Upload
aishwary-sharma
View
224
Download
0
Embed Size (px)
Citation preview
8/3/2019 12.10.07 - Deception Detection
1/37
8/3/2019 12.10.07 - Deception Detection
2/37
Centrality of deception
Interviews Importance of coming off as innocent
Miranda waivers Self-presentation of innocence, dont want to look
deceptive
Interrogation
Failure to convey innocence Presentation of false evidence
Maximization and minimization
8/3/2019 12.10.07 - Deception Detection
3/37
Deception Detection How doYOU stack up?
8/3/2019 12.10.07 - Deception Detection
4/37
8/3/2019 12.10.07 - Deception Detection
5/37
Might professionals do better?
Kraut and Poe (1980) - customs officials no more accurate thancollege students in detecting deceit in mock customsexaminations.
Kohnken (1987) - police officers no better than chance whenjudging videotapes of college students who had lied or been
truthful
Underlying question: Were there really any information on thevideotapes to suggest differences? We presume there are actual differences between groups, but no
one has independently verified and quantified them.
If we KNEW there were differences, might professionalsoutperform students?
8/3/2019 12.10.07 - Deception Detection
6/37
Who can catch a liar?
Ekman & OSullivan (1991) Examine ability to detect deception in statements
about peoples feelings
part of talks Ekman gave to various groups Ps include:
Students and Extension Class (general public) Forensic Services unit of Secret Service Polygraph experts from FBI, CIA, NSA, Marines, Army,
Air Force, and other federal agencies Judges and Police Psychiatrists
8/3/2019 12.10.07 - Deception Detection
7/37
Ekman & OSullivan
Judge ten 1-minute videos Told that about half showed lies
Lie re: feelings about a movie subject was
watching Happy feelings during nature/horror film
Independently-verified differences between
liars and truth tellers! Facial muscle movements (fake/real smiles) Voice (pitch)
86% accuracy when using both
8/3/2019 12.10.07 - Deception Detection
8/37
Ekman and OSullivan Ps
Notice, were not talking about spring chickens,nor inexperienced lie detectors!
8/3/2019 12.10.07 - Deception Detection
9/37
Ekman & OSullivan
Questions How well do groups detect deception?
Do certain individuals stand out, obscured by mean?
How does confidence factor in?
What cues do people use? Related to accuracy?
8/3/2019 12.10.07 - Deception Detection
10/37
Ekman & OSullivan (1991)
Results Accuracy
Supports past work: people tend to perform poorly Only the Secret Service performed significantly above
chance!
8/3/2019 12.10.07 - Deception Detection
11/37
Ekman & OSullivan,breakdown Accuracy, continued
29% of Secret Service went at least 8/10! Psychiatrists next group of high scorers: 12% went at least
8/10
8/3/2019 12.10.07 - Deception Detection
12/37
Ekman and OSullivan
Results Demographic predictors?
Gender didnt matter
Younger were better in S.S. and polygraph pros
All who went 8/10 were under 40 years old
Were people appropriately confident? Asked how confident they were
in general (before task) About their performance (after task)
Neither correlated with performance (r = .03 and .02)
8/3/2019 12.10.07 - Deception Detection
13/37
Ekman & OSullivan (1991)
Results: Confidence Before and after the task
Secret Service, law enforcement personnel, and
polygraphers are significantly more confident than are
judges and psychiatrists. Why? Secret Service may have reason to be confident
Experience? Training?
Does police performance on task merit the
confidence? 55.79% accurate, on average!
Implications for interviews? Innocents in interrogation?
8/3/2019 12.10.07 - Deception Detection
14/37
Ekman & OSullivan
What tactics are associated with accuracy? Nonverbal behavior
People who were more successful listed behavioral cuesas basis of their decision (after the fact) But overall, people who said beforehandthat they were
going to use behavioral cues werent better than those whosaid they would use verbal cues
Microexpressions For extension class participants ONLY Tested for ability to recognize emotions conveyed through
microexpressions Weak correlation between accuracy and microexpression
r= .27; 7% of variance
8/3/2019 12.10.07 - Deception Detection
15/37
Ekman & OSullivan (1991) Issues
What constitutes good performance? Without trying, you can hit 50% by answering in one
direction for all items So is Secret Services 64% reallythat great?
Role of age/experience Age seems to be a hindrance. Why?
Ekman & OSullivan suggest less frequent interrogation
Slipping in old age? Havent learned newest techniques? Negative response bias tendency to see guilt
Remember Kassin & Fong, 1999, from last class
8/3/2019 12.10.07 - Deception Detection
16/37
Ekman & OSullivan (1991) Issues What are they even judging?
Positive feelings while watching nature film or gruesome
depiction
External validity? Do we want to make claims about how well lawenforcement performs based on tasks only marginally
related to their actual job?
Internal validity? Are the types of differences seen between liars and the
truthful in this experiment the same that show up in
criminal justice settings?
8/3/2019 12.10.07 - Deception Detection
17/37
A better deception detectionmousetrap
Mann, Vrij, & Bull (2004) 99 British police officers
Clips ofREAL interrogation footage Veracity of each statement was independently verified Watch a tape with several clips from different interviews
Experience operationalized as interrogation
experience, not age/years on force
8/3/2019 12.10.07 - Deception Detection
18/37
Mann et al. (2004) Questions
How accurate are police at detecting lies?
Does experience interviewing increase accuracy?
Do good and poor lie detectors rely differentially on speech-related cues?
Do Inbau-recommended behavioral cues limit accuracy?
Gaze aversion, unnatural posture changes, hand over mouthor eyes when speaking
Accuracy related to confidence?
8/3/2019 12.10.07 - Deception Detection
19/37
Mann et al. (2001) Results
Lie and Truth detection Lie accuracy: 66.16%
Truth accuracy: 63.61% Both significantly better than chance
Better than all but S.S. in Ekman & OSullivan (1991)
Interrogation experience and accuracy Lie accuracy r=.18, p
8/3/2019 12.10.07 - Deception Detection
20/37
8/3/2019 12.10.07 - Deception Detection
21/37
Mann et al. (2001) Results
What do good v. poor judges rely on? Good
Illustrators (hand and arm gestures w/ story) Aspects of story
Poor Suspect gender (male seen as more guilty) More gaze aversion More head nods
Accuracy and confidence Not significantly related in truth or lie detection
8/3/2019 12.10.07 - Deception Detection
22/37
Ekman (1991) v. Mann (2004)
Why the differences in accuracy acrossstudies?
Different materials Higher stakes, more motivation
Maybe just easier to read these particular people No student comparison group in Mann, so we cant tell
British police officers cannot lie to suspects Need to be better at other abilities
8/3/2019 12.10.07 - Deception Detection
23/37
Why so poor? If chance is 50%, why do the most favorable
studies report only 60-65% accuracy amongpolice officers?
Reasons There is no ONE sign that someone is lying Differences result from withholding emotion, cognitive
load, and/or attempted control Can all work simultaneously
Can compete: eye blinking Nervousness increases blinking Cognitive load decreases blinking
Vrij (2004)
8/3/2019 12.10.07 - Deception Detection
24/37
Why so poor?
Reasons Heuristics Probing heuristic
If someone is questioned and isnt labeled as deceitful, they most
likely never will be thought of that way
Importance of the interview!!! Representativeness heuristic
Some behaviors are a stronger signal than others
Representative not necessarily accurate
Fidgeting erroneously believed to signal deception
FAE Over-reliance on dispositional attribution
Guilty before, guilty again bad apple theory
Come off as innocent, youre a good guy
Vrij (2004)
8/3/2019 12.10.07 - Deception Detection
25/37
Why so poor?
Reasons Failure to account for Individual Differences
Some peoples natural demeanor just makes them look
guilty/deceptive
Expressive people presumed innocent illustrators
Introverts
Lie less frequently, commit fewer crimes
More likely to look suspicious since they gesticulate less
Vrij (2004)
8/3/2019 12.10.07 - Deception Detection
26/37
Why so poor?
Individual differences and culture African-Americans display greater gaze aversion
than do those of European descent Same pattern among Turkish and Moroccan Dutch v.
White Dutch Power issues - SDO Eye contact polite in Western cultures, but extremely rude
in others, especially with someone of authority
Speech and behavior
South American immigrants to Netherlands make morespeech disturbances (ah speech) and self-manipulations (nonessential movement) Natural behavior is what White Dutch see as deceptive
Vrij (2004)
8/3/2019 12.10.07 - Deception Detection
27/37
Why so poor?
Feedback hypothesis (i.e. Training) Its not enough to simply make deception
judgments for a living
You need timely and accurate feedback about yourdecisions
8/3/2019 12.10.07 - Deception Detection
28/37
So who might do well?
How about criminals?
8/3/2019 12.10.07 - Deception Detection
29/37
Criminals and deceptiondetection
Environmental feedback Criminals may live in more deceptive environment
May need to be hyper-vigilant to keep from being
deceived Abused children living in institutional environment
significantly better at detecting lies (Bugental et al., 2001)
Recidivism
People tend to commit multiple crimes, get feedback onhow to be deceptive successfully Inmates can rapidly produce convincing false confessions
(Norwick, et al., 2002)
8/3/2019 12.10.07 - Deception Detection
30/37
Criminals and deceptiondetection Most lay people believe that deception is associated
with: Greater movement Less eye contact Less consistency in story, etc.
Criminals beliefs about the following were more inline with research than were students and prisonguards: Excessive consistency in story
Fishy if you make a lie and stick to it! Length of pauses in speech
Lying takes resources, need time to come up with something
Granhag, Andersson, Stromwall, & Hartwig, 2004
8/3/2019 12.10.07 - Deception Detection
31/37
Criminals and deceptiondetection
So do criminals actually make better liedetectors? (Hartwig et al., 2004) 52 Swedish prison inmates, 52 Swedish university
students
Evaluate video of eyewitness to a staged robbery Witness interviewed 3 hr, 4 d, and 11 d after attack
Told to tell what happened (or to lie and say the victim
initiated the knife attack and hurt himself in the process)
Decide whether witness was truthful and rateconfidence from 50% (guess) to 100% (certain) Provide cues used, if so inclined
8/3/2019 12.10.07 - Deception Detection
32/37
Hartwig et al (2004)
Types of cues provided Verbal
Witness confidence
Consistency across interviews
Degree of detail in account
Plausibility
Story sound rehearsed?
Nonverbal Body/trunk movements
Gaze Nervousness
8/3/2019 12.10.07 - Deception Detection
33/37
Hartwig et al (2004) Accuracy
Overall Criminals:65.4% - sig better than chance
Students: 57.7% - chance
If statement is Truth Criminals: 42.3%
Students: 50%
If statement is Lie
Criminals: 88.5% - sig better than chance Students: 65.4%
8/3/2019 12.10.07 - Deception Detection
34/37
8/3/2019 12.10.07 - Deception Detection
35/37
Cues to Deception What cues did criminals and students say they
used? Criminals relied primarily on plausibility
34.6% of truth judgments, 28.8% of lie judgments Students use it 9.3% & 15.7%, respectively
34.9% of all correct judgments used plausibility Used significantly more often in correct judgment than in incorrect
judgment
Students rely primarily on consistency 24.1% of truth judgments, 30% of lie judgments
Criminals use it 0% and 5%, respectively
25.4% of all correct judgments used consistency Not used significantly more for correct judgments, always cited
8/3/2019 12.10.07 - Deception Detection
36/37
Criminal detection deception
Review Criminals may be environmentally-trained to detect
plausible stories to avoid neg. consequences of
being deceived Criminals beliefs about the role of consistency
more in line with research findings than are
students and guards
Criminals tend to be more accurate Greater reliance on plausibility, less reliance on
consistency in story across time
8/3/2019 12.10.07 - Deception Detection
37/37