Experts become good at remembering information in their area of expertise
because the portions of their cognitive system that support the expertise will
have many enduring patterns of connections to represent and reconstruct that
information. For example, an expert on climate may be able to remember that
Seattle has a milder winter than Denver by activating the general principle
that oceans moderate climate. Much of what is involved in becoming an expert
is understanding the general principles that a body of information entails.
Oneoftheimportantpredictionsoftheconstructionistaccountofinterfer-
ence is that interference is not inevitable. Interference is expected when there is
no effective learning strategy for extracting the patterns that integrate increas-
ingly larger bodies of information. If such patterns can be extracted and used to
reconstruct the information, then increasing the amount of information should
not produce interference. Research confirms this prediction.
In one experiment, Jones and Anderson (1987) required subjects to memorize
varying numbers of facts about hypothetical characters. In some cases, the facts
were all related by a common theme. For example, subjects might learn that
John has a rifle, John is a hunter, and John is in the forest. In other cases the
facts were unrelated—for example: Jerry has a rifle, Jerry is a researcher, and
Jerry is at the beach. As in other experiments investigating the fan effect, the
subjects were then asked to verify whether particular facts were true (e.g.,
‘‘John has a rifle’’) or false (e.g., ‘‘John is a researcher’’). When the facts were
unrelated, the usual fan effect was observed. Subjects took longer to verify facts
about a character when there were six unrelated facts about that character to
memorize than when there was only one fact to memorize. But when the facts
were related by a common theme, the fan effect was greatly reduced. When the
facts were related by a theme, subjects took about as long to verify facts about a
character when there were six related facts to memorize about that character as
when there was only one fact to memorize. Similar results have been obtained
by Radvansky and Zacks (1991) and by Smith, Adams, and Schorr (1978). The
constructionist explanation is that when the facts are related by a theme, that
theme can be used to reconstruct whether a fact fits the theme and so must be
true, or does not fit the theme, and so must be false.
The constructionist theory of memory also implies that interference depends
on what kind of information subjects are asked to remember. If subjects in list-
memorizing experiments were asked to remember the general pattern of infor-
mation in the lists, rather than the unique details of each list, then presenting
subjects with several lists should promote better memory of the general pat-
tern, even while undermining memory for the unique details of each list.
Evidence consistent with the constructionist prediction is provided by Reder
and Ross (1983). They required subjects to memorize a varying number of facts
about hypothetical characters. Later, some subjects were required to indicate
whether a particular fact was explicitly on the memorized list, while other
subjects were asked to indicate whether a particular fact was similar to (implied
by) other facts on the list. For example, subjects might learn three facts about
Marvin: Marvin skied down the slope, Marvin waited in the lift line, and Mar-
vin waxed his skis. The usual fan effect was observed if the memory test required
subjects to judge whether a particular fact was explicitly on the list (e.g., Mar-
348 R. Kim Guenther