Deception Detection Part One: Behavioral and Biometric Tools

As an intelligence officer dealing with the legal and effective conduct of intelligence interrogations, I was more than once involved in discussions regarding various behavioral or biometric analysis techniques for detecting deception, such as voice or handwriting analysis, neurolinguistic programming, kinesiology, polygraph, etc. There was and is no uniform agreement among the intelligence interrogation professionals with regard to the utility of such techniques.  After an examination of the available data many of us had significant issues with including such techniques in our training and policy related to intelligence interrogation. For example:

– The polygraph test is the only deception detection apparatus with any significant attempt at empirical evaluation. Some assert success rates above 90% for the successful employment of the polygraph to detect deception. A study published by the National Academies Press challenges this claimed success rate. The discussion is lengthy and detailed, and the final analysis indicates that current attempts at empirical testing have not been sufficiently controlled to provide accurate assessments of utility. The problems in design and control notwithstanding, attempts at empirical research have failed to provide rates of accuracy even remotely approaching the 90th percentile. The folks who posit a 90+% rating on polygraphs are those who either perform polygraphs or sell polygraph equipment for a living. As a technical point, the polygraph does not measure deception, but physiological responses to questions. There is a huge amount of subjectivity in the interpretation of these reactions. The lack of reliability in results is why polygraph results remain inadmissible in court. However, the tool can be effectively implemented as part of an approach strategy to attempt to elicit a confession, which is how the tool is generally employed for law enforcement interrogations. In some cases a guilty-minded individual might be convinced to confess to something on the basis of a reading, but the machine cannot be relied upon to give an accurate reading on its own.

– Neurolinguistic programming (NLP) is a very broad subject, encompassing many individual sub-disciplines related to influencing or interpreting behavior. There are studies supporting certain aspects effectiveness, but not necessarily in detecting deception. Much of the available research has to do with evaluating treatment of mental disorders or hypnotherapy. (Hypnotizing a detainee into cooperation would clearly constitute coercion, and the implication of implanting suggestions would undermine the veracity of any information obtained.) When discussing NLP in the context of interrogation and deception, most often there is reference to a kinesiologic connection between certain eye movements and a uniform way in which humans access memory (looking up/left means “X” and looking down/right means “Y”). There is a study that specifically refutes the hypothesis that eye movements have a direct link to ongoing cognitive functions.  [Burke DT, Meleger A, Schneider JC, Snyder J, Dorvlo AS, amd Al-Adawi S. (2003), Eye-movements and ongoing task processing. Perceptual and Motor Skills 96, 1330–1338.]  There are several studies indicating that NLP strategies such as mirroring verbal and non-verbal cues can help foster rapport, but that is about as far as its utility will go in terms of interrogation. There could be utility in understanding an individual’s baseline physical response to questioning to assess for deviations (as will be discussed with Reid below), but otherwise there is no guarantee that a specific action means that a person is or is not lying.

– The Laboratory for Scientific Interrogation (LSI) is a company that markets a line of deception products and training called Scientific Content Analysis (SCAN).  SCAN focuses largely on cues found in written statements that indicate deception.  I have found no empirical study validating SCAN as an effective means of detecting deception.  LSI posits only anecdotal evidence, and little of that.  Out of the few specific behavioral analysis tools that I examined this one appears to have the least reinforcing data. That is not to say that a written statement could not be used in an effort to elicit a confession from a detainee if the written statement appears to be falsified (whether predicated on SCAN methodology or just logic), but it cannot be relied upon as an accurate way to detect deception in and of itself. People with literacy or cognitive issues or people under duress are highly likely to provide deception indicators per SCAN protocols while being completely honest.

– The Reid Technique of Interview and Interrogation is the product of John E. Reid & Associates Inc.  The focus is on establishing behavioral baselines during an interview to spot deception.  There is again no empirical evidence to support the effectiveness of this technique, though the web site offers a number of case studies and boasts a 99% rate of confessions obtained using the technique admitted in court. The rate of admitted confessions and the number of convictions from confessions can be a compelling metric; however, this fails to account for people who were lying and did not confess, and who did not go to trial on the strength of other evidence. Similarly, it excludes those who did not confess, but did go to trial and were not convicted, as well as those who did not confess and were later convicted anyway. By narrowly defining the parameters for evaluation Reid is able to provide a high and technically true metric that is simultaneously very misleading. Certain techniques taught by Reid are still useful tools for approach strategies in intelligence interrogation and for eliciting confessions for law enforcement; it just can’t be relied upon as effective for detecting deception in and of itself, as is the case with all of the other available techniques.

Ultimately, the inability to know for certain whether someone is lying without their admission or other evidence to indicate such is the key obstacle to objectively evaluating deception detection techniques. That does not mean that these techniques have no utility whatsoever. As indicated, all can be used as facets of a well-developed questioning strategy for identification and exploitation of potential areas of deception. The larger point is that, absent an admission or additional corroborating evidence, one cannot be certain of deception solely on the strength of a deviation from a known (or assumed) baseline.

The techniques mentioned herein, and other tools for detecting deception not specifically addressed, are traditionally marketed toward law enforcement efforts as opposed to intelligence interrogations. As noted, their primary purpose is assisting the interrogator in detecting deception to obtain confessions that will later be admissible in court.  This is not in and of itself supportive of intelligence interrogation efforts, as the purpose of intelligence interrogations is not confessions or convictions but information. Nonetheless, many of these techniques can be effectively employed as control measures, or incorporated into an ongoing approach strategy, as long as the intelligence interrogation is aware that the mere act of identifying indicators of deception will not result in information.

I have long advocated that reference to deception detection methods such as those described here for the purpose of intelligence interrogations should be kept minimal, and always focused on the context of how to leverage a deception indication into a strategy for answering an information requirement. Though it is made explicit in the Reid training, the interrogator must remember that for each of these techniques the establishment of a comprehensive baseline assessment of each individual subject is required for best effect. It is also important within the context of intelligence interrogations, which are strictly bound by federal legislation, executive order, and military regulations, that discussion of and instruction in these techniques should not distract intelligence interrogators from the doctrine on which they have been trained, or from the objective of intelligence interrogation – answering information requirements. Our recent ventures into the Middle East have in some cases led to a blurring of lines between intelligence interrogation and determinations of guilt or innocence. For intelligence gathering, whether or not a detainee is guilty of anything is immaterial to his ability to provide useable information. In that respect, understanding when and how a detainee is being deceptive is useful to know, but it is often not necessary to convince him to confess his lie. If we, as interrogators, employ a well-developed questioning strategy and use proper controls to identify misleading information we are more likely to get the information we need, even if the detainee maintains innocence to a specific role or activity.

Our next segment on deception detection will deal with effective strategies to employ during questioning to evaluate the veracity of the detainee’s information, and to help redirect a deceptive detainee toward compliance (without coercion). Stay tuned!