ABSTRACT
To behave effectively in a changing environment requires an information processing system that can easily adapt to new tasks and to new situations. This ability depends on the degree of flexibility of the system. A highly flexible system can adapt to a variety of new conditions, whereas an inflexible or rigid one is limited to a comparatively small set. At a very general level, to be flexible one needs to acquire new skills and relevant knowledge very quickly in unknown situations as well as to transfer and adjust already available knowledge so that it meets the demands of the new situation. In a strict sense, it may be necessary to modify strategies and procedures in every problem-solving situation. By definition, to solve a problem, contrary to just performing a task, requires more than just bringing to bear prior acquired knowledge. In many cases, finding a solution requires changing the point of view, as in the candle problem (Duncker, 1935), changing mental representations, as in the mutilated checkerboard problem (see Anderson, 1990, p. 245), or changing strategies, for example, switching from a goal-driven to a more data-driven procedure.
