Lies or Nerves? A Signal Detection Theory-Guided Approach to Deception Detection in Natural Speech

Creators: Jechle, Deinera and Faußer, Stefan A. and Gewald, Heiko
Title: Lies or Nerves? A Signal Detection Theory-Guided Approach to Deception Detection in Natural Speech
Item Type: Conference or Workshop Item
Event Title: 9th International Conference on Information Technology & Systems (ICITS)
Event Location: Roatán Island, Honduras
Event Dates: February, 16-18, 2026
Projects: IDI
Paper No. / Type of Paper: 312 / Full Paper
Date: February 2026
Divisions: Informationsmanagement
Abstract (ENG): The automated identification of deceptive human behavior is a longstanding research endeavor. A major challenge for detection is to distinguish between liars and truthful but nervous individuals, as the subconscious human cues are very similar. Misclassifications are the unwanted consequences. This Design Science Research study applies Signal Detection Theory (SDT) as an explanatory kernel theory to design and evaluate an AI-based artifact for automated deception detection. Using a specifically created dataset and a controlled evaluation set of truthful but nervous speech, a support vector machine classifier was trained on acoustic-prosodic and spectral features to distinguish deceptive intent from nervousness. The artifact achieved a training accuracy of 76.6% in detecting deception and a false alarm rate of 6.25% for nervous-but-truthful cases. The findings demonstrate that SDT-guided design reduces bias, improves fairness, and enhances the applicability of speech-based analytics in sensitivedecision-making domains.
Forthcoming: Yes
Language: English
Uncontrolled Keywords: signal detection theory; design science research; deception detection; speech analytics; machine learning
Citation:

Jechle, Deinera and Faußer, Stefan A. and Gewald, Heiko (2026) Lies or Nerves? A Signal Detection Theory-Guided Approach to Deception Detection in Natural Speech. In: 9th International Conference on Information Technology & Systems (ICITS), February, 16-18, 2026, Roatán Island, Honduras, Paper 312 / Full Paper. (Lecture Notes in Networks and Systems).

Actions for admins (login required)

View Item in edit mode (academic staff only) View Item in edit mode (academic staff only)