The Accuracy of Human Decisions and Objective Measurements in Psychophysiological Detection of Knowledge

I examined the relative detection efficiency of three measurements--skin resistance response (SRR) amplitude, respiration line length (RLL), and human respiration response evaluation (RRE)--in a guilty knowledge experiment. Thirty-two subjects were presented with lists of personal information in which only one of five items was correct. Subjects were instructed to try to avoid detection of the relevant items in the polygraph test. All three measurements discriminated better than chance between relevant and neutral items, with the objectively measured SRR and RLL superior to the subjective RRE. When SRR and RLL were compared for their relative efficiency, the former yielded better detection. The differentiation was even better with a measure that combined the SRR and the RLL. The results are discussed with respect to previous findings and practical implications.

Laboratory experiment with 32 subjects presented with multiple-choice questions containing personal information (one relevant item among five total). Subjects were instructed to avoid detection while three measurement approaches were compared: automated skin resistance response amplitude, automated respiration line length, and human examiner evaluation of respiration patterns.

Background & Context

This 1994 study by Eitan Elaad addresses a fundamental tension in polygraph science: whether objective, automated physiological measurements outperform subjective human interpretation. While the Guilty Knowledge Test (now called the Concealed Information Test or CIT) had demonstrated validity in detecting concealed information, the field debated the optimal balance between human expertise and computerized analysis.

The GKT framework tests whether examinees possess knowledge they wish to conceal by presenting multiple-choice questions where guilty subjects recognize the relevant item and show heightened autonomic responsivity, while innocent subjects cannot distinguish relevant from neutral alternatives. The question of measurement methodology was particularly pressing given the growing availability of computerized scoring systems in the early 1990s.

Elaad's research built on emerging evidence that objective measurements might reduce the interpretive biases inherent in chart reading. His position with Israel's Division of Identification and Forensic Science provided direct insight into operational challenges where examiner subjectivity could influence outcomes in high-stakes contexts.

Research Design & Methodology

The study involved 32 subjects who were presented with lists of personal information in which only one of five items was correct, and subjects were instructed to try to avoid detection of the relevant items in the polygraph test. This design used personally meaningful information rather than mock crime scenarios, making recognition natural and automatic while maintaining experimental control.

The research examined three distinct measurement approaches: skin resistance response (SRR) amplitude, respiration line length (RLL), and human respiration response evaluation (RRE). The first two represented objective, quantifiable metrics while the third captured traditional examiner judgment of respiratory patterns—one of the most difficult channels for human interpreters to score reliably.

Key methodological features included:

• Multiple-choice format with one relevant item among four neutral alternatives per question
• Use of personal information (names, dates, identifying details) ensuring genuine recognition
• Counterdetection instructions motivating subjects to conceal their knowledge
• Parallel comparison of automated metrics versus human subjective evaluation on identical data

Results & Key Findings

All three measurements discriminated better than chance between relevant and neutral items, with the objectively measured SRR and RLL superior to the subjective RRE. This represented a clear validation that automated approaches outperformed human judgment, at least for respiratory channel assessment.

When SRR and RLL were compared for their relative efficiency, the former yielded better detection. Skin resistance—the electrodermal channel—emerged as the single most effective measure. However, the study revealed an even more powerful approach: The differentiation was even better with a measure that combined the SRR and the RLL.

Key empirical outcomes:

• SRR amplitude showed the strongest individual detection efficiency
• RLL (objective respiration measurement) exceeded human RRE evaluation
• Combined SRR + RLL produced the highest overall detection rates
• Human evaluation of respiration proved the weakest of the three approaches tested

Discussion & Significance

These findings provided empirical support for computerized scoring systems and challenged the notion that human expertise necessarily enhances detection accuracy. The superiority of objective RLL over subjective RRE was particularly noteworthy since respiration is among the most complex physiological channels to interpret, involving multiple pattern features (amplitude, rate, baseline shifts, suppression).

The combined-measure finding aligned with signal detection theory principles suggesting that independent information sources, when properly weighted, yield better discrimination than single measures. This supported polygraph practices that integrate multiple physiological channels rather than relying on any single indicator. The study contributed to the broader meta-analytic literature on GKT/CIT validity, later incorporated into Ben-Shakhar and Elaad's 2003 meta-analysis showing an overall average effect size of 1.55 for electrodermal measures, with mock-crime studies producing the highest average effect size of 2.09.

Elaad's work emerged during a transitional period when polygraph shifted from purely analog chart interpretation toward hybrid and fully computerized systems. The results suggested that automated algorithms could capture physiological differentiation more reliably than subjective pattern recognition, at least for certain channels.

Limitations & Considerations

The laboratory context using personal information rather than mock crimes limited generalizability to field polygraph examinations where crime-related knowledge is tested. Field research has consistently shown lower detection rates for guilty suspects (65-76%) compared to laboratory studies, though innocent suspect accuracy remains high at approximately 94%.

The study focused specifically on respiration for the human judgment component, not testing whether subjective evaluation of electrodermal activity might perform differently. With only 32 participants, statistical power to detect smaller effect size differences was limited. The personal information paradigm, while ecologically valid for knowledge detection, created different motivational dynamics than criminal investigation contexts where stakes and stress levels differ substantially.

Practical Applications

This research provided empirical justification for polygraph programs to adopt computerized scoring algorithms, particularly for respiratory measurements where human interpretation is most vulnerable to inconsistency. The finding that combined measures outperform single channels reinforced multi-channel recording protocols and supported development of algorithms that optimally weight different physiological indicators.

For examiners, the results suggest that objective metrics should be preferred when available, though human judgment remains valuable for quality control, artifact detection, and contextual interpretation that algorithms cannot capture. The study influenced subsequent development of standardized scoring systems and highlighted the importance of validated, objective decision rules over subjective clinical judgment in forensic applications.