EyeDetect Accuracy: What the Research Shows (2026)

The Science Behind Ocular-Motor Deception Testing

EyeDetect is a credibility assessment technology developed by researchers at the University of Utah that uses high-speed infrared eye-tracking cameras to measure involuntary changes in ocular behavior when a person reads and responds to true/false statements on a computer screen [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The core scientific premise is that cognitive load — the mental effort required to maintain a deception — produces measurable changes in eye behavior that differ from those observed when a person is telling the truth [2]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational 2012 study introducing the ocular-motor deception test, published in Journal of Experimental Psychology: Applied, demonstrating pupil diameter and reading behavior changes during deception.

Unlike the traditional polygraph, which monitors autonomic nervous system responses such as skin conductance, blood pressure, and respiration, EyeDetect focuses exclusively on ocular-motor responses. These include pupil diameter changes, reading speed and patterns, fixation duration, blink rate, and other eye movement metrics [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. Research has consistently demonstrated that greater pupil dilation occurs during deception due to increased cognitive processing demands [4]Verified Effectiveness of Pupil Diameter in a Probable-Lie Comparison Question Test for Deception
Confirms pupil diameter is a significant predictor variable for deception and may replace traditional blood pressure measurement in polygraph testing. The technology captures these data points at approximately 60 measurements per second for each eye, generating thousands of data points per test session [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

How an EyeDetect Test Session Works

During a typical EyeDetect examination, the subject sits in front of a computer monitor equipped with a specialized infrared eye-tracking camera mounted beneath the screen [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The subject reads a series of true/false statements presented on the screen and responds using a trackpad or mouse. The entire process typically takes between 15 and 30 minutes, depending on the test protocol used [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

No human examiner is present during the automated data collection phase. Converus argues this eliminates examiner bias — a longstanding concern in polygraph testing. Once the session is complete, the data is uploaded to Converus's cloud-based servers, where proprietary algorithms analyze the ocular response patterns. The system generates a credibility score, with results categorized as credible (truthful) or not credible (deceptive). A result can also be classified as inconclusive if the data does not meet the threshold for either determination [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

Converus reports that test proctors can be trained in a single day and that one proctor using three stations can administer up to 30 tests daily [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries, making the technology significantly more scalable than traditional polygraph examinations.

Two Primary Test Formats

EyeDetect offers two main test formats, each designed for different use cases. Understanding these formats is essential for evaluating accuracy claims, because research findings may apply to one format but not the other.

The Relevant Comparison Test (RCT) is used primarily for screening purposes such as pre-employment vetting [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. This was the first test technique developed for EyeDetect, originally created by Kircher, Raskin, Gardner, Jewell, and Patnaik in 2012 [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The published mean accuracy for screening tests is 86% [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

The Directed Lie Comparison (DLC) test is a newer format introduced in January 2019 for diagnostic or single-issue investigations [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. An initial field study showed this test technique achieves approximately 90% accuracy [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. This format can be used for criminal investigations, civil cases, parole compliance, and similar specific-issue testing.

Converus also offers EyeDetect+, which combines the eye-tracking camera with traditional polygraph sensors (wrist sensors replacing the blood pressure cuff, and strain gauges replacing pneumographs) to create what they describe as the world's first automated polygraph [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The distinction between test formats matters because the underlying cognitive mechanisms differ, and published accuracy rates are not necessarily interchangeable between them.

Dr. John C. Kircher and Dr. Douglas Hacker

The idea for EyeDetect was born in 2002 when Dr. John C. Kircher and his colleague Dr. Douglas Hacker, an educational psychologist with expertise in the psychology of reading, were driving to Seattle to climb Mt. Rainier [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. En route, they wondered whether changes in eye movements and pupil size while reading and answering questions about a crime would reveal deception [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

In 2003, Kircher and Hacker, along with cognitive scientists Anne Cook and Dan Woltz, formed the core research team at the University of Utah [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. Dr. David C. Raskin, a highly respected figure in the polygraph community who co-invented the computerized polygraph with Kircher in 1991, joined the science team in 2009 [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. Kircher and Raskin first published research on polygraph technology in the 1970s [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries and have contributed extensively to the development of computerized polygraph scoring systems, including work relevant to the Control Question Test.

The initial validation research progressed through several key milestones. In 2006, the Osher Dissertation documented the first laboratory study demonstrating the effectiveness of an ocular-motor deception test [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. In 2008, the Webb Dissertation confirmed the technology's effectiveness [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The first peer-reviewed publication appeared in 2012 in the Journal of Experimental Psychology: Applied [2]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational 2012 study introducing the ocular-motor deception test, published in Journal of Experimental Psychology: Applied, demonstrating pupil diameter and reading behavior changes during deception.

The Formation of Converus, Inc.

In September 2013, the technology was given the brand name EyeDetect, and in April 2014 it was released to market at ExpoSeguridad in Mexico City [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The company Converus, Inc. (meaning 'with truth'), is headquartered in Lehi, Utah [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The company's management team includes chief scientist John Kircher, who has also served as a consultant for the CIA and Department of Defense [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology.

Converus claims more than 500 customers in over 40 countries using the technology in 40 different languages [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. Government and law enforcement users have included the U.S. State Department (which paid $25,000 to use EyeDetect for vetting local hires at the U.S. Embassy in Guatemala), as well as trials by the Defense Intelligence Agency and U.S. Customs and Border Protection [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology. For a detailed comparison of how EyeDetect is being evaluated alongside polygraph in specific applications, see our EyeDetect for security clearance guide.

The relationship between the technology's academic origins and its commercial development is relevant to evaluating the research. As a 2025 critical review noted, the vast majority of published studies originate from a small, closely linked academic-commercial group — primarily John C. Kircher, Anne Cook, Douglas Hacker, and David Raskin [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

Overview of Manufacturer-Published Research

Converus has published and funded the largest collection of research on EyeDetect's accuracy. As of mid-2024, the company cited 12 peer-reviewed publications validating the ocular-motor deception test (ODT) method [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. These studies, many conducted by or in collaboration with the technology's inventors, form the primary evidence base that the company uses to market EyeDetect to government agencies, corporations, and other potential clients.

Converus consistently markets EyeDetect's accuracy as 86 to 90 percent [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. More specifically, the company states that scientific studies show an accuracy range of 86 to 88% depending on the type of test administered [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries, with screening tests at approximately 86% mean accuracy and diagnostic tests at approximately 90% [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The company reports that for every 100 people tested, EyeDetect accurately classifies 88 in screening tests and 90 in diagnostic tests [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

Key Published Studies

Several published studies from the Converus research team deserve specific attention:

Cook et al. (2012) — "Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception," published in the Journal of Experimental Psychology: Applied, 18(3), 301–313. This foundational study introduced the ocular-motor deception test and used a mock crime paradigm with university participants [2]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational 2012 study introducing the ocular-motor deception test, published in Journal of Experimental Psychology: Applied, demonstrating pupil diameter and reading behavior changes during deception. It demonstrated that deceptive answers were preceded by greater increases in pupil diameter, fewer fixations, and shorter second-pass reading times [2]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational 2012 study introducing the ocular-motor deception test, published in Journal of Experimental Psychology: Applied, demonstrating pupil diameter and reading behavior changes during deception. This study established the scientific basis for EyeDetect's approach to detecting deception through cognitive load markers. Multiple independent research reviews cite this study as having found overall accuracies of around 80% [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

Kircher & Raskin (2016) — "Laboratory and Field Research on the Ocular-motor Deception Test," published in European Polygraph, 10(4), 159–172 [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. This comprehensive review examined the theoretical framework and empirical evidence for the ODT. The decision models yielded approximately 86% correct classifications in the original standardization sample and 83% correct when tested on independent samples (cross-validation) [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. The authors disclosed financial interests in Converus [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies.

Webb, Honts, Kircher, Bernhardt, & Cook (2009) — "Effectiveness of Pupil Diameter in a Probable-Lie Comparison Question Test for Deception," published in Legal and Criminological Psychology, 14(2), 279–292 [4]Verified Effectiveness of Pupil Diameter in a Probable-Lie Comparison Question Test for Deception
Confirms pupil diameter is a significant predictor variable for deception and may replace traditional blood pressure measurement in polygraph testing. This study found that pupil diameter was a significant predictor variable for deception and may serve as a possible replacement for traditional relative blood pressure measurement in polygraph testing [4]Verified Effectiveness of Pupil Diameter in a Probable-Lie Comparison Question Test for Deception
Confirms pupil diameter is a significant predictor variable for deception and may replace traditional blood pressure measurement in polygraph testing.

Patnaik et al. (2016) — "Generalizability of an Ocular-Motor Test for Deception to a Mexican Population," published in the International Journal of Applied Psychology, 6(1), 1–9 [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. This field study tested whether the ODT's effects generalize to native Spanish speakers in Mexico. With 147 university students participating in a mock crime experiment, cross-validation accuracy exceeded 80% for both truthful and deceptive participants [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants.

Webb et al. (2009) — "Eye Movements and Pupil Size Reveal Deception in Computer Administered Questionnaires," presented at the Foundations of Augmented Cognition conference [8]Verified Eye Movements and Pupil Size Reveal Deception in Computer Administered Questionnaires
Early 2009 conference paper demonstrating oculomotor test with 40 participants in mock crime paradigm, showing guilty participants exhibited speeded reading and increased pupil size. This early 40-participant study demonstrated that guilty participants showed speeded and accurate reading along with increases in pupil size when encountering crime-related statements [8]Verified Eye Movements and Pupil Size Reveal Deception in Computer Administered Questionnaires
Early 2009 conference paper demonstrating oculomotor test with 40 participants in mock crime paradigm, showing guilty participants exhibited speeded reading and increased pupil size.

Strengths and Limitations of Converus Research

The Converus-funded studies have several strengths: they use appropriate statistical methods, many have been published in peer-reviewed journals, and they provide a systematic theoretical framework linking cognitive load to ocular-motor indicators. The early validation work in Cook et al. (2012) was published in the prestigious Journal of Experimental Psychology: Applied [2]Verified Lyin' Eyes: Ocular-motor Measures of Reading Reveal Deception
Foundational 2012 study introducing the ocular-motor deception test, published in Journal of Experimental Psychology: Applied, demonstrating pupil diameter and reading behavior changes during deception, lending credibility to the foundational science.

However, these studies also have significant limitations. Most use mock crime paradigms where participants know they are in a research study, which may not create the same level of emotional stakes as real-world scenarios. The studies generally use university student participants, and the Patnaik et al. (2016) field study in Mexico found that reading proficiency significantly affected results — participants who struggled to understand test items did not exhibit expected patterns [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. The company itself acknowledged that it chose not to publish results of its first field experiment in Colombia, which appeared to show the technology working erratically [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology.

Why Independent Validation Matters

In scientific evaluation of any detection technology, independent validation — research conducted by scientists with no financial or professional ties to the technology's developer — is considered the gold standard. For EyeDetect, this distinction is particularly important because the financial and reputational stakes are high, and as one critical review noted, the proprietary nature of the scoring system prevents independent scientific scrutiny, replication, and legal auditability [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

Survey of Independent Studies

The body of fully independent EyeDetect research remains smaller than the manufacturer-funded literature. A 2022 study by Mundt et al. evaluated the ODT in a forensic setting with civilly committed sexually violent persons — one of the few studies conducted outside the core Converus research group [9]Verified Ocular-motor Deception Testing in Civilly Detained Sexually Violent Persons: An Alternative to Post-Conviction Sex Offender Polygraph Testing?
Confirms ODT tested in forensic PCSOT-relevant population, found accuracy rates not significantly less than published 80% benchmark, scoring algorithm modified post hoc with Converus input. The study tested whether ODT accuracy would be equivalent to or better than the 80% rate published by Kircher and Raskin (2016), and found results that were not significantly less than that benchmark [9]Verified Ocular-motor Deception Testing in Civilly Detained Sexually Violent Persons: An Alternative to Post-Conviction Sex Offender Polygraph Testing?
Confirms ODT tested in forensic PCSOT-relevant population, found accuracy rates not significantly less than published 80% benchmark, scoring algorithm modified post hoc with Converus input. However, the study noted that the scoring algorithm was modified post hoc based on initial data, with technical input from Converus [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

Andrew Potts' 2020 dissertation tested a multiple-issue screening ODT and achieved high AUC values in some specific comparisons (0.92 for one crime category), demonstrating the technology's potential in specific contexts [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns. However, the overall body of research from researchers outside the core Utah group has produced accuracy figures that cluster around 80% rather than the 86–90% range marketed by Converus [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

A recent independent study using AI models with eye-tracking data for concealed information detection achieved accuracies of up to 74% in binary classification — lower than the ODT figures, though using different methodologies [10]Verified Eye Movements as Indicators of Deception: A Machine Learning Approach
Independent AI-based deception detection using eye tracking achieved up to 74% binary classification accuracy; meta-analysis of cognitive lie detection studies found weighted mean of 74% overall accuracy. This provides useful context for understanding the broader state of eye-based deception detection research.

For comparison with other emerging deception technologies, see our guides on fMRI lie detection and Voice Stress Analysis.

The Proprietary Algorithm Challenge

A key concern raised by multiple researchers and a 2025 critical review is the 'black box' nature of EyeDetect's scoring system [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns. The proprietary algorithms, feature weightings, and decision thresholds are not accessible to independent researchers, legal professionals, or regulatory bodies. Unlike established polygraph scoring systems such as the Empirical Scoring System (ESS) or validated numerical scoring methods that have published their criteria, the ODT scoring model cannot be independently examined or replicated [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

This opacity has practical consequences: independent researchers cannot fully verify how the system weighs different ocular metrics or how scoring models may be adjusted for different populations. Reports suggest that Converus may adjust sensitivity thresholds based on group characteristics [11]Verified Why EyeDetect (Converus) Falls Short — and Why No Phone App Can Detect Lies
Confirms critics warn of internally conducted studies, artificial test conditions, and adjustable sensitivity thresholds raising bias concerns, which critics note could introduce bias in operational settings.

Mock Crime vs. Real-World Testing

Perhaps the single most important methodological consideration in evaluating EyeDetect accuracy research is the distinction between laboratory mock crime studies and real-world operational use. The vast majority of published EyeDetect studies use mock crime paradigms where participants are randomly assigned to guilty or innocent conditions [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

Critics argue that real-world deception involves significantly higher emotional arousal, more motivated deception strategies, and different cognitive processes than those engaged in a low-stakes laboratory simulation. The Converus researchers themselves acknowledge this limitation. Kircher and Raskin (2016) noted that laboratory effects are encouraging but questions can be raised about the generalizability of these effects to field settings [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. A meta-analysis of cognitive lie detection studies found a weighted mean of only 74% overall accuracy [10]Verified Eye Movements as Indicators of Deception: A Machine Learning Approach
Independent AI-based deception detection using eye tracking achieved up to 74% binary classification accuracy; meta-analysis of cognitive lie detection studies found weighted mean of 74% overall accuracy, highlighting how detection rates can vary substantially across different research designs.

Sample Sizes and Statistical Power

Many EyeDetect studies use relatively modest participant pools. The foundational Webb et al. (2009) conference study used just 40 participants [8]Verified Eye Movements and Pupil Size Reveal Deception in Computer Administered Questionnaires
Early 2009 conference paper demonstrating oculomotor test with 40 participants in mock crime paradigm, showing guilty participants exhibited speeded reading and increased pupil size, while Cook et al. (2012) and other laboratory studies involved similar numbers. The Patnaik et al. (2016) Mexico study used 147 participants — among the larger samples [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. The Mundt et al. (2022) forensic study included 74 participants in its primary experiment [9]Verified Ocular-motor Deception Testing in Civilly Detained Sexually Violent Persons: An Alternative to Post-Conviction Sex Offender Polygraph Testing?
Confirms ODT tested in forensic PCSOT-relevant population, found accuracy rates not significantly less than published 80% benchmark, scoring algorithm modified post hoc with Converus input.

For context, a study with 30 guilty and 30 innocent participants reporting 86% accuracy has confidence intervals wide enough that the true accuracy could plausibly be anywhere from roughly 75% to 95%. The APA's 2011 meta-analytic survey, by comparison, analyzed 38 studies involving 3,723 examinations and 11,737 scored results [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy — a substantially larger evidence base for polygraph accuracy conclusions.

Cross-Cultural and Demographic Considerations

EyeDetect requires subjects to read statements on a screen, which introduces potential confounds related to reading ability, literacy level, and language proficiency. The Patnaik et al. (2016) study was specifically designed to test whether effects generalize to Spanish-speaking populations in Mexico and found that cross-validation accuracy exceeded 80% [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants.

However, the researchers also disclosed that in an unpublished field study conducted in Colombia, they found almost no effect of deception on ocular-motor measures [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. They attributed this partly to inadequate reading proficiency among participants, noting that deceptive individuals who struggled to understand test items did not exhibit expected pattern changes [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. Reviews of laboratory and field studies show similar accuracies in English-speaking, Spanish-speaking, and Middle Eastern cultures, though achieving comparable accuracy for Arabic-speaking participants required modifying the number of statement repetitions [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies.

For more on how EyeDetect is being adopted in emerging markets, see our guides on EyeDetect in Africa and EyeDetect for African law enforcement.

The Inconclusive Rate Factor

When evaluating any credibility assessment technology's accuracy, it is important to understand how inconclusive results are handled. The Kircher & Raskin (2016) study reported the mean decision accuracy of EyeDetect at 86%, comprised of 0.89 for True Negatives and 0.83 for True Positives with no inconclusive results [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. This zero-inconclusive approach means every case received a definitive classification, which provides a clear accuracy picture but may also mean the system is forced to classify cases where confidence is low.

Consumers and organizations should understand that in operational settings, a system with a modest false-positive rate can have significant consequences when applied to large screening populations — a point acknowledged by EyeDetect's own chief scientist, who has advised against relying solely on any single screening technology [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology.

The Human Examiner Question

Experienced polygraph examiners have raised several concerns about EyeDetect's fully automated approach. The polygraph testing procedure includes a pre-test interview where a skilled examiner establishes rapport, evaluates the subject's suitability, and formulates questions. Polygraph professionals argue this human element is critical for identifying factors that might affect test validity, such as medical conditions, medications, psychological states, or question misunderstandings.

Converus counters that automation eliminates examiner bias and enables 100% scoring consistency — every time the same data is analyzed, the same result is produced [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. This is a genuine advantage: research on polygraph testing has shown that human scoring can introduce variability [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy. The question is whether the trade-off between consistency and contextual judgment ultimately serves examinees and organizations better.

The Base Rate Problem in Screening

A statistical concern relevant to any screening technology is the base rate problem. In pre-employment screening where the vast majority of applicants are truthful, even a modest false-positive rate produces a disproportionate number of innocent people flagged as deceptive. Kircher himself has written that even if a test is 90% accurate, about 10% of the tested population would fail it, and the vast majority of those who fail would be innocent [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology.

Converus states that EyeDetect's false positive rate of 10% is the lowest of any credibility assessment technology on the market, including polygraph [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology. This concern applies to polygraph screening as well — the APA's 2011 meta-analysis found multi-issue screening achieves 85% accuracy [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy, with comparable base rate challenges. For organizations considering screening programs, this mathematical reality means that either polygraph or EyeDetect should be used as part of a broader evaluation process, not as a standalone determination.

Transparency and Scoring Concerns

Unlike validated polygraph scoring methods, the EyeDetect scoring architecture functions as a proprietary system whose algorithms, feature weightings, and decision thresholds are not publicly accessible [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns. A 2025 critical review described this as a significant barrier to establishing the scientific and legal legitimacy required for high-stakes applications [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

The polygraph community's established scoring approaches — including numerical scoring methods with published criteria [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy — offer greater transparency. This distinction matters for legal proceedings, where defense attorneys may need to challenge the basis for a credibility determination. For more on how polygraph scoring has evolved, see our guide on polygraph history.

Known Limitations

EyeDetect has several practical limitations that consumers and organizations should understand. Any substance or medication that affects reaction time, eye movement, or pupil dilation will compromise test results [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. This includes stimulants, depressants, and various prescription medications. For those interested in how mental health conditions interact with credibility testing more broadly, see our guide on mental health and polygraph testing.

The technology requires adequate reading proficiency — the unpublished Colombia field study and the Patnaik et al. (2016) findings both demonstrate that poor readers may not produce valid results [7]Verified Generalizability of an Ocular-Motor Test for Deception to a Mexican Population
Confirms cross-validation accuracy exceeding 80% for Mexican population, reading proficiency issues in Colombia field study, 147 university student participants. Visual acuity and conditions affecting the eyes could also potentially influence test outcomes.

Regarding countermeasures, one interesting observation from a journalist's demonstration was that the system incorrectly identified the chosen number in a concealed-number test, demonstrating that EyeDetect is not infallible even in controlled demonstration settings [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology. However, the technology's proponents note that because it measures involuntary ocular responses rather than autonomic functions, traditional polygraph countermeasures (like controlled breathing) would not be effective.

Published Accuracy Comparisons

Comparing EyeDetect and polygraph accuracy requires careful attention to which studies and metrics are being compared. The APA's landmark 2011 meta-analytic survey analyzed 38 studies meeting rigorous inclusion criteria, encompassing 3,723 examinations and 11,737 scored results [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy. That analysis found event-specific diagnostic polygraph tests achieve 89% accuracy (with a confidence interval of 83–95%), while multi-issue screening tests achieve 85% accuracy [12]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 11,737 scored results; event-specific diagnostic tests 89% accuracy (CI: 83-95%); multi-issue screening 85% accuracy.

EyeDetect's published figures — 86% for screening and approximately 90% for diagnostic tests per Converus's claims [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries — appear broadly comparable at first glance. However, the independent research picture is more nuanced. Multiple studies have found overall ODT accuracies of around 80% [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns, with rates ranging from 78% to 86% based on logistic regression analysis [3]Verified Laboratory and Field Research on the Ocular-motor Deception Test
Confirms mean ODT accuracy of 86% in standardization sample, 83% on cross-validation, with accuracy rates ranging 78-86% across studies. Earlier reviews of polygraph accuracy reported accuracy exceeding 0.90 [13]Verified Review of Polygraph Accuracy Research
Reported polygraph accuracy exceeding.90, consistent with findings by Raskin (2002) and Raskin & Podlesny (1979)[14]Verified Accuracy of Polygraph Techniques
Reported polygraph accuracy exceeding.90 using the CQT format, establishing foundational standards for modern polygraph testing, and a comprehensive survey by Abrams (1989) found an overall accuracy level of 0.89 [15]Verified Survey of Published Polygraph Literature on Accuracy
Comprehensive survey finding overall polygraph accuracy of.89, with earlier review showing 83% lab accuracy and 98% field accuracy.

Converus itself has acknowledged that in screening applications, the accuracy of polygraph falls in a range of 65 to 85% [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries, positioning EyeDetect as comparable to polygraph for screening. For a detailed head-to-head comparison in specific applications, see our EyeDetect vs. Polygraph for PCSOT guide.

Complementary Use: Polygraph and EyeDetect Together

Converus positions EyeDetect as both a complement and alternative to traditional polygraph [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The company describes EyeDetect as the perfect add-on service for polygraph examiners, noting that studies show comparable accuracy levels for pre-screening and periodic evaluations [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

The Concealed Information Test (CIT) approach used in some EyeDetect protocols has a strong theoretical foundation in orienting response theory, as demonstrated by extensive CIT research [16]Verified Current Research and Potential Applications of the Concealed Information Test: An Overview
Comprehensive overview of CIT research with strong theoretical foundation in orienting response theory, false positive rate of approximately 5% across studies. Research on the CIT shows an extremely low false positive rate of approximately 5% across studies [16]Verified Current Research and Potential Applications of the Concealed Information Test: An Overview
Comprehensive overview of CIT research with strong theoretical foundation in orienting response theory, false positive rate of approximately 5% across studies, which provides a useful baseline for understanding how concealed-knowledge approaches perform. For polygraph examiners considering whether to integrate EyeDetect into their practice, the technology's scalability — one proctor can manage up to three stations simultaneously — offers a practical efficiency advantage [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

Current Legal Standing

EyeDetect results are generally not admissible in court proceedings, similar to polygraph results in most jurisdictions. The technology has not yet been subject to the type of legal challenges (under Daubert or Frye standards) that would establish admissibility precedent [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns. A 2025 critical review specifically identified the proprietary algorithm as a significant barrier to meeting Daubert standards for scientific evidence, which require that methodologies be testable, subject to peer review, have known error rates, and be generally accepted in the relevant scientific community [6]Verified The Ocular Motor Deception Test (ODT): A Critical Review of Scientific Validity, Legal Risk, and Operational Readiness
Confirms vast majority of ODT studies from small academic-commercial group, proprietary algorithm prevents independent scrutiny, lack of generalizable deception signature, Daubert admissibility concerns.

For organizations using EyeDetect in pre-employment screening, the Employee Polygraph Protection Act (EPPA) governs the use of lie detector tests in the private sector. The legal landscape varies by state and by application — law enforcement agencies and certain government positions have different rules. Organizations should consult legal counsel before implementing any credibility assessment program. For related federal requirements, see our PCSOT and Adam Walsh Act guide.

Government and Law Enforcement

According to Converus, EyeDetect is currently used by more than 500 customers in 40 countries [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. In the United States, this includes the federal government as well as 21 state and local law enforcement agencies [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology. The technology has been demonstrated to officials from the U.S. State Department, Secret Service, Department of Defense, National Security Agency, Department of Homeland Security, FBI, and members of the U.S. Congress [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries.

Law enforcement has used EyeDetect in criminal investigations, including a case where it was deployed during a child sex solicitation sting operation supervised by the Idaho Internet Crimes Against Children task force [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. The technology has also been used for document verification in Ukraine and corporate compliance in Latin America [5]Verified EyeDetect Marketer Converus Spends Hundreds of Thousands of Dollars Lobbying Federal Agencies
Confirms Converus lobbying efforts, 500 customers in 40 countries, U.S. Embassy usage in Guatemala, unpublished Colombia field study showing erratic results, Kircher's advisory against sole reliance on any single screening technology. For information on how polygraph and EyeDetect are being adopted in African countries, see our guides covering Namibia, Rwanda, Zambia, and Botswana.

PCSOT and Sex Offender Monitoring

One noteworthy application area is post-conviction sex offender testing (PCSOT). The Mundt et al. (2022) study specifically evaluated ODT as a potential alternative to traditional PCSOT polygraph testing, finding results that support its possible use in this context [9]Verified Ocular-motor Deception Testing in Civilly Detained Sexually Violent Persons: An Alternative to Post-Conviction Sex Offender Polygraph Testing?
Confirms ODT tested in forensic PCSOT-relevant population, found accuracy rates not significantly less than published 80% benchmark, scoring algorithm modified post hoc with Converus input. This is significant because PCSOT represents a major use case for credibility assessment technologies, with regular testing requirements under the Adam Walsh Act.

For polygraph examiners considering PCSOT certification, EyeDetect may offer a useful supplementary tool, particularly when examiner availability is limited or when testing multiple individuals in a short timeframe.

Emerging Developments

Converus has continued to expand the EyeDetect platform. In addition to the standard EyeDetect hardware, the company has launched VerifEye, a mobile app version that uses a person's phone camera as an eye-tracking device for a self-administered 10-minute test [1]Verified Converus EyeDetect Official FAQs and Product Information
Confirms Converus claims 86-88% accuracy for screening tests, 90% for diagnostic tests, technology invented in 2002, commercialized 2013-2014, 500+ customers in 40+ countries. While this promises dramatically increased scalability, the accuracy and reliability of smartphone-based eye tracking compared to specialized infrared cameras remains to be independently validated. For our perspective on why app-based approaches face significant challenges, see our analysis of fake AI lie detector tests online.

The broader field of eye-based deception detection continues to advance. Recent research has explored AI models using fixations, saccades, blinks, and pupil size for detecting deception in concealed information tests [10]Verified Eye Movements as Indicators of Deception: A Machine Learning Approach
Independent AI-based deception detection using eye tracking achieved up to 74% binary classification accuracy; meta-analysis of cognitive lie detection studies found weighted mean of 74% overall accuracy. While current results are modest compared to established methods, machine learning approaches may eventually improve classification accuracy. Researchers have also investigated the integration of pupil measures into traditional polygraph scoring — Webb et al. (2009) suggested that pupil diameter could potentially replace the blood pressure cuff component of polygraph [4]Verified Effectiveness of Pupil Diameter in a Probable-Lie Comparison Question Test for Deception
Confirms pupil diameter is a significant predictor variable for deception and may replace traditional blood pressure measurement in polygraph testing, which informed the development of EyeDetect+.

For the polygraph profession, EyeDetect represents both a potential complement and a competitive alternative. The technology's scalability and automation address practical limitations of traditional polygraph, while the polygraph's established legal precedent, transparent scoring methods, and deeper research base provide enduring advantages. Polygraph examiners interested in the broader landscape of credibility assessment can find additional resources in our polygraph research hub.