Concealed Information Test (Guilty-Knowledge Technique)

Description

Show one true detail among equally plausible false alternatives and test whether autonomic responses single out the true one, revealing recognition without asking for a verbal answer.

The concealed information test was designed as an answer to the weakness of the control-question polygraph, where a pointed accusation arouses innocent and guilty people alike. Instead of looking for lies, the examiner picks a detail of an event that only someone involved would know and presents it among several contrived alternatives that are equally plausible to an outsider. A person without the knowledge should respond similarly to every option; a person who recognises the true detail tends to show a larger skin-conductance response to it. In Lykken's mock-theft study, electrodermal activity alone identified most of the volunteers who had taken part and misclassified none of the uninvolved controls. Its rigour rests on genuinely equivalent foils, several items rather than one, and the recognition that it measures orienting to familiar information, not deception.

Attributes

Loss type
Unmigrated
Stage
Evaluation
Effort
Weeks

Steps

  1. Choose details of the event or material that only a person with prior exposure could know, and check with naive pilot participants that they cannot guess them.
  2. For each detail build an item with the true option and four or five equally plausible foils, and present the options one at a time in random order.
  3. Record skin conductance (and optionally heart rate or respiration) throughout, leaving about 15 to 20 seconds between options so responses can recover.
  4. Score the response to each option, rank the true option within its item, and repeat across several items to accumulate evidence.
  5. Classify a participant as recognising the material only when the true option reliably produces the largest response across items, and compare against an uninvolved control group.
  6. Debrief participants about what was measured, and treat results as recognition evidence, never as proof of wrongdoing.

Sources

  • Biopsychology (Pinel & Barnes, 2022)

Also known as

  • Guilty-knowledge test
  • GKT
  • Concealed information test
  • Guilty-knowledge technique

Categories

  • Analytics & Behaviour
  • Neuro, Biometrics & Affect