Raymond Nelson 11/22/04. Overview. National Academy of Sciences (2003) Polygraph Instruments. Introduction to the Polygraph

Transcription

1 Polygraph Science and Field Practice With Adult and Juvenile Sex Offenders a training for referring agents and consumers of the polygraph test Pyschophysiological Detection of Deception POLYGRAPH Blind Confidence or Responsible Science Provided for the Colorado Sex Offender Management Board December 7, 2004 Raymond Nelson, MA, NCC "I don't know anything about polygraphs, and I don't know how accurate they are, but I know they'll scare the hell out of people." - president Richard M. Nixon National Academy of Sciences (2003)... we conclude that in populations of examinees such as those represented in the polygraph literature... specific-incident polygraph tests for event-specific investigations can discriminate lying from truthtelling at rates well above chance, though well below perfection. (National Academy of Sciences, 2003) [p.214] Overview Intro to the Polygraph PCSOT Test Theory Basic Science Predictive Statistics Polygraph Accuracy Physiology The Brain Psychology Countermeasures Suitability for Polygraph Testing Professional Ethics Quality Assurance PCSOT Field Practices Introduction to the Polygraph Polygraph Instruments Computerized Polygraph Most common Computerized scoring for quality assurance Electronic Analog Polygraph Rarely used due to convenience and confidence of computers (good theatrics on TV and movies) Still considered highly accurate Analog Polygraph - mechanical No longer used Lack of electronic amplifiers = higher cuff pressures (increased discomfort to examinee) 1

2 Polygraph Components Pneumograph assembly Respiration patterns and changes Will also reflect the examinee's attempt to selfregulate reactions Respiration is autonomic when ignored Respiration is peripheral under attention Blood pressure cuff Changes in blood pressure (cardiovascular activity) Electrodermal sensors Skin conductance (or resistance) What the polygraph does Monitors and records sympathetic autonomic nervous system (ANS) changes that are empirically correlated with deception Allows for the measurement and evaluation of reactions through cut-scores or probability statistics Misinformation Contrary to popular belief, the polygraph does NOT measure any of the following Deep Breathing Sweaty Palms Rapid Heartbeat Nervousness Anxiety Emotional Reactivity Subconscious attitudes (or beliefs) Forgotten memories (an oxymoron) What About PNS Activity? Examinees are instructed to sit still during the test Non-cooperation is regarded as an indicator of deception Unique tracing signatures (patterns) will alert the trained examiner to PNS activity Movement sensors confirm that ANS activity is adultered with discrete PNS activity intended to alter or improve the test result PNS activity is correlated with the concealment of information Attempts to learn to defeat the polygraph test are correlated with deception Examination Methods Relevant Irrelevant Not used in PCSOT Primary use police and security applicant screening Probable Lie Comparison Question Test Most common method for criminal investigation Required method for PCSOT Directed Lie Comparison Question Test Used security clearance and monitoring of government employees and scientists Concealed Information Test (GTK, POT) Requires a known event about which the subject would possess some unique response potential Used as adjunct test in PCSOT Computerized Scoring Systems Developed for single-issue tests only Regarded as suitable for Quality Assurance purposes Examiner is responsible for the final decision (result) JHU-APL Polyscore - Atheoretical algorithm developed from 10,000 initial features from 1400 confirmed cases Axciton Chart Analysis - Proprietary Rank order analysis normed on an unknown number cases Identifi - Independently developed - Mirrors hand scoring rules Stoelting CPS - Kircher features (Kircher and Raskin, 1988) - Mirrors hand scoring rules 2

3 Polygraph Test Procedures Pretest Phase longest part In-test Phase actual testing Post-test Phase resolve any remaining inconsistencies Pretest Phase Obtain authorization and release (voluntary testing, dissemination of all information) Review health and medications (suitability) Review personal/background information Review case background and reason for investigation Review polygraph instrumentation and physiological basis Formulate test questions and answers In-test Phase Attach polygraph components Conduct a practice/sensitivity test Run the test 3 to 5 times Review charts for initial results In Test - Questions Test will include approximately 8 to 12 questions Examinee must pass all test questions Two to four target issues for comparison Personally relevant questions regarding honesty and integrity Known truth questions to gage truthful response baseline Other questions to conduct the test and condition the examinee Post-test Phase Examiner will provide the examinee an opportunity to explain any reactions or remaining inconsistencies Numerical scores and results are calculated Report is written and disseminated to referring professionals Examiner will not discuss results with examinee (or family) after the test is completed Polygraph Science 1. Measurable changes in autonomic functions occurs when an examinee responds to test stimuli (questions) in a deceptive manner 2. Reactions to test stimuli (questions) can be normed for distinct distributions of truthful and deceptive examinees 3. Individual examinee's test data can be evaluated for concurrence with decision thresholds or probability statistics for assignment to a deceptive or truthful group 3

4 Polygraph Theory 1. Deceptive test results will be confirmed through admissions or collateral information 2. Questioning examinees will not result in false confessions 3. Retesting of unresolved issues will not be impaired by prior testing experiences Polygraph Field Practice 1. Requirements for polygraph testing will yield a greater volume of information that is useful to the challenges of risk assessment and risk management 2. Requirements for polygraph testing will serve to deter some problem behavior from occurring 3. Requirements for polygraph testing will serve to motivate improved compliance with rules and expectations Question Formulation - targets Behaviorally descriptive of the examinee's involvement in an issue of concern (questions about knowledge, truthfulness, or another persons bx are considered less desirable) Time delimited (date of incident or time-frame) Simple and direct and easily understood by the subject Does not presuppose guilt or deception Be easily answered yes or no Avoid legal terms and treatment jargon Avoid mental state or motivational terminology Question Formulation - personally relevant questions Broader in scope Intended to test the limits of an individual's overall integrity Often broader in time, prior to or including the time period under investigation Used to rule out other issues that might interfere with the examination results Other Test Questions Known-truth questions Name Date (day of week) State/Country of residence Other known issues Ground-truth is unequivocal Not subject to investigation Used by examiner to conduct the test Non-testable Issues Intent Fantasy Thoughts Future behavior Beliefs Other individual s behavior 4

5 Polygraph Examination Results Significant Reactions (Deception Indicated) No Significant Reactions (No Deception Indicated) No opinion (Inconclusive) Purposeful Non-Cooperation (Intentional Distortion) Examinees must pass all test questions to achieve a completely resolved test result Polygraph Examination Results Significant Reactions (Deception Indicated) Numerical scores within decision range for the statistically significant presence of physiological responses correlated with deception The presence of reactions is unequivocal The meaning of observed responses is supported by empirically derived correlations Behavioral (PNS) indicators of attempts to alter or improve the test results (non-cooperation is correlated with deception) Polygraph Examination Results No Significant Reactions (No Deception Indicated) Numerical scores within decision range for the statistically significant absence of physiological responses correlated with deception Polygraph Examination Results No opinion (Inconclusive) Uninterpretable data (dampened, exaggerated, erratic, or inconsistent) Examinee exhibited significant reactions to other question/s during a mixed-issue (multi-facet) test Formulation and rendering of professional opinion about truthfulness or deception is not possible Polygraph Examination Results Purposeful Non-Cooperation (Intentional Distortion) Examinee has admitted to conscious and deliberate attempts to alter or improve the test result Includes both naïve and practiced countermeasure attempts Without admission of non-cooperation the examiner will score non-cooperative behavior/data to a deceptive conclusion Polygraph Techniques Zone Comparison Tests (primarily single issue diagnostic tests) Backster Zone Comparison Technique (3 questions) Integrated Zone Comparison Test (3 questions) Bi-Spot Zone Comparison Test / Backster U-phase) (2 questions) US Army Zone Comparison Test (3 questions) Matte Quadri-track Zone Comparison Test (2 questions) Modified General Question Techniques (often mixed issue or multifacet tests) Derived from the General Question Technique (3 or 4 questions) USAMPS Version (3 or 4 questions) US Air Force Version (2, 3, or 4 questions) US Secret Service Version (2, 3, or 4 questions) US Navy (2, 3, or 4 questions) 5

6 Other Polygraph Techniques Relevant-Irrelevant Technique Often used in police and security applicant screening Not numerically scored Test of choice when testing other polygraph examiners Concealed Information Test Considered most accurate and empirically sound investigative technique Requires a known incident Not numerically scored Test for Espionage and Sabotage (terrorism) Directed Lie Comparison Test Highly accurate screening test for government employees Has not generalized to other populations due to strict requirements on language and focus of target issues Who Uses Polygraph Tests PCSOT programs Sex History Investigations Maintenance of Compliance Monitoring Issues of Concern Specific Issue Investigations Law Enforcement Investigators Applicant Background Screening Criminal Investigation Internal Investigations Government Security Agencies Security Clearance Screening Monitoring Compliance with Security Policies Department of Energy (nuclear weapons programs) Private Security Agencies Armed security Pharmaceutical employees Reality Television Fishing Tournaments Cheating Spouses PCSOT Post Conviction Sex Offender Testing PCSOT - Objectives Disclosure - utility value Information gathering risk assessment / treatment planning reduction of denial improved engagement in treatment testing the limits of honesty Detection - accuracy value determining dishonesty and deviancy relieving truthful persons of unjustified concern Deterrence - monitoring value increased likelihood of detection / consequences improved behavior compliance PCSOT - Types of Tests Event Specific / Specific Issue (disclosure over the instant offense) instant offense / victim clarification new allegation or specific issue of concern Rule out unknown abuse prior to reunification with non-victim children or younger siblings Sexual History Disclosure other victims (excluding instant offense) paraphilic behaviors (sexual behavior patterns) Maintenance / Monitoring non-compliance, violations, dishonesty new offenses Specific Issue Polygraph - Target Selection Specific to date, location, known incident, accusation or allegation Details of allegations which the offender denies Use of violence to gain cooperation of or silence a victim physical force Physical restraint threats of harm Discrepancies (victim and offender statements) Testing the limits of behavior number of offenses or range of behavior especially when offender reports few incidents Penetrative / intrusive behaviors 6

7 Specific Issue - Sample Targets On or about January 1, 2004, did you engage in sexual contact with Jane Doe? On or about January 1, 2004, did you sexually touch Jane Doe s bare private parts? On or about January 1, 2004, did you physically force or threaten to harm Jane Doe for sexual contact? Besides what you already reported, did you engaged in any other sexual contact with Jane Doe? Besides what you reported did you sexually touch Jane Doe at any other time? Did you ever sexually touch Jane Doe s private parts? Did you ever rub your sexual organs against Jane Doe? Did you ever attempt to insert your penis into Jane Doe s vagina? Sex History Targets - Victim Selection Characteristics Sexual contact with underage persons (statutorily defined - under age 15 as adult, or four years younger as juvenile) Incest behaviors (legal family relationships) Forced/violent sexual contacts (physical force / threats of harm) Sexual contact with mentally / physically incapacitated persons (asleep, passed-out) Sex History - Sample Targets Since your 18th birthday, did you have sexual contact with anyone under age 15? Before you turned age 18, did you have sexual contact with anyone who was four or more years younger than you? Besides your two cousins, did you have sexual contact with anyone else who was related to you? Besides those three people, did you physically force or threaten to harm anyone else for sexual contact? Besides whom you already reported, did you have sexual contact with anyone else who was asleep or unconscious? Sexual Behavior Patterns (crossover / deviancy / compulsivity) Public indecency behaviors Peeping Voyeurism Frottage Public masturbation Sexual contact with animals Stealing or use of underwear or undergarments Use of non-human objects Child pornography Sexual Behavior Patterns - Sample Targets Did you ever show your bare sexual organs to an unsuspecting person in the community? Did you ever rub your sexual organs against an unsuspecting person? Did you ever look through someone s window in attempt to see their bare sexual organs? Did you ever engage in sexual contact with a non-human animal? Did you ever steal or use another person s underwear or undergarment for sexual purposes? Maintenance Polygraph Targets Unreported (secretive) sexual contacts Unauthorized contacts with vulnerable persons (physical contact, games, or being alone with underage persons) Use of pornography (sexually stimulating or sexually explicit materials) Masturbatory deviance (excessive or compulsive masturbation; thoughts of victims, young children, or violence) 7

8 Maintenance/Monitoring - Special Considerations Maintenance polygraphs targets high risk and noncompliance behaviors that can be signal the early phases of escalating risk level broader than reoffense questions increased sensitivity to life-style instability and sexual deviancy allows intervention before reoffense, without revocation or prosecution waiting to catch an offender after reoffense is too late Monitoring polygraphs are intended to investigate or rule-out unknown/unreported reoffense behaviors lower base-rates compared with non-compliance behaviors means differences in false positive index (FPI) for reoffense Qs Maintenance Polygraph - Sample Targets Since your last polygraph, did you have sexual contact with anyone that you have not reported (anyone besides your wife)? During the last six months, have you been completely alone with anyone under age 12? Since you were released from jail, did you view or use any X-rated or pornographic images? Since you returned home last January, did you masturbate to a thought or fantasy about a child? (All questions must use the same time-bar) Monitoring Polygraph Targets Forced sexual contacts Physical force Threat of force or violence Coerced sexual contacts Verbal or social pressure Manipulation, bribery, trickery Sexual contact with underage persons Sexual contact with relatives / family members Sexual contact with sleeping or mentally / physically incapacitated persons Public indecency behaviors Exhibitionism Voyeurism Frottage Monitoring Polygraph - Sample Targets During the last six months, did you force or coerce anyone to have sexual contact with you? Since your last polygraph, did you physically force or threaten to harm anyone for sexual contact? Since January 1, 2004, did you have sexual contact with anyone under age 12? During the last 12 months, did you sexually touch anyone who was asleep or unconscious? Since your last polygraph, did you have sexual contact with anyone who was related to you? Preparation Adequate preparation improves the signal value for both truthful and deceptive examinees Involves the timely, thorough and accurate disclosure of all information of interest to risk management and treatment planning processes Requires ability to recognize and define sexual behavior that is abusive or unlawful Depends upon an adequate conceptual vocabulary regarding sexual behavior Preparation requirements may vary across sub-populations of offenders (juveniles, adult, sexually compulsive persons) Preparation - Sexual History Identification of victims (behaviorally defined victim selection characteristics) Some examinees benefit from reviewing all sexual contacts Disclosure of all sexual behavior patterns (sexual compulsivity and deviancy issues) Disclosures should be reviewed with therapist, supervising officer, treatment group, family members (spouse/parents) as directed by the treatment and supervision team 8

9 Preparation - Monitoring / Maintenance Clearly defined boundary and safety plans re contact with children or younger siblings Clearly defined recording and disclosing procedures may assist truthful/motivated persons, and will improve the signal value surrounding issues of concern telephone reporting weekly or monthly reports (checklists) identifying high risk situations timely and accurate disclosure Prior issues of deception should be resolved through additional disclosure work prior to re-examination Question Formulation Behaviorally descriptive (refers to examinee s behavior vs. written summary of behaviors) Time delimited (date of incident, or time-bar to identify time of reference) Single interpretable meaning (simple and direct) Never presuppose guilt or deception Must be easily answered yes or no Avoid legal terms and treatment jargon (i.e., victim, rape, grooming) Avoid mental state or motivational terminology (i.e., purposefully, knowingly, intentionally) Preparation Adequate preparation improves the signal value for both truthful and deceptive examinees Involves the timely, thorough and accurate disclosure of all information of interest to risk management and treatment planning processes Requires ability to recognize and define sexual behavior that is abusive or unlawful Depends upon an adequate conceptual vocabulary regarding sexual behavior Preparation requirements may vary across sub-populations of offenders (juveniles, adult, sexually compulsive persons) Preparation - Sexual History Identification of victims (behaviorally defined victim selection characteristics) Some examinees benefit from reviewing all sexual contacts Disclosure of all sexual behavior patterns (sexual compulsivity and deviancy issues) Disclosures should be reviewed with therapist, supervising officer, treatment group, family members (spouse/parents) as directed by the treatment and supervision team Preparation - Monitoring / Maintenance Clearly defined boundary and safety plans re contact with children or younger siblings Clearly defined recording and disclosing procedures may assist truthful/motivated persons, and will improve the signal value surrounding issues of concern telephone reporting weekly or monthly reports (checklists) identifying high risk situations timely and accurate disclosure Prior issues of deception should be resolved through additional disclosure work prior to re-examination Test Theory 9

10 Screening and Diagnostic Tests Screening Test No known Incident No specific allegation or reason to suspect involvement in the issue/s under investigation Broad time-frame Multiple or single issues / of concern Diagnostic Test Known incident - identified issue or incident Reason to suspect involvement Specific date, time, and/or location Single behavior or issue of concern Single and Mixed Issue Tests SINGLE ISSUE TEST = All target issues are conceptually identical, though semantically different Inconceivable for a subject to lie to one question while telling the truth to another Alpha is assumed to be 1 (100) MIXED ISSUE TEST = Target issues are conceptually distinct Conceivable to lie to one or more questions while being truthful to another Alpha is assumed to be 0 (zero) MULTI-FACET TEST = A mixed-issue test involving various aspects of a known incident or allegation Screening Polygraphs Diagnostic Polygraph PCSOT Post-conviction risk assessment (sex history) Monitoring compliance with treatment and supervision rules Deterring non-compliance with treatment and supervision rules National Security Preemployment / preclearance background check (risk assessment) Monitoring compliance with security policies Deterring non-compliance with security policies Known incident - identified issue or incident Reason to suspect involvement Specific date, time, and/or location Single behavior or issue of concern May address multiple facets of a known incident Screening vs. Diagnostic Tests Screening Test May prioritize test sensitivity in order to minimize the likelihood of false negatives Positive results require further investigation to confirm Attempt to be cost-effective and efficient Negative result is considered conclusive Diagnostic Test May prioritize test specificity in order to eliminate screened subjects who do not express or represent the concern under investigation Used serially with screening tests Usually more costly and invasive procedures Positive result is considered a basis for action Sensitivity Determines the ability to detect the issue of concern Greater sensitivity = greater detection rate Influenced by decision threshold Sensitivity can be increased at a cost of more falsepositive results False positives are reduced at a cost of reduced sensitivity Governs the ability to rule out a diagnosis 10

11 Specificity Determines the ability to reject all cases that do not express the specific issue of concern Test that are overly broad and sensitive are not suitable for diagnostic purposes though may have great utility for screening situations Positive result (inability to confidently rule out the issue) may become a basic for action In medical care, multiple (serial) positive test results are often required as a basis for action Test Results Screening and Diagnostic Tests Pass Screening Test Incomplete need more information further investigation further screening or diagnostic testing (Krapohl & Stern, 2003) Pass Fail Diagnostic Test Not Sure Test Theory Take Home Point Diagnostic and Screening tests represent different testing principles with differing objectives Diagnostic tests attempt to achieve a diagnostic opinion by accurately ruling out everyone who does not express the issue of concern Diagnostic tests are biased for specificity to the issue of concern Require a known incident or allegation Negative results are considered conclusive Positive results may be a basis for further action Screening tests are intended to determine the possible presence of variety of issue Screening tests are biased for sensitivity to reduce falsenegative errors Negative results are considered conclusive Positive results will require further inquiry before action Basic and Applied Science Basic Science Polygraph research has proceeded in relative isolation from related fields of basic science and has benefited little from conceptual, theoretical, and technological advances in those fields that are relevant to the psychophysiological detection of deception. (National Academy of Sciences, 2003) Concerned with investigating the accuracy of theoretical principles and basic questions Knowledge is accumulated by disproving ideas that are not sound Basic science provides the foundation of knowledge for the development of applied science QUESTION: Are there physiological reactions that are reliably correlated with the act of being deceptive? 11

12 Applied Science Scientific principles (experiments and methods) are used to produce work products that solve realworld problems Experiments and projects are designed to verify that methods work as intended How can we determine when a person is being deceptive? Scientific Method Develop a theory or hypothesis There are significant measurable differences in physiological reactions of truthful and deceptive persons Define the null-hypothesis There are no significant differences in physiological reactions of truthful and deceptive persons Experiments intended to disprove a hypothesis (nullhypothesis) Develop better hypothesis, then prove it wrong Invite criticism and replication of findings Accept hypothesis only when cannot prove the nullhypothesis Reliability Sets the upper limit of Validity Internal Consistency How are we sure about what we are measuring Inter-rater Reliability Do different raters achieve the same results Test-retest Reliability What happens if we repeat the test later Validity Construct Validity (NAS's biggest criticism) Have the basic theoretical constructs been empirically validated? Psychological Set Contextual Salience Classical Conditioning Habituation to neutral and threatening stimuli Criterion Validity Do the test results correlate with an external criterion? (confessions) Convergent Validity What do other validated measures contribute to our understanding of polygraph results? Predictive Validity Polygraph itself is NOT a measure of risk Information and results from the polygraph may be incorporated into other validated risk measures Predictive Statistics Internal Consistency - Alpha Cronbach's alpha coefficient of reliability Measures how well a set of items (questions) measures a single construct Reliability sets the upper limit for validity Tests cannot be considered valid if we are unsure what they measure Alpha statistics over.8 are considered good in the social sciences Alpha coefficients close to 1 are not desired in most social science test designs (redundancy) Assumed to be 1 for single-issue tests Assumed to be 0 for mixed-issue (multi-facet) 12

13 Inter-rater Reliability Kappa This statistic is used to assess agreement between different observers (examiners) Tests with poor inter-rater agreement cannot be reliable and will have inherently poor validity Kappa considered superior to percent agreement for qualitative data (DI/NDI) Kappa statistics of <.70 are satisfactory Computers have Kappa score of 1.0 Reliability alone does not equal Validity Parsimony The ability to define the greatest range of phenomena with the simplest explanation Elegance simple and easily understood Occam's Razor Accept the simplest explanation that fits the facts (no more complicated than necessary) KISS principle Central Tendency (sameness) Median middle value of all scores Mode most common occurance of scores Mean sum of all scores divided by N Example: Five (5) deceptive polygraph subjects produce scores of (1,2,3,4, and 5) 1 Mean (average) / 5 = 3 N = 5 Z Scores Mean = 0 St. Dev = 1 For each individual score subtract the mean Divide by the St. Dev Mean = = = -1.4 St. Dev. = = = = = = = = = 1.4 Common Metrics Derived from Z values (scores) Multiply by the St.Dev Add the Mean Standard Scores: Mean = 100 (St. Dev. = 15) T- Scores: Mean = 50 (St. Dev = 10) Sten Scores: Mean = 10 (St. Dev. = 3) Score Z-Score Standard Score T-Score Sten Score 1) -1.4 * = 79 * = 36 * = 6 2) -0.7 * = 89 * = 43 * = 8 3) 0 * = 100 * = 50 * = 10 4) 0.7 * = 111 * = 57 * = 12 5) 1.4 * = 121 * = 64 * = 14 13

14 Normal Distributions and Outliers Normal Distribution resembles a bell-shaped curve 68 % of subjects score within 1 St. Dev. of Mean 95 % of subjects score within 2 St. Dev. of Mean NORMAL RANGE only 5% of subjects score out side this range 99 % of subjects score within 2.5 St.Dev. of Mean <1% (1 in 100) score outside this range 99.9% of subjects score within three 3 St. Dev. <0.1% (1 in 1000) score outside this range Scores over three standard deviations are outliers Measurement Theory An individual test score is always a combination of two elements The individual's true (universe) score True score refers to actual not accurate Be careful not to confuse validity with reliability Random measurement error (reliability) There is no such thing as a perfect test Standard Errors What happens if the same individual takes a test several times The average of all scores would be best estimate of the individual's true score Variations in score are called errors of measurement The Standard Deviation of one individual's varying scores is termed the Standard Error of Measure (SEM) SEM is an estimate of a test's reliability Standard Error of Measurement Defines the range of variability for individual scores Standard Error of Measurement St. Dev. multiplied by the SqrRt of (1 reliability statistic) Standard Error of Statistic St. Dev. multiplied by the SqrRt of N (N-1 for sample) Improves with the sample size upon which the measurement is based Mean = 3 5 * SqrRt(5) = 5 * 2.24 = 2.24 St. Dev. = 1.4 N = 5 if N = * SqrRt(100) = 5 * 10 =.5 Standard Error of Mean Defines the expected range of variability for the mean of a sample or population St Err Mean (pop) = St Dev Sqr Rt of N St Err Mean (sample) = St Dev Sqr Rt of N-1 Mean = 3 5 * SqrRt(5) = 5 * 2.24 = 2.24 St. Dev. = 1.4 N = 5 if N = 25 5 * SqrRt(25) = 5 * 5 = 1 if N = * SqrRt(100) = 5 * 10 = 0.5 if N = * SqrRt(1000) = 5 * 31.7 =

15 Polygraph Theory Truthful and Deceptive persons are found to produce different distributions of scores based upon the presence or absence of reactions to test stimulus questions Reaction features are physiological response that are correlated with deception Pearson's Correlation Coefficient - r Truthful and Deceptive persons are found to produce different distributions of scores based upon the presence or absence of reactions to test stimulus questions Reaction features are physiological response that are correlated with deception HYPOTHETICAL DATA EXAMPLE Deceptive and Truthful Groups 10 hypothetical subjects Reaction Features are counted and totaled (teaching example only - not an actual dataset) Truthful Group SUBJECT 1) 1 2) 2 3) 3 4) 4 5) 5 Mean = 3 St. Dev. = 1.4 Features Mean for Both Groups = 5.5 Deceptive Group SUBJECT Features 6) 6 7) 7 8) 8 9) 9 10) 10 Mean = 8 St. Dev. 1.4 Normal Distribution is Assumed For This Example TEACHING EXAMPLE - NOT AN ACTUAL POLYGRAPH SCORING SYSTEM Standard Error of the Mean Difference How much do the means of two groups differ? Standard Deviation of the difference between the two means St. Err. of Mean Difference - EXAMPLE Deceptive Group Truthful Group N = 5 N = 5 Mean = 8 Mean = 3 St. Dev. = 1.4 St. Dev. = 1.4 Equal variance is assumed Depends in part on the standard error of the to means (will improve with sample size) Square root of the sum the squared standard errors of the means for the two groups 95% CONFIDENCE INTERVAL OF DIFFERENCE BETWEEN TRUTHFUL AND DECEPTIVE MEANS WILL BE WITHIN 2 ST. ERRORS TEACHING EXAMPLE - NOT AN ACTUAL POLYGRAPH SCORING SYSTEM 15

16 T Statistic (two samples t-test) - EXAMPLE Divide the observed mean difference by the standard error of the mean difference DIFFERENCE BETWEEN GROUPS IS SIGNIFICANT TEACHING EXAMPLE - NOT AN ACTUAL POLYGRAPH SCORING SYSTEM Rejection of the Null-Hypothesis Null hypothesis can be rejected if the observed difference is unlikely NULL HYPOTHESIS: There are no significant differences in physiological reactions of truthful and deceptive persons Rejection of null-hypothesis allows acceptance of the hypothesis HYPOTHESIS: There ARE significant differences in physiological reactions of truthful and deceptive persons Would be unable to reject the null hypothesis if 95 % confidence interval included other other mean Polygraph Field Practice Does not involve complex statistics Decisions made by cut scores that assure accurate classification of deceptive and nondeceptive subjects Inconclusive range reduces erroneous results Statistics Take Home Point Development of accurate diagnostic polygraph tests is possible with a small set of physiological response features Development of good diagnostic lie-detection methods using a single physiological feature (voice stress) seems unlikely as it would depend upon the uniqueness of the feature to deception and upon statistically significant differences in strength of reactions between truthful and deceptive responses 16

17 National Academy of Sciences (2003) Polygraph Accuracy... we conclude that in populations of examinees such as those represented in the polygraph literature... specific-incident polygraph tests for event-specific investigations can discriminate lying from truthtelling at rates well above chance, though well below perfection. (National Academy of Sciences, 2003) [p.214] Accuracy of Polygraph Results Studies show 95% to 98% accuracy rates when a decision is reached Truthful results more accurate than deceptive results Accuracy issues represent differing considerations for truthful and deceptive test results Certainty of findings may be diminished in relation to breadth of examination target issues Accuracy - Criterion Validity (National Academy of Sciences, 2003) 59 laboratory and field studies ROC A Value =.70 to.99 IR.81 to.91 Median =.8 7 field studies A =.71 to.99 Median =.89 Field studies showed greater accuracy than laboratory studies, though the committee predicted field studies would underperform laboratory situations Accuracy - Motivation External motivation is positively associated with polygraph accuracy in comparison question and concealed information (Kircher et al. 1988; Ben-Shakhar and Elaad, 2003) Research in Colorado Dept. of Corrections indicates staff attitudes (endorsement and requirements) affect test outcomes field experience indicates that externalmotivation is associated with improved polygraph outcomes (i.e., more subjects tell the truth and pass) Motivation - Standard Colorado Sex Offender Management Board (1999) The sex offender community supervision team shall consult about the completion of treatment. This decision shall come after the evaluation and assessment, treatment plan, course of treatment sequence, and a minimum of a non-deceptive disclosure polygraph examination and two or more nondeceptive maintenance polygraph examinations, regarding compliance with court rules, compliance with supervision conditions, compliance with treatment contract provisions, including complete abstinence from grooming of victims and full, voluntary compliance with all conditions required to prevent reoffending behavior. [p.37] 17

18 Accuracy - Criterion Validity Does the test give the correct decision? Single Diagnostic Scale - raw score indicates the strength of evidence Signal Detection Theory - diagnosis depends on the degree of evidence favoring one or the other alternative Accuracy Signal Detection Theory Accuracy of the test how well does the test measure physiological responses correlated with deception (sensitivity and specificity) Decision threshold what volume or threshold of observed reactions is necessary to allow a diagnosis Base rate of incidence (external probability) the types of errors observed in the field will be influenced by the base rate of incidence Polygraph Errors False Negative - deceptive examinee passes the test primary concern to the community False Positive - truthful subject fails the test primary concern of the individual examinee Inconclusive - the test fails to work effectively Decision Theory (Signal Detection) Decision thresholds should be set to minimize the net costs of false positive and false negative errors False Negative Error: Increase risk to community Actual harm to victim False Positive Error: Further investigation Temporary restriction of activities Sensitivity and Specificity Sensitivity Underlies negative predictive value Determines the ability to rule out the issue of concern Specificity Underlies the false positive index Determines the ability to diagnose the issue of concern In medical care, serial positive tests are often required as a basis for taking action Diagnostic Accuracy - Predictive Value Sensitivity - influences the ability to rule out a diagnosis Negative Predictive Value = True Negatives / (True Negatives + False Negatives) Sensitivity = True Positives / (True Positives + False Negatives) Specificity - in medical care multiple positive tests are often required to arrive at a firm basis for action False Positive Index = True Positives / False Positives Specificity = True Negatives / (True Negatives + False Positives) 18

19 National Academy of Sciences (2003) Conditional Probability Outcomes - Sensitivity to deception set hypothetically at 80 percent Criminal Suspects - Base Rate: 1 in 2 N=10,000 Guilty Not Guilty Total Failed Test 4,000 (TP) 800 (FP) 4,800 (FPI =.16) Passed Test 1,000 (FN) 4,200 (TN) 5,200 (NPV =.81) Total 5,000 5,000 10,000 National Laboratory Employees - Base Rate: 1 in 1,000 N=10,000 Spy Not Spy Total Failed Test 8 (TP) 1598 (FP) 1,606 (FPI =.99) Passed Test 2 (FN) 8,392 (TN) 8,394 (NPV =.99) Total 10 9,990 10,000 PCSOT Linear Conditional Probability - Decision accuracy set hypothetically at.90 (N=1,000) Hypothetical Base Rate =.50 (50% or 1 in 2 offenders) Deceptive Truthful Total Failed Test 450 (TP) 50 (FP) 500 (FPI =.11) Passed Test 50 (FN) 450 (TN) 500 (NPV =.9) Total ,000 (Sens =.9) (Spec =.9) Hypothetical Base Rate =.10 (10% or 1 in 10 offenders) Deceptive Truthful Total Failed Test 90 (TP) 90 (FP) 180 (FPI =.5) Passed Test 10 (FN) 810 (TN) 900 (NPV =.99) Total ,000 (Sens =.9) (Spec =.9) Inter-rater Reliability Kappa Bakster system interrater agreement reported as not statistically significant Federal system Kappa =.57 Utah system percent agreeement = 95% (Krapohl & Dollins, 2003) Physiological Features and Scoring Backster System Original objective scoring system Cleve Backster (1962, 1984, 1990) 25 reaction features, 8 relief features, 21 rules (Matte, 2000) DoDPI Numerical Evaluation System Originally derived from Backster Features 23 physiological features, with (Swinford, 1999) Utah Scoring Rules empirically derived at the Univ. of Utah by Raskin, Barland, and Podlesny (1977) 8 to 10 features with simple rules Kircher Features 4 most robust and most productive features developed at the University of Utah (Kircher and Raskin (1988) Predictive Value PSCOT estimation Negative Predictive Value - how accurate are NDI test results? NPV ranges from.9 to.99 across base rates of.50 to.10 Specificity remains high at all base rates Positive Predictive Value - how accurate are DI test results FPI ranges from.9 to.5 across base rates of.50 to.10 Sensitivity remains high across varying base rates Conditional Probability Conditional probability tables are NOT accuracy tables Conditional probability tables reflect predicted EXPERIENCE in field situation ROC statistics are used to evaluate diagnostic accuracy because they are less subject to the influence of base-rates of occurrence The use of ROC measures of test accuracy does NOT eliminate observational base-rate phenomena in field experience situations Base rates of occurrence can be predicted externally 19

20 Accuracy and Utility What factors limit accuracy Suitability for testing (increases) Accuracy (specificity) (decreases) Narrower test offers increased specificity to a known incident or allegation Broader test offers increased sensitivity to various concerns Random measurement error Target selection (testability of the issue/s of concern) Question formulation (language and logic) (decrease - narrower) - Utility (broader - increase) (sensitivity) Improving Accuracy (question formulation) Question formulation behaviorally descriptive time delimited single unequivocal meaning Quality Assurance Procedures Narrowly focused set of questions may improve accuracy reduction in broad utility Assure the examinee understands the relevance of the test outcome potential rewards Potential consequences Testing the Limits of Admitted Bx Did you ever - any involvement - presumed most accurate Studies have not investigated polygraph accuracy when questions are intended to test the limits of admitted behavior Field experience suggests testing the limits offer great utility Besides what you reported... - could be tossing a bone Besides those three times, did you - preferred Could be troublesome for examinee's with CSB we'll never know everything Objectives may be satisfied by examinee's admissions Inconclusive Results Examiner is unable to render a professional opinion regarding the test results Inconclusive = No Opinion Possible Reasons for Inconclusive Results Confusion (inadequate discussion or preparation) Test results don t matter to examinee (no rewards / consequences) Fatigue (inattention) Ineffective target selection Improper question formulation Illness / medical health Side effects from recent onset/change in medications Unsuitability for testing (developmental / neurological) Noncooperation with examiner's instructions Inadequate data quality Strong reactions to other test questions (mixed issue tests) Inconsistent reactions (single issue tests) 20

21 Inconclusive Rates SINGLE ISSUE TEST = all target issues are conceptually identical Inconceivable that a subject could one question while being truthful to others Estimated as the proportion of the population that would score outside the normal range (2.5-5 %) Observed inconclusive rate may be higher due to suitability and situational factors (5 to 10 %) MIXED ISSUE TEST (multi-facet test) = target issues are related though conceptually distinct Alpha is assumed to be 0 (zero) Inconclusive rate will be compounded by the number of distinct issues under investigation Estimating Inconclusive Rates for Mixed Issue (multi-facet) exams Estimated as the inverse of the resolution rate raised to the exponent of the number of distinct target issues Resolution rate is the inverse of the expected inconclusive rate for single-issue tests EXAMPLE: INC-M = 1-(1-INCQ ) INC-M = inconclusive rate for mixed-issue tests INC = expected inconclusive rate for single issue tests Q = number of distinct target issues Estimated Inconclusive Rates - mixed issues (multi-facet) tests EXAMPLE: INC-M = 1-(1-INCQ ) INC-M = inconclusive rate for mixed-issue tests INC = expected inconclusive rate for single issue tests Q = number of distinct target issues Inconclusive Myths Inconclusive results means the examiner was afraid to make a call Inconclusive results occur because of poor pretest More that 10 percent inconclusive means the examiner isn't very good = Inconclusive results are almost as good as pass Inconclusive results are almost as good as fail Accuracy - Truth and Falsehood (epistemology) What types of things can be true? Statements about objects and events Beliefs and opinions are not epistemologically true unless supported by evidence (data) How do we know? Empiricism Measurement (descriptive and predictive statistics) Logic (classification) What does it mean to say something is true. Rational Truth things are really what they are Correspondence Truth accurate and complete descriptions Postmodern Truth projection, belief, opinion Pragmatic Truth absence of deception What is a Lie An attempt to deceive (secrets, twisting facts, half-truth, fabrication, denial, etc.) Polygraph and Other Tests Screening and Diagnostic tests were compared for Polygraph, Medical, and Psychological tests Overall Kappa for Poly., Med., and Psych and.79 Overall screening sensitivity levels were and.74 Overall screening specificity levels were and.78 Overall diagnostic sensitivity levels were and.72 Overall diagnostic specificity levels were and.67 Polygraph validity and reliability studies appear to report results consistent with achievement in other developed fields of scientific testing (Crewson, 2003) 21

22 Accuracy Take Home Points The polygraph test is very accurate (though imperfect) for both truthful and deceptive persons Accuracy experience rates differ across truthful and deceptive groups Truthful results appear more robust than deceptive Outcomes experience rates (in field situations) vary across base-rates of behavior subject to selection-bias across groups of offenders Accuracy rates may be influenced by the breadth of the target issues and test questions Accuracy/specificity is reciprocal with Utility/sensitivity Conclusions CANNOT be drawn anecdotally Psychophysiology Physiology Autonomic Nervous System - life maintaining Sympathetic (autonomic) Nervous System Cardiovascular activity Electrodermal activity Respiratory activity Parasympathetic Nervous System Peripheral ( Nervous System behavioral Major Muscle Activity Discrete Muscle Activity Voluntary Respiratory Activity Physiology and Deception Respiratory Activity Adrenergic (ANS) activation of the respiratory tract Cholinergic (ANS) activation of thoracic/intercostal and abdominal muscles Also subject to PNS (behavioral) activity Cardiovascular activity Andrenergic (ANS) stimulation of cardiac functions Electrodermal activity Cholinergic (ANS) activation of the sympathetic neurons in the skin Polygraph Components Pneumograph assembly autonomic activity (thoracic/intercostal, abdominal) respiration activity (rate, pattern, changes) also subject to voluntary activity (peripheral/somatic) Blood pressure cuff changes in blood pressure indicates adrenergic sympathetic activation Electrodermal sensors skin conductance (resistance) indicates cholingergic sympathetic activation Movement sensor peripheral/somatic muscle activity (behavioral efforts to disrupt the test outcome) Somatic/Peripheral Activity Faking Acting Noncooperation and faking are correlated with lying Executed voluntarily in attempt to falsify or distort the recorded data pertaining to the sympathetic reactions Monitored to determine whether observed responses are ANS or ANS + PNS Absence of activity assures authenticity of ANS responses Acetylcholine is activating neurotransmitter for peripheral nervous system (muscle) activity 22

23 Cardiovascular Activity Sympathetic - norepinephrine - speeds the heart and increases contractility Sympathic demands are workload demands Alpha Adrenergic (heart rate) passive coping (enduring) Beta Adrenergic (contractility) active coping (mental math) Delivers fuel to cells in brain, muscle, and organs Oxygen Adenosine Triphosphate (ATP) Glycogen Return CO 2 to lungs Recycle/filter blood through Renal / hepatic system Parasympathetic - acetylcholine - slows the heart and lowers blood pressure to baseline levels Respiratory Activity Normally regulated by Autonomic Nervous System (ANS) Chemoreceptors in brainstem (hypothalmus) monitor carbon dioxide Chemoreceptors in large vessels of heart monitor oxygen Stretch receptors in lungs monitor inflation Thoracic/intercostal and Abdominal muscle groups are enervated by sympathetic Acetylcholine (not adrenaline) Respiratory Activity Can easily brought under voluntary control through attention Peripheral Nervous System (PNS) activity Voluntary changes in respiration can produce changes in cardiovascular and electrodermal responses Monitored to determine whether observed reaction are self-induced artifacts Impossible to accurately mimic the autonomic respiratory patterns of a truthful person Electrodermal Activity Monitored through eccrine (not apocrine) glands Enervated by sympathetic Acetylcholine (not norepinephrine) No parasympathetic neurons in the skin All eccrine activity is sympathetic activity or resorbtion Thought to be a reflection of cholinergic activation of the sympathetic nervous system Emotions hypothalmus Temperature Reticular Activation System (RAS) prefrontal cortex enervated by acetylcholine Everything You Ever Wanted to Know About Sweating Eccrine glands located throughout the body Denser concentrations in palms and feet Enervated by Acetylcholine Regulated in part by the Hypothalmus Apocrine glands located mainly in private areas Cholinergic and hormonal (adrenergic) activation Hair folicles Foster production of bacteria (BO) Thought to play a role in mating for some species Three main types of sweating emotional sweating fear and anxiety stimulate cholinergic activity in the hypothalamus thermoregulatory sweating hypothalamic acetylcholine signals gustatory sweating spicy food Sweating Disorder Hyperhidrosis (Reynaud's) excessive sweating (axillary or palmar) Frey's syndrome facial sweating in response to salivatory stim The Brain 23

24 The Brain Weight is approx 2.5 % of body Consumes approx 20% of oxygen intake Brain is over 77% water Grey matter (cerebral cortex) approx 40% Cerebral cortex consums 94% of oxygen in the brain Consumes oxygen and glucose at 10 times the rate of the rest of the body The Brain - Neurons Neurons found primarily in brain and spinal cord Do not appear to regenerate on a normal programmed basis like other cells neuron is Greek for string Dendrites (tree) project from neurons Connected to an Axon which is covered in myelin Embryonic brain produces 250,000 neurons per minute Neurons are paired (apoptis) to remove bad connections and strengthen good connections Electrical signals are carried by neurotransmitter chemicals The Brain - Neurotransmitters Most of the nervous system's important chemicals are produced in the Brainstem Peptides Digestion and hormones (amino acids) Acetylcholine voluntary and involuntary muscle activity, memory (Alzheimers) Epinephrine / Norepinephrine hormone in body, neurotransmitter in brain, produced in adrenal gland and brain, FF response activator, stimulates massive release of glucose to muscles (strength) Cortisol stress hormone Endorphine (endogenous morphine) Dopomine pleasure, motor functioning (Parkinsons) Serotonin mood, memory, appetite, temperature The Brain - Brainstem Located at the base of the brain Damage to brainstem is highly life-threatening Three parts Mid-brain (upper part) Pons (center area) Medula oblongata (lower end) Primary purpose is survival Controls autonomic functions Breathing Heartbeat Blood pressure Reticular Formation The Brainstem Reticular Formation RF Occupies the core of the brainstem Monitors information from throughout the body Regulates basic life-support functions RF neurons and the Thalamus make up the Reticular Activation System The Brain Reticular Activation System Receives information from body regarding the environment Regulates level of autonomic arousal to maintain survival Damage to RAS results in permanent coma Thalamus filters stimuli from environment decides what is most important (threats) Tunes-out unimportant stimuli 24

25 The Brain Thalamus, Hypothalamus Thalamus Directs traffic between cortex and sensory organs Processes all sensory input except smell 99 percent of sensory input is discarded immediately Hypothalamus (below thalamus) Regulates ANS Temperature and perspiration Sex drive Aggression Pleasure seeking Sleep Fight or Flight Response Fight or Flight Massive, diffuse, and global (non-specific) adrenal and endocrine activation during emergency states Brain does not distinguish between physical and psychological danger Epinephrine and norepinephrine (adrenaline and noradrenaline) Sometimes act as neurotransmitters in the brain In fight or flight they activate endocrine, circulatory, muscular, and digestive systems Cortisol (secreted by adrenal glands) Endocrine activation Cortisol Emotional hormone component of FF response or stress Emotion becomes dominant over cognition New learning becomes difficult High concentrations of cortisol over long periods of time cause damage Hippocampus (difficulting learning and remembering new things) Immune system compromised Increased risk of disease Retardation of growth Stops digestion and other normal metabolic process during FF response (butterflies) Vomiting The Brain - Hippocampus Responsible for working (short-term memory Over time the hippocampus stores short term memories in long term memory in the Cortex Hippocampus not involved in long-term memory retrieval Damage to hippocampus results in difficulty learning and storing new memories Chronic gluco-cortical stress response Adrenaline enhances memory for the event that stimulated the FF response Emotional stimuli are better remembered than neutral The Brain - Amygdala Small - shaped like and almond Alarm system in brain Checks with hippocampus to see if stimuli represent a threat Triggers hypothalamus if stimuli are potentially harmful Some researchers suggest Amygdala forms unconscious emotional memories without dependence upon conscious recollection PTSD inability to deactive through conscious attention to other stimuli 25

26 The Brain Cerebral Cortex Grey matter Thin layer folder over outside of brain Cerebellum (little brain) on lower back Four main lobes named after underlying skull bones Occipital Lobes Temporal Lobes Parietal Lobes Frontal Lobes (prefrontal cortex) The Brain - Cerebellum Small lobe in back of brain Primitive has changed little over millennia Maintains balance Muscle tonus Posture Large 11% of brains weight Almost adult size by age 2 Depends on Motor Cortex for new skill acquisition Motor Cortex depends upon Cerebellum for coordinated movement Cerebellum performs complex motor tasks with almost no conscious awareness The Brain Occipital Lobes Back part of brain (above Cerebellum) Visual cortex Together with Wernicke's area and Broca's area are termed the Aphasia Cortex The Brain Temporal Lobes Located on either side of the brain Auditory stimuli Various Subdivisions Hearing Memory Language (Wernicke's Area) The Brain Parietal Lobes On top part of brain Somatosensory Touch Temperature Environment Proprioception (awareness of body position) The Brain Frontal Lobes Front part of brain Association cortex Human cortex Performs most complex functions Speech (Broca's area) Large Prefrontal Cortex (different than other mamals) 26

27 The Brain Prefrontal Cortex Executive functioning Concentration Awareness Emotional self-regulation Behavioral/social self-control Problem solving Lying Executive Functioning Deficits Poor planning and organization Difficulties generating and implementing problem solving strategies Poor initiation Disinhibition / impulsivity / erratic careless behavior Rigid or concrete processes-inflexibility in thoughts or attitude Perservation / problems in stopping (deficits in self-monitoring) Inability to correct errors or use feedback Poor self-control Deficient self-awareness Limited insight Tendency to be self-satisfied Tend to experience little or no anxiety Fail to appreciate the existence and practical implications of these deficits Problems or difficulties with attention Deficits in social awareness including rude or insensitive behavior Lack of apparent concern about social expectations (Anderson, 1998; Sami et al., 2003; Lezak, 1993; Pennington, 1997; Stuss & Benson, 1987; Tranel, Anderson, & Benton, 1994; Walsh, 1978) Development of Executive Functions Three Distinct Stages of Executive Skill Development Age 6 Ability to resist distraction Age 10 Organized search, hypothesis testing, impulse control (sometimes consistent with adult levels Age 12 Expanded verbal fluency skills, improved motor coordination skills, increased planning (not consistent with adult levels) (Anderson, 1998; Welsh, Pennington, and Groisser 1991) Memory, Perception and Attention Novelty the great attention getter Habituation getting used to it (outhouse syndrome) Hippocampus responsible for working (short-term) memory and eventually cedes memories to long term storage in the cortex Hippocampus is independent of long term memory Emotional interest enhances attention and memory Information goes first to the Thalamus Amygdala then determines the emotional significance Cortex makes rational sense of stimulus Iconic (brief, visual) / Echoic (auditory longer) 18 second holding pattern (15 to 20 seconds) Cocktail Party Effect Perception depends upon meaning (RAS) Why does it matter? Perception experience and meaning (attributed to James, 1982) 27

28 Perception meaning and projection Cocktail Party Effect Contextual Salience - Psychological Set In performing an experiment like this one on man attention car it house is boy critically hat important shoe that candy the man material car that house is boy being hat read shoe by candy the man subject car for house the boy relevant hat task shoe be candy cohesive man and car grammatically house correct boy but hat without shoe either candy being man so car easy house that boy full hat attention shoe is candy not man required car in house order boy to hat read shoe nor too difficult. (Wolfe, 2001) Construct Validity Psychological Basis (Sensitivity and Specificity) research has confirmed that the polygraph instrument measures physiological reactions that may be associated with an examinee s stress, fear, anger, excitement, or anxiety about deception, or with an examinee s orienting response to information that is especially relevant to some forbidden act. (National Academy of Sciences, 2003) [p.72] Construct Validity Refers to how well explanatory theories and concepts account for a test s performance A solid theoretical and scientific base can give confidence about the robustness of a test across examinees and settings and against the threat of countermeasures and can lead to its improvement over time. Polygraph Theory Examinees will respond differently when trying to hide something (physiological arousal, emotional leakage, orienting responses) Pursuit of explanation to anomalous responses will lead to explanations and additional information regarding the target issues under investigation (National Academy of Sciences, 2003)[p.101] 28