Transcript Chapter 14

+
Chapter 14
Eyewitness Testimony
+
Eyewitness Demos

Greased Lightning (Grease)


http://www.youtube.com/watch?v=wK63eUyk-iM
Simons & Levin – change blindness “door” study

http://www.youtube.com/watch?v=wK63eUyk-iM
+
3
Change Blindness


Simons and Levin (1998)
 Design:
 A stranger asks unwitting
participants for directions
 After 10–15 seconds, people
carrying a door pass in front
of the participant, blocking
their view
 During this time the
stranger is replaced by a
different person
 Results:
 About 50% of participants
failed to notice the switch!
Change blindness:
 Failing to notice apparently
obvious changes in a scene

Levin et al. (2002)

Change blindness blindness:


The unduly optimistic belief
that one is very rarely
affected by change
blindness
People underestimate the
importance of fixating in order to
detect changes

They mistakenly assume that
they fully process everything in
the periphery, too
+
4
Some Factors Reducing the
Accuracy of Eyewitness Testimony

Eyewitnesses generally are
taken off guard by the crime

Eyewitnesses are subject to:

Change blindness

Prior expectations

Pre/post-event information

Overblown confidence
The criminal actions are often
brief and swift

Unconscious transference

Verbal overshadowing
Criminals take steps to avoid
recognition

Weapon focus




They are often preoccupied
with their own thoughts and
plans
e.g. they wear disguises
+
5
Expectations Can Distort Memory
 Hastorf
and Cantril (1954)
 Design:
 Dartmouth
and Princeton students watched an
American football game between the two schools
 They were asked to detect violations of rules
 Results:
 Princeton students detected twice as many
violations by Dartmouth than did Dartmouth
students
+
Schemas for Bank Robbery
The schema for a
bank robber is:
male, wearing a
disguise and dark
clothes, making
demands for
money, with a getaway car and
driver.
+
7
Expectations Can Distort Memory

Lindholm and Christianson (1998)



Design:
 Swedish and immigrant students watched a simulated robbery, in
which a knife-wielding burglar was either:
 Swedish – blond with light skin
 An immigrant – black hair with brown skin
 The students were asked to pick the perpetrator from a lineup:
 Half were Swedish (4); half were immigrants (4)
Results:
 The correct person was identified 30% of the time.
 Both Swedish and immigrant students were twice as likely to
mistakenly select an innocent immigrant as an innocent Swede
Conclusion:
 The overrepresentation of immigrants in Swedish crime statistics
likely influenced participants’ expectations
+
8
Schemas Shape and Distort
Memory

Bartlett (1932)

Schemas structure our world
knowledge and influence
memory storage/retrieval


Eyewitnesses use schematic
information to assist in their recall

Tuckey and Brewer (2003a;b)

Eyewitnesses have better recall
for schema-relevant information
than for irrelevant information.

Eyewitnesses generally
interpreted ambiguous
information in a way that made
it consistent with the schema.
Contribute to memory
reconstruction:

Used to piece together
the details of an event in
terms of “what must
have been true”

Balaclava wearer was male.
+
Schema Intrusions Increase with
Ambiguous Stimuli
Mean correct
responses and
schemaconsistent
intrusions in the
ambiguous and
unambiguous
conditions with
cued recall. Data
from Tuckey and
Brewer (2003b).
+
10
Leading Questions

Loftus and Palmer (1974)
 Design:
 Participants watched a multiplecar accident and described what
happened
 Then answered specific
questions:
 “How fast were the cars going
when the cars smashed
into/hit/collided
with/bumped/contacted each
other?”
 After one week, they were asked
if they saw broken glass
 (There wasn’t any)

Results:
 Speed estimates depended on
the word used in the question
 Highest for “smashed” -- 40.8
mph
 Lowest with “contacted” -- 31.8
mph
 When “smashed” was used,
participants were more likely to
mistakenly claim they saw
broken glass (32% compared to
14% for “hit”)

Conclusion:
 Memory can be systematically
distorted by the way questions
are phrased
 Different schemas may be
activated
+
11
Interference and Memory Distortion
Retroactive Interference

Eakin, Schreiber, and SergentMarshall (2003)
 Eyewitness memory can be
impaired by misleading
information presented after
they have witnessed the crime.
 This is an example of
retroactive interference
 Memory is impaired even when
eyewitnesses were warned
about the presence of
misleading information after it
had been presented.
 Example – maintenance man
repairing a chair steals money
and a calculator, hiding it under
a screwdriver – not a wrench.
Proactive Interference

Lindsay et al. (2004)
 Design:
 Listened to a thematically
similar or dissimilar narrative
prior to seeing a burglary
 Results:
 Recall errors were more
frequent when the prior
narrative was similar to the
actual event
 This is an example of
proactive interference
 Conclusion:
 Eyewitnesses’ previous
experiences can shape what
they remember
 Example – school visit to
palace vs palace burglary.
+ Explaining Retroactive Interference
Distortion
Source Monitoring Framework
 The


Source Monitoring Framework
A memory probe activates related traces
 Including memories from other sources
One tries to determine the source, based on the information
the memory contains
 Sometimes source misattributions occur
 Especially likely when memories from different sources
are similar
12
+
A Recent Source Misattribution
Hillary Clinton in Sarajevo, 1996

"I remember landing under sniper fire," she said in Washington on
Monday. "There was supposed to be some kind of a greeting
ceremony at the airport, but instead we just ran with our heads
down to get into the vehicles to get to our base."
News footage of the event however showed her claims to have been
wide of the mark, and reporters who accompanied her stated that
there was no sniper fire. Her account was ridiculed by ABC News as
"like a scene from Saving Private Ryan".
http://www.youtube.com/watch?v=8BfNqhV5hg4

Is this motivated political opportunism or was the original
memory distorted by viewing other sources of information?
+
14
Explaining Retroactive Interference
Overwriting the Original Memories

Loftus (1979)



Design:
 Witnessed a pedestrian accident with a car stopping at either a:
 Stop sign
 Yield sign
 Two days later, participants were asked a leading question, referring to the
opposite type of sign to the one they had seen (stop vs yield).
 Forced-choice recognition test for snapshots from the original scene
 One photo had a yield sign; the other had a stop sign
Results:
 70–85% selected the sign they were falsely led to believe existed.
 This happened even when subjects were paid for correct answers to
reduce demand characteristics.
Conclusion:
 Information from misleading questions permanently alters the original
memory, which is overwritten and destroyed.
+
15
Misinformation Effect: Caveats


Bekerian and Bowers (1983)
 Misinformation effects can
be eliminated by
systematically questioning
participants in order -- from
earlier incidents to later ones
 This suggests that the
original memory trace
survives.
Loftus (1992) revised her view
to say that eyewitnesses can
come to accept misleading
information as time goes by.
 The original memory need
not be overwritten but added.

Memory distortions are more
common for peripheral/minor
details than central details.
 Heath and Erickson (1998)

In real-life criminal
investigations, eyewitness
memories can be quite robust
against misleading questions.
 Yuille and Cutshall (1986)
studied people 5 months after
a shooting they witnessed.
 However, there were only 13
subjects and the extent of
misinformation was modest.
+
16
Individual Differences
Age

Young children and elderly adults (age 60-80) are more susceptible to
misleading information.


Older adults had 43% false memory compared to 4% for younger adults.
Elderly adults tend to be very confident in their false memories compared to
younger adults.

Older adults are more likely to choose someone from a lineup, even
when the culprit is absent.

Own Age Bias (Wright & Stroud, 2002):


Accuracy of identifying someone is increased when the culprit is about as old
as the witness – older adults are more accurate for older culprits.
 Perhaps people focus on features of other people like themselves.
Steps to reduce age biases:


Make sure older adults and children aren’t exposed to misleading information.
Ask the elderly detailed questions to help them weed out source
misattributions.
+
Children’s Testimony

http://www.youtube.com/watch?v=vVh22znRd2Q
+
18
Children as Witnesses
Accuracy


Are traumatic events more memorable than
nontraumatic ones?
 Not terribly (Cordón et al., 2004):
 Both are influenced by age, delay, and
nature of the event
Are children more suggestible than adults?
 Yes (Ceci, Baker, & Bronfrenbrenner,
1988):
 Younger children are more biased than
are older children by leading
questions:
 Questions that carry with them an
implication as to the correct answer
 10- to 12-year-olds are no more
suggestible than adults
The Effect of Leading Questions
+
19
Children as Witnesses
Suggestibility

Thompson, Clarke-Stewart, and Lepore (1997) found that young children’s
responses are largely consistent with the view of their questioner.


The responses of 5 to 6-year-olds to questions about potential abuse when
questioned by:
 Neutral interviewers:
 Are generally accurate
 Accusatory interviewers:
 Are biased in favor of guilt
 Exonerating interviewers:
 Are biased in favor of innocence
Young children continue to reflect the prior influence even when:


Questioned by a new, non-suggestive interviewer.
Warned that the previous interviewer may have been mistaken.
+
20
Children as Witnesses
Suggestibility

Young children are suggestible because of their:



Social compliance
 They yield to authority figures
 They lack social support to stand up for their views
Cognitive incompetence
 They come to believe their distorted reports because of limitations
in:
 Processing
 Attention
 Language abilities
Inability to source monitor
 They often confuse real-life and television events.
+
21
Children as Witnesses
How to Maximize Accuracy

Reduce social compliance
 Avoid leading questions at any point in the questioning process
 Garven, Wood, and Malpass (2000)

Train effective source monitoring techniques
 Thierry and Spence (2002)

Reinstate the encoding context
 According to the encoding specificity principle, memory should be
maximal when the encoding context and the retrieval context match
 Priestley, Roberts, and Pipe (1999)

Use nonverbal recall techniques
 Asking children to draw what they remember before asking for a verbal
report can elicit idiosyncratic retrieval cues and nonverbal information
 Gross and Hayne (1999)
 Children remembered 30% more in the drawing condition, which
only increased (without adding errors) at longer delays
+
22
Eyewitness Confidence

Jurors tend to be influenced by the witness’s apparent confidence.

But confidence is NOT always a good predictor of accuracy
 Sporer et al. (1995) found that the correlation between confidence and
accurate identification is:
 Nonexistent for people who don’t make a positive identification.
 Moderate (+.4) for people who do make a positive identification.
 Confidence does, however, predict general knowledge accuracy.
 Perfect and Hollins (1996) suggest that the difference is due to:
 Having no reference point for the accuracy of eyewitness events.
 Having a good idea of whether their general knowledge is
more/less accurate than others’.
 Witnesses are often coached to be more confident than they are.
 They also receive confirming feedback from police investigators.
+
23
Eyewitness Confidence
Confirmatory Feedback

Bradfield, Wells, and Olson (2002)


Design:
 Participants were asked to identify a man they saw in a video
from a six-person lineup
 Confirming feedback condition:
 Regardless of whether they had picked the right person,
witnesses were given confirmatory feedback:
 “Good, you identified the actual subject”
 Neutral condition:
 No feedback
Results:
 Confirming feedback increased eyewitnesses’ confidence
more when they were incorrect than when they were correct.
 The correlation between confidence and accuracy was
significantly worse in the confirming feedback condition.
+
24
Influence of Anxiety and Violence

Effects of anxiety are hard to assess in the laboratory because
it’s unethical to expose participants to extremely stressful
conditions.

Typical laboratory experiments involve:


Presenting a film or a staged incident, in which a crucial event
occurs, which is either violent/nonviolent, not actual threat.
Common findings:

Memory for the central aspects of an incident are enhanced by
the presence of violence.


This is called the weapon-focus effect
Memory for the peripheral aspects of an incident are reduced by
the presence of violence.
+
Robbery Victims Frequently do not
Remember Appearance Details
Convenience stores with
frequent robberies train
their staff to notice
appearance details and
use aids such as height
strips to make this
easier.
+
26
Influence of Anxiety and Violence
Weapon-Focus Effect

The presence of a weapon causes eyewitnesses to fail to recall
other details.


Probably due to attention being naturally drawn to the weapon at
the expense of other aspects of the situation


Witnesses are less likely to accurately identify a target when a weapon
is involved (Loftus, 1979). 49% vs 33%
Participants spend more time looking at weapons than a nonweapon
substitute (Loftus, Loftus, & Messo, 1987)
In real-life crimes the presence of a weapon:

Did not affect the rate of identifying a lineup suspect (Valentine et al.,
2003), but we don’t know who was actually guilty (accuracy).

Did have an effect according to police reports (Tollestrup et al.,
1994).
+

27
Weapon-Focus Effect
Unexpected
Expected
Low
Threat
Gun pointed
at ground in a
baseball field
Gun pointed at
ground in a
shooting range
High
Threat
Gun pointed
at woman in a
baseball field
Gun pointed at
woman in a
shooting range
Pickel (1999)


Proposed two possible
reasons why a weapon draws
more attention:

It poses a threat

It is unexpected
Fully crossed threat and
expectedness in four videos
he showed to participants.


Tested memory for the
person holding the gun
Found that:

Expected settings
improved the ability to ID

Threat did not influence ID
From Pickel (1999).
+
28
Influence of Anxiety and Stress
 Deffenbacher
et al.’s (2004) meta-analyses revealed
that heightened anxiety and stress:


Negatively impact eyewitness identification accuracy – 54%
for low anxiety vs 42% for high anxiety conditions.
Reduce the ability of eyewitnesses to remember:
 Culprit details
 Crime scene details
 Actions of central characters

64% low anxiety vs 52% high anxiety
+
29
Remembering Faces


Prosopagnosia
 “Face blindness”
 A profound inability to
recognize faces
 The ability to make fine
discriminations among
objects is unimpaired
 Commonly results from
damage to the fusiform
face area of the brain
 This area responds more
to faces than objects in
normal people
Normal people’s face
recognition is poorer than we’d
think:
 Bruce et al. (1999) asked
participants to match the
faces:
Copyright © American Psychological Association.
Reprinted with permission.

Results:

Only 65% accurate when the correct face
was present

35% of participants still picked a face even
when the correct face wasn’t present

Video, in conjunction with the photograph,
did not improve recognition
+
30
Factors Influencing Face Memory
Patterson and Baddeley (1977)

Task:


Two groups were asked to categorize faces based on either:

Physical features (e.g. chin, nose, eyes, type of hair)

Psychological features (e.g. honesty, intelligence, liveliness)
Results:

Participants were better at recognizing faces they earlier categorized on
psychological dimensions rather than by physical features.

Adding a disguise (or removing one from the categorized face) reduced
recognition performance.

Faces seen in three quarter view are more recognizable than faces seen
in profile (side view).
+
Effects of Disguises
With a wig and beard, you
halve your chances of being
recognized as the guilty
party. Faces seen in threequarters view are much
more recognizable than
faces seen in profile. From
Patterson and Baddeley
(1977).
+
32
Holistic Face Processing

Holistic/Global Processing:


Processing the overall
structure of a face/object,
paying little attention to the
details.
Farah (1994) suggested that:

We process faces holistically.

We process objects in a more
detailed fashion.

Explains why Patterson
and Baddeley (1977) found
a benefit for psychological
categorization.

Inverted Faces

It is harder to recognize
the overall structure of a
face when it is inverted,
compared to an inverted
object.

Any difficulty recognizing
inverted objects quickly
disappears with practice.


Not true for inverted
faces.
This accounts for the
Thatcher Illusion
(Thompson, 1980)
+
33
Holistic Face Processing

The Thatcher Illusion
The Thatcher illusion. From Thompson, P. (1980). Margaret Thatcher: A new illusion. Perception, 9, 483–484.
Copyright © Pion Limited. Reproduced with permission..
+
34
Holistic Face Processing

The Thatcher Illusion
The Thatcher illusion. From Thompson, P. (1980). Margaret Thatcher: A new illusion. Perception, 9, 483–484.
Copyright © Pion Limited. Reproduced with permission..
+
35
Unconscious Transference

Unconscious Transference:


The tendency to misidentify a familiar (but innocent) face as
belonging to a culprit
Ross et al. (1994)

Eyewitnesses were three times more likely to select an innocent
bystander from a lineup than a stranger.

The effect was eliminated by informing them before the lineup that
the bystander was not the culprit.
+
36
Verbal Overshadowing

Verbal Overshadowing Effect for Faces:

Describing a previously seen face impairs recognition of that face


Schooler and Engstler-Schooler (1990).
Clare and Lewandowsky (2004)

Providing a verbal report of the culprit makes eyewitnesses more
reluctant to identify anyone in a subsequent lineup.


When forced to pick someone, the effect disappears.
Brief verbal descriptions are more likely to produce the effect than
detailed verbal descriptions.
+
37
Cross-Race Effect

People are more accurate in recognizing same- than crossrace faces
Expertise Hypothesis
Social-Cognitive Hypothesis

We are more experienced
distinguishing among same-race
faces

Thorough processing of faces only
occurs for individuals with whom we
identify (i.e. our ingroup).

Evidence:
 People with more cross-race
experience show smaller crossrace effects


Caveat:
 The effect of expertise is small
and fragile (Hugenberg et al.,
2007)
 Cross-race effect is eliminated
by asking white participants to
closely attend to facial
features distinguishing black
faces from each other.
Evidence:
 Shriver et al.’s (2008) white,
educated participants regarded:
 White faces in wealthy
contexts as ingroup
members
 Both white and black faces
in impoverished contexts as
outgroups
 Black faces in wealthy
contexts as outgroup
members
 Only ingroup faces were well
recognized
+
38
Police Lineups

Police Lineups:

A suspect is present along with
nonsuspects with broadly
similar characteristics


It is essential that the
suspect is not obviously
different from the other
members.
Valentine, Pickering, and
Darling’s (2003) analysis of 314
real lineups:

40% identified the suspect

20% identified a nonsuspect

40% failed to ID anyone

Methods to improve lineups:
 Warn eyewitnesses that the
culprit may not be present in
the lineup:
 Steblay (1997)
 Reduces mistaken IDs by
42%
 Only reduces positive IDs
by 2%
 Present members of the lineup
sequentially instead of
simultaneously
 Results in a more stringent
criterion:
 Reduces mistaken IDs by
about 50%
 Also significantly reduces
positive IDs
+
39
Police Interviews
Inadequate Technique
More Effective
Technique
Result
Close-ended questioning
•“What color was the car?”
Open-ended questioning
•“What can you tell me about
the car?”
Generates more complete
responses without leading
the witness
Interrupting the witness
Allowing the witness time to
finish responding
Doesn’t disrupt
concentration/retrieval cues
Asking questions in a
predetermined order
Ask relevant follow-up
questions
Takes account of previous
answers
+
40
Improving Police Interviews
The Cognitive Interview (Geiselman et al., 1985)
Retrieval Rule
Mental reinstatement
of context
Encouraging complete
reporting (even small
details)
Attempting to describe
the events in several
different orders
Reporting the incident
from different
viewpoints
Empirical Basis
Goal
Encoding Specificity
Principle
Improve the match
between encoding and
retrieval contexts
Memory traces are
complex and contain
various features
Prompting access to
multiple, different cues
improves recall
+
41
Improving Police Interviews
Enhanced Cognitive Interview (Fisher et al., 1987)

In addition to the four rules, investigators should:

Minimize distractions

Induce the subject to speak slowly

Allow for a pause between responses and new questions

Use appropriate language for the witness

Follow up responses with an interpretive comment

Try to reduce eyewitness anxiety

Avoid judgmental and personal comments

Review the eyewitness’s description of events/people under
investigation
+
42
Comparing Interview Methods

The Cognitive Interview:



Generally produces the most accurate information
Caveats:
 Yields slightly more false information than most
people given the standard interview
(Kohnken et al., 1999)
 It is most effective when conducted
immediately after the crime
 It is more valuable for recalling peripheral than
central details (Groeger, 1997)
 It is not yet clear how the individual guidelines
of the interview contribute to its effects
Hypnosis is controversial, as it increases:


People’s suggestibility
The amount of false information reported
Based on data in Geiselman et al. (1985).
+
43
Laboratory vs. Real-Life Settings

In contrast to real-life scenarios, laboratory conditions:

Tend to ask for information from individuals not directly involved in
the crime (i.e. not the victims themselves)

Are less stressful and anxiety-provoking

Provide the eyewitness with only a single, passive perspective

They cannot move around or interact with other participants in the
event

Typically grant witnesses far less time to view the event/people

Carry only minimal consequences for inaccurate information or false
identifications
+
44
The Role of Experts

Leippe’s (1995) review of mock-juror/trial studies


The use of experts in eyewitness testimony during a trial:

Makes jurors more skeptical of eyewitness testimony

Reduces guilty verdicts
Leippe et al. (2004)

The presence of expert testimony produced a sizeable reduction in
guilty verdicts, regardless of the overall strength of the case

Thus, it may make jurors unfairly weight potential pitfalls over
otherwise strong evidence.
+
45
The Case Against Expert Testimony
Ebbesen and Konecni (1997)

Conclusions experts might offer are likely debatable given highly
inconsistent evidence.

e.g. about equal numbers of studies report that high arousal increases
(or decreases) eyewitness accuracy.

Most research on eyewitness memory emphasizes situational
factors, largely ignoring individual differences.

Factors that influence eyewitness memory are interactive;
however, they are usually studied in isolation.

No empirical data convincingly demonstrates that the testimony of
defense experts can actually improve the accuracy of jury
decisions.

In fact, it is often prejudicial.
+
46
The Value of Laboratory Research

Lindsay and Harvie (1988)


Slideshows, videos, and live staged events all produce roughly
equivalent accuracy rates
Ihlebaek et al. (2003)



Memory for both live staged and videotaped robberies is:
 Exaggerated in terms of the duration of the event
 Largely similar in terms of what was remembered/not remembered
Yet, watching a videotaped version of events yields more information
 e.g. better estimates of robber’s age, height, weight, and weapon
used
Conclusions:
 Witnesses to real-life events are more inaccurate than those who
observe the events under laboratory conditions
 Memory distortions/inaccuracies in the laboratory provide an
underestimate of real-life memory deficiencies
 Thus, laboratory research can still be relevant
+
47
Do Jurors Need Expert Advice?

The same factors that are important in the laboratory influence
real-life police cases (Tollestrup et al., 1994)


Mock jurors cannot discriminate between accurate and
inaccurate witnesses (Leippe, 1995)


e.g. exposure duration, weapon focus, and retention interval
Inaccurate witnesses are mistakenly judged to be accurate by 40–
80% of mock jurors
Mock jurors do not adequately moderate their verdicts based
on factors influencing eyewitness accuracy, like lighting
conditions (Lindsay et al., 1986)
+
48
The Case For Expert Testimony

Cutler, Penrod, and Dexter (1989)
 Design:
 Mock jurors viewed realistic videotaped trial of an armed robbery
 A witness made an ID of a subject under good or bad conditions
 Expert advice was either offered or not
Good Conditions
Bad Conditions
Robber not disguised
Robber disguised
Weapon was hidden from view
Weapon was exposed
ID took place 2 days after robbery
ID took place 14 days after robbery
Lineup instructions not suggestive
Suggestive lineup instructions


Results:
 Expert testimony led jurors to better weigh the quality of the ID
conditions
Conclusion:
 Jurors not exposed to eyewitness testimony were largely insensitive
to the quality of witnessing and ID conditions
+
49
The Case For Expert Testimony

Jurors who hear expert testimony make more accurate
decisions than those who have not (Cutler & Penrod, 1995).

Experts only discuss findings from eyewitness research that
are generally agreed to be well established.


They are forbidden to discuss the research in relation to the
specifics of the case, leaving the jurors to decide what is relevant.
While the admission of expert advice isn’t perfect, it does
generally level the playing field and assist in making jurors
reasonably more skeptical of eyewitness testimony.