processing. One of the most important is the literature on expertise or expert–
novice differences. Still, as the MDL has taken shape and been repeatedly
tested in various academic domains, some of the basic attributes of the extant
literature on expertise have been brought into question.
TRADITIONAL EXPERT–NOVICE STUDIES
During the 1970s and 1980s, under the theoretical umbrellas of artificial intel-
ligence (AI) and information-processing theory (IPT), programs of research
formed around the problem-solving performance of experts. There were at
least two goals for this inquiry. For one, the intention was to capture the
characteristics of expert performance and to validate them over a variety of
problem-solving tasks so that those characteristics could be programmed
into nonhuman systems (Alexander, 2003; Ericsson & Smith, 1991). The re-
sult would be smart machines that approximated effective human behavior.
Another goal was to determine what cognitive attributes distinguished ex-
perts from novices so that those attributes could be trained in nonexpert hu-
man populations (Chi, Glaser, & Farr, 1988).
First Generation: Expertise as Generic Problem Solving
According to Holyoak (1991), at least two prior generations of theory and re-
search have shaped current understandings about expertise. The first genera-
tion, represented by research in AI, conceptualized expertise as the efficient
and effective solution of generic problems. Researchers of this generation,
such as Newell and Simon (1972), set out to isolate the search strategies that
experts employ to identify and then solve problems for which content knowl-
edge presumably would play an insignificant role. The classic cannibal–mis-
sionary problem that follows is representative of the knowledge-lean prob-
lems that served as the experimental stimuli for studying expert performance
and for documenting search heuristics (e.g., means–end analysis) in this initial
generation. Those tasks are called knowledge-lean because it is assumed that
all the information needed to answer them is given in the problem statement:
There are three missionaries and three cannibals on a river bank. The mission-
aries and cannibals need to cross over to the other side of the river. For this pur-
pose, they have a small rowboat that holds just two people. There is one prob-
lem, however. If the number of cannibals on either river bank exceeds the
number of missionaries, the cannibals will eat the missionaries. How can all six
get across to the other side of the river in a way that guarantees that they all ar-
rive alive and uneaten? (Sternberg, 1986, p. 57)- MODEL OF DOMAIN LEARNING 275