letter diagram representing the chess positions, his memory performance was
only half as good as his performance with a real board. But after only 16 trials,
his performance with the unfamiliar diagrams had improved to the level of his
performance with the real board. Charness (1991) provides a review of the
current research using different visual representations of chess positions.
There is some evidence that there are limits to the extent to which stimuli can
be abstracted. Gilhooly, Wood, Kinnear, and Green (1988) demonstrated that
the lack of superior memory performance by expert map users, as compared
with the novices studied by Thorndyke and Stasz (1980), could be attributed to
their use of schematic maps (mainly used by tourists) as stimuli. By studying
recall of both schematic maps and more advanced contour maps by expert and
novice map users, Gilhooly et al. (1988) found, as expected, superior memory of
contour maps by the experts, but no differences between experts and novices
for the commonly available schematic maps. The fact that superior perfor-
mances can be reproduced in the laboratory with schematic stimuli is important
not only for practical purposes but also for theoretical analyses of the media-
ting mechanisms.
The issues of how to design representative laboratory tasks are discussed in
many chapters in this volume. For example, Patel and Groen (1991) consider
the differences between medical diagnoses based on written texts presenting
medical cases and diagnoses based on interviews with real patients. Do ̈rner and
Scho ̈lkopf (chapter 9, this volume) report on the management of simulations of
very complex systems.
Summary. Theessentialfirststepofthestudyofexpertperformanceinvolves
identifying a collection of standardized tasks that can capture the superior
performance under controlled conditions. It is a necessary condition for further
analysis that superior performance by experts be reliably shown for the
designed tasks. In complex domains it is often especially difficult to identify a
population of tasks to capture the expertise; it may be possible to identify in-
stead a small number of representative tasks to elicit superior performance.
Nonetheless, it may be useful to think of expertise in terms of a corresponding
population of tasks. Various experts may, however, require different popu-
lations of tasks. Patel and Groen (chapter 4, this volume) show that with in-
creasing expertise in medicine, experts become more specialized in particular
areas of medicine. Similar specialization is to be expected in most complex
domains. To capture specialized expertise adequately, it is necessary to design
special populations of tasks appropriate for a small group of experts or even
individual experts (case studies). Superior memory performance by an expert is
a legitimate subject for study as long as we keep in mind that the processes
underlying the superior memory performance may only partially overlap with
those that generally underlie the superior performance of experts.
Thefactthatitispossibletoreproduceexpertperformanceinalaboratory
task has important theoretical implications. It reduces the significance of large
numbers of factors that influence complex real-life situations. Furthermore, it
indicates a fair degree of generalizability, especially concerning the detailed
stimulus representation. Let us now turn to further analysis of the processes
that mediate superior performance.
532 K. Anders Ericsson and Jacqui Smith