 Hello and welcome to this webinar from Converas. Today we will speak about deception detection technologies available on the market. Lying is, in a pure physiological sense, an unnatural act. For that reason, there are many technologies available. There have been many attempts to detect deception. We're going to talk about, in this webinar, an overview of lying, some other deception detection methods, traditional methods, up-and-coming methods, and then perhaps the impact of corruption and deception on our companies, organizations, and economies. Doctors Kirchner and Raskin and Hacker postulated that you could detect deception in someone by observing their eyes because lying requires more cognitive effort and an increase in cognitive load is manifested in eye behaviors. Why do we lie and how do we lie? Sometimes we lie to protect others. Sometimes we lie to protect ourselves, to enhance a story, to prevent embarrassment, to benefit someone or, excuse me, we lie to hurt others. There are those who are self-serving liars. There are those who are pathological, who are sometimes very persuasive. The dissociative liar has a departure from reality. And then the compulsive liar, the one who is habitual or chronic. Regardless of the type of liar or the purpose for lying, lying does increase cognitive load and certain deception detection technologies will detect that. There are physiological changes that occur when people lie. If you're familiar with polygraph, you know that there are physiological changes, such as heart rate, blood pressure, respiration, and even skin conductance. If you're familiar with eye detect, you know that there are changes in pupil dilation, eye fixation, cognitive load. If you're a user of CVSA, you know that voice fluctuation tends to be affected by lying. In all of these detection deception technologies, response rate is also affected. So, in essence, all of these things are measurements of emotional arousal. There are certain types of detection deception tests today, those for pre-employment. Typically in the United States, those are used for law enforcement or other federal or state positions. There are periodic examinations, which are done on current employees. And then the event specific, which are common to crime investigations, where there's a specific issue, one issue, and an issue that you want to discuss and discover. We're going to talk about right now seven different deception detection methods, ranging from intuition, interrogation, or observation, personality, or integrity test, polygraph, voice stress, brain scanning, and even ocular motor, which is eye detect. Today's technologies either predict the future to determine if you're going to do something in the future, or to identify the past. Those that identify the past intend to be objective to just get the facts about something that already has occurred. Intuition is a method commonly used by many organizations where you rely on interviewing candidates for positions to determine if they're lying. But studies say that humans have about a 54% accuracy rate in catching a liar based on a conversation. That's about as effective as flipping a coin. Now there might be one possible exception, which is if you know someone really, really well, you might have the ability to determine if they're lying. But in most cases, that's not the case, most organizational cases. Things, behaviors such as gaze aversion or touching your body or your face or covering your eyes or mouth while speaking really haven't found to be a reliable indicator of deception. And despite popular belief, there really are no nonverbal deception cues, especially in those who are expecting you to observe them for those cues. So you're back to the accuracy rate of 54%. There are personality tests out there, integrity tests. Personality tests really just try to determine if you are fit to work with a company or with a certain team or with others. Really is there a behavioral fit? Dr. Raskin did a review of this metadata analysis which was conducted in 2012 where they looked at over 100 different studies that looked at the validity and accuracy of integrity tests. So there's been a metadata study of integrity tests. Over 104 I just mentioned. From test publishers, those who have a vested interest in selling their test, and from non publishers. And in those tests, they try to determine if those tests could be used to determine job performance, training performance, whether or not you'll have counterproductive behavior such as substance abuse or theft and even employee turnover. From Dr. Raskin's review of this metadata analysis, really the accuracy rate of integrity tests are not very good. These are the validity estimates for job performance, training performance, counterproductive work behavior where you self-report, where employee records indicate what you've done and in terms of turnover. So integrity tests attempt to predict certain behaviors. Some from substance abuse to theft withdrawal. And they're moderately accurate. And really if employee records are kept, those are probably a better indicator of these counterproductive behaviors. They attempt to predict job performance, training performance, and turnover. But in this metadata study, their performance or accuracy was very poor in that regard. So in reality, integrity tests don't have a very high rate of accuracy for predicting the future. So they have a low predictive value. Self-reporting or attitudes are easily faked. I mean, if I were to ask you, hey, what's your worst personality trait or behavior? You're going to tell me that you're a workaholic and that causes you to get stressed out, but you're also very productive and people can do or say what they believe you want to hear. So these integrity tests really don't directly assess many disqualifying factors. And they're extremely difficult to validate for topics that could be assessed better by other measures or methods. Polygraph is another method which is used commonly today for detecting deception. Many of you know about polygraph. Many of you are polygraph examiners. Polygraph has been around and been used in the US since the 1930s. It's a measure of cognitive load, but more of a measurement of emotional arousal. You know that polygraph measures respiration, pulse, blood pressure, skin conductance. And according to the APA, the American Polygraph Association, when they conducted and published their metadata analysis in 2012, for event-specific questioning across a broad variety of topics, formats, sorry, formats, techniques, and examiners, polygraph can have a very high level of accuracy for event-specific questioning. When you talk about more general tests for pre-employment with multiple issues or periodic tests with multiple issues, the range in accuracy was a bit lower than specific event. Sometimes we've heard stories of the interpersonal interaction between the examiner and the examiner affecting the outcome because of how the examiner treated the examiner or how effective the examiner is in his or her skills or abilities at interpreting data. So there is a level of subjectivity is the point here. And someone who has a liar who's concerned about being believed often will come across as helpful and truthful in an interview simply to impress the interviewer. These are people who are perhaps accustomed to lying and they've practiced. And so sometimes it's difficult to even be able to detect a lie in this person like that, especially if they're trying to control their physiological reactions. And at times an innocent person who is under stress, feeling anxiety will manifest some of these typical lying behaviors simply because of those feelings being stressed. They might fidget, their speech might be slurred, they might not look at you directly, and this is really only related to their discomfort with the situation, not about their behavior. Voice stress analysis is a technology available on the market for a number of years. The basis for the method is that when you lie and you're speaking, your voice box will either tighten or loosen, which will change the sound of your voice. They're looking for small involuntary frequency modulations in the voice when under stress, basically. From the studies that we've seen from Hollyanne and Harnesberger in 2008, the true positive rates for voice stress range from 50 to 65 percent, and the false positive rate was just as high and often higher. In essence, the CVSA system operates according to Hollyanne and Harnesberger, operates at about a chance level. Other sources that drew the same conclusion, simply not just Hollyanne and Harnesberger, but there have been a couple of others from the Oklahoma Department of Mental Health, the National Research Council, etc. Let's talk for a moment about brain scanning, the electroencephalogram. As you can see from the photos, the electroencephalogram requires that electrodes be attached all over the head, that specific equipment be used, that a trained technician be monitoring the person. And it's been found that electroencephalogram measures familiar stimuli, unfamiliar stimuli, and probes on topics of truthfulness. So by measuring familiar stimuli, it creates a positive baseline, like when you see a friend. Unfamiliar stimuli can create a negative baseline when you see a stranger. So brain scanning is looking for these types of changes in brain function. Research has shown that the electroencephalogram can be as effective as 87 percent. The problem is it's very invasive, takes a long time, requires specialized equipment, and a specialized examiner. So it's too costly and very invasive to really be something used at mass production. Also talking about brain scanning or other types of scanning, the MRI, the Magnetic Resonance Imaging Device or Equipment, measures real-time, or creates a real-time 3D model of the body. So the MRI method is to create a whole brain map, to looking for pattern vectors trying to predict the cognitive state of the brain while a discussion is being had or an interview is conducted. This requires a radiologist, it requires a multi-million dollar piece of equipment, it can be highly accurate, but it's costly. It might be best, according to the data that we've seen, best for a pool of subjects, not for individuals or a specific case. There really isn't standardization, however, and it is a bit invasive. It is fairly accurate, according to the National Academy of Sciences, but again, costly, requires a radiologist, requires specialized training, may or may not be a good deception technology for mass production today. Then we get into the ocular motor deception detection test, which is what I detect is. Kirchner and Raskin sought for a method of deception, a new method for detecting deception. You know that Kirchner and Raskin developed the computerized polygraph in the early 90s, but they were looking for another way to monitor and check and determine deception based on other physiological things, and in this case, they found by their research and through observation that the eye exhibits involuntary changes when lying because of an increase in cognitive load. In the case of eye detect, an examinee's eyes are scanned, subtle changes are detected in movement, fixation, dilation, blink rate, the infrared camera or the infrared eye tracker shown at the bottom of the screen in the picture here, takes 60 measurements per second of the eye, and there are measurable increases in cognitive load when a person is lying, and that's basically the premise behind eye detect. These results were published in 2012 in the Journal of Experimental Psychology Applied. According to the study that they did that they published and wrote about, this is basically a graph, a line graph of pupil diameter response, basically changes in pupil dilation. On the left you see a graph of guilty subjects, on the right you see a graph of innocent subjects. The red line indicates the change in pupil dilation over time after asking a guilty person, again the line graph on the left, asking a guilty person about their behavior, knowing in a lab study, knowing which people or subjects are guilty, knowing which subjects are innocent. The pupil dilation was increased or higher in guilty subjects asked about the crime they committed. On the right you see that the red line is basically about the same amplitude and shape as the yellow line. The yellow line indicates pupil dilation when asking a subject about crimes that they did not commit. Again in a lab study you know which persons have committed the crime and which haven't. So you can measure, you do have absolute truth in these types of studies. So in essence pupil dilation increases when a person is lying. Some of the challenges with eye detect you must be able to read. It is a standardized test so if you want to customize tests it requires some modification. We do design tests based on cultural and local use of vocabulary based on topics as well and there are a few relevant issues. There are in a typical eye detect test you can have maybe three or four relevant issues. Now it is highly accurate but you have a broader base of questions to be asked. The ocular motor deception test ODT or eye detect as it's commercially called began to be developed in 2002 and has been under development for the last 14 years. There is ongoing development ongoing enhancement. In this case in the case of eye detect various eye characteristics are measured. The camera measures at 60 measurements per second. Responses are recorded, averaged, an algorithm is used to correlate and score the changes in our response and compared to true and false responses to the questions asked and it is has been published to be 85% accurate. It's not intrusive because there are no sensors attached to the subject. The test duration is about 30 to 35 minutes. Tests are scored within about 5 to 10 minutes. This chart was created by Dr. David Raskin. It is an attempt to compare eye detect polygraph, voice stress, brain scanning, behavioral observation, integrity test and linguistic analysis which we didn't talk about today. The green squares indicate a positive characteristic. Yellow indicates kind of a neutral characteristic, red is a negative characteristic and gray means we're not sure. On the left you see some of the characteristics ranging from accuracy, the potential range of application, the cost of the test, the duration of the test, if you can control the settings, how much training is required, if you can automate the test, if the decisions or scoring of the test can be automated, if the test is objective, how much does it cost overall, if you attach sensors, if there are any counter measures that can be applied and if language can be flexible in administering the test. As you can see from this, it's a subjective measurement of course because these are not numbers that we're looking at. We're looking at opinions from Dr. Raskin, but eye detect as you can see has more green squares or rectangles than any other method here based on these characteristics. Why do we need deception detection technologies? Because crime and corruption are rampant in the world. Even the United States finance industry could spend about 10 billion annually to combat money laundering, that's according to City Corp. Employees tend to steal from the employers, 50 billion dollars in the US is stolen by employees. The figure I like here on this chart is that 75% of all employees have stolen at least once from their employer. That's a very stark fact. Who steals the most? Men typically do, and those who have college degrees or high school degrees tend to steal more. The greatest obstacle to economic and social development is corruption and fraud. Corruption costs about 5% of global GDP or gross domestic product. In Russia corruption consumes about 44% of GDP, which is a horrible statistic. What is eye detect? As I mentioned in 2012, a formal peer reviewed study indicate that the technology works, that you can measure deception based on changes in cognitive load, which are manifested in eye behavior. The subject sits in front of a screen answering true and false questions while their eyes are being scanned and measurements are being taken. The data along with the true and false responses are uploaded to the cloud after being encrypted on a hard drive and then uploaded to the cloud, and proprietary algorithms in that web server will take all of those measurements, which are about 90,000 measurements in a 30 minute period along with the true and false responses and a score is produced. This webinar today was intended to give you a basic overview of a variety of different deception detection technologies on the market. This is intended to help you become more of an expert in a variety of deception detection methods so that you can talk more openly and freely about