Detecting deceptive behavior is a critical challenge across various domains, including security, recruitment, and criminal investigations. Traditional methods, such as polygraphs, rely on physiological cues and often lack reliability and scalability. This study introduces a deep learning-based methodology that enhances deception detection through the analysis of answer patterns derived from a Strategic Interview Technique (SIT) and publicly available datasets, including LIAR and Deceptive Opinion Spam. By integrating cognitive behavioral features such as response consistency, delay, and reactions to unexpected questions with textual embeddings generated from Bi-LSTM networks, the model provides a comprehensive framework for detecting lie tendencies. The proposed method demonstrates exceptional performance, achieving an accuracy of 89.5% and an F1-score of 88.9%, outperforming recent studies in the field. Comparative analysis highlights its robustness in distinguishing truthful and deceptive responses across structured and unstructured data. Error analysis reveals areas for refinement, including addressing false positives caused by ambiguous responses and false negatives in rehearsed deception. The model's reliance on cost-efective and non-invasive features makes it scalable and practical for real-world applications. This work lays the foundation for integrating multimodal data, such as audio and video, to further enhance the efectiveness of deception detection systems.
Lie detection is a critical area of research with applications in law enforcement, recruitment, and
psychological assessments. Traditional methods, such as polygraphs, rely on physiological signals but
face criticism for being invasive and susceptible to countermeasures and prone to manipulation [
Lie detection has been an essential focus of study, with traditional methods such as polygraph testing
relying on physiological signals like heart rate, skin conductance, and respiratory patterns [12, 13].
While widely used, these methods have several limitations, including invasiveness, high dependency
on instrumentation, and susceptibility to countermeasures [
This research aims to develop a robust deep learning-based framework for detecting deception by
integrating textual and behavioral features. The key objectives of this study are:
• Incorporating Behavioral Metrics: Utilize response consistency, delay, and unexpected
question reactions derived from SIT to detect cognitive strain indicative of deception [
By achieving these objectives, this work bridges the gap between traditional behavioral analysis and modern deep learning techniques, providing a scalable, eficient, and efective solution for deception detection.
This study employs a deep learning-based approach to detect lie by integrating textual and behavioural features derived from multiple datasets. The methodology includes data collection, feature engineering and the design of hybrid Bi-LSTM model that leverages the complementary strength of behavioural and linguistic analysis.
The model is trained and evaluated on a combined dataset consisting of three sources. The LIAR dataset [22], the Deceptive Opinion Spam dataset [20] and a custom Strategic Interview Technique (SIT) dataset.
SIT involves strategically farmed and repeated questions to elicit consistent or deceptive behavioral
patterns, where interviewees are subtly challenged to maintain consistency. Table 1 shows sample
interview questions, demonstrating both the repetitive nature and variations in phrasing used to
encourage cognitive consistency. The variety of topics and rephrasing within each category are aimed at
detecting changes in response consistency [
• Includes 12,836 labelled political statements with metadata (e.g., speaker, context, credibility). • Truthfulness levels: true, mostly true, half true, false, barely true and pants on fire. • Provides 1600 truthful and deceptive hotel reviews, categorized into positive and negative statements.
• Features: Textual content, word count and sentiment polarity.
The datasets were combined, as shown in Table 2, to train the model efectively. The combination of these datasets provides several advantages, enhancing the robustness and generalizability of the study. The benefits are
The success of any deep learning-based deception detection system relies heavily on the quality of features extracted from the input data. This study integrates three distinct datasets—responses collected via the Strategic Interview Technique (SIT), the LIAR dataset, and the Deceptive Opinion Spam dataset. Each dataset undergoes tailored preprocessing and feature engineering to ensure consistency and compatibility for training a unified deep learning model.
• Response Consistency (C): The calculation of the Response Consistency is shown in Equation 1: = ∑︀=1( − ¯)2 (1) Where, is the number of repeated responses, is the response at instance , ¯ is the mean response. A higher may indicate potential deception. • Response Delay (R): The formula to calculate the Response Delay is shown in the Equation 2: = ∑︀=1 (2) Where is the time taken for each response. Increased suggests cognitive processing, often associated with deception. • Unexpected Question Score (UQS): An Unexpected Question Score (UQS) is calculated by averaging response variances to unexpected questions. Higher UQS values indicate spontaneous inconsistencies, a potential indicator of deception.
Both the LIAR and Deceptive Opinion Spam datasets provide textual data labeled as truthful or deceptive. The following preprocessing steps are applied to ensure the extraction of meaningful semantic and syntactic features: • Text Cleaning: – Removal of punctuation, special characters, and stopwords.
– Conversion to lowercase to standardize input. • Tokenization and Lemmatization: – Tokenization splits sentences into individual words.
– Lemmatization reduces words to their base or dictionary forms, ensuring consistency.
• Word Embedding:
– Represent words as dense vectors using pretrained embeddings such as BERT. These
embeddings capture semantic and syntactic relationships between words [
To ensure uniformity and enhance model generalizability, the following techniques are used:
• Normalization: Numerical features (e.g., C, R, UQS) are normalized using Min-Max scaling [
max() − min() Where, represents the original feature value, min() and max() are minimum and maximum values of the feature in the dataset and ′ is the normalized value. • Data Augmentation: Synthetic samples are generated for the SIT dataset by simulating variations in response delay and consistency, ensuring balance between truthful and deceptive classes. Textual data augmentation includes synonym replacement and backtranslation techniques, particularly for small subsets of the Deceptive Opinion Spam dataset.
The deep learning framework developed in this study is designed to analyze both textual and behavioral features, leveraging their complementary nature to detect deceptive tendencies with high accuracy. The model employs a multi-branch architecture that processes distinct feature types—textual embeddings and behavioral metrics—through specialized neural network layers, culminating in a unified classification output.
• A textual branch that processes semantic information using a Bi-directional Long Short-Term
Memory (Bi-LSTM) network.
• A behavioral branch that analyzes numerical features using fully connected dense layers. These branches are integrated through a concatenation layer, followed by a classification layer that predicts the likelihood of truthfulness or deception. The modular nature of this architecture allows seamless incorporation of additional feature types, such as metadata or audio signals, if required.
The input to the model consists of: • Textual Data: Preprocessed text embeddings from the LIAR and Deceptive Opinion Spam datasets. Embeddings are generated using pretrained BERT models, capturing both semantic and contextual nuances. • Textual Data: Numerical features derived from SIT responses, including: Response Consistency (C), Response Delay (R), Unexpected Question Reaction (UQS).
Each input type is normalized and scaled to ensure compatibility with the subsequent layers.
The textual branch employs a Bi-LSTM network to capture sequential dependencies and contextual relationships in the input text: • Embedding Layer: Pretrained BERT embeddings are used to represent each word as a dense vector. These embeddings are fine-tuned during training to align with the deception detection task. • Bi-LSTM Layer: The Bi-LSTM network processes the sequence of embeddings, capturing both forward and backward temporal dependencies. The hidden states of the Bi-LSTM encode contextual relationships between words, which are crucial for detecting nuanced patterns of deception. • Dropout Layer: A dropout rate of 0.3 is applied to prevent overfitting, ensuring robust generalization to unseen data.
The output of the Bi-LSTM layer is a fixed-dimensional vector representing the entire input text, which is passed to the concatenation layer.
The behavioral branch processes numerical features through fully connected dense layers: • Input Layer: Accepts normalized behavioral features (C, R, and UQS) as input. • Dense Layers: Two fully connected layers, each with 128 and 64 neurons, apply non-linear transformations using ReLU activation. These layers enable the network to learn complex relationships among behavioral metrics.
• Dropout Layer: A dropout rate of 0.2 is applied after each dense layer to reduce overfitting. The final output of the behavioral branch is a feature vector summarizing patterns in the behavioral data.
The outputs of the Bi-LSTM and dense layers are concatenated into a unified feature vector. This layer enables the model to jointly analyze textual and behavioral patterns, leveraging their complementary strengths.
The concatenated feature vector is passed through a series of dense layers for classification: • Dense Layers: Two fully connected layers with 64 and 32 neurons, using ReLU activation. • Output Layer: A softmax layer outputs probabilities for two classes: truthful and liar. The final prediction is based on the class with the highest probability.
The model is trained to minimize the Categorical Cross-Entropy Loss, shown in Equation 4:
1 ∑︁ ∑︁ log(ˆ )
= −
=1 =1
(4)
Where is the true label for class of sample , ˆ is the predicted probability for class , is the
number of samples, and is the number of classes (truthful and liar). The textbfAdam optimizer is used
for eficient and adaptive gradient updates, with an initial learning rate of 10− 4 [
The combined dataset was divided into training (80%), validation (10%) and test (10%) sets. Cross validation was employed to ensure generalizability across diverse data domains. Training Set is Used to update the model weights., Validation Set Monitors during training for early stopping and Test Set Evaluates the model’s generalization performance on unseen data. A batch size of 32 is used, with training conducted over 50 epochs or until early stopping criteria are met. Performance is measured using accuracy, precision, recall, and F1 score. The overall workflow is illustrated in Figure 1, which provides a step-by-step depiction of how the inputs are processed, features are extracted, and predictions are made.
The proposed methodology integrates cognitive principles with deep learning models to efectively classify truthful and deceptive responses. The evaluation is based on behavioral features from SIT and textual patterns from LIAR and Deceptive Opinion Spam datasets. The model’s performance is analyzed using key metrics, comparative studies, and visual aids, ensuring a comprehensive understanding of its capabilities.
The performance of the proposed model was evaluated using standard metrics: accuracy, precision, recall, and F1-score. The results are presented in Figure 2 and Table 3. The proposed model achieves high accuracy (89.5%) and recall (92.4%), outperforming traditional techniques. Its superior F1-score (88.9%) highlights its balanced capability in identifying deceptive behavior while minimizing false positives and negatives.
The model’s performance is compared with other state-of-the-art methods, as shown in Table 4 and visualized in Figure 3. The proposed methodology achieves the highest accuracy and recall among the compared models. The integration of SIT behavioral features with textual embeddings gives it a competitive advantage.
The model’s discriminative capability is illustrated in Figure 4, which depicts the Receiver Operating Characteristic (ROC) curve. The Area Under the Curve (AUC) value of 0.75 confirms the model’s ability to reliably diferentiate between truthful and deceptive responses.
A detailed error analysis was conducted shown in Figure 5 to identify limitations: • False Positives: Truthful responses misclassified as deceptive, often due to ambiguous or overly concise answers. • False Negatives: Deceptive responses misclassified as truthful, typically observed in rehearsed or highly consistent responses.
To mitigate these errors: • Enhanced SIT Questions: Increase the variability and complexity of questions to induce greater cognitive load. • Augmented Training Data: Include more diverse examples of liar and truthful responses to reduce bias.
As shown in the Table 5, the proposed model achieves significantly higher accuracy than traditional techniques, such as the Polygraph Test, while requiring less time and no specialized equipment.
The proposed methodology demonstrates significant advancements in deception detection by integrating behavioral features from the Strategic Interview Technique (SIT) with textual data from the LIAR and Deceptive Opinion Spam datasets. By leveraging deep learning architectures such as Bi-LSTM and dense layers, the model achieves robust performance, with an accuracy of 89.5%, precision of 85.7%, recall of 92.4%, and an F1-score of 88.9%. These metrics highlight the model’s ability to efectively distinguish between truthful and deceptive responses across diverse datasets. A key strength of the methodology lies in its integration of cognitive and textual features, which provides a comprehensive analysis of deceptive behavior. Behavioral metrics such as consistency score, response delay, and unexpected question score add valuable insights into cognitive patterns that are dificult to capture using textual data alone. The inclusion of textual embeddings ensures that the model generalizes well across unstructured data, making it versatile for a range of applications. Compared to recent works, the proposed model outperforms methods such as the hybrid deep learning approach by Mendels et al. (2017), the multimodal neural model by Krishnamurthy et al. (2018) and a language guided deep learning method by Wang et al. (2020). These improvements are attributed to the innovative use of SIT-derived behavioral metrics and the efective design of the deep learning architecture. Challenges remain in addressing false positives caused by ambiguous truthful responses and false negatives in rehearsed deceptive responses. Refining SIT question design to increase variability and cognitive load, along with diversifying training datasets to include a broader demographic range, could further enhance model performance. Despite these challenges, the methodology’s reliance on simple inputs and short test durations makes it cost-efective, scalable, and practical for real-world applications. This work establishes a strong foundation for future research. The integration of additional modalities, such as audio or video, could further improve the detection of deception and expand the applicability of this approach in domains such as security, recruitment, and criminal investigations.
In this study, an innovative approach was introduced for detecting an individual’ tendency to lie by examining response patterns using an Artificial Neural Network (ANN) based on a Strategic Interview Technique (SIT). Unlike traditional lie detection methods that rely heavily on physiological responses or simplistic question-answering techniques, the proposed method analyzes subtle variations in answer consistency and response delay. The findings reveal that the ANN-based SIT model achieves a high accuracy rate of 89.5%, surpassing traditional methods such as Polygraph and previous ANN-based lie detection techniques like PVMANN and PVMMWANN. By reducing the dependency on specialized equipment and minimizing testing time, this model provides a practical and accessible alternative for lie detection, particularly in settings where traditional methods may not be feasible or afordable. The analysis of Consistency Score (CS) and Response Delay (RD) proved valuable in distinguishing between truthful and deceptive individuals. While truthful individuals typically exhibit stable patterns and shorter response times, deceptive individuals tend to show higher variability and delay, highlighting cognitive load diferences. However, certain limitations, such as the potential for manipulation by highly trained individuals and emotional influence, suggest areas for future improvement. Integrating additional biometrics, enhancing cultural sensitivity in questioning, and exploring adaptive question models could further enhance accuracy and applicability. In conclusion, the proposed ANN-based SIT model demonstrates significant progress in the field of lie detection by combining cognitive science principles with machine learning techniques. The results underscore its potential for practical use in criminal investigations, security assessments, and personnel evaluations, contributing a reliable, cost-efective, and less intrusive alternative to traditional methods.
We sincerely thank the Civic Volunteers of Siliguri Metropolitan Police for their enthusiastic participation in the interview sessions, which were crucial to this study. We are especially grateful to Mr. Sunil Yadav, IPS, Assistant Commissioner Police (Trafic), for his invaluable support and for facilitating this research. We also extend our gratitude to the academic community of the University of North Bengal—students, scholars, and faculty members—for their cooperation, guidance, and valuable feedback, which greatly enriched the quality of our work.
During the preparation of this work, the author(s) used OpenAI’s ChatGPT to assist with grammar and spelling checks. After using this tool, the author(s) reviewed and edited the content as needed and take full responsibility for the publication’s content. [2] J. Charles F. Bond and B. M. DePaulo, "Accuracy of Deception Judgments," Personality and Social
Psychology Review, vol. 10, pp. 214-234, 2006, DOI: 10.1207/s15327957pspr1003_2. [3] Wang, X., Peng, H., & Pan, S. (2020). Language-Guided Deep Learning for Deception Detection. IEEE Transactions on Knowledge and Data Engineering, 32(4), 667–677, DOI: 10.1109/TKDE.2019.2892408. [4] Zuckerman, M., DePaulo, B. M., & Rosenthal, R. (1981). Verbal and Nonverbal Communication of Deception. Advances in Experimental Social Psychology, 14, 1-59. DOI: 10.1016/S0065-2601(08)603697. [5] Vrij, A. (2008). Detecting Lies and Deceit: Pitfalls and Opportunities. John Wiley & Sons, ISBN: 978-0470516256. [6] Chakraborty, S., & Mandal, R. K. (2016). Pattern Variation Method to Detect Lie Using Artificial
Neural Network (PVMANN). National Conference on Computational Technologies, 57-60. [7] Mandal, R. K. (2016). Pattern Variation Method with Modified Weights to Detect Lie using Artiifcial Neural Network (PVMMWANN). AMSE JOURNALS, Modelling C, 77, 41-52, available at: https://iieta.org/sites/default/files/Journals/MMC/MMC_C/2016.77.1_04.pdf. [8] Vrij, A., Fisher, R. P., & Blank, H. (2017). A cognitive approach to lie detection: A meta-analysis.
Legal and Criminological Psychology, 22(1), 1–21, DOI: 10.1111/lcrp.12088. [9] Granhag, P. A., & Stromwall, L. A. (2001). Deception detection based on repeated interrogations.
Legal and Criminological Psychology, 6, 85-101. DOI: 10.1348/135532501168217. [10] Hartwig, M., Granhag, P. A., & Strömwall, L. A. (2007). Strategic use of evidence during police interviews: When training to detect deception works. Law and Human Behavior, 31(2), 233–247, DOI: 10.1007/s10979-006-9053-9. [11] M. J, B.-G. I, M. C, H. C and I. I, "Strategic Interviewing to Detect Deception: Cues to Deception across Repeated Interviews," Front. Psychol, vol. 7, pp. 1-17, 2016, DOI: 10.3389/fpsyg.2016.01702. [12] National Research Council 2003, The Polygraph and Lie Detection, Washington, DC: The National
Academies Press, 2003, DOI: 10.17226/10420. [13] A. Slavkovic, "Evaluating Polygraph Data," 2018, DOI: 10.1184/R1/6586598.v1. [14] Vrij, A., Mann, S., & Fisher, R. P. (2006). Information-gathering vs. accusatory interview style: Its impact on deception detection. Legal and Criminological Psychology, 11(1), 1–15, DOI: 10.1348/135532505X39099. [15] Hartwig, M., Granhag, P. A., & Strömwall, L. A. (2007). Strategic use of evidence during police interviews: When training to detect deception works. Law and Human Behavior, 31(2), 233–247, DOI: 10.1007/s10979-006-9053-9. [16] Walsh, D., & Bull, R. (2012). How do interviewers attempt to overcome suspects’ denials? Criminal
Behaviour and Mental Health, 22(2), 102–116, DOI: 10.1002/cbm.1829. [17] Vrij, A., Leal, S., Fisher, R. P., Warmelink, L.,& Mann, S. (2018). Drawings as an innovative and efective lie detection tool. Journal of Applied Psychology, 103(5), 501-513. DOI: 10.1037/apl0000298. [18] Y. Tsunomori, G. Neubig, S. Sakti, T. Toda and S. Nakamura, "An Analysis Towards Dialogue-Based Deception Detection," in Natural Language Dialog Systems and Intelligent Assistants, Springer, ChamSpringer, Cham, 2015, pp. 177-187, DOI: 10.1007/978-3-319-19291-8_17. [19] Krishnamurthy, S., Ramesh, S., & Elhabian, S. (2018). "Multimodal Neural Networks for Deception Detection." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW 2018), 1–7, DOI: 10.1109/CVPRW.2018.00009. [20] Ott, M., Choi, Y., Cardie, C., & Hancock, J. T. (2011). "Finding Deceptive Opinion Spam by Any Stretch of the Imagination." Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), 309–319, DOI: 10.3115/2002472.2002512. [21] Perez-Rosas, V., Kleinberg, B., Lefevre, A., & Mihalcea, R. (2018). Automatic Detection of Deception in Text: A Survey. Computational Linguistics, 44(4), 1–25, DOI: 10.1162/coli_a_00332. [22] Wang, W. Y. (2017). "Liar, Liar Pants on Fire: A New Benchmark Dataset for Fake News Detection." Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (ACL 2017), 422–426, DOI: 10.18653/v1/P17-2067.