Investigative Support October 28, 2024 Birmingham, UK 3 hour session

Witness Credibility Assessment

A detailed case study showing how 8‑channel BrainBit EEG P300 analysis achieved 93% internal classification accuracy in a high-value commercial dispute, while a parallel traditional polygraph protocol reached 48% accuracy under the same conditions. The assessment was used to inform investigative and settlement decisions, without any claim of court admissibility.

Important Use & Limitations Notice

Context: This case study describes an internal and investigative assessment carried out in the context of a high-value commercial dispute between two corporate parties. All names and identifying details have been anonymised or adapted.

No Court Claim: Nothing on this page should be read as stating that this protocol, or EEG-based assessment in general, is legally admissible in any court or legal forum. Admissibility is case- and jurisdiction-specific and requires independent legal advice.

Purpose: The information provided here is for educational, technical and investigative planning purposes only. It is not legal advice, does not guarantee results, and should not be used as the sole basis for any legal or commercial decision.

Commercial Background

A UK-based property investment company (“Meridian Holdings Ltd” – anonymised) engaged a regional construction contractor (“Blackstone Construction Group” – anonymised) for a mixed‑use development in Birmingham city centre. The dispute centred on whether an informal discussion between senior representatives had created additional verbal expectations about accelerated completion, beyond the written contract.

Both parties agreed that a key one‑to‑one meeting took place on 15 March 2024 but gave incompatible accounts of what was actually said. With substantial money at stake, both sides wanted a structured way to examine the consistency of their primary witnesses before finalising any negotiated outcome.

£4.2M
Value in Dispute
18
Month Project
2
Primary Witnesses
6
Months Delay Alleged

The Disputed Meeting

  • Date: 15 March 2024, 2:00 PM
  • Location: Contractor’s Birmingham office (meeting room)
  • Attendees: “Sarah Williams” (client CEO) and “David Chen” (contractor project manager)
  • Key Question: Did Chen agree to a verbal acceleration of completion milestones that was never written into the contract?
  • Practical Impact: If such a verbal understanding existed, one side considered the subsequent delays a serious breach of trust and expectations.

The Credibility Challenge

From an investigative point of view, the situation presented classic credibility problems:

  • No independent record: No audio recording, no minutes and no third‑party witnesses were available.
  • Direct conflict: Both individuals gave detailed and confident but mutually incompatible accounts.
  • High incentives: Both sides had strong financial and reputational reasons to be believed.
  • Memory delay: Several months had passed between the conversation and the assessment.
  • Clean history: Neither individual had an obvious prior record of dishonesty at work.

The instructing parties wanted something more structured than subjective “impressions” of credibility. Traditional polygraph testing was suggested as one option but raised immediate concerns:

  • Variable performance in high‑stakes environments.
  • Sensitivity to general stress and anxiety rather than specific deception.
  • Known vulnerability to certain countermeasures.
  • Mixed reputational status in many professional settings.
For both sides, the priority was to understand which version of events was more internally supported by objective physiological data, without pretending that any single tool could deliver an infallible “truth verdict”.
— Internal Assessment Brief (Extract)

EEG-Based Assessment Approach

With agreement from both parties, DeceptionDetection.co.uk was asked to design and conduct a combined EEG and polygraph assessment focused on comparative performance and internal consistency, not on producing a legal “truth label”. The 8‑channel BrainBit EEG system was used to monitor recognition‑related P300 responses under carefully controlled conditions.

Why Consider EEG for Credibility Work?

  • Neurophysiological basis: P300 responses and recognition‑related event‑related potentials (ERPs) are widely described in neuroscience literature.
  • Structured metrics: EEG allows quantitative analysis of timing, amplitude and reproducibility of responses to specific stimuli.
  • Complementary to autonomic data: EEG provides a different information channel than heart rate, respiration and skin conductance used in polygraphy.
  • Pattern‑focused: Emphasis is on relative patterns (baseline vs critical items) rather than single‑trial readings.
  • Documentable protocol: Every stage of stimulus design, presentation and analysis can be documented and independently reviewed.

Scope and Boundaries of the Assessment

  • Objective: To compare the internal performance of EEG and polygraph in differentiating each witness’s own “neutral”, “accepted” and “disputed” statements.
  • Not a lie detector verdict: Results were framed as probabilistic indicators and patterns, not as a categorical declaration that any individual “lied”.
  • No legal guarantee: The protocol was not presented or represented as automatically acceptable in any legal setting.
  • Two‑party agreement: Both sides agreed in writing to the protocol and to how results would be reported back to them.

Structured Assessment Protocol

Pre‑Assessment Phase (1 week prior)

Each participant completed medical and psychological screening to rule out obvious contraindications (e.g. active neurological disorders, certain medications). Written informed consent was obtained, explaining the aims, limitations and non‑legal nature of the assessment. Participants were reminded that they could withdraw at any time.

Baseline Establishment (≈45 minutes)

Neutral, non‑controversial questions about work history, role descriptions and undisputed project facts were used to establish each person’s typical EEG and polygraph response profiles. This step produced an individualised baseline against which later responses were compared.

Control Question Block (≈30 minutes)

Each witness answered questions on details both sides accepted as accurate (e.g. project start date, agreed written milestones, known meeting locations). These items were used to model “agreed‑truth” response signatures for each individual.

Disputed Meeting Block (≈60 minutes)

Carefully structured questions explored specific elements of the 15 March conversation: proposed acceleration dates, phrases allegedly used, and follow‑up commitments. EEG focused on recognition and consistency patterns between these items and baseline/control responses.

Concealed Knowledge Block (≈45 minutes)

Lists of short statements were presented, mixing details that only a person present at the meeting would be expected to recognise with plausible but incorrect alternatives. The goal was to examine involuntary recognition responses rather than verbal claims alone.

Polygraph Comparison Block (≈90 minutes)

A standard polygraph protocol using similar question themes ran in parallel with the EEG assessment. This allowed a like‑for‑like internal performance comparison under the same overall conditions.

Comparative Results & Internal Analysis

Credibility Pattern Assessment Results (Internal Metrics)

8‑Channel EEG P300

93%

High internal discrimination between responses consistent with baseline “accepted” items and those showing atypical attenuation or recognition patterns (p < 0.001 in this protocol).

Traditional Polygraph

48%

Low internal discrimination with strong, non‑specific stress responses across multiple question types, reducing clarity for decision‑making.

EEG Findings by Participant

  • Witness A (“Sarah Williams” – client CEO): Displayed stable P300 amplitudes during baseline and control questions (mean 8.7 μV at Pz). Across questions about the disputed meeting, her responses remained within the expected confidence interval for “accepted” items. No systematic attenuation associated with the key elements of her narrative was observed.
  • Witness B (“David Chen” – contractor PM): Showed typical baseline P300 activity (mean 9.2 μV) but marked amplitude reduction (to ≈3.1 μV on average) in response to several critical statements he verbally rejected. At the same time, certain phrases he denied using triggered enhanced recognition‑type responses (peaks ≈12.4 μV).
  • Recognition effects: In Witness B, items labelled as “never discussed” nevertheless produced clear P300‑style recognition signatures, suggesting internal familiarity inconsistent with his verbal responses.
  • Polygraph overlay: In both participants, polygraph channels indicated sustained arousal across large portions of the protocol, limiting their usefulness for fine‑grained credibility discrimination.

Statistical Overview

  • P300 latency window: Both participants showed typical peak latencies around 280–320 ms post‑stimulus for recognised items.
  • Amplitude variance: Witness B exhibited >60% reduction in P300 amplitude for specific disputed elements, relative to their own baseline.
  • Consistency index: Witness A’s responses remained 97.3% consistent across baseline, control and disputed blocks. Witness B’s overall pattern consistency dropped to 23.7% when disputed elements were included.
  • Internal false‑positive estimate: <2% under the specific thresholds and analysis rules calibrated for this protocol, based on independent validation data.

Key Assessment Takeaways (Non‑Legal Interpretation)

  • The EEG protocol achieved 93% internal accuracy in differentiating consistent vs inconsistent responding across predefined item categories in this case.
  • Witness A’s physiological pattern aligned closely with their stated version of events across both neutral and disputed items.
  • Witness B’s pattern included classic markers of internal conflict or inconsistency, including P300 attenuation and recognition responses to verbally denied details.
  • The polygraph component, at 48% internal accuracy, provided limited additional clarity in a high‑arousal context.
  • EEG‑based assessment offered the instructing parties a more structured basis for understanding relative credibility patterns, without claiming absolute “truth detection”.
  • All interpretations were presented as probabilistic and contextual, emphasising limitations and the need to integrate other evidence and professional judgment.

How the Results Informed Decisions

The two corporate parties reviewed the combined EEG and polygraph report with their advisers. While exact negotiation details remain confidential, several practical outcomes were reported back to us:

Reported Commercial Outcomes

  • Re‑evaluation of positions: The clearer pattern in favour of Witness A’s account prompted a re‑examination of risk exposure for the contractor.
  • Accelerated negotiations: Having structured, quantitative data reduced the temptation to prolong the dispute purely on the basis of “I’m more believable than you”.
  • Financial settlement: The parties ultimately reached a negotiated financial resolution significantly closer to the client’s original claim than initial offers.
  • Cost‑benefit reflection: Internal estimates suggested that the assessment cost (≈£15,000 including both EEG and polygraph components) was substantially lower than the projected cost of fully litigating the dispute.

Individual and Organisational Implications

  • Internal review: The contractor initiated an internal review of project governance and verbal commitment practices.
  • Leadership perception: The client’s leadership team reported increased confidence in their CEO’s handling of contentious negotiations.
  • Future conflict handling: Both organisations explored adding structured assessment options into internal dispute‑resolution policies.
From a risk‑management perspective, the EEG data did not “decide the case” for us, but it helped quantify which narrative was better supported by consistent physiological patterns. That clarity was a major factor in reaching a commercial resolution.
— External Risk Consultant (Post‑Matter Debrief)

Technical Methodology Details

EEG Equipment Specifications

  • Device: 8‑channel BrainBit EEG system (assessment configuration).
  • Sampling rate: 250 Hz per channel with 24‑bit resolution.
  • Electrode layout: 10–20 system positions: Fz, Cz, Pz, P3, P4, T7, T8, Oz.
  • Signal processing: Real‑time artefact handling with Independent Component Analysis (ICA) and post‑hoc quality control.
  • Data handling: Encrypted storage with tamper‑evident logging of acquisition and analysis steps.

P300 & ERP Analysis Parameters

  • Latency window: 250–500 ms post‑stimulus for P300‑like responses.
  • Amplitude metrics: Peak‑to‑peak voltage at Pz and related sites, averaged across repeated presentations.
  • Significance threshold: p < 0.01 for key comparisons after correcting for multiple tests.
  • Trials per condition: Minimum of 50 valid epochs per critical item category.
  • Analysis approach: Pre‑registered criteria for what counted as “consistent” vs “atypical” responses, with independent analyst review.

Quality Assurance Measures

  • Cross‑system check: Selected segments were compared with recordings from a hospital‑grade reference EEG system to confirm signal fidelity.
  • Inter‑rater agreement: Three analysts independently scored key segments, with 97.3% agreement before discussion.
  • Test–retest checks: Short repeat blocks were run within the same session to verify stability of observed effects.
  • Blinded interpretation: Analysts evaluated responses without knowing which party each witness represented or what outcome either side preferred.

Practical Lessons for Organisations

This case study suggests several ways structured EEG‑based assessment can be used carefully and realistically:

In Commercial Disputes

  • Decision support: Provides an additional reference point when deciding whether to escalate, negotiate or settle.
  • Scenario testing: Helps explore how different narratives align with measurable patterns of recognition and consistency.
  • Evidence mapping: Encourages parties to identify which factual elements are most central to credibility disputes.

In Internal Investigations

  • Fraud and misconduct: Can be integrated into broader investigation frameworks where witness accounts are heavily disputed.
  • Policy breaches: May assist HR and compliance teams in complex cases when used alongside documentary and digital evidence.
  • Training effect: Awareness that such tools exist may encourage more careful record‑keeping and clearer communication.

Important Caveats

  • Not a stand‑alone answer: EEG and polygraph data must be combined with other evidence and professional judgment.
  • Context‑dependent: Performance figures (such as the 93% internal accuracy in this case) depend heavily on protocol quality and participant characteristics.
  • Legal variability: Laws and practice regarding scientific and technical evidence vary by jurisdiction; you must obtain independent legal advice before relying on any such method in legal processes.

Future of EEG in Credibility Assessment

Although this case study is focused on one specific protocol, it points towards broader developments in the field of credibility and deception assessment:

Short‑Term Developments

  • Refined protocols: Continued optimisation of stimuli design, timing and analysis rules to reduce variability.
  • Combined metrics: Integration of EEG with behavioural, linguistic and digital‑trace data for richer assessment models.
  • Training and standards: Growing emphasis on training assessors and developing best‑practice guidelines.

Longer‑Term Possibilities

  • AI‑assisted analysis: Use of machine learning to identify subtle multidimensional patterns beyond simple single‑channel metrics.
  • Cost reduction: Hardware and software advances are likely to reduce cost per assessment over time.
  • Broader awareness: Increased understanding among organisations that such assessments can inform—but must not replace—human and legal judgment.