EYEWITNESS
FACE RECOGNITION AND EYEWITNESS ACCURACY IN APPLIED CONTEXTS
face recognition task -> cognitive processing
• Eyewitnesses
• Variables affecting accuracy
• Changes to police procedure
• Developing areas
• other applied contexts for face perception
• Perception is not as reliable as we might expect: illusions are hard to be perceived accurately
Memory -> School Play, when remembering you remember it as the audience saw you: mis-
reconstruct memory, memory is reconstructed falsely
• Attention is quite limited
• Demonstrate that perception, attention, memory
Systems that we cannot rely on to the extent we would like to
Eye witness processing-> requires accurate and reliable process
•
Filler identification rates in actual cases
10 faces line up
one is the suspect
*covered in black, definitely not the perpetrators of the crime, certainly innocent
Verifiable misidentification: wrong selection
19.9% of 1561 cases
21% of 843 cases
21.6% of 119 cases
THE INNOCENT PROJECT
Organisation's purpose is to exonerate people that were wrongly convicted of crimes
Numerous miscarriages of justice
Victims serving time wrongly due to incorrect eyewitnesses.
Death penalty in US- wrongly executed
Advances in DNA evidence to prove that the person could not have committed the crime
Eyewitnesses ways of responding when picking someone from lineups
a) pick face
b) declare he is not there
Target is n7, if he is present If target is not present in line-up
Target present line-up : Selected target = “hit “he’s not there” = correct rejection
response”; “It’s number 9” = false positive
other target selected = “misidentification”
“He’s not there” = a “miss”
, Accuracy in this task
Target present 70% hits
Target-absent 30% false positives; 70% correct hits
* accuracy not perfect
DIFFICULTIES IN FACIAL PERCEPTION
Face is not a static object: it changes
Jenkins & Burton, 2011
• Is this low accuracy necessarily bad? Orders this represent fairer test conditions
Thing used to be worse
Touch the perpetrator -> very intimidating and unpleasant situation
• Line up itself is limited by the people who are available to make up the line-up (other prisoners +
guards) Unfair line-up when people are similar
Some of the variables
Distinction seen in the literature
Things police can control Cannot control
“System variables” “Estimated variable” cant control them thus the ef-
fects have to be estimated afterwards
- Eyewitness instructions - Attention
- content - Lighting
- Presentation Method - Viewing distance
- Behavioural Influence - Disguise
- Questioning Method - sensory capabilities
- Presence of weapons
- race of perpetrators
- State of eyewitness (stress, intoxication etc…)
Race of perpetrator -immediately evident (e.g. skin colour, ethnicity)
Eyewitness accuracy is affected by instructions
Malpass & Devine, 1981
-> Informed or not that the culprit might or might not be present in the line up
Informed
Providing instruction (Steblay, 1997)
Minimal reduction (1.9%) of accurate identifications
Large reduction (41.6%) of mistaken identifications
Importance of language
* Instruction to eyewitnesses is fundamental; Accuracy is affected by instructions
Content of line-up: Lineup construction
Not fair test (look totally alike or totally different)
System variables: content
• Selection of fillers is important
• Ideal target should not “stand out” but also should not “blend in”
• Fillers can be based on physical similarity to the target
• fillers can be based on verbal description of eyewitnesses (Wells et al., 1994)
• Remains an active issue and unclear what is the best method (Wells & Olson, 2003)
Tredoux (1998; 1999)
FACE RECOGNITION AND EYEWITNESS ACCURACY IN APPLIED CONTEXTS
face recognition task -> cognitive processing
• Eyewitnesses
• Variables affecting accuracy
• Changes to police procedure
• Developing areas
• other applied contexts for face perception
• Perception is not as reliable as we might expect: illusions are hard to be perceived accurately
Memory -> School Play, when remembering you remember it as the audience saw you: mis-
reconstruct memory, memory is reconstructed falsely
• Attention is quite limited
• Demonstrate that perception, attention, memory
Systems that we cannot rely on to the extent we would like to
Eye witness processing-> requires accurate and reliable process
•
Filler identification rates in actual cases
10 faces line up
one is the suspect
*covered in black, definitely not the perpetrators of the crime, certainly innocent
Verifiable misidentification: wrong selection
19.9% of 1561 cases
21% of 843 cases
21.6% of 119 cases
THE INNOCENT PROJECT
Organisation's purpose is to exonerate people that were wrongly convicted of crimes
Numerous miscarriages of justice
Victims serving time wrongly due to incorrect eyewitnesses.
Death penalty in US- wrongly executed
Advances in DNA evidence to prove that the person could not have committed the crime
Eyewitnesses ways of responding when picking someone from lineups
a) pick face
b) declare he is not there
Target is n7, if he is present If target is not present in line-up
Target present line-up : Selected target = “hit “he’s not there” = correct rejection
response”; “It’s number 9” = false positive
other target selected = “misidentification”
“He’s not there” = a “miss”
, Accuracy in this task
Target present 70% hits
Target-absent 30% false positives; 70% correct hits
* accuracy not perfect
DIFFICULTIES IN FACIAL PERCEPTION
Face is not a static object: it changes
Jenkins & Burton, 2011
• Is this low accuracy necessarily bad? Orders this represent fairer test conditions
Thing used to be worse
Touch the perpetrator -> very intimidating and unpleasant situation
• Line up itself is limited by the people who are available to make up the line-up (other prisoners +
guards) Unfair line-up when people are similar
Some of the variables
Distinction seen in the literature
Things police can control Cannot control
“System variables” “Estimated variable” cant control them thus the ef-
fects have to be estimated afterwards
- Eyewitness instructions - Attention
- content - Lighting
- Presentation Method - Viewing distance
- Behavioural Influence - Disguise
- Questioning Method - sensory capabilities
- Presence of weapons
- race of perpetrators
- State of eyewitness (stress, intoxication etc…)
Race of perpetrator -immediately evident (e.g. skin colour, ethnicity)
Eyewitness accuracy is affected by instructions
Malpass & Devine, 1981
-> Informed or not that the culprit might or might not be present in the line up
Informed
Providing instruction (Steblay, 1997)
Minimal reduction (1.9%) of accurate identifications
Large reduction (41.6%) of mistaken identifications
Importance of language
* Instruction to eyewitnesses is fundamental; Accuracy is affected by instructions
Content of line-up: Lineup construction
Not fair test (look totally alike or totally different)
System variables: content
• Selection of fillers is important
• Ideal target should not “stand out” but also should not “blend in”
• Fillers can be based on physical similarity to the target
• fillers can be based on verbal description of eyewitnesses (Wells et al., 1994)
• Remains an active issue and unclear what is the best method (Wells & Olson, 2003)
Tredoux (1998; 1999)
