Psychology 323: Deception

1
Psychology 323 Deception

• J. P. Rosenfeld, Ph.D.
• jp-rosenfeld_at_northwestern.edu

2
We should spend an equal time on phenomenology
and MECHANISMS of deception.

• as we do on DETECTION of deception. But there is
  a lot more research-based (real) knowledge about
  the latter, so most of the time, thats what we
  will cover.

3
Regarding detection there are two approaches

• 1. Behavioral
• (a) non-verbal
• (b) Verbal
• 2. Physiological Nervous system activity.
• .We start with Behavioral

4
Are the following notions True?
1. People typically reveal their lies by
fidgeting, acting nervous, avoiding eye contact,
etc. 2. Therefore, we are rather good lie
detectors (unless we are very stupid). 3. This is
especially true when we detect lies in those
close to us. 4. Criminals, con-men, professional
crooks, and such, however, are harder to spot
for us lay people. 5. Fortunately, trained
professionals (police, FBI) are superior lie
detectors, so they protect us against pros.

5
We wish we had a Pinocchio's nose indicator
6
But we dont

• NONE of those assertions is true!

7

• People think they are great lie detectors. But
  they miss many lies. Why?

8

• (1a) If someone thinks you look great, why
  argue? (Vrij Ostrich effect)

9

• (1b) Ill ask no questions(and hope you tell no
  lies!)
• Bill to Hillary Ah did not have sex with that
  woman..

10

• (2) Lie detection is tough!
• There is no behavioral or physiological specific
  index of deception, though things are not as bad
  a Vrij suggests P300, fMRI

11
Beware of internet Media claims!

• Pavlidis thermal imaging not so great
• BEOS Brain Electrical Oscillations Signature is
  Bogus!
• Brain Fingerprinting?! (See my critique.)

12
How common is lying?

• A) once a year?
• B) once a month?
• C) once a week?
• D) More than daily?

13
Definition of lying

• OK, give one

14

• Intentional, without warning, misinform or
  mislead another.
• Liar must believe information is false.
• Lie could fail.

15
Taxonomy of Deceptions

• (Types of Lies)

16
From Bad-est to least bad

• (1) Outright Lie (total B.S.)
• John Edwards I am not the father of that child,
  no way, Ill happily take a DNA test.
• Stating I didnt do it while knowing one did.

17
(2)Exaggerations (overstatements) or
Minimizations(understatements)

• I am the most eminently famous researcher in the
  Psych. Department. (but almost)
• This is an important discovery of mine not yet
  published (which was actually rejected by one
  journal and now in review at another.)

18
(3) Subtle Lies (Omit details)

• Clinton Ah did not have sex with that woman,
  Miss Lewinski
• (Not according to the usual definition
  intercourse.)

19
(4) Concealment

• How was your day?mine was as usual
• Heaven forbid she asks me about my day!

20
Why do people Lie?
21
Many reasons

• (1) Material Gain
• (2) Avoid loss/punishment
• (3) Avoid embarrassment
• (4) Make good impression ( get the job)
• All these involve gain for oneself.

22
The other kind of (Other-directed) Lie

• Altruistic Lies
• You made a great impression
• Telling a gravely ill person (child) that she
  will be fine..

23
Why is it hard to know for sure how often people
lie?

• Self report fails because people dont want to
  recall.
• But 2 times a day is a reasonable occurence.

24
Whom do people lie to?

• Strangers more than close contacts. (Why?)

25
Attributes on which to classify Liars

26
(1) Gender

• Men tell more self-oriented lies.
• Women tell more other oriented lies

27
(2) Age

• What is definite minimum age for lies in
  children? (2,4,6,8,10)
• What motivates youngest childrens lies? (Gain of
  say cookies? Avoid punishment? Make a good
  impression?)

28
(3) Personality Type

• In romantic relations
• (a) attachment style avoidant (lack of trust,
  keep people at distance)..these folks lie to
  keep others at a distance.
• Versus
• anxious attachments lack of self-esteem. These
  people lie to look good as they feel bad.

29
(b) Psychopathy

• The types have no empathy or sympathy for others.
  They regard others as pieces of themselves, and
  objects of manipulationlike their own limbs.
  These are big time liars to everyone they might
  wish to manipulate. Superficially charming, they
  get away with it for a while

30
(c) Extraverts and introverts

• Outward/sociable vs. reserved.
• Which are the bigger liars?

31
(d) Self-consciousness

• These are very concerned with impression they
  make, so how do they behave with respect to
  deception?
• In this connection, one speaks also of social
  anxiety they lack self confidence, so they tell
  tales.

32
Are there non-verbal, behavioral lie signals?

• (Are we all on same page? What are these?)

33
We wish we had a Pinocchio's nose indicator
34
DiPaolos group emphasizes 3 theoretical
perspectives

• During deception, 3 phenomena occur
• 1) emotional reactions guilt, fear----and
  delight. (delight?)
• Guilt ? low eye contact
• Fear ? stuttering, blinks, squeals.
• Delight (of duper)? excitement, inappropriate
  laughter. (Ekman)

35
What else does liar experience?

• 2) mental demand, also called cognitive load.
  Liars will also tend to look away, so as to
  concentrate on made up details. Except when they
  monitor you to see if you believe.
• 3) Attempted self control of behavior to avoid
  tells. Not easy. Voices break, facial expressions
  occur Ekmans micro-expressions.

36
Remember

• These are theories of what should happen.
  Moreover

37
these effects are affected by certain factors,
e.g.,

• GUILTThe degree of guilt felt depends on
  personality. Consider Psychopathy. Consider
  Stakes Consider sense of righteousness or duty
  felt by CIA agent.
• FEAR Experienced liars of whatever breed have
  less fear.
• EXCITEMENT helped by audience. Also, if its a
  big conquest, or else, no.

38
More

• COGNITIVE LOAD--- also depends on verbal
  skill-intelligence and experience. (Vrij is wrong
  about Psychopaths feeling less load.) Degree of
  rehearsal also makes a difference (practice makes
  perfect!)
• SELF CONTROL depends on Psychopathy.

39
Vrij is a pain, but best we have..(E.g.Box 3.1 p.
47).

• and his best statement is on the top of p. 49.
  the relationship between lying and non-verbal
  behavior is complex. The implications of this
  statement regarding development of non-verbal
  tells is what?! (Life is not Poker.)

40
Research Methods for studying behavioral
non-verbal signs of lying
41
Field Studies.(Define.)

• We want to compare liars vs. truth-tellers, but
• Difficulties
• 1)Getting videos (Are cameras always running?)
• 2) Establishing Ground Truth.
• 3) Controls only possible in lab to have
  unconfounded conditionsso not many field studies
  re nonverbal cues are out there

42
OK, how about lab studies?

• Pluses Ground truth easy to get by design.
  Likewise, un-confounded control conditions.
• Minuses Lies are instructed. So what happens to
  guilt? fear? What happens regarding stakes? You
  can offer lots of money, but not punishment (as
  in real life).
• DECEPTION HARD TO STUDY!

43
Bottom Line

• Vrijs Appendix 3.1 makes very clear that there
  is NO reliable non-verbal sign of deception.
  There may be trends and even significant group
  effects (explain), which support theoretical
  views noted above..
• But in D of D, the key statistic is INDIVIDUAL
  HIT RATE (as in d from SDT).

44
Dramatically

• Is there such a thing as lyin eyes? Are gaze
  or eye contact helpful?
• Are liars more emotional than truth tellers? Why
  or why not?

45
Vrijs Group Differences (p.67).

• Very suspect!
• As is rest of chapter about Clinton, Saddam
  Hussein, Huntley.
• Amazing that he says look for the cluster of
  c(l)ues to deception, not single c(l)ues. Which
  cluster? There is no straightforward answer.

46
Verbal signs ("cues" clues?) about deception

• That is.speech signs

47
We wish we had a Pinocchio's nose indicator
48
(No Transcript)
49
Vrij description of possible verbal signs
(cues)

• .as in his box 4.1 list is mostly pathetic,
  i.e., wrong (imho).
• On the other hand, sometimes one has to say,
  duhhh! as when he states that direct statements
  are more credible than evasive ones.
• Obviously plausibility helps! As does consistency
  (lack of contradiction).

50
But Vrij best statement is

• As I will demonstrate, a verbal cue uniquely
  related to deception, akin to Pinocchio's growing
  nose, does not exist. (p. 103)
• Again, an appendix(4.1) shows some trends group
  effects as supporting some views in Box 4.1.
• Again, clustersgt singles. But which cluster? No
  one knows.

51
Vrij indulges himself with many I think.. type
ideas.

• Thats fine. Be sure you do not confuse these
  hypotheses for facts!
• Note people researchers always start out with
  what appears obvious, like twitches speech
  signs. These all are ultimately controllable,
  unlike physiology (looking ahead).

52
Henceforth.

• We will be looking at more specialized and
  developed approaches to detecting deception. Vrij
  still likes more immediate (non-physiological)
  tools, because they are allegedly quick, easy,
  cheap. Imho, he stretches things by emphasizing
  how laborious some methods are...

53
OK,

• fMRI requires a 1M machine and an annual budget
  of gt1M
• But when Vrij states (p. 189) that EEG/ERP
  recording usually requires 10 or more scalp
  electrodes, he exaggerates 2 (actually 1) will
  do, and it takes only 5 minutes to hook up.
• But Vrij we will go on with SVA, RM SCAN,
  imho, unphysiological extensions of what we just
  did

54
More complex behavioral methods

• Reality Monitoring, Statement Validity
  Assessment, Scientific Content Analysis.

55
Reality Monitoring

• RM started not as a DoD method, but rather as a
  discipline in Cognitive Psychology to study on a
  theoretical basis, the differences between
  perceived versus imagined experience as with
  false (but honestly believed) memories.

56
What are the characteristics of Real versus False
Memories..

• with regard to
• Sensory Information?
• Contextual Information?
• Affective information?
• Think of dimensions of clarity versus vagueness?
• Give examples of what you might actually remember
  about an event.

57
This Q A system of M. Johnson sorta works, but

• What happens as time passes?
• So what do you remember better, last week or when
  you were 3?

58
Actually

• Actually, it turns out that Physiology (ERPs)
  come to the rescue again regarding false
  memories
• as we will see later.

59
If RM is not a perfect false memory detector,
then why..

• would we think this strategy would be a good
  deception detector?
• Are there differences between false memories and
  deceptions?
• HINT What do subjects believe about veracity of
  f.m. and lies?

60
RM Criteria (after Sporer 97) with some overlap
with CBCA

• From Vrij Table 9.1. The first 7 should be there
  in true statements.
• Clarityclear sharp, vivid statements
• Perceptual info.details of smell, sound, etc.
• Spatial info.details of location, seating
  arrangement, object location.
• Temporal Info When things happened, and in what
  order.
• (continued)

61
Four more

• 5. Affect detailHow subject felt I was
  disgusted when I saw the body
• 6. Reconstructability of the testimony in
  detail.
• 7. Realism Is the story plausible, realistic,
  logical?
• 8. Cognitive operations. Should be less in true
  story It shouldnt be necessary to make
  inferences He appeared nervous.

62
The RM method usually involves 2 raters scoring
tape.

• Inter-rater reliability should be and is found by
  research to be high. Good as CBCA, anyway.
• Vrij thinks its easier to learn and teach,
  because
• There are fewer criteria which are more concrete
  and thus less subject to interpretation.

63
Vrijs review of RM research in Appendix 9.1 has
limitations

• Not all reviewed studies are peer-reviewed
  publications.
• Not all studies are given!
• Scoring not standardized.
• Some listings lack inter-rater data.
• Only recent events are studied.
• RM is problematic in children.
• Etc, etc. maybe not important .

64
...because

• Appendix 9.1 shows a mostly mixed pattern.
• Average accuracy is 68.8 overall, which mean
  many false alarms (about 29 in truth tellers)
  and misses (about 34 in liars). I.e., group but
  not individual results.Yes, RM beats flipping a
  coin, but not by much. More work is needed.
  (Physiology wins.)

65

• End of RM

66
Scientific Content Analysis

• SCAN

67
Israeli tool

• Vrij Widely used, though
• Not much researched, nor endorsed by researchers.

68
Basic assumption

• A memory based on actual experience will differ
  in content and quality (?) from an invented or
  fantasized memory. (Duhhhh!)
• Also problematic Not all lies are invented out
  of whole cloth. Some are substituted actual
  memory, as in most alibis.

69
Proceedure

• Suspect writes in handwriting a statement of
  events in absence of investigator (to avoid
  contamination).
• Statement is then scored on several criteria

70
The Criteria ( seen in truth tellers. V means
vs. CBCA)

1. Direct denials
2. Specific identifications
3. Cross-outs(against instructs.) V
4. Vague memories V
5. 20-50-30 structure
6. Emotions
7. Subj/obj time correspondence

71
More.

• 8. Events in chronological order
• 9. Missing info A bit later.. -
• 10. 1st person, present tense
• 11. Pronouns I Ownership.
• 12. Change in language(???) Change from
  conversation to Discussion. - if not
  justified. Too subjective (imho)

72
Research results with SCAN

• 1. Field study .77 TTs, .88 Liars at best with
  Denial criterion.
• .But, as with almost all field studies, no
  ground truth!!!!! So, fuggedaboudid!
• 2. Lab Study(So no ground truth issue) Three
  crirteria utterly failed to discriminate liars
  and truth tellers!! So, fuggedaboudid!

73
Final field study

• Again no ground truth!!
• Though they claim75 TT accuracy, 80 liar
  accuracy.
• Great lack of consistency among SCAN users So,
  fuggedaboudid!
• Bottom SCAN line Little empirical support,
  little standardization.
• Thus the widespread use is stupid.

74
Protocols for Physiologically based Deception
Detection

• Based for openers on Polygraph.

75
Polygraph

• Does not mean lie detector.
• Means a machine with many (poly) channels of
  visually presented (graphical) information in
  inked paper or laptop display format.
• A Polygraph machine can be and is used in many
  fields of Psychophysiology

76
2 Major Protocols currently used

• Comparison (Control) Question Test (CQT)--
  beloved by enforcers, cuz its easy to use
  (Didja do it?) and tends to elicit
  confessionsbut hated by academics who prefer the
• Guilty Knowledge Test (GKT better known as the
  Concealed Information Test (CIT), since knowledge
  cant be guilty (just perps.)
• Both utilize autonomic arousal.

77
Why do us professors frown on CQT?

• Pre-test interview cannot be standardized.
• It compares arousal responses to relevants.
  controls.There can never be true comparison
  control questions in the scientific sense of
  control.(Yet there could be better versions.)
• High false positive rate (which CIA et al. dont
  care about.)

78
Why do we love CIT?

• It compares arousal (?) responses to critical
  (probe) and non-critical (irrelevant) stimuli.
• There are usually 6 or more multiple choice
  questions with 1 5 choices per question. (Throw
  away 1st.) Thus for each item, the P(chance hit)
  1/5.2
• For 6 INDEPENDENT items, probability of subject
  hitting on all 6 by chance P .26 .2 x .2 x
  .2 x .2 x .2 x .2. Use Bernoulli for chance hits
  on lt 6. Thus you can reduce the false positive P
  as low as you like by adding independent items
  (not always so easy).

79
Event Related Potential Averaging..

• and bootstrapping

80
Here is what spontaneous EEG looks like

• ..from 4 sites on the scalp

81
(No Transcript)
82
(No Transcript)
83
Here is a single sweep
84
(No Transcript)
85
(No Transcript)
86
Here are 3 trials.
87
(No Transcript)
88
(No Transcript)
89
Reminder for prof to take

• the class to the moviecalled, ERP Averaging

90
Since averages are so much cleaner than single
sweeps,

• and show the true stimulus-evoked event that is
  time-locked to the eliciting event, and are more
  noise free,
• .it obviously makes sense to compare averages
  rather than single sweeps, that is, to do
  analysis, like t-tests on averages.
• People did that in comparing group ERPs or grand
  averages.

91
For example,

• The schizophrenic group average versus the normal
  average
• or the well-trained group average P300 vs. that
  of the untrained group.
• Remember, in a group, each subject has an average
  ERP.
• .but within a single subject, there are only
  single sweeps to compare

92
In Bootsrapping

• ..the original set of single sweeps is
  repeatedly randomly sampled but with
  replacement
• yielding multiple averages in a single subject.
• Lets say there are 6 repetitions of sampling of
  18 single sweeps

93
Each set of 18 single sweeps is averaged yielding
6 averages

• .that look like real average of original set but
  with variations

94
(No Transcript)
95
Correlation Example Let's assume that we want to
look at the relationship between two variables
Person Height Self Esteem
1 68 4.1
2 71 4.6
3 62 3.8
4 75 4.4
5 58 3.2
6 60 3.1
7 67 3.8
8 68 4.1
9 71 4.3
10 69 3.7
11 68 3.5
12 67 3.2
13 63 3.7
14 62 3.3
15 60 3.4
16 63 4.0
17 65 4.1
18 67 3.8
19 63 3.4
20 61 3.6
96
Finally, we'll look at the simple bivariate
(i.e., two-variable) plot
97
Now we're ready to compute the correlation value.
The formula for the correlation is
98
(No Transcript)
99
(No Transcript)
100
(No Transcript)
101
Remember 3ST protocol

• Probe (P) is the guilty item (stolen item)
• Irrelevants (Ir) are non-stolen items
• Target (T)is a IR with unique response

102
How P300 amplitude is supposed to catch Liars
1)PgtI (BAD) 2)P-TR corr gtP-I corr(BC-AD)
1)PI 2)P-I corr gtP-TR corr