EyeDetect vs a Traditional Polygraph Examination: Why the Procedure Matters

EyeDetect and polygraph are frequently compared as rival technologies. Dr Keith Ashcroft, investigative psychologist and principal forensic polygraph examiner, explains why the most important difference is not between two instruments but between an automated test and a multi-stage professional examination.

Introduction

EyeDetect and polygraph are frequently compared as if they were rival instruments competing for the same role. Articles, marketing material and online discussions often frame the question as “Which machine is more accurate?” This framing, while understandable, misses the most significant distinction between the two approaches.

EyeDetect is primarily an automated ocular-motor credibility test. A professionally conducted polygraph is a multi-stage examination incorporating case analysis, suitability assessment, interviewing, question formulation, physiological data collection, transparent numerical scoring, post-test procedures and professional reporting.

EyeDetect may provide a rapid automated credibility classification. A traditional polygraph examination seeks to understand the case, define the evidential proposition, assess the examinee’s suitability, formulate appropriate questions, record physiological data and interpret the result within a wider professional process.

This is not a categorical superiority claim. The appropriate procedure depends on the purpose, setting and evidential question. However, anyone comparing these methods should understand that the comparison is not simply between two sets of sensors. It is between two fundamentally different levels of procedural depth.

What EyeDetect Measures

EyeDetect, developed by Converus, is a computer-administered credibility assessment system that uses an infrared eye tracker to record ocular-motor and behavioural variables while the examinee reads and responds to statements presented on screen. The recorded variables may include:

• Pupil diameter changes
• Gaze position and gaze tracking patterns
• Fixation duration and frequency
• Reading and rereading behaviour
• Response latency (reaction time)
• Response accuracy
• Blink rate and blink-related variables
• Other computer-derived behavioural measures

The proposed rationale draws on theories of cognitive demand. The underlying premise is that deception requires additional cognitive effort — including inhibition, strategic processing, memory retrieval and working-memory engagement — and that this cognitive load produces measurable changes in ocular-motor behaviour. The system uses a proprietary algorithm to classify the examinee’s responses.

EyeDetect has genuine practical advantages. These include:

• Rapid administration — a test may be completed in approximately 15–30 minutes.
• Standardised delivery — the computer administers the same procedure to every examinee.
• Limited variation between scorers — the same algorithm applies consistent rules.
• Suitability for high-volume screening — multiple examinees can be tested efficiently.
• Relatively simple deployment — the system requires a computer, eye tracker and controlled environment.
• Prompt availability of results — a classification may be returned within minutes of completion.

These are not trivial advantages. For certain applications — particularly large-scale screening or rapid triage — speed and standardisation are valuable.

What a Traditional Polygraph Measures

A modern polygraph examination normally records physiological activity through several channels:

• Thoracic respiration — upper-chest breathing patterns
• Abdominal respiration — diaphragmatic breathing patterns
• Electrodermal activity — changes in skin conductance reflecting sympathetic nervous system activation
• Cardiovascular activity — blood pressure, pulse rate and pulse amplitude
• Movement or activity data — seat-sensor or accelerometer recordings

These are indirect physiological correlates influenced by attention, cognition, emotion, behavioural control and stimulus salience. Importantly, a polygraph does not detect anxiety as such. Modern polygraph interpretation does not depend on observing whether someone “looks nervous.” It depends on structured comparisons — specifically, the differential physiological responses to different categories of question within a validated examination format.

The Difference Between a Test and an Examination

This distinction is central to understanding the comparison between EyeDetect and polygraph.

A complete polygraph examination includes substantially more than physiological recording. Each stage contributes information, safeguards and professional judgement that an automated classification cannot replicate independently.

Case Review

Before any testing, the examiner must assess the referral to identify:

• What conduct is genuinely disputed
• Whether there is a testable proposition
• Whether the examinee had an opportunity to engage in the conduct
• Whether the examinee possesses the relevant knowledge
• Whether the allegation is too broad to test validly
• Whether multiple issues require separate examinations
• Whether independent evidence is available
• Whether the examination request is ethical and defensible

This stage may determine that the proposed issue cannot be examined in the form initially requested, or that the referral is not appropriate for polygraph testing at all.

Suitability Assessment

A responsible examiner conducts a suitability assessment before proceeding. Considerations may include:

• Whether participation is informed and voluntary
• Medical conditions affecting physiological recording
• Current medication
• Psychological presentation and mental health
• Comprehension and cognitive capacity
• Language proficiency
• Fatigue, pain or physical discomfort
• Neurodiversity
• Memory limitations relevant to the target period
• Safeguarding considerations
• Coercion or external pressure to attend
• Ability to remain seated and engage with the procedure

Technically readable physiological data cannot repair an unsuitable or coercive examination. The fact that data were collected does not mean they were collected appropriately.

Pre-Test Interview

The pre-test interview is not a brief administrative introduction. It is a substantive professional interaction that may be used to:

• Obtain the examinee’s account of events
• Clarify the allegation and identify misunderstandings
• Define relevant terminology
• Explain the examination procedure
• Assess comprehension
• Review rights and consent
• Discuss medical or psychological factors
• Formulate the final test questions
• Ensure that every question is understood exactly as intended

This stage may reveal that the proposed issue cannot be validly tested, that the questions need reformulating, or that additional background information materially changes the assessment.

Question Formulation

The client’s original allegation should not simply be copied into a test. Effective question formulation requires:

• Behaviourally specific questions about clear past actions
• Defined time periods
• Simple, unambiguous language
• Avoidance of compound questions
• Avoidance of vague or subjective concepts
• Avoidance of questions based only on belief, opinion, intention or interpretation
• Separation of distinct allegations into individual questions or separate examinations
• A limited number of relevant questions consistent with the selected technique

The principle that “How many questions can I ask?” is less informative than “What type of questions can be validly examined?” is important. Adding more questions does not necessarily increase the amount of reliable information obtained. It may dilute the focus and reduce the precision of the assessment.

Acquaintance or Demonstration Test

This stage can:

• Familiarise the examinee with the procedure
• Demonstrate physiological responsivity
• Check the quality of the recordings
• Identify movement or recording artefacts
• Reinforce the examination instructions
• Indicate whether further explanation or accommodation is required

Physiological Data Collection

The relevant and comparison questions are presented through a structured and repeated sequence. The examiner monitors data quality throughout, checking for artefacts, question timing and adherence to the selected validated technique. This active monitoring is a significant difference from a fully automated procedure.

Numerical Scoring

Recognised polygraph scoring methods use published physiological features, numerical transformations, decision rules and cut scores. Not every examiner uses the same system, but validated approaches — such as the Empirical Scoring System — use documented criteria that another qualified examiner can review and, where appropriate, replicate.

This raises an important distinction: transparent or replicable scoring rules are not the same as an automated but proprietary algorithm. Both approaches have merits, but they offer different kinds of assurance.

Scoring Consistency Versus Scoring Transparency

EyeDetect offers strong procedural consistency. The same algorithm applies the same rules to similar inputs, producing classifications that are not affected by inter-scorer variability. This is a genuine strength.

However, consistency does not automatically mean transparency. Publicly available descriptions of EyeDetect may explain the broad type of variables recorded and the general statistical modelling approach without necessarily providing all coefficients, weightings, transformations or operational details required for an independent expert to reproduce an individual classification from the raw data.

A properly documented polygraph examination, by contrast, may allow another qualified examiner to review:

• The exact questions asked
• The chart recordings
• The specific physiological features scored
• The assigned numerical scores
• The decision rules applied
• The examiner’s report and reasoning
• Audio or video recordings, where retained
• Relevant case information

This level of transparency supports independent quality-control review and professional accountability.

It is important to note that polygraph scoring still involves professional judgement and is not free from human variability. Independent quality-control review and computer-assisted scoring can mitigate this, but it would be misleading to suggest that human interpretation is always superior to automated classification. Each approach has distinct strengths and limitations.

The Post-Test Interview and PENOR

A completed polygraph examination does not end when the physiological data have been collected. Where significant reactions are identified, best practice includes a post-examination notification of result and opportunity to respond (PENOR).

The examinee should have a reasonable opportunity to explain any significant reactions. Possible explanations may include:

• A misunderstanding of the question wording
• An incorrectly defined time boundary
• A related but different incident
• Partial involvement in the alleged conduct
• An unreported qualification or context
• A memory problem
• Previously undisclosed relevant conduct
• Concern about an associated issue rather than the precise conduct under examination

An explanation is not automatically accepted as true. Its significance must be assessed against the physiological data, the case context and the available evidence. However, providing this opportunity is a safeguard against misclassification and a procedural-defensibility measure.

Properly conducted, a post-test interview is not an interrogation. It is a fairness mechanism that allows additional information to be considered before a conclusion is finalised.

A rapid automated classification without this opportunity may produce an outcome but provides less contextual information than a complete examiner-led process.

Reporting and Quality Control

A full professional polygraph report may document:

• The referral purpose and background
• Consent and suitability considerations
• The examinee’s account
• The questions asked
• The technique used
• Data quality observations
• Numerical scores and decision rules
• The result and its basis
• Relevant explanations or admissions
• Limitations and caveats
• Quality-control review, where undertaken

This is not to suggest that a longer report is automatically a better report. The value lies in transparency, traceability and the ability to understand how the conclusion was reached. A brief automated classification may be entirely appropriate in a screening context, but it does not provide the same level of professional accountability as a documented examination.

Concise Comparison: EyeDetect and Traditional Polygraph

Can EyeDetect and Polygraph Be Combined as “Successive Hurdles”?

A proposed model suggests that EyeDetect and polygraph can be administered together as “successive hurdles.” Under this approach, the examinee completes both procedures concerning the same target issue. Where both methods reach the same conclusive classification, it is argued that confidence in the joint result should be higher than confidence in either test alone.

Converus has described the two tests as independent or largely independent on the basis that EyeDetect records ocular-motor and behavioural variables while polygraph records respiratory, electrodermal and cardiovascular activity. The argument is that because different physiological channels are recorded, errors in one test should not predict errors in the other.

This reasoning deserves careful examination.

Different Sensors Do Not Establish Statistical Independence

Statistical independence does not simply mean that two tests record different parts of the body. For errors to be independent, the factors that increase the chance of error in one test must not materially increase the chance of error in the other.

Both EyeDetect and polygraph may be influenced by overlapping psychological and cognitive processes, including:

• Attention and alertness
• Autobiographical memory retrieval
• Recognition and stimulus salience
• The orienting response
• Language comprehension
• Working memory capacity
• Response inhibition
• Motivation to appear credible
• Concern about consequences
• Fatigue
• Deliberate attempts to influence the assessment

EyeDetect and polygraph can reasonably be described as different and potentially complementary measurement systems. That is not the same as demonstrating that their errors are statistically independent.

The Mathematical Problem with Combined Accuracy Claims

Consider a simple hypothetical example. Test A has a 17% false-negative rate. Test B has a 20% false-negative rate. If the errors are statistically independent, multiplying 0.17 by 0.20 gives 0.034, or 3.4% — implying a 96.6% chance that at least one test correctly identifies the deceptive case.

This multiplication is appropriate only if the relevant errors are genuinely independent. If the same features — such as poor comprehension, unusual cognitive processing, weak memory, fatigue, high motivation or countermeasure preparation — increase the risk of error in both procedures, the joint error rate may be materially higher than 3.4%.

Without valid joint data from studies in which both methods were administered to the same participants under controlled conditions, any combined accuracy figure should be treated as an estimate based on an unverified assumption, not as an established fact.

Base Rates and Posterior Confidence

Sensitivity and specificity are not the same as the probability that an individual result is correct. The predictive value of any credibility assessment — including two agreeing results — depends on additional factors:

• The prior probability or base rate of deception in the population being tested
• Whether the objective is screening or specific-issue diagnosis
• The relative costs of false-positive and false-negative errors
• Whether inconclusive outcomes are excluded from the reported accuracy figures
• How disagreement between the two methods is handled
• The quality and comparability of the two examinations

These considerations apply to any credibility assessment method. They are not specific criticisms of EyeDetect or of the successive-hurdles concept.

Order Effects and Lack of Blinding

Additional methodological considerations arise when two assessments are administered in sequence:

• The second assessment may not be independent because the examinee has already encountered substantially similar questions.
• Familiarity with the target issue may affect attention, reading behaviour or physiological responses.
• The examinee may adapt strategies between tests.
• If the second examiner knows the first result, confirmation bias may influence the conduct or interpretation of the second examination.
• The same case assumptions or poorly formulated questions may contaminate both assessments.

Any genuine successive-hurdles protocol should, where possible:

• Use independently formulated but proposition-equivalent questions
• Ensure the second examiner is blind to the first result
• Predefine how inconclusive and conflicting outcomes will be handled
• Document the order of administration
• Validate the complete combined process rather than multiplying accuracy figures from unrelated studies

The Required Evidential Standard

The independence assumption should ideally be examined through joint validation studies in which:

• Both methods are administered to the same participants
• The same underlying proposition is examined
• Ground truth is independently established
• The order of testing is controlled or randomised
• Examiners are blinded to the other method’s result
• False-positive and false-negative error correlations are reported
• Inconclusive results are retained and reported
• The complete combined decision rule is tested prospectively

Separate studies reporting accuracy for each method individually cannot alone establish the accuracy of the combined process.

Practical Value of Successive Hurdles

The successive-hurdles concept should not be rejected entirely. It may still offer practical value where EyeDetect is used as:

• A preliminary screen or triage tool
• A way to identify subjects or topics requiring fuller investigation
• An additional source of information
• A scalable first-stage assessment

A full polygraph examination may then add:

• Detailed case analysis
• Suitability assessment
• An examinee narrative
• Individually formulated questions
• Multi-channel physiological recordings
• Transparent numerical analysis
• An opportunity to explain significant responses
• Professional reporting
• Independent quality control

The value of the second procedure should arise from the additional information and procedural safeguards it contributes — not merely from using two headline accuracy percentages to produce a larger combined number.

Balanced Conclusion

EyeDetect has legitimate strengths in speed, standardisation and scalability. It offers a genuine contribution to the field of automated credibility assessment and may be well suited to high-volume screening applications where rapid classification is the primary objective.

A traditional polygraph examination offers greater procedural depth because it incorporates professional interviewing, suitability assessment, question formulation, multi-channel physiological data collection, transparent numerical scoring, post-test safeguards and contextual reporting. These elements are not incidental additions. They are integral to producing a result that can be professionally defended, independently reviewed and meaningfully interpreted.

EyeDetect may complement polygraph in an appropriately designed assessment pathway, but it should not automatically be regarded as a procedural substitute for a complete examiner-led examination. Equally, a polygraph examination conducted without proper case review, suitability assessment and question formulation does not necessarily offer more than an automated test.

The central question is not which machine “detects lies” more effectively. It is whether the referral requires a rapid automated classification or a complete professional examination capable of defining the issue, assessing the examinee, testing a properly formulated proposition and explaining how the conclusion was reached.

Scientific and Legal Cautions

• Neither EyeDetect nor polygraph should be treated as conclusive proof of truthfulness or deception.
• Results from either method should be considered alongside independent evidence.
• Suitability and voluntary participation must be established before any credibility assessment.
• Legal admissibility and employment use of credibility assessment methods vary by jurisdiction.
• In the United Kingdom, polygraph results are not generally admitted as evidence of guilt or innocence in criminal proceedings, although they are used in statutory offender management and in a range of civil, investigative and therapeutic contexts.
• This article is general information and does not constitute legal advice.

Frequently Asked Questions

This article was written by Dr Keith Ashcroft, investigative psychologist and principal forensic polygraph examiner at the Centre for Forensic Neuroscience. Dr Ashcroft is a Chartered Psychologist registered with the British Psychological Society and a full member of the American Polygraph Association.