*Md Kowsar Hossain Sakib, Neethiahnanthan Ari Ragavan, Shanjita Akter Prome,
Md Rafiqul Islam, Goh Wei Wei, David Asirvatham, Bharati Rathore

Abstract:

Deception detection (DD) presents a critical challenge across high-stakes domains such as security, finance, and education, necessitating accurate, non-invasive, and portable solutions. Traditional unimodal DD systems, which often rely solely on facial expressions or basic physiological signals, exhibit inherent limitations in both accuracy and complexity. To overcome these shortcomings, multimodal approaches have become the state of the art, integrating diverse data streams, including facial cues, vocal characteristics, body language, and advanced physiological metrics, to capture deeper insights into deceptive behaviour. This literature review systematically analyses contemporary multimodal DD methodologies. First, we provide an overview of current models for deception detection, focusing on methods that use computers to analyse multiple types of information. It explains how combining behavioural, physiological, and contextual evidence improves accuracy.

Keywords: Deception detection, multimodal approach, artificial intelligence, deep learning, machine learning, facial expression