Inquest,

Welcome to the message board.  The currently and most widely used polygraph formats (control question tests)  seek to allow a polygraph examiner to make determinations about the presence or absence of examinee deception (actually not a determination of guilt).  The generally offered polygraph community explanation of how this works is based on the concept of "fear of detection."  I have previously posted material (part of a thread on underlying theory) for why I believe this to be an inadequate/incorrect explanation for what transpires and for what I believe actually does occur.  I will repost here that post and a preceding one that was posted by Anonymous (Addressed to Dr. Gordon Barland, formerly Chief of Research at the Department of Defense Polygraph Institute (DoDPI)).  I hope you will find this material useful in your analysis.  I would be happy to discuss it further when you have had an opportunity to read it.  I apologize for its length, but I’m afraid this is only the beginning…


First Post By Anonymous:

Gordon,

While I have enjoyed our exchanges regarding countermeasures/counter-countermeasures involving the CQT polygraph exam and don't in any way want to discourage you from continuing that thread (responding to the numerous outstanding questions/comments put to you as well as adding your own additional thoughts), I thought I'd introduce another fundamental line of discourse regarding the CQT. That would be to take a look at the fundamental theory of its practice. Although many theories have been offered over the years, it appears that polygraphers seem to return over and over again to the notion involving "Fear of Detection" on the part of both guilty and innocent (of relevant issues) examinees. Theory has it that, following the proper "setting" of test questions (a very subjective pre-test procedure worthy of its own discussion--another post) guilty subjects, upon in-test presentation, will respond most strongly to relevant questions and innocent subjects to control questions, thereby giving a basis for the general overall and spot scoring used (scoring, yet another future topic for our consideration) in evaluating CQT polygraphy.

The notion that I would like to raise at this point is a diametrically opposing theory for why CQT polygraphy would NOT be expected to work (would be expected to result in false positives): "Fear of Consequences." I am not suggesting that there are not other reasons (anger, surprise, revulsion, etc.) for why the "test" might be confounded on any given use, but I believe the suggested general mechanism is why one would question its working in a large number of situations.  I think it is quite reasonable to believe that anyone would react to being asked about involvement in certain crimes, not because of fear of being caught in a lie, but because of a fear of the consequences of having been so branded, independent of and regardless of whether that one (the polygraph examinee) had actually told the truth or told a lie during a polygraph exam regarding matter(s) under investigation.

Is it not reasonable to expect an examinee to respond to the relevant question "Did you rob the bank?" physiologically because he knows that deception-indicated polygraph results may lead to prosecution and conviction, denial of freedom, access to family, friends and worldly resources, etc.?  Furthermore, is this not even successively more and more likely in the case involving a capital crime, and one in which the examinee's being subject to the death penalty may have been mentioned in the course of the investigation prior to an exam? Perhaps the worst real-case scenario relates to a well-publicized case in which a CQT polygraph exam was given to a death row inmate within hours of the administration of his death penalty sentence. Is there any sane rationale for such an action, and, with the obvious psychological pressures commensurate with such a scenario, is there any way one could reasonably expect an innocent polygraph examinee, under such conditions, to focus on control questions (via some fear of detection mechanism) as opposed to the relevant questions that directly relate to his imminent demise (via a fear of consequences mechanism)? I think not.

Admittedly, I have gone from a rather commonplace criminal investigation (bank robbery) to what to me is one of the most egregious uses of CQT polygraphy on record, but one can imagine the same issues applied to a job applicant having an offer of employment riding solely on a polygraph exam or even look to a rather routine non-criminal setting to see evidence of what I am suggesting.  It is not uncommon for an individual who is having his blood pressure measured to respond with a sudden increase in systolic blood pressure (as much as 30 mm Hg), not because he is necessarily hypertensive but because he may well be hyper-reactive. The nature of this hyper-reactivity, the so-called "white coat effect," may well stem from a fear of consequences, i.e., the inability to maintain a certain job, inability to get or maintain life insurance, inability to participate in certain athletic activities, etc., following a determination of hypertension.  Is not this sort of phenomenon that we see fairly routinely evidenced with blood pressure readings (perhaps occurring through the suggested mechanism via the same autonomic response system in play with CQT polygraphy) actually what may well be happening with polygraphy? I would appreciate your thoughts on these issues.

Note: I expect to be preoccupied with other matters over the next several days, but do look forward to continuing our discussion(s).


Second Post by Drew Richardson:

The theory of "fear of consequences" for why a polygraph exam does NOT work may have some additional important ramifications beyond what Anonymous has already provided us with.  If the real-life consequences of relevant question material are what is driving polygraph results, this would tend to make the test more accurate for guilty people (who would be identified through present polygraph scoring if they are focused and respond more strongly to relevant-question consequences than control) and less accurate for innocent (whose focus and response to relevant-question consequences would be a confound for present polygraph formats and scoring stratagems).  It could well be that polygraph accuracy and paradigm validity is greater for guilty subjects than innocent subjects (perhaps the latter being accurately claimed by Lykken and Iacono to be somewhere near fifty percent (random chance) and zero (or completely lacking), respectively).

This would also explain why criminal polygraph examiners might be at least partially blind to the error they propagate.  In a real field criminal setting polygraph examiners do not develop their own cases.  Investigating case agents present them with cases and polygraph examinees.  A high percentage of those cases brought are from case agents who believe the suggested examinee is guilty of a crime and, in fact, are likely correct as evidenced through their investigation(s).  This phenomenon (a disproportionately high number of guilty subjects), the "fear of consequences" effect, plus any bias developed by the polygraph examiner as a result of having the case agent's investigative hypothesis would all tend to produce a higher degree of accuracy for the preponderance of criminal tests (those given to guilty subjects), but would leave examiners completely blind to and surprised by the error they produce with innocent criminal suspects and the legions of screening examinees who are largely innocent of relevant matters examined.

The theory of "fear of consequences" also has great implications for polygraph research.  Any paradigm which did not have sufficient consequences to invoke this phenomenon (assuming it to be a valid explanation) would have very little external validity and would tend to underestimate false positive results.  A typical polygraph screening study of the past would be a simulated crime scenario having college students or military recruits play the roles of programmed guilty or innocent subjects.  I strongly believe that neither the crimes nor the polygraph exams would have sufficient consequences to invoke such a reaction and physiological response and therefore be largely meaningless.  Let me explain.  A typical crime would be something like a theft of $10.00 from a desk drawer (again of no particular consequence--I have witnessed participants even laugh at the notion of such a crime and that an FBI agent would be investigating such).  Typically, examinees would be given some sort of financial reward (perhaps even $100.00) if they could pass the polygraph exam related to the simulated crime.  This latter exercise not only does not represent the consequences associated with real life polygraph exams, but also in fact is quite the opposite--a reward for some sort of achievement.  Following the polygraph exam, both guilty and innocent subjects leave the suite not to be further contacted, investigated or to have any further involvement with those who conducted the exercise---again, all this having been accomplished in the absence of any true consequences.  

All of this put together would lead to a variety of absurdities--not only the reward (as opposed to punishment/consequences) associated with polygraph decisions but the notion of having the would-be criminal committing a crime, not for the spoils of the crime, but for the privilege of entering the polygraph lottery where the rewards might be ten times as great as those associated with the original crime.  People do not commit crimes for the opportunity to profit from polygraph exams.  Aside from all the humor involved in such an exercise, the serious downside and flaw is that the lack of any consequences would result in an underestimate of the true rate of false positives in the real world and be severely lacking in external validity as a test paradigm.  Bill Iacono has quite correctly pointed out how field-based polygraph research that utilizes confessions for determining ground truth about subject guilt would also lead to the underestimation of the rate of false positives occurring in real life polygraph exams.  Polygraph research and how we fundamentally view polygraphy may well need to go back to the drawing board...

Inquest wrote on Jul 31^st, 2002 at 7:59pm:


It should be noted that such claims of accuracy are themselves highly suspect.  As is pointed out in TLBTLD, you have to consider the basis for the claim: if you have 100 subjects, with 99 being innocent and one guilty, you can attain a "99% accuracy rate" by simply declaring all subjects innocent.

You need to be very careful how accuracy claims are determined, as the above statements aren't terribly specific.  If you are innocent or guilty, what are the chances of being correctly categorized by the polygraph?  If the chances are low for either (for a given test group), or vary depending upon the percentage of innocent or guilty subjects in the test group, you have a little problem with the test itself.  Unfortunately, that seems to be the case with the polygraph.  When I get the time, I'll look up the link you have provided.

I would suspect that the majority of people who are "caught" by the polygraph are actually those who are persuaded by the polygrapher's interrogation skills to make voluntary confessions.  Such confessions have nothing to do with the accuracy of the polygraph itself.

Additionally, what is the basis for your claim that "the majority of people will have a greater response if their is some concrete association with the question"?  What do you mean by a "concrete association", and why would you not consider "fear of consequences" to fall into this category?

Skeptic

Drew Richardson,

Thank you for your replies. I see that there is some room for argument here which it seems could only be resolved by psychological research. My own background is in physics and, happily from my point of view, most of the signal processing aspect of polygraph can be easily verified including the statistical accuracy figures.

However, you have certainly made some valid arguments which I believe cannot be easily refutted or verified though, as psychology is not my field, I may be wrong on this point. If there is any literature you could recomend on this point I would be interested.

Inquest

Quote:


Yes, exactly. In fact, with the CQT test, the better one quantifies responsiveness the worse off a totally honest people is. The GKT (much more accepted in Japan, BTW) conversely, would benefit from a lowered variance. It seems to me that in the current anthrax investigation that GKT tests might be more revealing and accurate yet they likely are falling back to what they are used to, CQT's.

The basic problem is how does the FBI hire anyone? Few would fail to do minimal internet research and if they did, how would they answer the question about telling the truth knowing the examiner expects them to lie to the control questions?   It doesn't take much reading to understand quite intuitively how that test works.

-marty-

Public Servant,

Thank you for the time you expended on your last post.  You asked about the thread I previously referred to and provided two posts from.  That thread was begun approximately 15 months (on 5/11/2001 by Anonymous) ago and has not yet been substantively responded to by Dr. Barland; it is for that reason that I did not include his preliminary responses.  That thread can be found at http://antipolygraph.org/cgi-bin/forums/YaBB.pl?board=Proc&action=display&num=98....

With regard to control question testing and art, I, although nary a producer of art, am an admirer of such and those whose talents and inspiration produce it.  I hope to spend some time this weekend at galleries in Washington D.C. enjoying various examples.  The ability to enjoy that art comes from an ability to witness it and to critique it, if not to share in its production.  

This is where I believe your analogy with polygraphy and control question setting falls apart.  Although you may have the best of intentions in the "art" of setting control questions, neither you nor anyone else can objectively and quantitatively describe what constitutes this process.  Even more problematic is the fact that no one can objectively describe when it has been accomplished or whether it has been accomplished or offer evidence of either.

For a given relevant question/control question comparison that has resulted in a greater relevant than control question response, you can not know whether (1) this is indicative or a deceptive response to a relevant question, (2) an innocent individual reacting to the consequences of the relevant question, or (3) an innocent examinee for whom you did not "set" control questions properly (again, an unspecified activity) or as you had hoped to do.  Neither you, nor I, nor anyone else can know by looking at the charts which has occurred, yet this portion of an exam would contribute to an overall deceptive score.  It is this phenomenon which makes control question test polygraphy wishful thinking at best and hucksterism in the hands of the unscrupulous.  I'm sorry that I don't see the connection with the world of art.  The absence of science is not the evidence of art.  Perhaps you might try again to convince me.  Until such time as you can clearly describe what it is you hope to do with your art of control question setting and offer objective evidence that it has been accomplished, I'm afraid I see no basis for recognizing it as such nor as a basis for making important decisions that affect either individuals or society.

Yes, we are in agreement about concealed information tests.  I believe they do have great potential--their results can be statistically defended; they will not be viewed by examinees as offensive as is the case with the control question "setting" process in a CQT exam; they will not invade the purview of judge and jury; and, they can be followed with the same strong interrogations of deemed “to be knowledgeable” individuals as is the case with those deemed to be deceptive individuals following a CQT exam.  You are also correct inasmuch as these tests do require (amongst other things) that  investigators to collect, preserve, and properly protect probe information.  This, of course, should be standard procedure now and clearly a responsibility of every investigator and those entrusted with investigative information.  Regards,

Drew Richardson

Public Servant,

I would agree with you that both crime scenes differ as well as do polygraph examinees/polygraph examinations.  That is where I believe the analogy ends though with CQT examinations (there are considerably greater parallels with concealed information testing).  What a crime scene technician does, whether it be the collecting of DNA evidence or latent fingerprints is very intuitive, is easily documented and verified, and can be so sufficiently described in detail, as to be evaluated in terms of results and compliance with proper procedure.  I do not believe this to be the case with the setting of control question material in a polygraph exam (in spite of the fact that virtually every polygraph school would claim that it teaches its students how to do this).  

The problem with the setting of control questions in a CQT exam is not that it differs for various examinees, but that it largely cannot be described, defined, and defended for any one examinee.  Again, I repeat, what is necessary to convince me is not to suggest that this is art, but to describe for me and others what constitutes the proper setting of a control question with objective standards for allowing a neutral observer to evaluate whether and when this process has been completed for a given examinee.  And speaking of neutral observers, do you believe, for the purpose of an examinee who claims to have been wrongly found deceptive, that he/she and his attorney should be given audio/video tapes (you do record yours, yes?) of his/her pretest to allow for the expert evaluation of whether control questions were properly set (as well as a host of other issues which might be evaluated)? 

I am glad we agree on the possible interpretations for a polygraph result which includes a greater physiological response for a relevant question in a given relevant/control question comparison.  I see no reason though to accept your probability assessment for the likelihood of the various explanations.  I suspect that "the consequences of a deceptive polygraph finding" is what has caused the plethora of false positives testified to on this site with regard to polygraph screening.  I don't even find likely that this outcome has anything in particular to do with polygrapher error (the improper/incomplete setting of control question material, etc), although because this process is not sufficiently well defined it would be hard to disprove this explanation.  And no, I don't believe polygraphers to be universally or even largely disingenuous (let alone unscrupulous), but largely sufficiently lacking in relevant formal education as to be able to adequately evaluate the assumptions and theoretical underpinnings of their practice.

I am now working with Dr. Larry Farwell, who is the examiner/researcher who conducted the exam that was featured on the CBS 60 Minutes piece you referred to.  Both he and I would be happy to provide cites from the peer reviewed literature regarding the many issues you might care to explore regarding P300 evaluations or to discuss the principles of "brain fingerprinting" and its utility for specific issue testing and in narrowly focused screening applications (e.g. which of a group of a thousand detainees might have connections to al Qaeda, etc).  He and I have in recent months spoken to several groups of investigators, chiefs of police, etc. regarding this technology.  We would be happy to do any of the aforementioned for you and your organization as well.  Perhaps you might care to begin with the following article published in the peer-reviewed Journal of Forensic Science and authored by Dr. Larry Farwell and Ms. Sharon Smith (FBI Supervisory Special Agent and Instructor, Behavioral Science Unit, FBI Academy).  The bibliography of this article will provide you with numerous additional literature cites. 

cite:

Farwell, L. A. and Smith, S. S. (2001).  Using Brain MERMER Testing to Detect Concealed Knowledge Despite Efforts to Conceal.  Journal of Forensic Sciences 46,1:1-9.

You can find an on-line preprint version (does not include figures) of this article at

http://www.brainwavescience.com/FBILFSmithAsPubFigs.htm

Mark, et al.,

Speaking from direct experience with specific-issue CQT testing, I certainly feel that "fear of consequences" and "desire to succeed" are both very real, not to mention problematic, factors. It's beyond me how the polygraph community simply dismisses the fact that the examinee (even one with no polygraph knowledge) knows full well what the most important question(s) is (are) in the exam, and that this would translate to enhanced anxiety transcending any that might be produced during control questions.

I can directly attest to this, as when the relevant question came up on my poly, I would think "ok, this is the one, the most salient question, here we go". I could even feel myself getting tense when it came up. I sooooo wanted to show/prove that I was being forthright. Yet in the back of my mind I knew full well there's absolutely no way of physically proving the issue. All this totally contributed to my anxiety, and, you know what, it's a completely natural human reaction, manifested I'm sure by multitudes of other polygraph victims.  Of-course, having truthfully answered all control questions certainly didn't help.

False +