Brain Fingerprinting (BFP) is an event-related potential (ERP) based technology that detects concealed knowledge utilising electroencephalography (EEG) to detect a distinct brainwave response pattern called the P300. In a forensic context, BFP is designed to identify when a suspect emits a brain response consistent with recognition of a stimulus comprising significant, specific information pertaining to a crime. The aim of the present study was to independently replicate the Brain Fingerprinting methodology and investigate whether this tool is as reliable and accurate as its inventor and other proponents claim.

Over the past century, and especially in recent decades, police and intelligence agencies have applied various technologies in an attempt to detect deception. Favoured forensic tools such as the Control Question Test (or “polygraph”) have been widely criticised by scientists, and these tools are struggling under the weight of comprehensive empirical evidence demonstrating them to be unreliable and inaccurate. Neuroscientists revolutionised “lie detection” by shifting focus away from the constructs of guilt or fear, and instead sought to detect concealed knowledge by measuring recognition of crucial crime-relevant information rather than the traditional method of making inferences about a suspect’s honesty based on their presumed emotional state.

While the scientific foundations of the ERP-based Concealed Information Test (CIT) are well-established, supported by decades of research, and generally perceived as legitimate by experts in the field, rival scientists have put forth a variety of different methods for applying the technology, and thus far, no consensus has been reached regarding which method is the most accurate and reliable. Brain Fingerprinting has consistently outperformed other modes of ERP-based concealed knowledge detection in all published research to date, with its proponents claiming the technique is over 99% accurate, whereas other similar methods have often produced higher error rates. However, BFP and those who promote the technology have been consistently criticised and challenged on various grounds, most notably on the basis that their highly accurate results have not been independently replicated thus far. The inventor of Brain Fingerprinting, Dr. Lawrence Farwell, cites 20 Scientific Standards (20SS) that distinguish the BFP method from other ERP-based CIT paradigms, and he posits that researchers who have previously attempted to replicate BFP, and reported low accuracy, have not applied the technique correctly according to these standards.

The present study is the first instance of a direct and faithful replication of Brain Fingerprinting, which stringently adhered to the 20SS. It is also the first reported instance of a false positive classification where the BFP method has been applied. BFP was 96.7% accurate at classifying university students as Information Present (possessing concealed knowledge) or Information Absent (no concealed knowledge) after they underwent a BFP test relating to either a significant event in their life, which they knew the specific details of, or a significant event in the life of someone else about which they knew no specific details. The BFP method utilises a bootstrapped cross-correlation mode of analysis known as the Classification CIT (ClaCIT) which resulted in a mean bootstrap probabilty of 98.8% (for correct determinations).

Another method of data analysis (bootstrapped amplitude difference, also known as the Comparison CIT or ComCIT), which is favoured by other ERP-based CIT experts and researchers, was also applied to the same data. This method yielded three false positives, but correctly classified the one participant who was misclassified using the BFP ClaCIT method. ComCIT analysis produced lower bootstrap probabilities overall, with only 14 of 28 tests resulting in a ComCIT determination that was both correct and valid (bootstrap probability over 50%). Implications for the applicability of Brain Fingerprinting, limitations, and generalisability of the findings are discussed. Further independent validation of the accuracy of BFP is recommended.