The Future of Malingering Detection: Integrating Neuroscience, Psychology and Artificial Intelligence

This thesis examines how the interdisciplinary integration of psychology, neuroscience, and artificial intelligence can enhance the detection of malingering, which is the intentional feigning or exaggeration of physical or psychological symptoms for personal gain. After assessing the limitations of traditional lie-detection techniques and the inconsistencies in behavioural indicators, the work systematically reviews three key domains: (1) Psychological approaches, which encompass symptom validity tests (such as SIRS-2, SIMS, and M-FAST), performance validity tests (including TOMM, WMT, and VSVT), embedded indicators (for instance, Reliable Digit Span), and emerging cognitive-load and verifiability techniques; (2) Neuroscientific methods, covering neuroimaging (fMRI-based paradigms like the Guilty Knowledge Task and dlPFC modulation), electrophysiological measures (EEG/ERP, often referred to as “brain fingerprinting”), and hemodynamic methods (fNIRS); and (3) AI and machine learning, demonstrating how these models can refine traditional tests through pattern-based classification, extract meaningful features from complex neuroimaging data, and implement real-time signal-processing pipelines for EEG and physiological data streams. A chapter dedicated to multimodal integration illustrates how the fusion of behavioural, neural, and computational metrics can address the vulnerabilities and countermeasures inherent in single-domain methodologies. Case examples include support-vector models that combine response times and error patterns, hybrid EEG-polygraph systems, and deep-learning frameworks that integrate facial, vocal, and autonomic signals. Ethical, legal, and social considerations are discussed, emphasizing the importance of transparency, interpretability, and the potential for biases. In conclusion, the thesis proposes a roadmap for future research: rigorous cross-validation of multimodal classifiers in ecologically valid contexts, the development of lightweight neuro-AI tools for forensic applications, and the establishment of guidelines to safeguard data privacy and admissibility. Overall, this work presents a comprehensive framework for next-generation malingering detection that balances scientific rigor with real-world applicability.