Expertise as mental set: The effects of domain knowledge in creative problem solving

JENNIFER WILEY ) 0 1 0 This article is based on a dissertation submitted to the University of Pittsburgh in partial fulfillment of the requirements for a Ph.D. The au thor thanks her advisor , James F. Voss, and her committee, Kevin Ash ley, Robert Glaser, and Jonathan Schooler , for their suggestions and support; Mara Georgi for her assistance in data coding; and Colleen Seifert, Steven Smith, and Barry Stein for their comments on an earlier Department of Psychology, University of Massachusetts , Amherst, MA 1 University ofPittsburgh , Pittsburgh, Pennsylvania Experts generally solve problems in their fields more effectively than novices because their wellstructured, easily activated knowledge allows for efficient search of a solution space. But what happens when a problem requires a broad search for a solution? One concern is that subjects with a large amount of domain knowledge may actually be at a disadvantage, because their knowledge may confine them to an area of the search space in which the solution does not reside. In other words, domain knowledge may act as a mental set, promoting fixation in creative problem-solving attempts. A series of three experiments in which an adapted version of Mednick's (1962) remote associates task was used demonstrates conditions under which domain knowledge may inhibit creative problem solving. - When Nonexperts Outperform Experts A number of studies have demonstrated that novices or individuals at an intermediate level ofknowledge may be seen to outperform experts. A first group of studies, re lated to the domain-specific nature ofexpertise, indicates that novices may outperform experts in conditions in which experts cannot make use of their domain knowledge. For example, chess experts' recall ofrandomized chess boards is slightly worse than that of novices (Chase & Simon, 1973). Similarly, Voss, Vesonder, and Spilich (1980) found that individuals with low baseball knowledge recalled a significantly greater number of baseball-irrelevant propo sitions from a text passage describing one half-inning of a baseball game than did individuals with high baseball knowledge. Further, studies have found no advantage for experts on tasks in which the presentation of domain related stimuli does not correspond to typical structures ofthe domain, such as memory for novice-generated pas sages about baseball (Voss et aI., 1980). Experts can also be worse at accurately judging their own performance within their domain. Glenberg and Epstein (1987) showed that physics and music experts overestimated their com prehension ofa text about their area of expertise, whereas nonexperts showed a more accurate calibration. Even though the experts did comprehend the texts better than did the nonexperts, the experts were overconfident in their abilities and were less accurate than nonexperts in their estimation of their understanding of the text. A second group of studies indicates that novices or in termediates surpass experts on tasks that require memory for the surface structure ofpresented information. In med ical diagnosis tasks, the intermediate effect has been well documented (Patel & Groen, 1991; Schmidt & Boshuizen, 1993). The basic finding is that, although doctors are more accurate at diagnosis, they are worse than interme diates, such as third-year interns, at recalling or recog nizing the exact information that they were presented with to make their decisions. Likewise, baseball experts are worse than novices on recognition tasks that require verbatim memory for baseball stories (Arkes & Freedman, 1984). When asked to select only the exact sentences that were actually in the text they had read, experts were more likely than novices to choose synonymous distractors or inferences based on the presented material. Experts in computer programming (Adelson, 1984) are worse at rec ognizing programs they have analyzed than are novices. And, although expert radiologists have better memory for atypical (and hence diagnostic) features of X rays that they have seen, they are noticeably worse than novices at recognizing normal X rays (Myles-Worsley, Johnston, & Simons, 1988). In these cases, experts are at a disadvantage as a result ofthe more abstract or principled nature of their processing, which, in most domain-related circumstances, leads to better or more efficient performance. A third group of studies indicates that experts can be outperformed by novices when a new task or context runs counter to highly proceduralized behaviors. Experts per form worse than novices when a shift from a standard means of representation is required or when a standard re sponse is inappropriate. For example, waitresses and bar tenders perform surprisingly poorly at Piaget's water-level task, as compared with groups without intensive liquid filled container experience, presumably because the rep resentation required to correctly solve the task is not the one that is offunctional importance in their fields (Hecht & Proffitt, 1995). Instead of shifting to an appropriate way of framing the task, waitresses and bartenders continue to use the representation that is usually relevant to their job performance. Likewise, Marchant, Robinson, Anderson, and Schadewald (1991) found that expert accountants were less likely than novices to correctly apply new infor mation they had just read about a tax law that disqualified certain standard business deductions. In another study, Frensch and Sternberg (1989) found that expert bridge players had a harder time than novices adapting to a new version of the game that changed the bidding procedure. In these studies, highly proceduralized domain knowl edge, which is usually advantageous in processing do main-related information, may have led the experts to act less flexibly in new contexts. The present study investigated another condition in which people with less domain knowledge may outper form people with a large amount of domain knowledge namely, creative problem solving. Experts have been seen to solve problems in their fields more effectively than novices because a well-structured, easily activated knowl edge base allows for efficient search of a solution space and possibly for automatic access to promising solution paths. But what happens when a problem requires a broad search for a solution outside the usual scope of the do main? One possibility is that subjects with expertise in the domain might actually be at a disadvantage, because their knowledge may confine them to an area of the search space in which the solution does not reside. In other words, domain knowledge may act as what traditionally has been called a mental set in the problem-solving literature, constraining search and promoting fixation in the cre ative problem solving of experts. Mental Set and Fixation in Creative Problem Solving A number of classic examples of mental set and fixa tion in problem solving have come from the Gestalt psy chologists in their studie (...truncated)