Facial skin surface temperature changes during a "concealed information" test.

When individuals who commit a crime are questioned, they often show involuntary physiological responses to remembered details of that crime. This phenomenon is the basis for the concealed information test, in which rarely occurring crime-related details are embedded in a series of more frequently occurring crime-irrelevant items while respiratory, cardiovascular, and electrodermal responses are recorded. Two experiments were completed to investigate the feasibility of using facial skin surface temperature (SST) measures recorded using high definition thermographic images as the physiological measure during a concealed information test. Participants were randomly assigned to nondeceptive or deceptive groups. Deceptive participants completed a mock-crime paradigm. A focal plane array thermal imaging radiometer was used to monitor SST while crime-relevant and crime-irrelevant items were verbally presented to each participant. During both experiments, there were significant facial SST differences between deceptive and nondeceptive participants early in the analysis interval. In the second experiment, hemifacial (i.e., "half-face" divided along the longitudinal axis) effects were combined with the bilateral responses to correctly classify 91.7% of participants. These results suggest that thermal image analysis can be effective in discriminating deceptive and nondeceptive individuals during a concealed information test.

Comparison of polygraph data obtained from individuals involved in mock crimes and actual criminal investigations.

In a preliminary attempt to determine the generalizability of data from laboratory mock-crime studies, the authors examined the similarities and differences among the cardiovascular, electrodermal, and respiration responses of deceptive and nondeceptive individuals elicited to crime-relevant and crime-irrelevant questions. Participants in the laboratory group were randomly assigned to nondeceptive (n = 28) or deceptive (n = 27) treatment groups, and a mock-crime scenario was used. The field participants were confirmed nondeceptive (n = 28) or deceptive (n = 39) criminal suspects who underwent polygraph examinations between 1993 and 1997. The results indicated that there were salient differences between field and similarly obtained laboratory polygraph response measures. However, accuracy of laboratory participants' classifications using logistic regression analysis was not significantly different from field participants' classification accuracy.