PSC 153 Lecture 4: Lie Detection aka Credibility Assessment

43 views3 pages
Published on 6 Oct 2016
of 3
PSC 153 Lecture 4 Lie Detection (Credibility Assessment)
Lie Detection
Long been part of the legal system
Trial by ordeal in the Middle-Ages holding hot iron, if wound heals, then not guilty
Acquittal rate 50% of those suspected of crime
Bocca della Verita in Rome
Now more sophisticated techniques to detect lies
Truth about Lying
It is extremely common
Lies of commission saying something not true
Lies of omission not saying something true
Lying is psychologically sophisticated
It requires counterfactual thought
It is evolutionarily adaptive
Enables cheating for survival
The legal system puts a lot of faith in jurors and interrogators ability to detect lies
Additional factors (time to rehearse lie, observing same person multiple times) make it
easier/harder but only by 2%
Research shows professionals are actually worse than lay people in detecting lies but
were more confident about their judgements
Training people to detect lies only increases confidence
70% of people hold liar stereotypes but behavioral cues are not reliable
A dangerous combination of factors:
Presumption of innocence in the legal system innocent until proven guilty
Confirmatory bias tendency to seek out info that confirm ones expectations and
dismiss info that contradicts it particularly ambiguous information such as behavioral
cues for lying
Overconfidence linked to lie detection training
What does innocent/guilty person look like? Either too unemotional or too overly
Presumption of Guilt expectation makes a big difference, interrogators really try to get
person to plead guilty
Interrogators with guilt expectations more aggressive when interrogating particularly
if suspect is actually innocent
find more resources at
find more resources at
Suspect interrogated by those with guilt expectations appeared more defensive, were
more likely perceived as guilty (self-fulfilling prophecy)
The Polygraph
Machine that records a subjects physiological responses to questions asked by an
Developed by William Marsten in 1917 and he also created Wonder Woman and her
Lasso of Truth
It includes sensors for heart rate, blood pressure, and galvanic skin response
Based on theory that lying causes physiological arousal, changes in heart rate and
breathing, increases in BP, and skin moisture
Measures physiological arousal, not lying per se
Similar to fMRI measures blood oxygenation, not necessarily brain activation
Who uses polygraphs?
Employers private ones until 1980s, now only public ones (NSA, FBI, CIA)
Law enforcement banned from court in 23 states
Admissible if judge decides it meets Daubert Standard (relevant/reliable)
Admissible if both sides agree pre-trial
Always admissible in New Mexico
Polygraph Procedures
To be used as a lie detector, the polygraph must be combined with systematic questioning
Relevant-Irrelevant Test (RIT) Compare responses to irrelevant vs relevant questions
If response to relevant is greater than irrelevant, then person is guilty
High rate of false alarms
Comparison Question Test (CQT) comparison questions aka known lie questions,
anyone who answers no is assumed to be lying
Relies heavily on examiners theatrical skills
Problems with Polygraph
Guilty suspects might be controlled or nonreactive and might use countermeasures to
mask relative differences in arousal (bite tongue, press toes against floor)
Innocent suspects might have strong reactions to relevant questions even if they are not
Lack of standardization subjectivity of rating system, content/numbers of questions,
appearance/demeanor of examiners
find more resources at
find more resources at
Accuracy of Polygraph
Review of 109 tightly-controlled studies using CQT in lab simulations
Results potentially inflated
Artificially low stakes = low motivation for lying convincingly
Real perpetrator of crimes might have experience or training in countermeasures
Ability to detect simple lies under controlled circumstances might overestimate results
Statistically speaking NOT BAD correctly identified 74% of guilty, and exculpated
68% of innocent
Practically speaking NOT SO GOOD incorrectly exculpated 21% of guilty, and
incorrectly accused 16% of innocent
Alternatives to Polygraph
Guilty Knowledge Test (GKT) detect whether someone knows facts only the
perpetrator knows
If guilty recognizes scenes/events from crime
If innocent fail to recognize
Recognition should cause physiological arousal
Series of questions about details of the crime with many possible answers
If innocent equal reaction to all questions
If guilty strong constant reaction to correct choice
Promising results from lab simulation
Limitations lots of crime details must be available, crime details must not be
publicized, perpetrator must remember crime details, innocent person might have
discovered the crime but not committed it, only 13-18% of FBI cases using CQT would
accommodate GKT, resistance of professional polygraphers
Cognitive approach to lie detection
Lying = cognitively demanding as one must inhibit truth and make up a story
Limiting availability of cognitive resources should make lying harder (ask to tell story
backwards, ask to maintain eye contact, ask to draw picture of events)
find more resources at
find more resources at