Deception detection based on neuroimaging: Better than the polygraph?

Two Cases of Malingered Crime-Related Amnesia

Amnesia is a frequent claim in major crimes, and it is estimated that the complete or partial absence of memory following a crime ranges from 25% to 50% of total cases. Although some cases may constitute a genuine form of amnesia, due to organic-neurological defects or psychological causes, and possibly combined with a dissociative or repressive coping style after an extreme experience, malingering is still fairly common in offenders. Therefore, one of the main goals in medico-legal proceedings is to find methods to determine the credibility of crime-related amnesia. At present, a number of lie and memory detection techniques can assist the forensic assessment of the reliability of declarative proof, and have been devised and improved over the past century: for example, modern polygraphs, event-related potentials, thermal imaging, functional magnetic resonance imaging, kinematic, and facial analysis. Other ad hoc psychological tests, such as the so-called Symptom Validity Test (SVT) and Performance Validity Test (PVT), as well as the autobiographical Implicit Association Test (aIAT), can also be used. To date, however, there is little evidence or case reports that document their real usefulness in forensic practice. Here, we report two cases of crime-related amnesia, whereby both defendants, who were found guilty of homicide, appeared to exhibit dissociative amnesia but where the application of SVTs, PVTs, and aIAT detected a malingered amnesia.

Modeling and Comparing Brain Processes in Message and Earned Source Credibility Evaluation

Understanding how humans evaluate credibility is an important scientific question in the era of fake news. Source credibility is among the most important aspects of credibility evaluations. One of the most direct ways to understand source credibility is to use measurements of brain activity of humans who make credibility evaluations. This article reports the results of an experiment during which we have measured brain activity during credibility evaluation using EEG. In the experiment, participants had to learn source credibility of fictitious students based on a preparatory stage, during which they evaluated message credibility with perfect knowledge. The experiment allowed for identification of brain areas that were active when a participant made positive or negative source credibility evaluations. Based on experimental data, we modeled and predicted human source credibility evaluations using EEG brain activity measurements with F1 score exceeding 0.7 (using 10-fold cross-validation). We are also able to model and predict message credibility evaluations with perfect knowledge, and to compare both models obtained from a single experiment.

What to Believe? Impact of Knowledge and Message Length on Neural Activity in Message Credibility Evaluation

Understanding how humans evaluate credibility is an important scientific question in the era of fake news. Message credibility is among crucial aspects of credibility evaluations. One of the most direct ways to understand message credibility is to use measurements of brain activity of humans performing credibility evaluations. Nevertheless, message credibility has never been investigated using such a method before. This article reports the results of an experiment during which we have measured brain activity during message credibility evaluation, using EEG. The experiment allowed for identification of brain areas that were active when participant made positive or negative message credibility evaluations. Based on experimental data, we modeled and predicted human message credibility evaluations using EEG brain activity measurements with F1 score exceeding 0.7.

Whom to Believe? Understanding and Modeling Brain Activity in Source Credibility Evaluation

Understanding how humans evaluate credibility is an important scientific question in the era of fake news. Source credibility is among the most important aspects of credibility evaluations. One of the most direct ways to understand source credibility is to use measurements of brain activity of humans performing credibility evaluations. Nevertheless, source credibility has never been investigated using such a method before. This article reports the results of an experiment during which we have measured brain activity during source credibility evaluation, using EEG. The experiment allowed for identification of brain areas that were active when a participant made positive or negative source credibility evaluations. Based on experimental data, we modeled and predicted human source credibility evaluations using EEG brain activity measurements with F1 score exceeding 0.7 (using 10-fold cross-validation).

Look Who’s Talking: Modeling Decision Making Based on Source Credibility

Understanding how humans evaluate credibility is an important scientific question in the era of fake news. Source credibility is among the most important aspects of credibility evaluations. One of the most direct ways to understand source credibility is to use measurements of brain activity of humans who make credibility evaluations. Nevertheless, source credibility has never been investigated using such a method before. This article reports the results of an experiment during which we have measured brain activity during source credibility evaluation using EEG. The experiment allowed for identification of brain areas that were active when a participant made positive or negative source credibility evaluations. Based on experimental data, we modelled and predicted human source credibility evaluations using EEG brain activity measurements with F1 score exceeding 0.7 (using 10-fold cross-validation).

How good are we at detecting deception? A review of current techniques and theories

The task of discerning truth from untruth has long interested psychologists; however, methods for doing so accurately remain elusive. In this article, we provide an overview and evaluation of methods of detecting deception used in the laboratory and the field. We identify and discuss three broad approaches to detecting deception: measurement of non-verbal behaviour, verbal interview methods, and statement evaluation by humans and computers. Part of the problem in devising good methods for detecting deception is the absence of a sound understanding of deception in human lives. We thus consider three theories of deception – leakage, reality monitoring, and truth-default – and conclude that although promising, they do not yet provide an adequate foundation. We review 10 extant methods of detecting deception in the second part of the article, focusing at greatest length on the most widely used method in South Africa, the polygraph test of deception. Our conclusion is that non-verbal methods that work by inducing anxiety in interviewees are fundamentally flawed, and that we ought to move away from such methods. Alternate methods of detecting deception, including statement analysis, are considered, but ultimately our view is that there are currently no methods sufficiently accurate for practitioners to rely on. We suspect that a precondition for developing such measures is a coherent and validated theory.

Do impression management and self-deception distort self-report measures with content of dynamic risk factors in offender samples? A meta-analytic review

Self-report measures provide an important source of information in correctional/forensic settings, yet at the same time the validity of that information is often questioned because self-reports are thought to be highly vulnerable to self-presentation biases. Primary studies in offender samples have provided mixed results with regard to the impact of socially desirable responding on self-reports. The main aim of the current study was therefore to investigate-via a meta-analytic review of published studies-the association between the two dimensions of socially desirable responding, impression management and self-deceptive enhancement, and self-report measures with content of dynamic risk factors using the Balanced Inventory of Desirable Responding (BIDR) in offender samples. These self-report measures were significantly and negatively related with self-deception (r = -0.120, p < 0.001; k = 170 effect sizes) and impression management (r = -0.158, p < 0.001; k = 157 effect sizes), yet there was evidence of publication bias for the impression management effect with the trim and fill method indicating that the relation is probably even smaller (r = -0.07). The magnitude of the effect sizes was small. Moderation analyses suggested that type of dynamic risk factor (e.g., antisocial cognition versus antisocial personality), incentives, and publication year affected the relationship between impression management and self-report measures with content of dynamic risk factors, whereas sample size, setting (e.g., incarcerated, community), and publication year influenced the relation between self-deception and these self-report measures. The results indicate that the use of self-report measures to assess dynamic risk factors in correctional/forensic settings is not inevitably compromised by socially desirable responding, yet caution is warranted for some risk factors (antisocial personality traits), particularly when incentives are at play.