Eyewitness Memory
natalia.na
EYEWITNESSMEMORYPOWERPOINT.pptxOpen-ended mugshot searches
The more photos are shown, the more likely it is that someone will resemble the perpetrator
Selecting someone may reinforce memory for later recognition
Cross-race effect may increase the likelihood of faulty recognition
“Dynamic mugshots” possibly more accurate
1
2
Issues with the Bundy identification
Witness was in a high stress situation (preceded by a moderate stress for about 10 minutes)
Yerkes-Dodson law: low & high arousal/stress lead to poorer memory than moderate arousal
Attentional tunneling: high stress narrows focus on central details (e.g. an unexcepted weapon)
3
Issues with the Bundy identification
Victims/survivors tend to have poorer memory than bystanders
Memory deteriorates with time; facial memory lasts a long time for familiar faces but not necessarily for strangers’ faces
Unconscious transference/source confusion: showing two different photos of the same person can become a false memory of the person
The photos were followed by a photo-biased lineup
4
Eyewitness memory
Eyewitness testimony is the leading contributing factor to false convictions (Source: Innocence Project)
Eyewitness memory can be pretty good with proper procedures, but the legal system may rely on contaminated eyewitness evidence
5
Misinformation effect
Exposure to misleading information can change existing memories
Press reports
Conversations with other people (e.g. other witnesses)
Investigators asking suggestive questions about the event
6
Loftus & Palmer (1974):
Subjects see a video of a car accident
Subjects are asked: How fast were the cars going when they smashed/collided/bumped/hit/contacted each other?
| Verb | Estimated Speed |
| smashed | 41 |
| collided | 39 |
| bumped | 38 |
| hit | 34 |
| contacted | 32 |
Misinformation effect
7
One week later, subjects were asked:
Did you see any broken glass?
Stronger when misleading info conforms to expectations
| Yes | No | |
| "smashed" | 32% | 68% |
| "hit" | 14% | 86% |
Misinformation effect
8
How should a lineup be conducted?
Double-blind with no reinforcement (Garrioch & Brimacombe, 2001: Witnesses most likely to choose who the administrator thinks is the suspect)
Not a forced choice (target-absent lineup)
Sequential lineups better than simultaneous (absolute vs. relative judgment)
Suspect should not stand out; fillers must fit the description (but not be “clones”)
An option: a blank lineup first (fillers only)
9
The cross-race effect/own-race bias
Better recognition of members of own race/ethnicity than others (estimate: 48% of wrongful convictions!)
Social contact leads to processing similar faces holistically?
Prejudice leads to superficial processing of members of “outgroups” as more homogeneous than “ingroup” members?
Encode “race” as a feature for outgroups which requires memory resources (own race is “default”)?
Shown even in 9-month-old infants
Instructions might decrease the bias
10
Additional findings about lineups
Own-age bias (weaker than own-race bias)
Own-gender bias (maybe stronger for women)
Children and the elderly less reliable as witnesses; more feature-based before age 8-10
Mixed evidence for attractive faces: harder if attraction based on “prototypicality”
Poor illumination, long distance (50+ feet) and small changes (wearing a hat) impair recognition
Show-ups are worse: suggestive, no safeguard for innocent person
Fast recognition is more accurate
11
How to get a reliable eyewitness testimony?
The initial memory test provides the least contaminated information
Witness must report their confidence level; high confidence after the initial identification may correlate with accuracy, but not for later identifications
Open-ended questions
Cognitive interview that mentally recreates the context
12
Encoding specificity (e.g. context-dependent learning)
Deep sea divers learned word lists on land or under water (Godden & Baddeley, 1975)
Later the divers had to recall these words either on land or under water
13
Cognitive interview: methods to reduce the influence of misinformation and expectations
Mentally reinstate the context of the original event
Free recall and in-depth reporting: investigators should not interrupt the sequence of recall for even “unimportant” details
Report events from different perspectives: memory retrieval improved by using multiple retrieval pathways
Discourage guessing (“I don’t know” is better)
14
Polygraph: the control question test (CQT)
Changes in heart rate, blood pressure, respiration and sweating (skin conductance)
Relevant, control and irrelevant yes/no questions
Relevant questions relate to the crime, and control questions are upsetting & similar to the crime but not directly related to the crime (irrelevant questions not measured)
15
Polygraph: the control question test (CQT)
Relevant: “Did you rob a bank last Friday?” “Did you forge your mother’s signature?”
Control: “Have you ever taken something that does not belong to you?” ”Have you ever betrayed anyone?”
Greater response to relevant questions: Fails the test (“guilty”)
Greater response to control questions: Passes the test (“innocent”)
Same response to both: Inconclusive
16
P300 brain response
P300 brain wave is elicited when subjects recognize a meaningful stimulus (significant/relevant) in a sequence of stimuli
Probes: items guilty person is suspected of knowing (e.g. murder weapon)
17
Guilty Knowledge Test (GKT)
Detects an “orienting response” (electrodermal or neural) to information that only a guilty person is expected to know
Questions with about 5-6 answer options shown one at a time to isolate the response to each; only one option is true
18
Guilty Knowledge Test (GKT)
How did the burglar get into the house?
Through the:
Front door
Basement window
Back door
Chimney
Bedroom window
Kitchen window
19
Guilty Knowledge Test (GKT)
The victim was killed with…
A knife
A gun
A hammer
An axe
A rock
Poison
20
Challenges of the control question test
Detects intensity of emotion, not lying
Based on the idea that deception causes anxiety (but this is not necessarily so)
Subjects can “beat” the test (controlled breathing, muscle tensing, tongue biting, complicated mental processes during the control questions, etc.)
Answering “control questions” honestly may decrease anxiety to them (look guilty)
Innocent people may also find relevant questions more threatening (e.g. due to fear of being suspected)
21
Challenges of the GKT
Need multiple items that are only known by the guilty person and the investigator
The critical items must not be significant in other ways to the person being tested
All the items must be “equal”; none can stand out to innocent people
The critical items must be memorable to the guilty person
“Memories” detected may be false/subject to the misinformation effect
22
Poor GKT questions:
What color was the door of the house?
How many chairs were in the living room?
What was the name of the owners shown at the front door?
What items were on the kitchen table?
23
Facial recognition
24
Facial recognition
25
Facial expression recognition
26
Facial expression recognition
27
Facial expression recognition (McKone et al., 2013)
28
Facial expression recognition (McKone et al., 2013)
29
Face recognition in the brain
Harder to recognize identity and local feature changes or distortions with inverted faces
Face recognition depends more on configurational/holistic perception than recognition of features
Fusiform face area (FFA) selectively activated by upright faces
30
(FFA)
Fusiform face area (FFA)
31
Traditional ways of testing memory for faces
Verbally describe the face to a forensic artist who sketches it
Facial composite is constructed by selecting individual facial features (feature-based composite system)
The sketch or composite is used for recognition
32
How should facial memory be tested?
Present features in the context of other features on a complete face with a holistic face program (Image: EvoFIT)
Select a small set of best matches from computer-generated complete faces
Recombine them to a new set, repeat till the best match is found
33
How should facial memory be tested?
Verbally describing a face may impair recognition (verbal overshadowing): often involves describing individual features
Make holistic judgments about personality, attractiveness, intelligence, emotions, aggression, “masculinity/femininity”, health
Mentally reinstate the context in which the face was originally seen
34
Super-recognizers
35
Super-recognizers
1-2% of people with superior ability to recognize and remember faces
No correlation with IQ scores
Differential eye fixation patterns when viewing faces
Focus on the center of face (regular facial recognizers focus more on the upper face, prosopagnosics on lower face & avoid the eyes)
Different from having a superior memory in general (only super-recognizers are good at the Before They Were Famous test)
36
Super-recognizers
Before They Were Famous Test
37
Super-recognizers
Before They Were Famous Test
38
Super-recognizers
Before They Were Famous Test
39
Super-recognizers
Before They Were Famous Test
40
