Roebers, C.M. & McConkey, K.M. (2003). Mental
reinstatement of the misinformation context and the misinformation effect in
children and adults. Applied Cognitive Psychology, 17, 477-493.
This is a paper about the misinformation effect. In the misinformation effect people witness
and event and then are presented with either misleading or neutral information
(in some experiments there’s a third condition where people get consistent
information). On later memory tests
misleading information tends to reduce accuracy.
People have tried to explain the misinformation effect in a
bunch of ways. For instance, Loftus provided a storage based account where the
suggested information ‘destructively updates’ memory. McCloskey and Zaragoza argued that the misinformation effect
happens when people can’t remember the correct answer but reason that the
experimenter must know what happened, and so they deliberately go along with
the experimenter. Lindsay explained the
misinformation effect as being due to a source monitoring error. According to fuzzy trace theory the
misinformation effect can occur either due to gist or verbatim memory for the
misinformation. Anyway, lots of
possibilities.
The authors attempt to understand the misinformation effect
in terms of an associative network model that Ayers and Reder came up
with. Here’s the idea, when people
encode an event it gets represented in an associative network where the details
of the event, including contextual details get represented. When people are presented with misleading
information the misleading information and the context in which the misleading
information were encoded gets integrated with the original event
representation. Importantly, at time of
test, when concepts get activated, people don’t become consciously aware of the
source of the activation. All this
leads the authors to predict that reinstating the context of the postevent
session should increase the size of the misinformation effect. Pretty counterintuitive if you ask me, let’s
see how it turns out.
Five year olds, seven year olds and adults watched a brief
film. One week later they received a
misleading interview that included some misleading questions about central details,
some misleading questions about peripheral details, and a few filler
questions. Finally one week later they
took a two alternative forced choice recognition test. Before taking the test they were either
asked to mentally reinstate the context of the earlier interview or they were
not. No robot monkeys were harmed in
this experiment.
There was a strong age effect on the acquiescence to
misleading information during the first interview. Children were much more likely to incorrectly accept the
suggestions in those interviews.
On the final forced choice recognition test, children were
less accurate than adults. Overall,
reinstatement of context decreased accuracy for the adults but not for the
children. The effect of reinstatement of
context appeared to be limited to the peripheral questions, and when peripheral
questions are looked at reinstatement of context had an effect on both 7 year
olds and adults.
Reinstating the context of the postevent questions decreased
how accurate adults and 7-year olds were with regard to peripheral
information. Reinstatement of context
did not have the same effect on the youngest children.
Here’s what the authors think is going on: When
people witnessed the original event it got represented in an associative
network. When people were asked the
misleading questions, the misleading information, as well as the context in
which that misleading information was presented, became integrated with the network
representing the original event. When
the misinformation context was reinstatement, activation spread to the
misleading information. Because people
don’t have conscious access to the cause of the activation (i.e. that it was
from misleading information) they incorrectly accepted the items that had been
previously suggested. The authors
suggest that the age differences might be because as kids get older they’re
able to build bigger, more interconnected networks, with stronger associations
between the elements.
Another interesting finding was that kids were more likely
to yield to questions during the postevent interview, but then to change their
answer to the correct one at the time of the final test.