Knowledge Issues
To what extent is memory
reliable?
To what extent do our
senses give an accurate picture of the world?
Analytical Approaches
Examining the validity of drawing generalized conclusions from specific experiments
Examining the impact of language – questions in particular – on recall.
Determining cause and effect
Examining methodological issues
What are the ramifications of this research for establishing the validity of eye witness testimony in court?
How easily can memory be
altered?
Study 1: 'Reconstruction of automobile
destruction' Loftus and Palmer (1974)
BACKGROUND
There
is much support for the idea that most people, when they are witnesses to a
complex event, such as a traffic accident, are very inaccurate when reporting
numerical details like time, distance, and especially speed, even when they
know that they will be questioned on them (e.g. Marshall, 1969). As a
consequence, there can sometimes be large variations in estimates between
witnesses and so it seems likely that such inaccurate testimony could easily be
influenced by variables such as the phrasing of questions or 'leading'
questions.
AIM
Loftus
and Palmer, therefore, aimed to investigate the effect of leading questions on
the accuracy of speed estimates in, and perceived consequences of, a car crash.
EXPERIMENT ONE
Subjects: 45 students,
tested in groups of different sizes.
Design: Laboratory
experiment.
Procedure:
7
films of traffic accidents, ranging in duration from 5 to 30 seconds, were
presented in a random order to each group.
After
each film, the subjects had to give a general account of what they had just
seen and then answer more specific questions about the accident. The critical
question, 'About how fast were the cars going when they hit each other?' acted
as the independent variable, since it was manipulated in five conditions.. Nine
subjects heard the sentence with the verb 'hit' in it, and then an equal number
of the remaining subjects were asked the same question but with the verb
'smashed', 'collided', 'bumped' or 'contacted' instead of 'hit'. The estimated
speed was the dependent variable.
Results
Speed estimates for the verbs of experiment one
Verb
Mean speed estimate
Smashed
40.8
Collided
39.3
Bumped
38.1
Hit
34.0
Contacted
31.8
|
Significance of result
Results were significant at the P < .005
level, according to analysis of variance of the data.
|
Accuracy of subjects' speed estimates
In 4 of the 7 films the speed of the cars was
known.
Actual
speed of collision Mean speed
estimate
Film 1
20 mph 37.7 mph
Film 2
30 mph 36.2 mph
Film 3
40 mph 39.7 mph
Film 4
40 mph 36.1 mph
|
Discussion
The
results indicate that not only are people poor judges of speed, but they are
systematically and significantly affected by the wording of a question.
However,
it is unclear if this finding could be attributed to either response-bias
(the subject remembers accurately but is pressured by the word to increase or
decrease the estimate) or a genuine change in the subject's memory of the event
(the word makes the subject recall the event as worse than it was).
Therefore,
this study suffers from a lack of validity. There is a lack of evidence that
suggests memory is being altered as the subjects could simply be changing their
account to suit the needs of the researcher but the memory remains accurately
intact.
Therefore,
further testing was necessary into whether the subject might be led into
recalling details that did not occur. The second experiment was designed to
specifically test memory accuracy.
EXPERIMENT TWO
Subjects: 150 students,
tested in groups of different sizes.
Design: Laboratory
experiment.
Procedure:
A
film lasting just less than a minute was presented to each group which featured
four seconds of a multiple traffic accident. After the film, the subjects had
to give a general account of what they had just seen and then answer more
specific questions about the accident. The critical question concerning the
speed of the cars was the independent variable, and it was manipulated by asking
50 subjects 'About how fast were the cars going when they hit each other?',
another 50 'About how fast were the cars going when they smashed into each
other?' and another 50 acted as a control group who were not asked the question
at all. One week later the dependent variable was measured - without seeing the
film again they answered ten questions, one of which was a critical one
randomly positioned in amongst the ten questions, asking 'Did you see any
broken glass? Yes or no?'. Although there was no broken glass it was expected
that some might be falsely remembered if the leading question of a week ago had
changed the memory of the event to seem worse than it was.
Results
Verb
Mean estimate
Smashed
10.46 mph
Hit 8.00 mph
|
Response
Smashed Hit Control
Saw broken glass 16 7 6
Did not see glass 34 43 44
|
Probability of seeing broken glass with speed
estimate
Verb
1-5 mph 6-10 mph 11-15 mph 16-20 mph
Smashed
.09 .27 . 41
.62
Hit .06 .09 .25 .50
|
Discussion
The
authors conclude that the results show that the verb 'smashed' not only
increases the estimates of speed, but also the likelihood of seeing broken
glass that was not present. This indicates that information from the original
memory is merged with information after the fact, producing one distorted
memory. This shift in memory representations in line with verbal cues has
received support from other research.
EVALUATION
Methodological: A well operationalised
and controlled experiment, but lacked the ecological validity of having real
life events and involved witnesses.
However,
there is a key problem with internal validity (the results may have an
alternative explanation than what the authors claim). For example: the results
show the higher the speed, the more likely the participants claim they saw
broken glass. This may be due to a desire to appear consistent with their
earlier responses: If they had estimated a higher speed then it is logical for
them to assume there will have been broken glass. They 'took a chance' that
broken glass was present which was a logical assumption after estimating a
higher speed but the study does not necessarily show a failure in memory.
In
the introduction of their study the authors claim to be measuring memory by
asking:
"How
accurately do we remember the details of a complex event, like a traffic
accident, that has happened in our presence?" (1974: 585, italics mine)
However,
in their discussion they acknowledge memory may not be at fault but the answer
given:
"To
reiterate, we have first of all provided an additional demonstration of
something that has been known for some time, namely, that the way a question is
asked can enormously influence the answer that is given." (1974:
588, italics mine)
Therefore
they implicitly acknowledge the difference between answers and memory but
they do not distinguish between them in the study.
However,
it is not discussed further.
Therefore,
the author's claim of reconstructive memory can be challenged by suggesting
that rather than memory being altered it is the answers that are changing. This
is a moot point in courts of law but it is a key point of validity for social
scientists: What is being manipulated, the actual memory, recall
or responses of participants under experimental conditions?
Study 2: The effects of coherence and
culture on reconstructing memory
War of the Ghosts
Aim:
To investigate the effect of schemas on memory (Bartlett, 1932)
Method:
Bartlett asked English participants to read The War of the Ghosts, a Native
American folk tale containing details about hunting seals in canoes.
The
participants’ memory for this story was tested by, serial reproduction and
repeated reproduction. In serial reproduction, the first participant reads the
original story and then reproduces it on paper. The first participant’s
reproduction is read by the second participant who also reproduces it for a
third participant. This procedure continues until six or seven reproductions
are completed by an equal number of participants. In repeated reproduction, the
same participant writes all six or seven reproductions over many years.
Results:
The two methods led to very similar findings; the story became increasingly
shorter and changed to suit the English cultural background of the
participants. For example, ‘hunting seals’ became ‘fishing’ and ‘canoes’ became
‘boats’. What started as a very strange story to the participants became a traditional
English story.
Study 3: Age effects on
eye witness testimony.
Parker and
Carranza (2005)
Research focusing on the accuracy of elderly eyewitnesses has
addressed two separate components of eyewitness memory: the ability of
eyewitnesses to describe details of the event, and their ability to recognize
the crime’s perpetrator
Aim: To investigate the effects of age of witness and age of suspect
on eyewitness testimony
Method: Forty-eight elementary school children and 48 college students
viewed a slide sequence of a mock crime. This was followed by target-present or
target-absent photo identification.
Findings: In photo identification, child witnesses had a higher rate of choosing
the correct crook than adult witnesses, suggesting that children have more lax
criteria of responding and adults made significantly more 'don't know' choices.
This supports the notion of schema development with age: Older people have more
fixed expectations over what a crook should look like and so tried to fit what
they remembered with their existing schema.
Concluding comments:
Memory
is not a straightforward recall of the world as 'is' It is influenced by
leading verbs, schemas and context. However, in all of these studies the 'gist'
of the situation (e.g. car crash, office objects, burglary) was recalled
correctly. Memory was influenced when researchers deliberately set out to
influence it. Generalizability to everyday life is difficult. But some broad
conclusions can be drawn: Memory is an active reconstructive process (as shown
by War of the Ghosts), rather than a passive reproductive one
