using the information implied by the question to reconstruct the details of the
accident. If the question falsely implied that the car stopped at a stop sign, then
subjects reconstructed a stop sign in their recollections of the accident.
Exactly what would such a reconstruction be based on? One possibility is
that mentioning a stop sign effectively erased or somehow undermined the
connection between the accident and the yield sign and replaced it with a con-
nection between the accident and the stop sign (Loftus & Loftus, 1980). There is
another possibility, though. Maybe subjects do remember that the film con-
tained, say, a yield sign and that the subsequent question mentioned a stop
sign. But when given the choice between a yield and stop sign, the subjects
figure that the experimenter wants them to say that they saw a stop sign in
the film (otherwise, why would the experimenter ask the question?). In other
words, maybe subjects ’memories are just fine in this paradigm; maybe they are
just responding to the demands characteristic of the experiment; maybe this
research is not supportive of a construction approach to memory (McCloskey &
Zaragoza, 1985).
To see if the question about the stop sign really erased the information about
the yield sign (or, more generally, if misinformation erases previously acquired
information), McCloskey and Zaragoza (1985) devised a somewhat different
experimental paradigm (this paradigm, the Loftus paradigm, and a couple of
other paradigms that I discuss below are all illustrated in figure 14.4). Subjects
first saw a film that contained details like the yield sign, and then read a text
that contained misinformation, such as a description of a stop sign, and then
were asked to decide if the original film contained a yield sign or, say, a caution
sign. Again, the actual paradigm includes several pieces of information, and
not just information about traffic signs. If the misinformation really wiped out
memory of the yield sign, subjects should be just as likely to choose the yield
sign as to choose the caution sign. Instead, subjects overwhelmingly selected
the correct alternative, the yield sign in this case.
So does this mean that the misinformation has no effect on eyewitness mem-
ory at all? No. In other research (see Lindsay, 1993), subjects were first shown a
film (or slide show) of a crime or accident, then read a text that contained some
misleading information, and then were asked of each piece of information
whether the information was presented in the film, in the text, in both places, or
in neither place. The memory test in this case (see figure 14.4) asks subjects the
source of the information. The idea is to see whether source memory is worse
for a detail in the film when there is misinformation in the text than when there
is not misinformation in the text.
The main finding is thatsource memorytends to be good (i.e., subjects cor-
rectly remember that the stop sign was in the text and not in the film) when it is
easy for subjects to discriminate between the experience of seeing the film and
reading the text (e.g., Zaragoza & Lane, 1994; see Lindsay, 1993). One way to
make the discrimination easy is to present the film on one day, wait until the
next day to present the text, and then immediately follow the text with the
memory test. Source memory tends to be poor (i.e., subjects think that the stop
sign was in the film) when it is hard for subjects to discriminate between the
experience of seeing the film and the experience of reading the text. For exam-
ple, a way to make the discrimination hard is to present the text right after the
Memory 335