2.2 Problem solving, reasoning and related constructs

Problem solving as defined above is quite similar to some other constructs in modern psychology. Among them are "critical thinking" (Ennis, 1996; Norris, 1989), which encompasses judging the credibility of arguments, and naturalistic decision-making (Zsambok and Klein, 1997), defined as the use of knowledge and expertise to act under complex and uncertain conditions. Each of these constructs describes some kind of intellectual activity, based on reasoning and the application of knowledge. Therefore, they are closely linked to the construct of intelligence that many modern psychologists understand as a generalized capability to acquire, integrate, and apply new knowledge. Intelligence in turn is linked to more basic features of the human information processing system such as working memory capacity or mental speed (Neisser et al., 1996).

In the tradition of psychometric research, the core of general intelligence is called reasoning (Carroll, 1993) or information processing capacity (Süß, 1999). It is operationalized by tests using mathematical word problems, number series (e.g. 1, 2, 4, 7, 11, ... ?), and analogical reasoning, in particular by figural analogies like "/ is to \ as # is to ... ?". All these may be subsumed under the broad concept of problem solving as it was defined above — with the exception of rare cases in which highly trained persons solve such tasks using special algorithms. Thus, whatever indicator for problem-solving competence we use, it will to a certain degree be correlated to psychometric measures of reasoning ability. How strong this correlation is, and hence the extent to which problem-solving competence can actually be distinguished from reasoning, is an open question in cognitive-psychological research. Even with respect to complex, dynamic, computer-based problem-solving tasks (Frensch and Funke, 1995; see section 2.3 below) several studies suggest that inter-individual performance differences can be explained to a large extent by reasoning ability and basic features of the human information processing system (Süß, 1999).

In recent publications in the area of Differential Psychology, Robert Sternberg and his colleagues (see, e.g., Sternberg and Kaufman, 1998) have supported a very broad concept of intelligence, basically equating it with problem-solving abilities. Sternberg identifies three subcomponents of intelligence: a) analytical abilities such as "identifying the existence of a problem, defining the nature of a problem, setting up a strategy for solving the problem, and monitoring one's solution process", b) creative abilities "required to generate problem-solving options", and c) practical abilities needed to apply problem-solving strategies to real-life tasks. Sternberg assumes that practical intelligence is clearly discernible from analytical intelligence as assessed by means of the classic psychometric measures (IQ). However, methods for measuring creative problem-solving abilities and practical aspects of intelligence independently have yet to be devised. The procedure Sternberg proposed to measure practical intelligence cannot be regarded as a performance test: He presents respondents with descriptions of real-life or jobrelated problem situations and asks them to evaluate different response alternatives. If the evaluations made by the respondent correspond to those of a reference group ("experts" in an occupational field or representatively selected control groups for real-life problems), the respondent is said to have tacit knowledge, which Sternberg sees as the core of practical intelligence (Sternberg and Wagner, 1986).

The assessment of the third aspect in Sternberg's triarchic concept of intelligence — creativity — appears to be just as difficult. As problem solving involves new situations that cannot be dealt with routinely, it always requires a certain degree of creativity. Attempts to measure creativity independently as originality, flexibility and fluency of ideas (see Krampen, 1993) or to assess it as a distinctive feature of problem-solving performance (Mumford, Supinski, Baughman, Costanza, and Threlfall, 1997) have, however, yet to yield convincing results.