Expertise 513skills may develop in part because time limits in chess penal-
ize long search times and encourage sacrificing; this factor
could limit the generalizability of the characterization of
chess expertise to other skills.
Studying Expertise
The original expertise approach consists of three steps. The
first step is to produce and observe outstanding performance
in the laboratory under relatively standardized conditions
using tasks that are representative of the skills possessed
by the expert. The second step is to create a detailed picture
of expert performance by analyzing and describing the
processes critical to the production of an outstanding perfor-
mance on the tasks. The third and final step is to examine
critical cognitive processes and propose explicit learning
mechanisms to account for their acquisition. In other words,
the object is to develop an account of the expert’s knowledge
structures and processes.
A variety of knowledge acquisition techniques has been
used to analyze expert knowledge structures and processes.
Some techniques, such as hierarchical card sorting and gen-
eral weighted networks, are based on judgments of similarity.
In card sorting tasks, cards containing one piece of domain
knowledge each are sorted into categories and subcategories.
The resulting hierarchical structure is presumed to reflect the
way the expert structures his or her actual knowledge. A lim-
itation of this technique is that the requirement to make cate-
gories may force the expert to create a different structure
from that which actually exists. The Pathfinder algorithm
(Schvaneveldt, 1990) poses fewer constraints. Experts sim-
ply rate pairs of domain terms for their similarity, and then
the algorithm is applied to find the network structure under-
lying the knowledge. Protocol analysis has also been exten-
sively used to study expert behavior. In its most general
application, experts are asked to think aloud while they solve
a problem or perform a task. In this method, experts are sim-
ply asked to verbalize any thoughts that come to mind as they
are performing a task. Ericsson and Simon (1993) maintain
that this technique, unlike requiring people to explain their
thinking, does not seem to cause any restructuring of the cog-
nitive processes involved in task performance. On the other
hand, it is limited to knowledge of which the expert is aware.
Actions that are performed automatically or very quickly are
likely to escape the notice of the performer.
Characteristics of Expertise
As mentioned above, changes in working memory span for do-
main information are often cited as a characteristic of
expertise.Skilled memory theorystates that at the time of
encoding, experts form a set of retrieval cues that are associ-
ated in a meaningful way with the information to be stored.
Retrieval then occurs via these cues. Rather than just chunking
information so that more information can be stored in short-
term memory, experts develop memory skills that enable them
to store and retrieve information in long-term memory more
quickly and efficiently, thus circumventing the limits of short-
term memory (Chase & Ericsson, 1982). Chase and Ericsson
thus argue that extensive practice develops skills that lead to
qualitative, and not just quantitative, differences in memory
performance for the practiced type of information.
In the area of problem solving, several generalizations
about expert performance can be made. As Anzai (1991) has
shown in studies of physics expertise, experts work forward,
novices backward. Novices are also apt to change problem
representations more frequently than experts; they seem
unable to decide which representation is best for solving a
problem. Experts, on the other hand, generate and update
a representation of a problem as they read it. By the time a
question regarding the problem is presented, they are often
able to retrieve a solution plan from memory based on this
representation (Larkin, McDermott, Simon, & Simon, 1980).
In other words, the representation cues the expert’s knowl-
edge. Novices may lack both the organized knowledge base
and the ability to build a representation that can act as a cue.
Physics experts seem to possess multiple modes of represen-
tation for solving problems as well as the procedural knowl-
edge for effective use of these multiple representations. This
provides the basis for the formation of abstract or simpler
representations from less abstract or more complex ones.
In the area of motor skill expertise, a distinction has been
made between knowing and doing (Allard & Starkes, 1991).
Knowing, in this context, involves directing the intake of en-
vironmental information in the appropriate way. The doing
component is essential for the execution of actions, sport
techniques, and motor-control programs. Knowing (consist-
ing primarily of search processes) dominates in open skills,
such as football, in which the environment (often including
an opponent) is important and performance is directed to-
ward an external goal. Doing (which can be characterized as
skilled memory) predominates in closed skills, such as figure
skating, in which the skill is performed in an invariant envi-
ronment and has the production of a particular motor pattern
as its goal.Skill and ExpertiseThe fact that practice seems to be the most important deter-
minant of the acquisition of expertise means that learning