Emotion Recognition Training and Deception Detection: Introduction Method

advertisement
Emotion Recognition Training and Deception Detection:
Importance of the type of lie told
Mircea Zloteanu (mircea.zloteanu.11@ucl.ac.uk)
& Daniel C. Richardson
University College London
Introduction
Method
Emotion cues relating to the true emotions of a deceiver leak out
during deception (Ekman & Friesen, 1969). Research attempting to
use such cues has resulted in poor results. A potential issue is the role
of moderating factors that determine cue production, such as the
stakes to the liar for deceiving. The study investigates the effect of
emotion recognition training and bogus training on accuracy in
detecting lie where the stakes to the deceiver are either low or high.
Participants
106 participants (22 male, 84 female), with a mean age of 20.9
(SD=4.7). Emotion Recognition Training (n=39), Bogus Training (n=38),
and No training (n=29).
Hypotheses
1. Emotion recognition training will result in improved truth detection
but will hinder lie detection, for Low-stakes videos.
2. Emotion recognition training will produce overall improvements in
deception detection in High-stakes video.
3. Training will increase confidence in veracity decisions as well as
affect the veracity bias.
Stakes: the motivation that liars receive to perform well. It relates to
the rewards and punishment of deceiving
Design
3(ERT vs. BT vs. NT) X 2(Lie or Truth) X 2(High or Low stakes) mixed
design. DVs: Truth and Lie accuracy scores, and Confidence.
Stimuli
20 videos (10 lies) low-stakes (BDS; Street, et al., 2011). 20 videos (10
lies) high-stakes. High-stakes had two subsets: Emotional and
Unemotional lies (Warren, Schertler & Bull, 2009).
- Lie type
□ Low-stakes
 not motivated to lie
□ High-stakes
 told performance would be rated
 they could win £10
• In low-stakes lies, emotion cues may be limited and not very intense
• in high-stakes lies, where arousal is higher, cue production should
increase
 (HS) Emotional
 lied while watching negatively arousing videos
 (HS) Unemotional
 lied while watching neutral videos
Results
• Training had no effect on accuracy, confidence or bias.
• Participants had higher accuracy on Low-stakes compared to Highstakes videos.
 METT training task
 SETT training task
70
*
Overall Accuracy %
Procedure
- No training (NT)
- Emotion recognition training (ERT)
- Bogus training (BT)
60
 METT neutral expressions
 3 Blocks: slow (1s), medium (.75s), fast (.5s)
ERT
BT
NT
50
40
+
+
What was the person’s eye colour?
1.
2.
3.
4.
Blue
Black
Brown
Green
30
*p<.05
Low-stakes
High-stakes
Type of lie
• High stakes: Emotional vs. Unemotional
(HS) Emotional > (HS) Unemotional
• Truth accuracy: Low-stakes > High-stakes
• Lie accuracy: Low-stakes > High-stakes
70
50
40
30
70
60
Overall Accuracy %
60
*
Lie detection %
Truth detection %
70
50
40
30
*p<.05 Low-stakes
High-stakes
Stakes
60
50
40
30
20
20
20
*
p=.06 Low-stakes
High-stakes
*p<.05
Emotional
Unemotional
Type of lie
Stakes
Conclusions
 Emotion recognition training has no effect on accuracy, confidence, or bias.
 Emotional information can both hinder and improve accuracy; Low-Stakes > High-Stakes, but Emotional > Unemotional.
 The distinction between suppressed and faked emotions seems to explain the accuracy difference in High-stakes lies.
References: Ekman, P., & Friesen, W. V. (1969). Nonverbal leakage and clues to deception. Psychiatry, 32, 88–106; Street, C.N.H., Tbaily, L., Baron, S., Khalil-Marzouk, P., Wright, K., Hanby, B., & Richardson, D.C. (2011). Bloomsbury deception set.
Paper presented at the Division of Forensic Psychology Conference, Portsmouth, UK; Warren, G., Schertler E., & Bull, P. (2009). Detecting deception from emotional and unemotional cues. Journal of Nonverbal Behavior, 33, 59-69.
Download