Task 5 - Polygraph testing

Question 1

Controlled Question Test

• “history”
• assumptions
• administration

Answer 1

• before CQT: control questions were psychologically insignificant to questions of interest
  (Are you from Leipzig?)
• CQT replacing the irrelevant query with a comparison/control (accusatory) question
  > control questions are likely to elicit lies = “probable lie test”
  > probes the integrity of a suspect by asking about past misbehaviours
• is the most commonly used in US
• truthful people will fear/respond more strongly to control question because it is likely that they misbehaved in the past but the did not “misbehave” now (specific crime)
• liar will fear/respond more strongly to relevant questions
• -> greater physiological response to what is feared more
• -> stress/fear response can be measured

1. Pre-test:
   o interview and formulation of questions
   –> goal: convince subject that control questions are as important as relevant questions
2. Asking questions while physiological reactions are recorded (skin conductance, heart rate, blood-pressure, fingertips, respiration)
   o 3 mins, 10 questions (repeated 3 times)
   > Control questions: “Did you ever steal something?”
   > Relevant question: “Did you steal the diamond?”
3. If deemed deceptive: confronting to gain admission/confession

–> is not so much one test and more a collection of procedures which have in common that control and irrelevant questions are asked

Question 2

Controlled Question Test - Criticism

• efficacy
• countermeasures
• accuracy
• theory

Answer 2

• unknown efficacy = conclusive result
  > around 90+% –> BUT only claims or field studies –> n valid is lacking!!
• can be cheated pretty simply by countermeasure (e.g.: biting your tongue)
• accuracy claimed by practitioners = near 100% if correctly used
  > BUT is over-estimated because no knowledge of ground truth + also, circular argument!
• estimates around 60% = better than chance level but not at all good (especially, if you want to hinder miscarriage of justice)
• efficacy depends on the ability of the practitioner to make the suspect take the control questions as serious
• No theoretical basis
• fear of detection is indistinguishable from the fear of false detection
• biased against innocent suspects (problem if you want to prevent miscarriage of justice)
  > if don’t lie in pre-test, machine can’t detect if lying = you are guilty
  –> if key ingredient is examiner (make up the questions) = problem because of bias

=> not (very) accurate, not valid, not standardized, nor objective!!!

Question 3

Controlled Question Test - Why we still use it?

• pros
• other reasons

Answer 3

• more info once you show results to suspects – new disclosure of evidence
• some people with empirically sound approaches find a good validity
• helped occasionally in solving important cases = remembered and used as anecdotal evidence
• risk to the innocent is perceived as minimal = presumption of guilt
• it is confession inducing
• practitioners are blind to criticism (claim they do it correctly)

Question 4

Offe, H., & Offe, S. (2007). The comparison question test: Does it work and if so how?

Aim
Method
Results
Limitations

Answer 4

• examine the suitability of the CQT for determining guilt
• 60 student pps in a mock crime study (innocent vs. guilty
• pps motivated through monetary reward to be perceived as innocent
• half of pps were informed about CQT and its importance in determining behaviour and the other half was not
• effects of comparison questions were again discussed with half of the participants
• then polygraph examination (CQT) was performed one week and five months after the crime
• interpreted by a law enforcement practitioner
• overall high identification rates (90%) for guilty and innocent people
• consistently quite high identification for innocent suspect (important): lowest ca. 80% in no explanation/no discussion condition
• different identification rates for guilty suspects (less important): lowest 35% in no explanation/no
  discussion condition
  => highly successful in identification (win for CQT)
• relatively small sample size

Appendix?

Question 5

Guilty Knowledge Test

• What is it?
• Japan

Answer 5

= series of multiple-choice questions, each having one critical alternative (true feature of crime) and several noncritical alternatives – need to deny each alternative

• if psychophysiological responses to several critical alternatives are consistently larger than to neutral
• -> knowledge is inferred =detects guilty knowledge
• around since 50s in Japan
• used on ca. 5000 criminal cases
• arousal usually measured by respiration (around the chest and abdomen), electrodermal activity (skin resistance response; SRR) and plethysmograph
• has restrictions: not to use with people with heart conditions, mental and/or intellectual handicap, substance abuse disorder, sleep problems
• should be carefully monitored with knowledge is out in the media
• relies on guilty knowledge/memory so things that are more central work better

Question 6

Guilty Knowledge Test

• procedure

Answer 6

1. Explanation of what is going to happen
2. Attach sensors
3. Card test - to practice the procedure of denying
4. Presenting GKT questions –> no psychophysiological recording made yet
   - orienting responses (OR) is reactions evoked by any novel stimulus -> repeated presentations of stimuli, involuntary ORs undergo habituation/response declines
   - suspects (if guilty) identifies critical alternatives and recognise critical alternatives beforehand, thus enhance voluntary orientating response by stimulus significance during the polygraph test
   - if stimulus meaningful greater orientating response
5. Actual GKT
6. “Double Check” if no item had a special meaning
7. Suspect leaves, no post-hoc interrogation to not give away critical knowledge

Question 7

Guilty Knowledge Test

• pros
• cons

Answer 7

+ not as many false positives as CQT (study found 6% GKT vs. 16% CQT)
=> protects innocent suspects (rather 10 guilty free than 1 innocent imprisoned!!!)

• less applied in field, so less data
• memory problems = GKT problems
• salient details of the crime are available in the media (try to counteract that with pre-GKT but still!!)
• less efficient (more timely and more work)

Question 8

Why is the “Guilty Knowledge Test” better than the “Controlled Question Test”?
(Ben Shakhar et al., 2002)

Answer 8

• Daubert criteria better: testability, known error rate, peer review and publication, and general acceptance
  –> GKT fulfils it better
• GKT has discriminant validity (CQT is lacking)
• GKT has more theoretical basis
  > orientating response
  > measures the present of certain knowledge, not
  guilt per se
• GKT can and is done blinded –> better to avoid bias and contamination (CQT is not really possible to do blinded because of the nature of the questions)
• can be less abused and is not biased against innocent –> no knowledge = no response (GKT)
  –> nervous = response –> assumed guilt (CQT)
• countermeasures are more difficult with GKT
  > electrodermal activity (CQT) require 20 s (need to return to baseline) -> enough time for countermeasures
  > EEG P300 signal (GKT) only requires 2 s -> not
  enough time for countermeasures

Question 9

Spence, S. A., Kaylor-Hughes, C. J., Brook, M. L., Lankappa, S. T., & Wilkinson, I. D. (2008). ‘Munchausen’s syndrome by proxy’or a ‘miscarriage of justice’? An initial application of functional neuroimaging to the question of guilt versus innocence.

Aim 
Case study
Method
Results 
Limitations

Answer 9

• explore possibility of functional neuroimaging as lie detection
• woman poisoned her child – she denied it
• served prison sentence but continues professing her innocence
• ‘Munchausen’s syndrome by proxy’ or ‘miscarriage of justice’?
• scanned her while she affirmed her version and the accusation with modified fMRI protocol
• subject was scanned 4 times at 3 Tesla
• significantly longer response times and relatively greater activation of ventrolateral prefrontal and anterior cingulate cortices when she endorsed her accusers’ version
  => neural correlates of making false statements/lying (relating to orienting response)
  => NOT proving her innocence but demonstrating that her behavioural and functional anatomical parameters behave as if she were

Not very conclusive because of many limitations:
- very indirect measure of lying
- if your convinced your right –> cannot be detected
- if so often heard may became automatisms –> cannot be detected
- many countermeasures:
> could move, thus introduce noise
> could deliberately slow response during the putative “lie”
> use alternative cognitive strategy

Question 10

Should polygraph testing be used with sex offenders?
- Experiment

Grubin et al., 2004

Aim
Method
Results
Limitations

Answer 10

• examined whether polygraph testing would result in sex offenders engaging in fewer high-risk behaviours
  (treatment benefit)
• 50 adult male sex offenders taking part in community treatment programs
• assigned to 2 groups:
  > “Polygraph Aware” - told they would receive a polygraph examination in 3 months regarding their high-risk behaviours
  > “Polygraph Unaware” - told their behaviour would be reviewed in 3 months
• baseline interviews (established relevant behaviour)
• both groups were polygraphed at 3 months
• again at 6 months
• at 3 moments offenders could disclose information:
  directly before polygraph, in pre-test, after polygraph
• 1st polygraph: 97% disclosing an average of 2.45 high-risk behaviours (previously unknown to supervisor or probation officer)
• 2nd polygraph: 71 % disclosing an average of 1.57 behaviours
  –> no significance btw. the groups
• mostly disclosed during the pre-test or when having failed the polygraph
• majority failed the first do everybody did a second
  => concluded that polygraph testing resulted in offenders engaging in less high-risk behaviour
• more reporting to supervisors afterwards (but not recoded only mentioned by supervisor)
• no control group with no polygraph (so saying high risk behaviour was lowered is a bit of stretch)
• also the reported behaviours got lower (maybe the suspects got better in counter measures)
• small sample
• selection bias: there was a high drop out rate so it could be that the motivation was the factor which lowered the high-risk behaviour and not the polygraph testing

That is really my least favourite experiment of all time!!!

Question 11

Should polygraph testing be used with sex offenders?
- The statement: NO

Ben-Shakhar, 2008

Answer 11

-> see critique of the CQT

• in case of sex offenders it is even less applicable because the nature of the questions
  > normally: relevant question about known crime
  > sex offenders: relevant questions about possible or hypothetical, future crime (event free)
  –> more vague, less valid
• cognitive distortions of sex offenders may lead to inconclusive results because they are not nervous about these acts
• repeated examination of sex offenders may result in habituation
• discourse are hard to verify (no ground truth)
• can elicit false confession (to explain wrong failing)

Question 12

Should polygraph testing be used with sex offenders?
- The response: YES

Grubin, 2008

Answer 12

• pro post-conviction sex offender testing (PCSOT)
  - criminal record often underestimated
  - could help for treatment
  - could help to decide for/against parole etc.
  
  • -> can facilitate disclosure
• not unethical = voluntary
• bogus pipeline effect is a good thing
• 4 types of tests employed
  > sexual history disclosure
  > maintenance (to see whether the person stays cooperative)
  > denial, or index offence disclosure (to get a better idea of circumstances which led to offenses)
  > specific issue / monitoring exams (monitors single issue)
• different to suspect interrogation
  > not an interrogation but a interview
  > regularly applied –> rewards progress
  > not so impactful (no one is going to jail for failing it)
  > accuracy rates in the region of 80% are sufficient to inform treatment and management
  > CRUCIAL: false positives are not detrimental (different criteria)
  > false confession rates are lower (less than 10%) and do not have such a bad impact (already in jail or under supervision)

Pros:

• Verifying info
• Compliance with parole
• Low rate of reoffending –> caution – can fake
• Large increases in self-disclosing info
• Reductions in high risk behaviour
• Increased reporting of relevant behaviours to supervising probation officers
• Report it to be helpful

Question 13

Should polygraph testing be used with sex offenders?
- Critical Review

Meijer et al., 2008

• research shortcomings
• Bogus Pipeline

Answer 13

=> main take away: the research with supports the beneficial effect of using polygraph test on sex offenders is either unsound or lacking altogether

methodological pitfalls:

• no controls
• sampling bias (only highly motivated)
• no ground truth
• no/little research if it reduces recidivism (which should be the main goal)

Bogus Pipeline Effect:
= expectation of an upcoming polygraph examination is enough to make offenders disclose information
- offenders are required to be completely honest for successful completion of their treatment program, BUT CQT relies on deception toward the examinee