Changes in Galvanic Skin Response

1 downloads 0 Views 1MB Size Report
Comparisons of the data set were performed using the Student's t-test and bootstrap randomization tests. .... Liar Liar, Pants on Fire! A. Physiological Study of ...
RUNNING HEAD: Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Aabid Shamji Grinnell College



1

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

2

Introduction Although the ruled in 1998 that polygraphs were inadmissible in court, the US department of Justice has used the test approximately 12,000 times in 2014 (Gershman, 2015). A polygraph is a test that measures heart rate, respiration, blood pressure and skin conductivity while the subject is being asked specific sets of questions. The uses of polygraphs range from criminal and civil cases, to pre-employment screening to applications concerning homeland security and counterterrorism amongst others. While they are inadmissible in courts, they still do have an impact of people’s lives in terms of job opportunities, investigations and security. Thus, polygraphs are an important field of study in psychology (Engler et al, 2011). This paper will focus on one physiological response measured in polygraphs; skin conductivity. We will measure changes in skin conductivity as changes in Galvanic Skin Response (GSR).

GSR is variation in the electrical characteristics, such as resistance and conductance, of the skin. The skin varies in resistance because due to the activity of the sweat glands which are mostly controlled by a branch of the autonomic nervous system called the sympathetic nervous system. Skin conductance is accepted to be an indication of both psychological or physiological arousal. The flight-or-flight phenomena is provoked when the stressor in the environment is experienced and catecholamines released from adrenal gland to produce the appropriate physiological response to the stimuli. Sweat gland activity increases when the sympathetic branch of the autonomic nervous system is aroused (Carlson, 2013) and thus can provide a measure of emotional and sympathetic responses.



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

3

Therefore, I would hypothesize that the GSR response of the subject would increase when they are lying as compared to when they are telling the truth because at that point they would not be consciously controlling their sympathetic nervous system. Changes in the autonomic response would have been prompted as a result of the changing emotional state (CleveLabs, 2011), thus lying would provide a higher change in GSR.

Methods Students in an into psychology class were placed in groups of three. Each group assigned one member to be the study subject, one to record the question times and the last to be the experimenter. Both the experimenter and the subject were given separate color-coded envelopes. The experimenter was instructed to make sure that the participant was seated comfortably and facing away from the monitor. The subject was instructed to wipe the GSR plates with an alcohol wipe and wipe their fingers with a hand wipe to get off all the oils so that we get more accurate GRS readings. Inside the subject’s envelope there was a card instructing them to answer all questions with a ‘yes’ or ‘no’, look at the recorder when answering and answer all questions about a certain topic (i.e. food, music or animals) wrong. The experimenter was then instructed to make sure that the software was running and completely configured to the default protocols. The software used was ToneTrend with Optical Isolation (bio-medical, 2007 edition). The subject was then placed two fingers on each of the GSR plates on the GSR2 Biofeedback Relaxation System (bio-medical) and the recorder began taking note of the times. The experimenter then asked the subject a series of predetermined questions twice in the two different prescribed orders waiting for each response to play out before asking the next. Each time the experimenter asked a question, they marked the graph just before they began the



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

4

question. The recorder also recorded the time of the mark. After the questioning, the software was paused and the graph adjusted to for clarity and axis manipulation to ensure that the entire wave was on the graph. The graph was then printed and the changes in GSR was computed by subtracting the lowest point on the wave from the highest within the given question interval. The process was repeated twice with two distinct set of questions and every group member playing a different role each time.

Statistical Analyses Comparisons of the data set were performed using the Student’s t-test and bootstrap randomization tests. Bootstrap randomization tests of difference in mean were completed using 1000 simulations and random base generator and were conducted due to the relatively small sample size. P values < 0.05 were considered statistically significant. Minitab 17 Statistical Software (Penn State College, 2010) was used for these analyses.

Results Figure 1 shows the difference between GRS when subject 2 was telling a lie and when they were telling the truth. By looking at the bars alone, there seems to be a greater change in GSR when the subject was lying and when they were telling the truth. There is a difference of 192 between telling a lie and telling the truth. However, using p-value generated from the students t-test (p=0.1546) and bootstrap randomization (p=0.1510) (Appendix B) under the alternative hypothesis that the change in GSR would be higher when lying indicating that there is no statistical significance between the two means and we fail to reject our null hypothesis. Thus,



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

5

this data does not support our original hypothesis as it indicates that we can say with 95% accuracy that there is no difference in GSR when telling a lie and telling the truth.

Looking a figure 2, there seems to be the opposite pattern. The bars on figure 2 indicate that there may be a higher GSR difference when telling the truth. The difference in the average here is -127.95. However, as with the first case, the extremely large p-value generated from the student t-test (p=0.8584) (Appendix C) under the alternative hypothesis that the change in GSR would be higher when lying indicates that there is no statistical significance between the two means and we also fail to reject our null hypothesis in this instance. This this data also does not support our initial hypothesis.

The data for the 3rd subject, shown in figure 3, tells a very mixed story. Looking at the bars alone, the first round showed that GSR was higher when telling the truth and the second round showed the opposite; that GSR was higher when telling a lie. This put the average difference between the two to be 86.25. However, as with the previous two, the large p-value generated from the student t-test (p=0.3892) (Appendix D) under the alternative hypothesis that the change in GSR between the two groups in not equal indicates that there is no statistical significance between the two means and again we also fail to reject our null hypothesis. This this data also does not support our initial hypothesis.

In a last-ditch attempt to compare the means, I combined all the data (figure 4) and ran another t-test which also showed not statistical significance between the two group means (p=4003) (Appendix E).



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

6

Discussion Although on the surface some of these finding seem to have confirmed out hypothesis, none of the data was statistically significant and thus we are left to conclude that there is not statistical difference between the two group means. This means that there is no significant difference in the change in GRS when subjects were telling the truth or a lie.

This study employed a compressive method that allowed adequate test of this hypothesis by having repeated the set of questions that were asked. This ensures that the ordering of the questions did not confound the change in GSR and that we had two values for each question asked so that our data had more breadth. However, the sample size that was used may have been too small to draw accurate inferences. This created a small power of approximately 0.3842 and consequently a higher change of a type II error. To fix this, next time we would use a larger sample size in our analysis by combining all the data from everyone in the class. Moreover, the questions that were asked in this study had little to no significance on the lives on the subjects and thus they did not have higher alterations in stress levels and emotions to cause a larger change in GSR when lying. For instance, lying about your favorite pizza topping would have less emotional stress then a criminal lying about committing a crime. Therefore, the low stake in this experiment would result in lower changes to GSR. Thus, it would be advisable to ask more personal questions that the subject would identity with and would put in more effort to deceive the experimenter. Lastly, analyzing the area under the curve could help us to see if the subject took longer to return to baseline when lying. This could also help prove our hypothesis.



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

7

Additionally, our initial hypothesis could also be flawed because although lying is can be stressful and thus cause changes to the autonomic nervous system and thus the increase autonomic arousal could have larger GSR changes, these changes could also have been caused by increased nervousness in using the technology or not being believed. Moreover, because it everyone’s level of autonomic arousal is different, it would be hard to find an absolute measure of a presence or absence of a lie (Gazzaniga et al, 2016). Thus, autonomic stress does not necessarily mean that a person is lying or is guilty. In his article entitled “Polygraph and integrity tests. Current Directions in Psychological Science” Leonard Saxe (1994) makes the interesting conclusion that the aim of these physiological measurements is to trick the subject into believing that they can detect deception in a hope that that would make them more honest. Thus, calculating GSR would be an effective way of increasing honesty my triggering other physiological responses to the environment that the subject is in.



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

8

References Carlson, N. R. (2013). Physiology of behavior. Boston: Pearson.

CleveLabs. (2011). CleveLabs Laboratory Course, Biomedical Engineering and Health Science Course. Retrieved September 17, 2017, from http://www.clevemed.com/ CleveLabs/overview.shtml

Engler, L., Lloyd, W., Martin-Koob, J., & Naze, S. (2011). Liar Liar, Pants on Fire! A Physiological Study of Deception. University of Wisconsin--Madison. Physiology, 435.

Gazzaniga, M. S., Heatherton, T. F., & Halpern, D. F. (2016). Psychological science. New York, NY: W.W. Norton & Company, Inc.

Gershman, B. L. (2015). Lie Detection: The Supreme Court's Polygraph Decision.

Saxe, L. (1994). Detection of deception: Polygraph and integrity tests. Current Directions in Psychological Science, 3(3), 69-73.



9

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Appendices Appendix A: Figures 700.00 600.00

Change in GSR

500.00 400.00 300.00 200.00 100.00 0.00 Round 1

Round 2 Truth

Average

Lie

Figure 1: Comparison of GRS when telling the Truth and When Lying for Subject 2 (green). Error bars represent SEM, n = 9, p = 0.1510 800 700

Change in GSR

600 500 400 300 200 100 0 Round 1

Round 2 Truth

Average

Lie

Figure 2: Comparison of GRS when telling the Truth and When Lying for Subject 1 (blue). Error bars represent SEM, n = 9, p = 0.8584



10

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying

900.00 800.00 700.00

Change in GSR

600.00 500.00 400.00 300.00 200.00 100.00 0.00 Round 1

Round 2 Truth

Average

Lie

Figure 3: Comparison of GRS when telling the Truth and When Lying for Subject 3 (yellow). Error bars represent SEM, n = 9, p = 0.3892 700.00 600.00

Change in GSR

500.00 400.00 300.00 200.00 100.00 0.00 Round 1

Round 2 Truth

Average

Lie

Figure 4: Comparison of GRS when telling the Truth and When Lying for All Subjects. Error bars represent SEM, n = 9, p = 0.4003



Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Appendix B



11

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying



12

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying



13

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Appendix C



14

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Appendix D



15

Changes in Galvanic Skin Response (GSR) Under the Stress of Lying Appendix E



16