Fun with Polygraph Exams

The polygraph is a sometimes controversial method of attempting to ascertain whether a subject's testimony is true or deceptive. Most states do not allow polygraph evidence to be admitted in court, and there is no accepted scientific validation of this practice. However, certain agencies of the Federal Government require polygraph exams of their employees and contractors, which seems in conflict with judicial practice and scientific research. I recommend the recent summary of the situation published in Skeptical Inquirer (Jan/Feb 2005 pp. 34-39) if you want to know the details.

But that kind of stuff is pretty boring. Read this article, check the references, blah, blah, blah. I thought I'd describe the polygraphs that I have personally taken, to provide a more interesting account. No, the FBI doesn't suspect me of being a serial killer or terrorist, I've taken the more mundane employment-related exams. But they are still kind of interesting: amusing in some ways and frightening in others.

The impetus for writing this account, though, is that after my most recent polygraph I'm pretty sure I figured out some of the mysteries of the polygraph examiners' black art. Specifically, I think I know how they look at those squiggly lines and decide whether you are telling the truth or not. Unfortunately, I'm afraid that this method does not work as advertised, but I'll let you be the judge of that presently.

Employment Polygraphs in the 1970's

My first polygraph occurred when I was 17 years old and looking for a job repairing electronic equipment. It's a little hard when you're under 18 to convince people that you know your stuff, so I studied a few books and then took the FCC 2nd class telecommunications license test. I was awarded license # P2-19-15604 (isn't it amazing how you can remember these kinds of things years later!) and armed with this credential, went looking for a job. [The 2nd class license enabled you to adjust CB and other two-way radios legally. The 1st class license enabled you to adjust commercial radio station transmitters. I had no real interest in either, but it was a cheap certification which included basic electronics knowledge. And I did later work on a few CB radios.]

My first attempt was with Allied Electronics Corp in Willowick, Ohio. They gave me their own test which I passed, and then everything was supposed to be set. They just had to call me to finalize everything and give me a starting date. Unfortunately, they didn't do it and when I finally called them, there was no job any more. I moved on to the next possible employer, Olson Electronics of Willoughby, OH. [The funny part about this story was six months later when Allied called me twice at my new employer, evidently trying to hire me. You guessed it, I never returned their call!]

Olson's had a policy of polygraphing their employees both pre-hire and yearly. When you're 17 with very little to hide you don't mind that, so I drove to downtown Cleveland and took the polygraph. I don't remember all the details, but the principle was to ask some questions that you would tell the truth on, and then some that you would lie on to get a baseline. I wasn't in the mood for lying, so I doubt if that went according to plan but after a fairly brief test of maybe an hour, as I recall, I was evidently cleared for employment and hired. [I actually wasn't an Olson's employee, but rather a subcontractor. Nothing was withheld from my paychecks; I had to deal with the tax stuff myself, a bit of a challenge for a teenager but I managed.]

Several months later, a substantial amount of merchandise was stolen from the store by the device of removing it from the boxes and then re-sealing the boxes so everything appeared to be intact. The number that sticks in my mind for the loss is $70,000. Olson's response was to order everyone to take a polygraph test except me, since I had been hired so recently. Now Olson's paid their salesman next to nothing, essentially minimum wage (or maybe even less) plus commissions, which might or might not amount to a living wage. One employee, Mike, used to open his weekly paycheck and complain loudly, "Look at that paycheck--I could make more than this on welfare!" So whenever a dozen or so such employees were ordered to take such a test under suspicion of being criminals, not all would agree.

In this case, I believe that three employees declined the test, and were thereby let go. One of them was Mike, and another was Jim, the other technician who I believe was doing more sales and less service since I arrived. Jim was a veteran of army electronics school and seemed to know some stuff, but I never figured out if he knew as much as me in the short time that I interacted with him. I heard nothing else about the investigation; it seemed that the assumptions was that one or more of those who declined the test were probably the guilty parties, but nothing could be proved.

Later that spring, I was walking in downtown Cleveland when suddenly I was hailed by a cab driver. It was Mike, in his new job. He readily confessed to being one of the culprits, and he may even have named an accomplice, but that's a little hazy. Anyway, he told me that if I could steal phono cartridges, he could fence them for me and give me 50% of the retail price! [For those of you who can't remember, music used to be distributed on LP records which were played by needles which vibrated around the grooves to reproduce the music. The cartridge was the whole assembly which converted the vibrations into electricity, generally electromagnetically. They usually sold for from $10 to $75 or so.]

I was even less inclined to go into a life of crime than I was to lie to polygraph examiners, so I declined. And that was that, until I contemplated my yearly polygraph exam. I realized that although I had nothing to do with any criminal activity, I was now in possession of information that might be of interest to the polygraph examiner!

A few months later I had my yearly polygraph re-examination. It went pretty much like the first one. There were questions about whether or not I had stolen anything, but no questions about whether I knew about anyone else involved in theft. A couple of times the examiner asked me what I was thinking about, and at least once I said that I was thinking about the heist six months before, but that seemed to satisfy him. I certainly could testify with a clear conscience that I stole nothing.

It seemed that the polygraph examiner missed an opportunity here. Could better questions have elicited any information from me about Mike's involvement in the crime? If such questions were asked, and my responses deemed inadequate and deceptive, would I have considered turning Mike in? Should I have reported my encounter with Mike to the authorities at any rate? Possibly so, but at the height of the Vietnam War the government wasn't the favorite organization of many young people my age so I never really considered it.

My last polygraph at the behest of Olson's started like the others, but at the very last minute, after I had driven all the way to downtown Cleveland, the polygraph examiner refused to conduct the exam. That was because of my age; he wanted me to be 21, as I recall! Naturally I protested that they were completely wasting my time. [In those days gas was $.379 per gallon, so it wasn't a big cost in gas.]

To fill in this security gap, they decided to polygraph me over the phone. They called me at work with a new fangled voice-analysis machine which was supposed to analyze my "yes" and "no" responses to see if I was lying. The baseline questions were a little silly such as whether I ever broke the speed limits. A few chuckles at these questions elicited stern rebukes from the examiner who wanted absolutely no sounds except "yes" or "no". Anyway, he went on to the real questions, and evidently I passed those OK.

That ended my early personal experience with the polygraph. I gave up the Olson's subcontract soon thereafter since it was interfering with my engineering studies, and just ran a smaller scale service business out of my home.

Aldrich Ames Defeats the Polygraph

Fast forwarding 20+ years the Aldrich Ames news story broke. Here was a guy who was high in counter-intelligence in the U.S. intelligence community, and was polygraphed every year, and yet was probably the most damaging traitor of the Cold War. He personally exposed many critical agents in the Soviet Union who were quickly executed. The whole story certainly would make you doubt the efficacy of the polygraph exam. If you aren't familiar with this sordid tale, you really should read about it. There must be some good on-line resources on such a critical intelligence failure.

My 21st Century Polygraph Experience

I recently took another polygraph test myself. There is a recommended pre-brief, at which the procedures are explained, ostensibly to make you comfortable with the procedure, but in reality to make you believe more in the effectiveness of the procedure. That way you should be more relaxed if in fact you have nothing to hide, and more nervous if you do. Of course your state of agitation is really all the polygraph examiner will be monitoring, although I'm pretty sure that many if not most of them are bigger believers in their system than I am.

At the pre-brief, you are warned that if you have committed small violations (stealing the paper clips, as it were) you should confess that before the exam, because it will inevitably come out anyway and that way they can exclude that confession from the question: "have you stolen anything from your employer besides the paper clips?" Of course it will only come out if you are nervous about it or if the polygraph is able to detect lies not related to any nervousness of the subject. Certainly to the degree that people believe in the polygraph they are likely to confess small violations, so that could be a positive. If you admit clip kleptomania and they keep you, you may be less likely to do it again!

They polygraph is billed as possibly taking a complete morning or afternoon, so you are urged to clear your schedule for it. The polygraph examiner did her best to keep me a believer, but since I don't believe in lying, even to polygraph examiners, it didn't take long for her to figure out that she was dealing with a lost cause there. The conversation went something along these lines:
Ms. P.E.: You've probably heard some stuff about the polygraph on the internet, etc...
Me: Oh yes, I know all about the polygraph and unfortunately it's bunk.
Ms. P.E.: I've been doing polygraphs for years and it is a very successful method of...
Me: Two words: Aldrich Ames
Ms. P.E.: Well, those polygraphs weren't done right at all; I've seen those charts...
Me: Please, you can always re-interpret the data to show that it should have been some other result. If anyone should have gotten the full treatment it's this guy. I'm very concerned that the government is placing undue faith in the polygraph. It would be much better to require the tax returns of everyone, and we'd have a lot more security.
Ms. P.E.: Well they are taking some more steps along those lines.
Me: I'm happy to hear it.

Well, I didn't want to create a hostile environment for Ms. P.E., and I certainly didn't try to say anything in a mean way. She had her job to do and she did it. It wasn't her fault that she got an incorrigible subject on this particular day! We got along fine for the rest of the afternoon.

The polygraph equipment seemed to consist of only a blood pressure cuff, a skin resistance hookup, and a diaphragm sensor to detect breathing, I assume. I was a little surprised that there wasn't more stuff, but I have to admit that I can't recall exactly what I was hooked up to 30 years ago. Of course there would be voice recording, as well.

Most polygraphs start with some kind of baseline procedure to try to get some lies and some truths from the subject to aid in interpreting the later, real questions. That's potentially a problem with someone like me who is unusually averse to lying, but non-liars are actually a pretty good segment of the population once you're hooked up to the machine! So I was curious how that would go. Ms. P.E. asked me to pick a number between 3 and 8, but not to pick 3 or 8. This wasn't a secret number; I wrote it on a card and she posted it on the board in front of me. Then she would ask me if certain numbers were the correct number and I would deny it for all of them, even for 4 which was the number I picked. I had to inform her that if we agreed that yes meant no, then saying no was not lying, which is defined as intentionally trying to mislead!

We went through the procedure, and I'm sure there was no distinguishable difference between my recitations of "no" in either my voice, my blood pressure, or my skin resistance, but she said that everything was fine, and my responses were good. Of course this would be more reinforcement for the idea that the polygraph can't be beaten, but was more said just as part of the routine in my case! (See my explanation of the polygraph art below for more on this).

For the real polygraph test, I was asked to think of some event in my personal life which contradicted a statement of hers, like "I have never made anyone really angry in my life." Once I had that incident in mind, I was told to reply "yes" to that statement while thinking about the contradictory episode. There was also an innocuous question to which I could answer truthfully. Interspersed were the real questions about whether I could be trusted to work in my current job. What struck me the most was the repetition of the questions. I had to answer both the control questions and the real questions several times. There also was a significant delay after each answer, to enable them to record maybe 10 or 15 seconds of data.

I had some questions in which Ms. P.E. said that I was breathing too hard, and she repeated them. Now that wasn't because of anxiety; I think it was mainly because I tend to breathe through my mouth more than the average person. If you have a lot of allergies you might develop this habit, but it seemed like it could affect my data. Halfway through the first hour session I developed a method which seemed to satisfy the examiner. While still breathing through my mouth shallowly I waited until I had inhaled to answer each question.

I completed the first session and then the second session, taking a total of about two hours. Ms. P.E. consulted with her fellow examiners on my data, and then told me that unfortunately my breathing was a little too uneven on the first hour, so we would have to repeat that. That's why they require you to free up a lot of time for this! I repeated it with my new, approved method of answering at the same phase of my breathing and completed it without incident. That didn't mean that I had passed. I was told that the final decision would be made back at headquarters after someone else reviewed the data, but in the opinion of Ms. P.E. I didn't look like a security risk. This correct conclusion could be viewed as support for the efficacy of the polygraph procedure (but not by me).

The last thing was a questionnaire about whether you felt the exam was an invasion of privacy (no), whether you would take another exam as part of a criminal or security investigation of some kind (yes) and whether you thought the use of the polygraph improved security at your facility (no). The whole thing took several hours, more time than I had spent in three polygraph tests 30 years prior.

How it Works (Secrets of the Polygraph)

After this experience I, being an engineer, naturally thought about how the whole thing worked (or was supposed to work). After each session, the examiner would be looking at squiggly lines representing blood pressure and skin resistance. Amusingly enough my job at work involves the automated analysis of squiggly lines, although not any related to biological measurements. If there was an identifiable feature that indicated a lie, such as a sudden dip in blood pressure or gradual increase in skin resistance, I could write a computer program to detect that, and we could dispense with the polygraph examiners altogether, doing the whole thing by computer with just a technician to attach the bio sensors. So that's not it.

The repetition of the questions is a big clue. That enables the PE to get several squiggly lines representing a "true" response and several representing a "false" response. I would assume that the heartbeat and the breathing provides reference points to which the PE might associate hypothesized lie-detector features. Then all they need to do is find some feature which distinguishes the true from the false. This will in general be different for every subject. One guy might show a a decrease in skin resistance at the second breath after the lying response, while another might show a blood pressure spike right after a lie.

Sounds pretty good, doesn't it. All our expert PE has to do is be alert for nervousness, do as much reading of the subject as possible, try to convince them that the polygraph is infallible, and be ready to accept any confessions that fall into their lap. Unfortunately the squiggly line stuff is voodoo. This is just exploiting the human ability to find patterns in white noise, like when the "face on Mars" was pointed out in pictures returned by one of the early Mars probes. Later, higher resolution data showed it was no face at all. The fact that the imagined patterns vary from subject to subject is due to the fact that we're trying to make randomness deterministic and the resulting "determinism" will itself be random!

If the PE thinks that the subject is telling the truth, he or she can group the responses to the real questions with the true control question, and then look for some feature supporting that partitioning of the data. If one set of several answer traces doesn't seem to fit, they could consider that a possible lie. For any partitioning at all, you could probably come up with some kind of plausible argument, just not a true one. But the main thing is that the overall state of nervousness as revealed by heart rate, skin resistance related to sweating, etc. should be constant and not showing increased nervousness around the key questions. I guess the ideal is increased nervousness around the control lies as well as some of the real questions, but I don't know how often that occurs.

The present sorry state of affairs

Unfortunately, spies of the ilk of Aldrich Ames are likely to know how to beat the polygraph. The most important thing is to convince the PE that you have nothing to hide, are being completely straightforward, and are not nervous. The PE will be easily able to find features of the squiggly lines that support that thesis, and a general lack of nervousness will support that. If the innocent employee is very nervous about the whole procedure, they may easily be flagged as possibly deceptive or worse.

I'm totally convinced that tracking the money is a better idea; Aldrich Ames had an ostentatious lifestyle that he never could have explained to CIA auditors. Nothing is foolproof, however, since some spies do keep their standard of living reasonable as the recent case of FBI agent Robert Hanssen demonstrates.

So in the midst of the war on terrorism, is there any hope that the polygraph can improve national security? Not in its present form, but research continues on better methods, such as described in this Wired Magazine article Check it out!

Page last updated 9 February 2006