If we asked members of that pioneering generation of expert–novice re-
searchers to judge Jason, Evie, and Bruce’s expertise, they would not be inter-
ested in any of the history-specific information provided in the opening sce-
nario. Such descriptions (e.g., Jason’s understanding of history or Evie’s
volunteering as a docent) would merely distract from critical determinations.
Rather, these first-generation expertise researchers would assess Jason, Evie,
and Bruce’s ability to tackle demanding but generic problems with speed, ac-
curacy, and efficiency.
Second Generation: Expertise as Knowledge-Rich
Problem Solving
Soon there followed a second generation of expertise researchers who dem-
onstrated that general problem-solving strategies did not adequately distin-
guish experts from nonexperts (Holyoak, 1991). This second generation, like
the first, continued to focus on problem solving as the mechanism for opera-
tionalizing expertise. However, the second-generation researchers were not
interested in general search strategies applied to knowledge-lean problems.
Instead, they targeted tasks within particular problem-solving contexts, such
as playing chess, typewriting, waiting tables, or solving physics problems, for
which certain knowledge was expected to matter (Anderson, 1983; Chi, 1978;
Chi, Feltovich, & Glaser, 1981). Careful task selection allowed the research-
ers to document that domain-specific knowledge and associated problem per-
ceptions were significant determiners in expert performance (Ericsson &
Smith, 1991).
Take the pioneering research of de Groot (1946/1978) and Chase and Si-
mon (1973; Simon & Chase, 1973) as a case in point. Because these research-
ers wanted to discern the nature and defining attributes of expertise, they
chose to study chess. Chess was ideal as a task domain for this research be-
cause the game has a limited number of performance rules. However, there is
great diversity in the execution of those rules between very inexperienced and
highly skilled players. Further, problem-solving moves are externalized in
chess. Thus, researchers could record and analyze the perceptions and rea-
soning that instigated various moves by prompting players to verbalize their
actions. Finally, the procedural nature of chess and other selected problem
domains allowed researchers to create simulations or laboratory versions of
real-life tasks that could be studied without the contextual influences present
in everyday performance (Ericsson & Smith, 1991).
Second-generation expertise researchers would likely judge the expertise
of Jason, Evie, and Bruce by examining their performance of a carefully
crafted history task. They would select or construct a task that would rely on
solution procedures that could be represented as a series of production rules
or solution steps (Anderson, 1987). It is likely that our three individuals
276 ALEXANDER