesian models of uncertainty demand precise up-front assessments of all problem elements and discourage the dynamic evolution of problem understanding. They handle missing or conflicting data by mathematical aggregation, while real-world decision makers regard gaps in knowledge and conflicting evidence as problems to be solved. Finally, they produce as output a statistical average rather than a coherent picture of the situation. Another approach to decision making, based on pattern-matching, does not address the ways in which situation pictures are evaluated and modified. A third approach, however, called the Recognition / Metacognition model, treats decision making under uncertainty as a problem-solving process that starts with the results of recognition, verifies them, and improves them where necessary. Critiquing strategies identify problems of incompleteness, conflict, and unreliability in situation models, and lead to correcting steps that retrieve or collect additional information or adopt assumptions. Training methods based on this model have been developed and tested with active-duty Naval officers.
Stored only when both papers are in the corpus — 17 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.