How EyeDetect Works: The Pupillometry Science Explained

The Evolution of Credibility Assessment Technology

For over a century, the polygraph has been the dominant instrument for physiological lie detection. The traditional polygraph measures peripheral autonomic responses — blood pressure, respiration, electrodermal activity, and cardiovascular output — that correlate with the emotional arousal associated with deception [1]Verified A Realistic Perspective of the Art and Science of Forensic Psychophysiology
Provides perspective on the balance between evidence-based polygraph methodology and examiner skill in forensic psychophysiology. While the polygraph remains a powerful and validated tool, the scientific community has long sought complementary technologies that could measure deception through different physiological pathways.

EyeDetect, developed by Converus, Inc. and based on decades of research at the University of Utah, represents a fundamentally different approach [2]Verified EyeDetect Product Page — Converus
Confirms EyeDetect's 86-88% accuracy, 15-30 minute test duration, 60 Hz eye-tracking camera, and automated scoring within 5 minutes. Rather than measuring peripheral autonomic arousal, EyeDetect measures the central nervous system manifestations of cognitive load — specifically, the way the eyes behave when the brain is working harder to construct and maintain a deceptive response [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

The technology was born from research conducted by Dr. John C. Kircher and Dr. Doug Hacker, who conceived the idea in 2002 while driving to climb Mt. Rainier in Washington state [4]Verified History of Converus — Converus
Confirms EyeDetect conceived in 2002 by Kircher and Hacker, science team formed in 2003, Raskin joined in 2009, and computerized polygraph created in 1991. Kircher and Hacker, along with cognitive scientists Anne Cook and Dan Woltz, formed the Converus Science Team in 2003 [4]Verified History of Converus — Converus
Confirms EyeDetect conceived in 2002 by Kircher and Hacker, science team formed in 2003, Raskin joined in 2009, and computerized polygraph created in 1991. Dr. David C. Raskin joined in 2009 [4]Verified History of Converus — Converus
Confirms EyeDetect conceived in 2002 by Kircher and Hacker, science team formed in 2003, Raskin joined in 2009, and computerized polygraph created in 1991. Notably, Kircher and Raskin had previously co-invented the first computerized field polygraph system in 1991 [5]Verified Renowned Polygraph Expert Dr. John Kircher Joins Converus Staff
Confirms Kircher and Raskin developed the first computerized field polygraph system in 1991 and that Kircher conceived EyeDetect in 2002. Their fundamental insight was that deception is not merely an emotional event but a cognitive one — and that the eyes, as a direct extension of the brain, reveal cognitive processing in ways that are both measurable and involuntary.

For a broader view of how different credibility assessment technologies compare, see our EyeDetect vs. polygraph accuracy comparison.

Why Pupillometry Matters in Deception Detection

The pupil of the eye has been called a window to the mind by cognitive scientists. Unlike heart rate, blood pressure, or galvanic skin response — which are influenced by a wide range of emotional, physical, and environmental factors — pupillary responses are tightly coupled to cognitive processing demands. When the brain works harder, the pupils dilate. This relationship, known as the task-evoked pupillary response (TEPR), was first systematically documented by Eckhard Hess and James Polt in their landmark 1964 study published in Science, which demonstrated that pupils dilated when subjects performed increasingly difficult mental arithmetic [6]Verified Pupil Size in Relation to Mental Activity during Simple Problem-Solving
Landmark 1964 study establishing that pupil dilation is a direct measure of mental activity during cognitive tasks. The finding has since been replicated by a large body of studies spanning six decades [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system.

EyeDetect leverages this well-established psychophysiological phenomenon by presenting subjects with true/false statements on a computer screen and measuring their ocular responses with an infrared eye-tracking camera that captures approximately 60 measurements per second of each eye [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. The result is a credibility assessment system that operates on a fundamentally different measurement axis than the polygraph, making it both a standalone tool and a powerful complement to traditional methods.

To understand how this fits into the broader landscape of psychophysiological detection of deception (PDD), it helps to recognize that EyeDetect represents a cognitive approach to deception detection, while the polygraph primarily measures autonomic arousal responses [8]Verified Intelligent Eye-Tracker-Based Methods for Detection of Deception: A Survey
Reviews cognitive approach to lie detection using eye tracking, including oculomotor feature analysis for forensic application.

Understanding Cognitive Load in the Context of Lying

Cognitive load theory, originally developed by John Sweller in educational psychology, describes the total amount of mental effort being used in working memory. In the context of deception detection, cognitive load theory provides the theoretical foundation for understanding why lying is neurologically harder than telling the truth — and why this difficulty produces measurable physiological signatures [8]Verified Intelligent Eye-Tracker-Based Methods for Detection of Deception: A Survey
Reviews cognitive approach to lie detection using eye tracking, including oculomotor feature analysis for forensic application.

When a person tells the truth, they engage in a relatively simple cognitive process: they access a genuine memory, formulate a response based on that memory, and communicate it. Deception, by contrast, requires the brain to perform multiple simultaneous operations that dramatically increase cognitive demand:

Truth suppression — the genuine response must be identified and actively inhibited. Fabrication — a plausible false response must be constructed, requiring creative generation rather than simple memory retrieval. Consistency monitoring — the liar must track prior statements and ensure consistency, placing significant demands on working memory. Believability assessment — the deceiver must continuously evaluate whether the lie sounds plausible. Emotional regulation — stress, guilt, or fear of detection must be managed and concealed.

This multi-tasking during deception creates what researchers call a cognitive load differential between truthful and deceptive states. Neuroimaging studies using fMRI have shown that deception activates the prefrontal cortex, anterior cingulate cortex, and other executive function regions far more intensively than truth-telling [9]Verified Using Brain Imaging for Lie Detection: Where Science, Law and Research Policy Collide
Reviews 76-90% accuracy range for brain imaging deception detection under controlled conditions but notes translational gaps between lab and field settings. The cognitive approach to lie detection aims to amplify this gap by designing test conditions that make lying more difficult while leaving truthful responding relatively unaffected [8]Verified Intelligent Eye-Tracker-Based Methods for Detection of Deception: A Survey
Reviews cognitive approach to lie detection using eye tracking, including oculomotor feature analysis for forensic application.

How Cognitive Load Manifests in the Eyes

The critical insight behind EyeDetect is that cognitive load has direct, measurable effects on the oculomotor system. Many ocular behaviors — particularly pupil dilation — are governed by the locus coeruleus-norepinephrine (LC-NE) system, a brainstem nucleus that modulates arousal, attention, and cognitive effort [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system.

When cognitive load increases, the locus coeruleus increases norepinephrine release, which has a direct dilatory effect on the pupil via the sympathetic pathway to the iris dilator muscle [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system. This is an involuntary, reflexive response — the subject cannot consciously prevent their pupils from dilating when their brain is working harder [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features.

Beyond pupil dilation, increased cognitive load also affects reading speed (deception-related text requires more processing time), fixation patterns (eyes dwell longer on words relevant to the deceptive topic), saccadic movements (rapid eye jumps between fixation points become less efficient with more regressions), and blink rate (cognitive load typically suppresses blinking during high-demand processing) [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. EyeDetect's measurement approach captures all of these parameters simultaneously, building a comprehensive picture of the subject's cognitive state at each moment during the test.

Research on cognitive lie detection training has shown that cognitive approaches to detecting deception, when properly trained, can produce modest but meaningful improvements in accuracy [11]Verified Train the Trainers: A First Step towards a Science-Based Cognitive Lie Detection Training Workshop
Demonstrates that cognitive approaches to deception detection can produce improvements in detection accuracy when properly trained.

The Science of Pupil Dilation and Cognitive Processing

Pupillometry — the precise measurement of pupil diameter changes — is the primary measurement channel in the EyeDetect system and the most extensively studied indicator of cognitive load. The foundational research traces back to Hess and Polt's 1964 study in Science, titled "Pupil Size in Relation to Mental Activity during Simple Problem-Solving" [6]Verified Pupil Size in Relation to Mental Activity during Simple Problem-Solving
Landmark 1964 study establishing that pupil dilation is a direct measure of mental activity during cognitive tasks. Their work demonstrated that pupil dilation could be interpreted as a direct measure of mental activity, a finding that launched an entire sub-field of cognitive psychophysiology [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system.

Subsequent decades of research have refined our understanding of the pupillary response to cognitive load. The Task-Evoked Pupillary Response (TEPR) describes how when a cognitively demanding task is presented, the pupil begins dilating approximately 200-500 milliseconds after stimulus onset, reaches peak dilation proportional to the difficulty of the task, and then returns to baseline once the task is completed [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system. The magnitude of dilation is remarkably sensitive — differences of up to 0.5 millimeters can be discretely correlated with mental effort in increasingly complex tasks [6]Verified Pupil Size in Relation to Mental Activity during Simple Problem-Solving
Landmark 1964 study establishing that pupil dilation is a direct measure of mental activity during cognitive tasks.

The dose-response relationship is well established: the more difficult the cognitive task, the greater the pupil dilation. This has been demonstrated across memory retrieval, mathematical calculation, language processing, and decision-making tasks [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system. In deception contexts, the additional cognitive operations required for lying produce measurably larger pupil dilations than truth-telling.

Critically, the pupillary response to cognitive load operates through the autonomic nervous system and cannot be voluntarily controlled [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. While a person can choose to look away or close their eyes, they cannot consciously modulate the dilation of their pupils in response to cognitive demand. This involuntariness is central to EyeDetect's validity as a credibility assessment tool.

Deception-Specific Pupillary Patterns

Research specifically examining pupillary responses during deception has identified several consistent patterns that EyeDetect's algorithm is designed to detect. When subjects respond deceptively to relevant questions, their peak pupil dilation is significantly larger than when they respond truthfully [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. The effect is small in absolute terms but highly reliable and statistically significant when measured with precision eye-tracking equipment.

Deceptive responses also tend to produce pupillary dilation that begins slightly later than truthful responses, reflecting the additional time required for the brain to initiate the deceptive cognitive process before the autonomic response follows. The pupil remains dilated for a longer period during deceptive processing, as the brain continues to engage in maintaining the lie even after the response has been given. After a deception-related question, the pupil takes longer to return to baseline compared to a truthful response.

These patterns are not assessed on a single question but across an entire test session, allowing the algorithm to identify consistent differential responses between categories of questions — a statistical approach conceptually similar to how polygraph scoring compares reactions to relevant versus comparison questions [12]Verified A Field Polygraph Examination: Science or Art?
Examines how field polygraph integrates scientific methodology and professional artistry in question formulation and data integration. Understanding the different polygraph test formats helps illustrate how EyeDetect's question structure parallels established credibility assessment methodologies.

Saccadic Eye Movements and Fixation Patterns

EyeDetect tracks multiple oculomotor channels beyond pupil size. Saccades — the rapid, ballistic eye movements that shift gaze from one fixation point to another — provide rich diagnostic information. During normal reading, the eyes make a series of quick jumps (saccades) interspersed with brief pauses (fixations) where actual visual information processing occurs [8]Verified Intelligent Eye-Tracker-Based Methods for Detection of Deception: A Survey
Reviews cognitive approach to lie detection using eye tracking, including oculomotor feature analysis for forensic application.

EyeDetect monitors saccade velocity (speed of eye movement between fixation points), saccade length (distance covered by each jump), and regressive saccades (backward movements revisiting previously read text). Under increased cognitive load, saccade velocity can decrease as the oculomotor system becomes less efficient, and regressions increase as subjects struggle to process deception-relevant content [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

Fixation duration — the amount of time the eyes remain stationary between saccades — is one of the most robust indicators of cognitive processing depth in reading research. During normal reading, fixation durations average approximately 200-300 milliseconds. Subjects spend more time fixating on words and phrases in questions related to their deceptive topic, reflecting deeper semantic processing and the additional cognitive work of evaluating the question in the context of their intended deception.

Blink Rate, Reading Behavior, and Response Timing

Eye blinks carry significant cognitive information. Research has consistently shown that blink rate is suppressed during periods of high cognitive demand and increases during lower-demand periods. EyeDetect captures blink frequency, blink duration, and blink timing relative to stimulus presentation [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

Because EyeDetect presents its stimuli as text that the subject must read, the system captures a rich set of reading behavior data that standard pupillometry alone cannot provide. This includes reading speed (words per minute on different question types — deceptive subjects typically read relevant questions more slowly), re-reading behavior (how often subjects return to re-read portions of questions), and response latency (the time between finishing reading a question and providing a response) [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

As Dr. John Kircher explained in a 2017 interview, subjects under mental load show specific patterns: their pupils dilate, they take longer to read statements, fixations are closer together, and they do more re-reading [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. These multiple channels of measurement work together to create a comprehensive cognitive load profile that the scoring algorithm analyzes holistically.

Infrared Illumination and Camera Technology

EyeDetect uses a commercial-grade infrared eye-tracking system that operates at approximately 60 Hz — meaning it captures 60 images of the eyes per second [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. The system uses near-infrared light (typically around 850-940 nanometers wavelength) to illuminate the eyes without causing discomfort or interfering with the subject's vision. Near-infrared light is invisible to the human eye, so the subject is unaware of the illumination during testing.

The infrared illumination creates two key optical phenomena that the eye-tracker uses to determine pupil size and gaze direction. Pupil detection occurs because infrared light passes through the pupil and is absorbed by the retina, making the pupil appear as a distinct dark region in the infrared image. Corneal reflection (glint) occurs when a portion of the infrared light reflects off the cornea's curved surface, creating a bright spot whose geometric relationship to the pupil center allows the system to compute gaze direction independent of head position.

The hardware setup is deliberately simple: a standard computer monitor, the eye-tracking camera mounted below the screen, and a chin rest or head stabilizer to minimize movement [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. This simplicity means EyeDetect can be administered in virtually any office environment without specialized facilities. As Converus has noted, one test administrator with three EyeDetect stations can run up to 30-40 tests daily [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. For a look at how this compares to polygraph hardware complexity, see our review of the Stoelting CPS II Polygraph System.

Calibration and Continuous Adjustment

Before each test, the eye-tracking system undergoes a brief calibration procedure where the subject looks at a series of points on the screen. This calibration establishes the mathematical mapping between eye image features and screen coordinates for that specific individual, accounting for variations in eye anatomy, corneal curvature, and seating position. The calibration typically takes less than 60 seconds.

The system continuously adjusts for small head movements using the corneal reflection as a reference point, maintaining measurement accuracy even if the subject shifts slightly during the test. Converus has noted that the calibration process also helps detect attempts to manipulate results — eye liner or pupil-dilating eye drops can be identified during calibration, and the test can be postponed until these factors are resolved [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features.

How a Test Session Unfolds

Understanding the test protocol is essential to understanding why EyeDetect measures what it measures. The test is a carefully structured sequence of stimuli designed to elicit differential cognitive load responses from truthful versus deceptive subjects.

The primary stimulus format is a series of true/false statements presented on screen, one at a time, to which the subject responds by pressing a key [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. Statements fall into distinct categories: relevant statements that directly address the topic under investigation, comparison statements that establish a baseline of cognitive and physiological arousal, and neutral statements that serve as cognitive baselines. Tests take 15 to 30 minutes depending on the format [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

The subject hears and sees instructions onscreen, and a practice test helps them get comfortable with the process [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. No examiner is in the room during the actual test questions, which eliminates examiner bias — a significant advantage over traditional polygraph methods [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. As one Canadian law enforcement professional noted, the automated nature means the test is 100 percent unbiased since no human examiner can influence the outcome [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

EyeDetect Test Formats: Screening vs. Diagnostic

EyeDetect currently supports multiple test formats, each optimized for different applications. The Relevant Comparison Test (RCT) was the first format developed, originally designed for screening at border entry points [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. In the European Polygraph Journal of December 2016, Dr. Kircher and Dr. Raskin published the mean accuracy of the RCT at 86% [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method.

The Directed Lie Comparison (DLC) test protocol, originally developed for polygraph, was adapted for EyeDetect and is used primarily for diagnostic single-issue tests in criminal or civil investigations. The mean accuracy is 90% according to field data gathered and analyzed by Kircher and Raskin in December 2018 [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. The Multi-Issue Comparison Test (MCT), developed by Dr. Andrew Potts at the University of Utah, achieved 88% accuracy and introduced the capability of scoring up to four relevant issues in a single test [14]Verified FAQs for EyeDetect — Converus
Confirms 88% accuracy for screening, 90% for diagnostic, hybrid MCT at 91%, and details on MCT and DLC test formats.

In June 2021, Dr. John Kircher published results on Hybrid test formats that combine ocular-motor and polygraph physiology measures. The Hybrid DLC achieved 88% accuracy and the Hybrid MCT achieved 91% accuracy [14]Verified FAQs for EyeDetect — Converus
Confirms 88% accuracy for screening, 90% for diagnostic, hybrid MCT at 91%, and details on MCT and DLC test formats. These different examination varieties mirror the range of formats available in traditional polygraph testing.

How the Algorithm Classifies Truthful vs. Deceptive Responses

One of EyeDetect's most distinctive features is its fully automated scoring system. After test completion, raw oculomotor data are uploaded to a cloud-based server where proprietary algorithms process the data and produce a credibility score within approximately five minutes [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

The scoring system uses logistic regression models that combine scores on a subset of diagnostic measures to compute the probability of deception [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method. If the probability of deception exceeds 0.5, the system classifies the person as deceptive; if below 0.5, it classifies the person as truthful [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method. Credibility scores range from 50 to 99 for truthful classifications and 1 to 49 for deceptive classifications — the closer to 99 or 1, the higher the confidence in the classification [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features.

Because the scoring is entirely computer-based, EyeDetect achieves 100% reliability in the technical sense: the same data will always produce the same score [14]Verified FAQs for EyeDetect — Converus
Confirms 88% accuracy for screening, 90% for diagnostic, hybrid MCT at 91%, and details on MCT and DLC test formats. This eliminates inter-rater variability — a known challenge in traditional polygraph scoring where different examiners may reach different conclusions from the same charts [15]Verified A Field Polygraph Examination: Science or Art?
Discusses the role of validated techniques and scoring systems in field polygraph examinations. The test administrator cannot change test results, and test data is encrypted and tamper-proof [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

Key Published Research

The foundational peer-reviewed study for EyeDetect technology is Cook, Hacker, Webb, Osher, Kristjansson, Woltz, and Kircher's 2012 paper "Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception" published in the Journal of Experimental Psychology: Applied [16]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational peer-reviewed study establishing that oculomotor measures during reading can differentiate truthful from deceptive subjects. This study established that oculomotor measures during reading could differentiate between truthful and deceptive subjects.

In 2016, Kircher and Raskin published "Laboratory and Field Research on the Ocular-motor Deception Test" in the European Polygraph Journal (Volume 10, Number 4), reporting field accuracy estimates [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method. Using the five-fold cross-validation method, they estimated the test detected truthful responses 88% of the time and deceptive responses 86% of the time [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method. The authors disclosed their financial interests in Converus, Inc. and noted an approved conflict management plan at the University of Utah [13]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Reports RCT mean accuracy of 86%, truthful detection at 88%, deceptive detection at 86%, using five-fold cross-validation method.

Also in 2016, Patnaik, Woltz, Hacker, Cook, Francke-Ramm, Webb, and Kircher published in the International Journal of Applied Psychology, showing EyeDetect correctly classified approximately 85% of truthful and deceptive participants in a Mexican population, confirming the technology's cross-cultural generalizability [17]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms 85% accuracy with Mexican population, demonstrating cross-cultural generalizability of the ODT.

Converus reports that as of 2024, 12 peer-reviewed publications have validated the ocular-motor deception test methodology [18]Verified 12th Peer-Reviewed Publication Further Validates Eye Behavior Deception Detection
Confirms 12 peer-reviewed publications validating EyeDetect technology as of 2024. However, it should be noted that most published research originates from the University of Utah team associated with Converus [19]Verified Lie Detector Firm Lobbies CIA, DOD on Automated Eye-Scanning Tech — The Intercept
Reports that independent peer-reviewed replication of EyeDetect accuracy claims is limited; notes advisory board member's skepticism about small database from one laboratory. A 2020 member of Converus's own advisory board observed that the database remained relatively small and came mostly from one laboratory, recommending replication by other researchers [19]Verified Lie Detector Firm Lobbies CIA, DOD on Automated Eye-Scanning Tech — The Intercept
Reports that independent peer-reviewed replication of EyeDetect accuracy claims is limited; notes advisory board member's skepticism about small database from one laboratory.

For further context on how scientific validation works in this field, see our guide on psychophysiology for polygraph examiners.

Accuracy in Context

Converus states that EyeDetect accurately classifies 88 out of 100 people in screening tests and 90 out of 100 in diagnostic tests [14]Verified FAQs for EyeDetect — Converus
Confirms 88% accuracy for screening, 90% for diagnostic, hybrid MCT at 91%, and details on MCT and DLC test formats. Screening tests use the RCT format and are approximately 86-88% accurate, while the DLC diagnostic format achieves approximately 90% accuracy [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. With the newer Hybrid formats combining ocular-motor and polygraph measures (EyeDetect+), accuracy reaches 89-91% [20]Verified About Converus
Confirms 500,000 tests administered, EyeDetect+ 89-91% accuracy, and customer deployments in 13+ countries.

When EyeDetect is used in conjunction with traditional polygraph as a successive-hurdles approach, and both tests agree on the same result, the combined confidence outcome increases substantially [21]Verified EyeDetect Frequently Asked Questions — Global Data Fusion
Confirms EPPA applicability to EyeDetect and 99% confidence outcome when combined with polygraph. Converus has stated that when both tests indicate deception on the same issue, the probability of lying reaches 99% [22]Verified Tech Co. Says its Advanced Lie Detector Will Save U.S. Gov't $29B Annually — Converus
Confirms 1,000+ customers, 100+ U.S. law enforcement agencies, 50,000+ State Department tests, and 99% confidence when both tests agree.

The American Polygraph Association (APA) has noted that research studies published since 1980 report average polygraph accuracy rates ranging from 80 to 98 percent [23]Verified The Polygraph and Lie Detection — National Academy of Sciences
The APA cites accuracy rates of 80-98%; the NAS concluded CQT accuracy claims were difficult to verify and research was of low quality, though the 2003 National Academy of Sciences report concluded that polygraph accuracy claims were difficult to verify and that the extant research was of low quality [24]Verified Current Status of Forensic Lie Detection With the Comparison Question Technique
Reviews NAS 2003 conclusions that polygraph CQT scientific basis was weak and error rate unknown, though accuracy exceeds chance. Independent researchers have found published EyeDetect accuracies of around 80% in academic analyses, compared to the higher figures cited by Converus [19]Verified Lie Detector Firm Lobbies CIA, DOD on Automated Eye-Scanning Tech — The Intercept
Reports that independent peer-reviewed replication of EyeDetect accuracy claims is limited; notes advisory board member's skepticism about small database from one laboratory.

Measurement Axis and Methodology

The most fundamental difference between EyeDetect and the polygraph is what they measure. The polygraph records peripheral autonomic nervous system responses — cardiovascular activity, respiratory patterns, and electrodermal response — using physical sensors attached to the subject's body [1]Verified A Realistic Perspective of the Art and Science of Forensic Psychophysiology
Provides perspective on the balance between evidence-based polygraph methodology and examiner skill in forensic psychophysiology. EyeDetect records central nervous system manifestations of cognitive load through oculomotor behavior, using a contactless infrared camera [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

This difference in measurement axis means the two technologies are genuinely complementary rather than redundant. Polygraph excels at detecting the emotional arousal associated with deception, while EyeDetect excels at detecting the cognitive effort. The comprehensive comparison of EyeDetect and polygraph accuracy, cost, and admissibility explores these differences in detail.

The field continues to evolve, as reflected in research on the art and science balance in polygraph examinations, which highlights how both examiner skill and scientific methodology contribute to effective deception detection [25]Verified A Field Polygraph Examination: Science or Art?
Examines how field polygraph integrates validated techniques and professional artistry. Meanwhile, the development of standardized terminology in psychophysiological detection helps bridge the gap between traditional polygraph and newer technologies like EyeDetect [26]Verified Terminology Reference for the Science of Psychophysiological Detection
Standardizes terminology across the polygraph profession with 94 scientific references.

Practical Advantages and Speed

EyeDetect offers several practical advantages: tests take 15-30 minutes compared to 2-4 hours for a standard polygraph exam [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. The system requires no sensors attached to the participant [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. Proctors can be trained in a day, compared to the extensive training required for polygraph examiners [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception. One proctor with three stations can manage up to 30-40 tests in an 8-hour workday [3]Verified EyeDetect Lie Detector Uses Eye Tracking — CNBC
Confirms Kircher's description of how cognitive load affects pupil dilation, reading speed, and fixations during deception.

The U.S. State Department has facilitated over 50,000 EyeDetect tests across various countries since 2018 [22]Verified Tech Co. Says its Advanced Lie Detector Will Save U.S. Gov't $29B Annually — Converus
Confirms 1,000+ customers, 100+ U.S. law enforcement agencies, 50,000+ State Department tests, and 99% confidence when both tests agree. As of the latest reports, EyeDetect has been used by more than 1,000 customers worldwide, including over 100 U.S. law enforcement agencies [22]Verified Tech Co. Says its Advanced Lie Detector Will Save U.S. Gov't $29B Annually — Converus
Confirms 1,000+ customers, 100+ U.S. law enforcement agencies, 50,000+ State Department tests, and 99% confidence when both tests agree. Federal agencies in countries including Singapore, Poland, Ukraine, Jamaica, Mexico, Guatemala, Peru, Colombia, Panama, and South Korea have adopted the technology [20]Verified About Converus
Confirms 500,000 tests administered, EyeDetect+ 89-91% accuracy, and customer deployments in 13+ countries.

However, EyeDetect has limitations that should be understood. Current technology does not work well for examinees with poor reading skills, as reading difficulties can overshadow the effects of deception on ocular-motor measures [27]Verified Your Lying Eyes — University of Utah
Kircher's own account of EyeDetect's development, noting reading skill limitations and that the ODT is faster (30-40 minutes) and completely automated. For a thorough examination of these constraints, see our guide on EyeDetect limitations: what the technology cannot do.

EPPA and Regulatory Considerations

The Employee Polygraph Protection Act of 1988 (EPPA) is the primary federal law governing lie detector use in private employment in the United States [28]Verified Employee Polygraph Protection Act — 29 USC §2001
Confirms EPPA defines 'lie detector' broadly to include polygraph, deceptograph, voice stress analyzer, psychological stress evaluator, or any similar device used for honesty assessment. The EPPA's statutory definition of "lie detector" is broad: it includes "a polygraph, deceptograph, voice stress analyzer, psychological stress evaluator, or any other similar device (whether mechanical or electrical) that is used, or the results of which are used, for the purpose of rendering a diagnostic opinion regarding the honesty or dishonesty of an individual" [29]Verified Employee Polygraph Protection Act — 29 USC §2001 (Full Text)
Statutory text defining lie detector and polygraph under EPPA.

This broad definition means EyeDetect likely falls under the EPPA's prohibitions when used by covered private employers [30]Verified EyeDetect Frequently Asked Questions — Global Data Fusion
Confirms that although EPPA specifically mentions polygraph, the law has been applied broadly to any lie detection solution in practice. The EPPA does not cover federal, state, and local government agencies [28]Verified Employee Polygraph Protection Act — 29 USC §2001
Confirms EPPA defines 'lie detector' broadly to include polygraph, deceptograph, voice stress analyzer, psychological stress evaluator, or any similar device used for honesty assessment, meaning government use of EyeDetect for law enforcement screening, national security, and similar purposes is not restricted by the Act.

To address state-level regulatory compliance, Converus developed EyeDetect+ in 2020, which adds a "Physio Tracker" device gathering cardiovascular, respiratory, and electrodermal data — the same channels measured by a standard polygraph [31]Verified New Version of EyeDetect Legally Compliant in All 50 States — Converus
Confirms EyeDetect+ with Physio Tracker meets functional definition of polygraph for compliance in 17 states and DC. This allows EyeDetect+ to meet the functional definition of a polygraph in the 17 states and the District of Columbia where lie detector laws specifically reference traditional polygraph channels [31]Verified New Version of EyeDetect Legally Compliant in All 50 States — Converus
Confirms EyeDetect+ with Physio Tracker meets functional definition of polygraph for compliance in 17 states and DC.

For more on how federal requirements intersect with credibility assessment technology, see our guide on PCSOT and the Adam Walsh Act.

Court Admissibility: The Daubert Standard

In 2018, a New Mexico judicial district court judge granted a defendant's Daubert motion and allowed EyeDetect test results to be admitted as evidence for the first time [32]Verified US District Court Allows EyeDetect Lie Detector Test Results As Evidence for First Time
Confirms 2018 Daubert motion granted in New Mexico 8th District Court for EyeDetect evidence in criminal sexual assault case. The accused had passed the EyeDetect test with a "credible" score in a criminal sexual assault case in the 8th District Judicial Court, County of Taos, New Mexico [32]Verified US District Court Allows EyeDetect Lie Detector Test Results As Evidence for First Time
Confirms 2018 Daubert motion granted in New Mexico 8th District Court for EyeDetect evidence in criminal sexual assault case.

In other cases, an Ohio judge considered EyeDetect results among other evidence when dropping charges in a weapons possession case, and a Utah judge considered EyeDetect results when requesting an investigation into police officer conduct [32]Verified US District Court Allows EyeDetect Lie Detector Test Results As Evidence for First Time
Confirms 2018 Daubert motion granted in New Mexico 8th District Court for EyeDetect evidence in criminal sexual assault case. A Texas judge also ordered EyeDetect testing in a drug distribution case [32]Verified US District Court Allows EyeDetect Lie Detector Test Results As Evidence for First Time
Confirms 2018 Daubert motion granted in New Mexico 8th District Court for EyeDetect evidence in criminal sexual assault case.

However, court admissibility remains jurisdiction-dependent and evolving. The Daubert standard requires that scientific evidence be based on sufficient facts or data, produced by reliable principles and methods, and reliably applied to the facts of the case [33]Verified Daubert Standard — Wikipedia
Explains the Daubert standard for admissibility of expert witness testimony in federal court. The relatively limited body of independent peer-reviewed research on EyeDetect — compared to the much larger literature on polygraph — may present challenges in Daubert hearings as the technology seeks broader judicial acceptance.

Why EyeDetect Is Resistant to Common Countermeasures

A significant advantage of EyeDetect over traditional polygraph methods is its resistance to common countermeasures. Because pupillary responses are governed by the autonomic nervous system through the locus coeruleus-norepinephrine pathway, they cannot be voluntarily controlled [7]Verified Pupil dilation as an index of effort in cognitive control tasks: A review
Comprehensive review confirming that pupil dilation increases with task demands and is linked to the locus coeruleus-norepinephrine system. Even highly motivated subjects cannot simultaneously control their reading behaviors, response speed, response accuracy, and pupil dilation [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features.

Converus states that examinees must answer questions rapidly and have little time to attempt physiological countermeasures [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. Some examinees may attempt to use makeup (eyeliner) or pupil-dilating eye drops, but these are easily detected during the pretest eye-tracking calibration, and the test can be postponed [10]Verified Credibility Assessment for Public Safety Agencies — Converus Whitepaper
Details EyeDetect test protocols, scoring ranges, DLC accuracy of 90%, RCT accuracy of 86%, and countermeasure resistance features. Similarly, attempts to close eyes or squint are detected by the system.

This contrasts with the traditional polygraph, where certain countermeasures can, under some conditions, enable a deceptive individual to appear nondeceptive [24]Verified Current Status of Forensic Lie Detection With the Comparison Question Technique
Reviews NAS 2003 conclusions that polygraph CQT scientific basis was weak and error rate unknown, though accuracy exceeds chance. For a historical perspective on how countermeasure vulnerabilities have played out in real-world intelligence cases, see our article on CIA polygraph failures: spies who passed every test.

Global Deployment and Use Cases

EyeDetect was commercially released in April 2014, with its first major public showing at ExpoSeguridad in Mexico City [34]Verified Converus Releases First Lie Detection Technology That Reads Eye Behavior
Confirms EyeDetect commercially released April 2014, initial validation trials at 85% accuracy. The technology initially gained traction in Latin America, where banks, financial institutions, and government agencies adopted it for pre-employment screening [35]Verified The Eyes Have It — Deseret News
Reports Converus presence in 60 countries with 600+ clients by 2022, and 90%+ success rate on numbers test demonstrations.

As of 2022, Converus reported more than 600 customers in over 40 countries, with the number growing to over 1,000 customers by 2025 [20]Verified About Converus
Confirms 500,000 tests administered, EyeDetect+ 89-91% accuracy, and customer deployments in 13+ countries[22]Verified Tech Co. Says its Advanced Lie Detector Will Save U.S. Gov't $29B Annually — Converus
Confirms 1,000+ customers, 100+ U.S. law enforcement agencies, 50,000+ State Department tests, and 99% confidence when both tests agree. The technology is available in over 40 languages [31]Verified New Version of EyeDetect Legally Compliant in All 50 States — Converus
Confirms EyeDetect+ with Physio Tracker meets functional definition of polygraph for compliance in 17 states and DC. U.S. adoption includes over 100 state and local law enforcement agencies [22]Verified Tech Co. Says its Advanced Lie Detector Will Save U.S. Gov't $29B Annually — Converus
Confirms 1,000+ customers, 100+ U.S. law enforcement agencies, 50,000+ State Department tests, and 99% confidence when both tests agree, and the U.S. State Department has used EyeDetect for vetting local hires at embassies [19]Verified Lie Detector Firm Lobbies CIA, DOD on Automated Eye-Scanning Tech — The Intercept
Reports that independent peer-reviewed replication of EyeDetect accuracy claims is limited; notes advisory board member's skepticism about small database from one laboratory.

International military customers include the Air Force of Colombia and the Jamaica Defence Force [36]Verified New Lie Detection Technology Can Make Military Bases Safer — Converus
Confirms 500+ customers in 40 countries including Air Force of Colombia and Jamaica Defence Force. EyeDetect has also been used in therapeutic settings, including sex offender treatment programs as an alternative to post-conviction sex offender testing (PCSOT), and in marriage counseling contexts [19]Verified Lie Detector Firm Lobbies CIA, DOD on Automated Eye-Scanning Tech — The Intercept
Reports that independent peer-reviewed replication of EyeDetect accuracy claims is limited; notes advisory board member's skepticism about small database from one laboratory.

For information on how EyeDetect is being adopted across the African continent specifically, see our guide on EyeDetect for African law enforcement. And for broader historical context on how credibility assessment technology has spread globally, explore the history of polygraph in China and polygraph in Turkey.