Investigation in Brainwaves of EEG Signal for Lie Detection System Based on Power Spectral Density

Background & Context

Power Spectral Density (PSD) analysis represents a foundational approach to understanding brain activity patterns associated with deception. This 2025 study by Hamza and Al-Hamadani addresses a critical gap in lie detection technology by exploring how frequency domain features of EEG signals can differentiate truthful from deceptive cognitive states.

The theoretical basis rests on the principle that lying requires more mental effort than telling the truth, as the liar must work hard to make the lie seem believable, and this increased cognitive load manifests in altered physiological responses including changes in brain wave patterns. In recent years, lie detection techniques have advanced beyond polygraphs to include methods such as electroencephalography, offering non-invasive access to neural correlates of deception.

PSD analysis decomposes EEG signals into constituent frequency bands (delta, theta, alpha, beta, gamma), allowing researchers to identify which spectral features are most discriminative for deception detection. This approach offers advantages over traditional polygraph methods by directly measuring brain activity rather than peripheral physiological responses, potentially making it more resistant to countermeasures and less influenced by anxiety or fear.

Research Design & Methodology

Based on the research group's established methodology, this study likely employed a laboratory-based experimental design using the OpenBCI Ultracortex "Mark IV" EEG Headset, which acquired 14 channel of EEG data from participants. The research team has developed a novel dataset and protocol based on video to elicit deceptive and truthful responses in controlled conditions.

The Power Spectral Density methodology involves several key analytical steps:

• Signal preprocessing: Bandpass filtering (typically 0.1-64 Hz), notch filtering at 50/60 Hz to remove line noise, and artifact removal
• PSD computation: Application of Fast Fourier Transform (FFT) or Welch's method to extract frequency-domain features from each EEG epoch
• Frequency band extraction: Calculation of absolute and relative power in delta (1-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30 Hz), and gamma (>30 Hz) bands
• Statistical analysis: Comparison of spectral features between truthful and deceptive conditions across different brain regions

The study follows the authors' established approach of comparing truth-telling and lie-telling conditions, with participants likely responding to questions where they must either provide honest answers or instructed lies. Two conditions were analyzed truth or lie, with participants rewarded with a gift for their collaboration after the experiment.

Results & Key Findings

While the specific numerical results from the PSD analysis are contained in the Springer chapter, the research likely demonstrates that specific frequency bands show significant differences between truthful and deceptive states. Based on similar EEG lie detection research, key findings typically include:

• Frontal lobe activation: Increased power in theta and alpha bands during deceptive responses, reflecting elevated executive function and working memory demands
• Temporal and parietal changes: Altered beta band activity associated with memory retrieval and response inhibition during lying
• Gamma band modulation: Higher frequency oscillations showing differences in cognitive binding and attention during deceptive versus truthful responses
• Channel-specific patterns: Particular electrode sites (likely frontal channels like FP1, FP2, F3, F4) demonstrating the most discriminative spectral features

The authors' related work has demonstrated that CNN techniques achieve accuracy of 99.96% on their EEG dataset when using their novel protocol, suggesting that spectral features extracted through PSD analysis provide highly discriminative information for automated classification systems.

Research in this area has shown that the highest accuracy achieved in EEG-based lie detection using traditional machine learning approaches reaches 98.8% using SVM, indicating that PSD features can support robust classification when combined with appropriate machine learning algorithms.

Discussion & Significance

This investigation advances the field by providing systematic analysis of how brainwave frequency characteristics differ between truthful and deceptive cognitive states. The PSD approach offers several theoretical advantages: it captures sustained changes in neural oscillations that may be less susceptible to moment-to-moment variability than event-related potentials, and it provides interpretable features that can be linked to known cognitive processes.

The study contributes to a growing body of evidence that lie detection techniques have advanced beyond polygraphs to include methods such as electroencephalography, with researchers employing Artificial Intelligence algorithms (Deep learning and machine learning) for EEG signal processing for lie detection. By focusing specifically on PSD features, this work helps identify which aspects of the EEG signal are most informative for deception detection.

The findings align with cognitive neuroscience models suggesting that deception requires executive control, working memory, and response inhibition—processes known to modulate specific EEG frequency bands. This neurophysiological grounding provides theoretical support for the PSD approach beyond purely empirical pattern recognition.

Limitations & Considerations

Several methodological considerations warrant attention. The OpenBCI Ultracortex Mark IV is a research device used in multiple applications, but this device wasn't medical grade devices, and its signal quality may be restricted in comparison to medical devices, which may introduce measurement noise that could affect PSD estimates, particularly in higher frequency bands.

Laboratory-based lie detection studies face inherent ecological validity challenges:

• Low stakes context: Participants have little motivation to successfully deceive, unlike real-world forensic scenarios
• Instructed lies: Experimentally directed deception may differ neurologically from spontaneous, motivated lying
• Sample characteristics: Studies typically use healthy volunteers rather than populations of investigative interest
• Individual variability: PSD features show substantial inter-individual differences, requiring careful normalization and potentially personalized baselines

Given that this topic represents a challenge to privacy for individual volunteers, most researchers collect their data themselves and analyze it, limiting the availability of standardized datasets for validation and comparison across studies.

Practical Applications

The PSD-based approach offers practical advantages for operational lie detection systems. The indicated manner is non-surgical, effective, and powerful, with least time complication, consequently appropriate for real-time applications. Unlike complex time-frequency analyses, PSD calculation is computationally efficient and can be implemented in portable systems.

For practitioners, this research suggests that frequency-domain features provide a complementary or alternative approach to traditional event-related potential methods like P300-based concealed information testing. The technique could be particularly valuable in scenarios requiring continuous monitoring rather than response-locked analysis, such as security screening interviews or investigative questioning. However, no EEG method has yet proven to be entirely reliable for lie detection, and any operational application must acknowledge the significant gap between laboratory performance and field validity. The technology remains experimental and should not be used as standalone evidence in legal or security contexts without extensive validation and appropriate safeguards.