A New Approach to Detecting Deception Using Learning Theory
Abstract
The scientific literature on the detection of deception indicates that the use various physiological signals and testing approaches such as the guilty knowledge, or control question tests, yield results better than chance though lacking in sensitivity, specificity, and resistance to countermeasures (Committee to Review the Scientific Evidence on the Polygraph, 2003, "The polygraph and lie detection." Washington, DC: National Academy Press). Recent approaches that use brain imaging and other new technologies still rely on the emergence of a natural lie response that is presumed intrinsic to all people. While some people do intrinsically emit anxiety during deception, data do not support the ubiquitous nature of such a response. While serving on the National Academy of Sciences Committee to review the scientific evidence for the validity of the polygraph, we developed an alternative analytic approach to the detection of deception. The approach differs from previous approaches in two fundamental ways. First, we proposed to use Pavlovian conditioning techniques to instill a unique but innocuous physiological response (e.g., a micro-eye blink) when they are exposed to an untrue statement. Second, we proposed to develop a sensitive and specific digital signal processing algorithm for each person individually based on the pattern (e.g., timing, frequency components, symmetry across the right and left ocular regions) of responses that best discriminated that individual's perception of a true (e.g., I kick a ball with my leg") versus untrue (e.g., "I kick a ball with my arm")statement. If no such response template is found, evidence is secured that one cannot test for deception. If signal detection analysis suggests a response template is apparent, this template is used to evaluate whether subsequent test items (e.g., "I was born in June") are true or untrue. (Test items are personally relevant questions for which we have ground truth.)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 31, 2006
- Accession Number
- ADA455533
Entities
People
- John T. Cacioppo
Organizations
- University of Chicago