KHA312 Lecture Notes - Lecture 4: Cognitive Interview, Ecological Validity, National Institute Of Justice
Psychology and Law week 4: Eyewitness Recall and Interviewing
Reliability of Eyewitness testimony and recall: interview
- Memory:
oStored information
oOutput on memory test
Ability to retrieval and volunteer information
oDifferent ideas have different implications
Neither is better than the other
Different implication of interviewing practice
Different expectations
oActive vs. passive warehouse
- Initial considerations:
oThe actual reliability of eyewitness testimony / recall
This is difficult
Hard to define our keys terms
Need to give some thought to our definitions
oDefining accuracy
Accuracy vs. quantity
Differences between telling the whole truth vs. nothing but the
truth
More interested in the reliability of what people do offer
Not recalling lots of information that is inaccurate
Which of these measures are we interested in in
different settings
oAccuracy across items doesn’t really make sense:
Need to distinguish between:
Global accuracy vs. individual item accuracy
A single error doesn’t tell us anything useful about a witness’
overall reliability
Does mean they are unreliable
Doesn’t mean their other details are unreliable
Major general issues: strategic regulation of memory outputs
- two key processes at play:
oMonitoring
oControl processes: what we report and what level of detail we provide
oThe model: developed for semantic memory but can be used for episodic,
eyewitness memory
We have input question memory search best candidate
answer
This is monitored: thinking about whether this is a good answer
Will we propose best answer
Need a threshold level of confidence in order to propose
our best candidate answer
Caries depending on:
Situational demands and pay off:
oReport option instructions
find more resources at oneclass.com
find more resources at oneclass.com
oAccuracy incentives
oBeliefs about memory efficacy
oIf we doubt quality of our memory; might with-
hold more
Reponses criterion
If probability accuracy (confidence) outweighs response
criterion then we offer the answer
oIf not, we don’t
oWe may then go back and try again: initiate another
retrieval
oOr, reduce level of specific detail, make the answer
more vague
- How do witnesses know if they are correct?
oMonitoring process
oMetacognitive processes: most rely on heuristics
Reliance on heuristics that are often, but not always, reliable
Learned through experience
Most of the time give accurate
But can be vulnerable to distortion
Eg. Amount and strength of information retrieved that
supports the answer
oConsistent other information, or vivid highly
detailed memories
oInfer that that memory is strong and reliable
oPatchy memories = memory is weak
Eg. Fluency: the easier we find things the more likely we
think it is reliable and accurate and reflect a strong memory
oEast with which information comes to mind
oControl processes:
oRegulating report option:
Volunteering vs. withholding information
Regulating grain size: or precision
Detailed vs. vague responses
In formativeness vs. accuracy threshold
There are requirements for both
Maximize the reliability
Strategic regulation: balances repeated demands for in
formativeness and accuracy
Tradeoff is very robust
The more information you provide the more likely you are
to make an error: need to strike happy medium
- What determines what witnesses report?
oInfluenced by:
Level of confidence in information we have
By the constraints in placed at test by interviewer
oWitnesses can choose their report option: respond or withhold
- What determines the level of detail in reporting?
find more resources at oneclass.com
find more resources at oneclass.com
oRegulate the response of detail
oGrain size = coarseness of information provided
oIf memory is vague (not confident in this level of detail) we can be less
specific
oPeople actively regulate grain size of reporting
Role of confidence
oGrain size also influenced by interviewer
oIn lab and real world, recall accuracy shaped by grain size
The more detailed of a response required, the more likely it is it
will contain an error
In the lab, when we score witness recall report, the accuracy score
will depend on the level od detail required to code an information
as accurate or vague
Only starting to consider this properly
Only a small number of studies looking at this
If we don’t consider this, we can miss important effects of
memory quality of events
oEg. Proportions of information accurately reported will depend on grain
size required by scoring protocols
Variables associated with testimonial accuracy: some influences on testimonial
accuracy
- need to consider how these factors affect regulation of report option and grains
size
ospecificity of detail might decrease accuracy
obut less accuracy over time might not reduce accuracy
- Interview format:
oCan take a number of different shapes
oClosed questions:
Series of pointed, directed questions
oOpen ended formats:
Tell me everything you can remember
oTraditionally it was focused on closed questions
Now focused on open, free recall
Granted witness can create on cues
oNumber of interviews the witness does:
Over time, filtration process occurs
Final story doesn’t match story of the first interview
Over time: freezing
Key details reinforced by interviewer stay in the story
Other get pruned off
Details not included early on, don’t get included later
oDelay to first interview and intervals between interviews:
Quality they recall is influences: how much and detail
Not necessarily the accuracy of the information
oReport option and grain size may vary across interviews:
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Psychology and law week 4: eyewitness recall and interviewing. Memory: stored information, output on memory test. Ability to retrieval and volunteer information: different ideas have different implications. Different expectations: active vs. passive warehouse. Initial considerations: the actual reliability of eyewitness testimony / recall. Need to give some thought to our definitions: defining accuracy. Differences between telling the whole truth vs. nothing but the truth. More interested in the reliability of what people do offer. Not recalling lots of information that is inaccurate. Which of these measures are we interested in in: accuracy across items doesn"t really make sense: different settings. A single error doesn"t tell us anything useful about a witness" overall reliability. Doesn"t mean their other details are unreliable. We have input question memory search best candidate answer. This is monitored: thinking about whether this is a good answer. Need a threshold level of confidence in order to propose our best candidate answer.
