MERMER

MERMER - Memory and Encoding Related Multifaceted Electroencephalographic Response

P300 response derived from the EEG data at different sites. The main component of a MERMER is the P300 wave, an evoked response that has been well studied in the scientific literature as a potential indicator of recognition. The MERMER response is not present in subjects who lack specific knowledge about the word, phrase, or picture presented. MERMER is utilised in brain fingerprinting

Contents

1 MERMER Methodology 2 Paradigm of MERMER 3 Comparitive MERMER Methodologies 4 Deception Method 5 Sampling Methodologies 5.1 Sample Design 6 Conceptual Variables

MERMER Methodology

The procedure used is similar to the Guilty Knowledge Test; a series of words, sounds, or pictures are presented via computer to the subject for a fraction of a second each. Each of these stimuli are organised by the test-giver to be a “Target,” “Irrelevant,” or a “Probe.” The Target stimuli are chosen to be relevant information to the tested subject, and are used to establish a baseline brain response for information that is significant to the subject being tested. The subject is instructed to press on button for Targets, and another button for all other stimuli. Most of the non-Target stimuli are Irrelevant, and are totally unrelated to the situation that the subject is being tested for. The Irrelevant stimuli do not elicit a MERMER, and so establish a baseline brain response for information that is insignificant to the subject in this context. Some of the non-Target are relevant to the situation that the subject is being tested for. These stimuli, Probes, are relevant to the test, and are significant to the subject, and will elicit a MERMER, signifying that the subject has understood that stimuli to be significant. A subject lacking this information in their brain, the response to the Probe stimulus will be indistinguishable from the Irrelevant stimulus. This response does not elicit a MERMER, indicating that the information is absent from their mind. Note that there does not have to be an emotional response of any kind to the stimuli- this test is entirely reliant upon recognition response to the stimuli, and relies upon a difference in recognition- hence the association with the Oddball effect.

Paradigm of MERMER

The brain MERMER testing is a paradigm change that has resulted from the discovery of MERMER by Dr. [[Lawrence Farwell]. This discovery was brought about by the failure of polygraphs and other attempts at lie detection to solve the issue of a universal detector of knowledge of a crime. This inability to understand the mechanisms of lying or the ability to omit pertinent facts undetected has been acknowledged as a crisis by the forensic science and criminal science communities. Dr. Farwell discuss that is unclear what the impact will be of this new technology upon how crimes are investigated. However, if this technology proves legitimate and is accepted by the forensic science community, it will require entirely new ways of collecting evidence from a crime scene that is “specifically suited to identifying the perpetrator through brain MERMERs.” Farwell also suggests a societal paradigm shift with the advent of MERMER testing, wherein an innocent person can easily prove their innocence through a simple scan, which is harmless, and requires no action or testimony from them.

With this paradigm comes a reverse problem—when testing is so easy and harmless, arbitrary and random testing becomes acknowledged as a societal sacrifice that one must perform in order to constantly prove that one is not a threat. A good example of this is Airport Screening, where MERMER testing could easily be applied to the task of terrorist detection. In this era of fear and the dogma of the war on terror, it is easy to imagine a relinquishing of one’s cognitive freedom in exchange for proving ones’ innocence.

Comparitive MERMER Methodologies

Three other experimental methodologies are discussed in the introduction in order to address problems of the validity of the experiment. The study performed in this paper draws heavily upon earlier methods in prior studies.

MERMER was utilised to identify seventeen FBI agents using FBI-centric probes and targets, with four control (non-FBI) subjects. [1] The stimuli consisted of words, phrases, and acronyms of internally used FBI terminology. Each of the controls and FBI agents were correctly discerned.

The second study was commissioned by the CIA and consisted of three experiments that examined the interchangeability of indicators. The first consisted of pictorial stimuli with probes and targets directly related to a simulated espionage enacted by some of the subjects. The second used verbal stimuli of military medicine probes and targets to identify U.S. Navy officers with military medicine training. The third used verbal stimuli to identify subjects with knowledge of actual crimes. All 79 subjects in these three trials were identified correctly as “knowledge-present” or “knowledge-absent”.

The weight of evidence in this body of experimentation would imply that systematic replications had been performed and had been found consistent with the conclusion. Visual and pictorial stimuli were both shown valid methods of presentation. Live situation engagement, mock acting, and academic knowledge were all proven valid test criteria.

Deception Method

In the J. Forensic Sciences study [2], three couples were tested, each couple having some prior knowledge of one another and an established relationship outside the scope of the test. One of the subject’s results were eliminated from the final results due to lack of understanding of the test, although the authors specifically state that a subject can be tested again accurately after another explanation.

The equipment used was a computer with two monitors with appropriate input and processing hardware, a four-channel EEG amplifier system, a custom electrode headband, and custom software for data acquisition and analysis. In the future, I will refer to this equipment set as the MERMER system.

Brain responses were recorded from the midline frontal, central, and parietal scalp locations, (Fz, Cz, and Pz respectively) referenced to linked mastoids (behind the ear), and from a location on the forehead to track eye movement.

Each pair of subjects were interviewed about one specific actual event that they knew about the other person. Person A was interviewed about something that only A & B would know. Person B was interviewed about a different event that only person B & A would know. Both subjects were interviewed separately and were instructed not to discuss the subject of the interview with the other participant.

Nine “Probe” stimuli were constructed for each participant based on the interview of the other subject. The examples given in the paper are related to a Birthday Party, with the relevant Probes being the location, “Bosco’s”, another attendee, “Jim Jones,” the nature of the celebration, “a Birthday Party,” and an action the test subject might have participated in, “lit candles.”

Irrelevant stimuli were based off of the Probe stimuli. For example, a restaurant name “Henry’s” could be presented, or an attendee “Bill Johnson” would be introduced. This would help to create a baseline between Probes and Irrelevant stimuli. The stimulus consisted of names, words, or phrases up to 20 characters presented on a computer screen to the subject under computer control. For each Probe stimulus, there were four Irrelevant stimuli, and these were structured such that the Probe and Irrelevants were indistinguishable for someone that did not have specific knowledge of the event being tested for. In addition to the Probes and Irrelevants, 1/6th of the stimuli were Targets, one per Probe. The test subjects were given a list of Target stimuli, and were advised that they had to be able to recognise and identify these items during the test. Therefore, the Targets were distinct and significant for all subjects.

In all, there were 9 Probes, 9 Targets, and 36 Irrelevants for each event. These were divided into three subsets, with each subset containing 3 Probes, 3 Targets, and 12 Irrelevants.

Each stimulus presentation was provided to the test subject for a three-second duration. In each trial, a stimulus is presented on the screen and the ERPs are recorded by the MERMER system. The tests are conducted in blocks of 72 stimuli, which lasts roughly three and one half minutes. Each block consists of one stimulus subset presented 4 times. Trials that included eye movement or other muscle-generated noise were immediately discarded. The order of the presented stimulus was randomised during each presentation.

The subjects were instructed to use a mouse to relay their recognition of Probes, Irrelevants, and Targets, by pressing the left-hand button for Target stimuli (which they had studied) and the right hand button for Probes and Irrelevants. Even if the subject recognised the Probe, they were to register it via the button press in the same method as they would register an Irrelevant. Thus, the tested subject “concealed” their knowledge of the investigated event. There was no overt confirmation of their knowledge of the Probe, and therefore, the only way to discern their knowledge was via the MERMER system.

The hypothesis was that the Probes would elicit a MERMER for the knowledgeable subjects who had participated in the event under investigation, and the Probes would not elicit a MERMER for the subjects that were not knowledgeable because they did not participate in the event.

Sampling Methodologies

In every published study by Dr. Farwell, he has used a non-probability sampling method with which to select his subjects. in the J. of Forensic Science Deception Study, The three chosen pairs of subjects were between the ages of mid-20s to early-40s, and each pair had a pre-established relationship before the study onset. The sampling method appears to be reliant upon available subjects; however, the only statement made about the subjects is that “Pairs were not randomly selected” (p.4). This is vague and gives no insight to the selection criteria or method. During the course of the experiment, the testers replaced one subject with a secondary subject that was known to the original partner in the pair. The experimenters stated that the replaced subject failed to understand of the directions. This matter seems unclear, as the title of the study states that the goal was to “detect knowledge despite efforts to conceal”. In spite of such a small number of subjects, a misunderstanding was enough to exclude and replace a subject entirely. In their discussion section, the experimenters use this isolated population to generalise to all populations, stating that this technology is widely applicable to all people. However, it should be clear that with such a carefully selected, screened, and exclusive group, this is hardly provable.

Sample Design

All Brain Fingerprinting research has used an experimental methodology primarily using the counterbalanced quasi-experimental design described by Campbell and Stanley. The subjects of these studies were not randomly chosen, and were measured and tested through a series of experiments and observations. A typical subject would be pre-screened and trained for the experiment, and not observed before the experiment. The experiment consisted of a series of stimuli: targets, probes, and irrelevants. These stimuli were names, words, or phrases up to 20 characters in length. The irrelevants were stimuli chosen that the subject will have no prior knowledge of. The target would be something that the subject recognised and knew about, and knew that the testers knew about. The probe stimulus was something that the test subject had not told the testers about, and was instructed to hide the knowledge of. The probe stimulus was acquired from the subject’s partner in the study, who offered up this information for the purpose of its possible use as a “probe”. After the presentation of each stimulus, irrelevant, probe, or target, the experimenters would record the subjects EEG waves after each. The order of each type of stimulus was different for each of the six subjects, creating an incomplete and non-through quasi-experimental counter-balanced design. The deception experiment consisted of 73 stimuli divided into three subgroups. Each subgroup consisted of three probe stimuli, three targets, and 12 irrelevants for a total of 18 stimuli per subset. The stimulus presentation was in a randomised order, which is not as systematically through as the described counterbalanced test. However, it could be surmised that is in an acceptable approximation due to the inadequate number of subjects to complete such a study in an internally valid way. A properly conducted counterbalanced design is robust and valid in all but one aspect of validity: interaction of selection and maturation. History, testing, maturation, instrumentation, regression, and selection will not contaminate the main effects of Xs because of the Latin-square arrangement of stimuli arrangement. The experimenters arranged their stimuli arrangement by random selection, which could contaminate significant interactions. In terms of external validity, this experiment design is sorely lacking. Because the interaction between the treatment (presentation of stimulus) and the test is not observable or verifiable, it remains an unknown variable. The selection bias and test interaction is a very serious concern, particularly in light of the experimenter’s decision to exclude 16.6% of the subject data due to an inability to follow directions. The interaction between selection and X provides the “possibility that the effects validly demonstrated hold only for that unique population from which the experimental and control groups were jointly selected” (Campbell & Stanley). The artificiality of the experimental condition creates a detrimental reactive arrangement that is ungeneralisable to the human population that the experimenters apply this study to in their discussion. The nature of this study lends itself to a quasi-experimental counterbalanced design, however, the use of the Latin-square, and a through number of randomly selected subjects would be required before the test could acquire adequate external and internal validity. A control group in which no response to probes or targets could occur could also be used to create a more valid Solomon four-group design or posttest-only control group design.

Conceptual Variables

The researchers are investigating “information present” and “information absent” variables. These variables can be operationalised by how accurate the probes and targets were in contrast to the irrelevants. The predictor variables are the “information present”, and the outcome variables are the % accurate. The scale that is used it the microvolt similarity between probes and targets during an EEG, in contrast to irrelevants in response to the associated stimuli. If a target is similar to an irrelevant stimuli when compare to the probe stimuli, then the information is considered “absent”. If the target stimuli are closer in microvolts to the probe than the irrelevant stimuli, the information is considered “present” in the brain. Thus far, these measures have proven to be very accurate, and false positives are highly unlikely. However, false negatives are possible to invoke purely through mental re-enactment of upsetting events .