Professor Grubin, thank you very much for joining us this afternoon and taking the trouble to come here. We are extremely grateful to youQ . Perhaps you could start by introducing yourself to the Committee in the context of your academic background, and, in particular, your work on polygraphs.
I am a professor of forensic psychiatry, so I am a psychiatrist and not a polygraph examiner. I became interested in polygraph testing about 20 years ago in relation to work with offenders. What I found was that polygraph testing was being used very widely in the United States to monitor offenders. The people using it said, “This is fantastic. If they took it away, I would quit.” They would make comments like that, but the academics felt that there was no evidence for it and a lot of what we are hearing today is that it is not reliable. A lot of those issues were repeated. I became interested in that difference. We began to run some studies here in the UK. Gradually over time, a lot of evidence accumulated to show that it was a very effective means of monitoring and managing offenders in the community.
Q You have a long academic background, principally at Newcastle University, and you have been studying polygraphs for 20 years. Can you describe the evidence you have seen in the last 20 years about the role that polygraphs can usefully play in the criminal justice system? Feel free to comment on evidence from overseas as well as the United Kingdom.
The first thing to say is that there is a lot of misunderstanding about polygraph testing. We heard a lot of that earlier today, and I get very frustrated, because those same comments get repeated and repeated. There is also a lot of confusion about polygraph testing—what it is, what it does and how it is used.
In essence, all polygraph testing does is provide additional information—information gain—and it does that in two main ways. One is the test outcome, which people often get tied up in—is somebody lying or telling the truth?—but it is also about disclosures. The two are complementary. What every study ever carried out on polygraph testing has found is that when people are having a polygraph, they make disclosures. All the studies we have done here, and indeed the implementation of polygraph testing here with sex offenders, has found the same thing.
There was a comment that this has not been piloted, but we have now run about 5,000 tests in probation, with mandatory tests on sex offenders. We have tested over 2,000 individuals and the police, with voluntary testing, have tested about 1,000 individuals and run about 2,000 tests. We have a lot of information, and again we find that about 60% to 70% of tests result in new information that was not known before and is important to management.
The other aspect, of course, is test outcome. People always want to know how accurate it is, and we know —we have very good estimates. The best study was a comprehensive review carried out by the National Research Council in the United States about 20 years ago, when it was being raised for security vetting in Government agencies. It looked at all the available evidence and found it was between 80% and 90% accurate. That means it gets it wrong about one in five or one in 10 times, but that is a lot better than we can do.
The main question then is: is that accurate enough for the application you want to put it to? What we are talking about is post-release, post-conviction testing as part of monitoring offenders, and in that capacity it is being used alongside a number of other aspects of offender management. You are not relying on the polygraph either to clear someone or to send them back to prison or anything like that; it is not used in that way. It is just additional information that can be added. If you think about different sorts of results that you might get, if somebody, say, passes a polygraph test—I do not like to use the term pass, but I will for simplicity’s sake—and they do not make any disclosures and there are no other concerns about the individual, that provides reassurance that you are not missing anything; it is an agreement with everything else. If, on the other hand, you get some disclosures, that is something that can be investigated further. If somebody fails the polygraph, so they are thought to be lying, and there are already concerns, again, that reinforces that, but if there are not, the polygraph may be wrong—it may be one of the one in five or one in 10 times we have gotten it wrong—but it may also suggest that you need to look at it a bit closer and investigate further.
There were comments before about how if somebody fails a polygraph they are brought back to prison or brought before the courts. That is just not the policy, and we have heard that in the legislation that just does not happen. It is simply a warning sign that you had better take a closer look. Again, we have a lot of evidence from the testing we have done in this country—as I said, over 7,000 tests have been run—to show that that is in fact how things are working.
Can I say one last thing? We often present polygraph testing as if it is something that offenders do not like and is being imposed on them. That is true for some, but others actually find it useful. You have to remember that sometimes you catch people telling the truth, and where you have an individual who is being monitored, because a risk is a great cause of concern and there is a suspicion of them all the time, and they can demonstrate that they are actually not doing anything wrong and their risk is static or decreasing, that is very useful for them. We have anecdotal evidence of offenders saying they found that part of the testing helpful, and they like polygraph tests for that reason—because they can prove that they are following the rules.
Q That is extremely helpful, thank you. What I think you are saying is that the context in which we are looking to use polygraphs for terror offenders, as for sex offenders at the moment in England and Wales, is as a prompt which may, in some circumstances, stimulate further investigation. Just to make sure I have understood you; your evidence is that all the studies you have seen say that that is a safe way of using polygraphs—as a prompt for further investigation—and that public protection is enhanced by doing it. Is that a fair summary of your evidence?
It is fair, except that I would say it is more than a prompt and that it actually uncovers information. You have to remember that a lot of this management relies on self-reports, so it is a way of saying, “We’re asking these questions anyway, only now we want you to tell the truth and we have a way of trying to determine whether you are telling the truth.” The other aspect, of course, which is often overlooked is its deterrence effect. If you know you are going to have a polygraph test, you are going to pay a lot closer attention to your activities, your actions and your behaviour. Again, we have a lot of anecdotal evidence—it is very difficult to prove—that people do modify their behaviour, because they know they are going to have a polygraph test.
Q We have heard very different views on the use of polygraphs. The assistant chief constable, Ted Jacques, said this morning that maybe a trial would be a good idea before it is rolled out in this particular piece of legislation, and Les Allamby said it is untested in this environment, which I suppose is one and the same thing. Is 80% accuracy good enough to recall somebody to prison?
No. The sex offender work is, in effect, a pilot for this, because even though the risks are different the underlying principles are the same: there are individuals who are a cause for concern and you have time to intervene if you are picking up warning signs. If they are making disclosures that indicate that the risk is increasing, that would be grounds for recalling them to prison, but that is because of something they have told you. If they told you in another setting, if they said it in an interview with a probation officer, they would be recalled on that basis as well. If they simply fail a polygraph test but they do not make any disclosures, nothing happens to them. The questions on which they failed are explored further and to say, “Maybe this is wrong, maybe one in five times it is wrong, but maybe there is something there that we have missed and we have to have a closer look.” That is followed up by further interviewing with the offender. There may be other investigations that are put in place. We have a lot of examples, with the sex offender work, where that has happened. I would say, in a way, that the sex offender work is just a very large pilot for this application.
Q It is interesting that you should major so much on disclosures, because, as you know, the Bill takes away the role of the Parole Board in determining sentences. That is the source of a tremendous amount of data for the authorities. Do you have a view on that?
I am not quite clear what you are asking. In terms of the disclosure, this is after they have been released so the tests are not being run in prison, they are being run in the community, so any issues with the Parole Board I do not think are directly relevant to the polygraph testing.
Q Okay. Finally, on young people, I have reservations about the Bill applying to young people in the same way and the same applies to polygraph tests. I wonder if you would like to comment on that directly in relation to young people, bearing in mind more general issues about mental health and the effects of such a regime.
There are two aspects: one is mental health and one is young people. I share your concerns regarding young people. It depends on what sort of age we are talking about. Certainly, I have had discussions about what an appropriate age might be. I am very clear that certainly any individual below the age of 16 should not be subject to a polygraph test.
You say “subject”—that is probably not the right word. The reasons why you would not want to use it under 16 are, first, we are not sure that brain development means the polygraph will work in the same way as it does with adults. We know there is a big change in brain development around the time of puberty. Around the age of 16, I think things are adult-like enough to mean that polygraph testing will be valid.
It is actually a bit older than that—I have seen 29. It is not a question of full maturation; it is a question of whether the brain has matured enough so that the polygraph works in similar way to how it works with adults. Again, there is a lot of confusion about what a polygraph detects. It does not detect lies; we know that. It detects activity within the autonomic nervous system that reflects cognate processing in response to questions. By the time somebody has reached the age of around 16, that looks similar to an adult’s.
That is an issue for training and oversight. There is an important thing for me with polygraph testing. A lot of the criticisms of it are not about polygraph but bad practice and the limitation of polygraph. It is very important that examiners understand issues around mental health and mental illness. If there are problems, they can either adapt their testing to take that into account or not do the test, depending on what the circumstances are.
Q That is helpful, but what do we need to do to improve the Bill to make sure that the issues you have just mentioned about mental health, mental health capacity and illness are taken into consideration?
I am not sure that that is something you can legislate for, apart from saying that there needs to be proper training and proper supervision. My concern always is that, being Government, one day somebody will want to save a little bit of money and will say, “We don’t really need this supervision quality control. They can just get on with it.” That is where I think danger lies. Provided that there is proper supervision, I do not know how much further you can legislate.
Q Professor Grubin, your evidence is fascinating. I think the reference to the pilot project earlier might have been in relation to jurisdictions where polygraph testing is not currently used. You will gather from my accent that my jurisdiction is Scotland—I have a legal background—and we do not use it there. You say it is part of a suite of risk management measures, so it is not pivotal but part of a suite. The previous witness pointed out that Jonathan Hall has written about Scotland’s very highly respected Risk Management Authority, and at present it does not use polygraph testing. If it were to be introduced in Scotland, it would require a pilot and various steps to be taken before it could be rolled out. I think that that is what he was referring to.
I was a member of a risk management authority for a number of years, so I know how they work and what they look at. When you talk about piloting, are you looking to get disclosures that will have the same levels of accuracy? There is no reason why a Scottish offender should be any different from an English or American one. The polygraph should work in the same way. There is a lot of experience now on how to implement. From my point of view, this is one of the few things where we have been able to scale up from pilot studies to actual implementation and to continue to keep its integrity and keep it working. I do not see why any of that would be any different in Scotland. I appreciate there are resource and training issues, but that would not be a reason not to pilot it. That would be a reason to get the training and implementation issues in place.
Q While we have you here, can I ask a couple of other questions to aid my own understanding? Sometimes people call polygraph tests “lie detectors”, in common parlance, but, as I understand it, that is not entirely correct. It does not measure lies; it measures the physiological changes in the central nervous system when somebody is asked a question. Is that right?
Q And you have to ask a very closed question such as, “Have you accessed the internet?”
Q Or, in an Irish context, if someone asks me, “Have you been to Dublin recently?” I have to confess that I have, so perhaps I would fail a lie detector test. Joking apart, it is not a lie detector test. It measures physiological changes. There is some scientific dubiety as to whether those central nervous systems are under the conscious control of the subject. What is your view on that?
They are not under the conscious control of the subject. We know that. Also, you get those responses not just from being deceptive; there is a range of things that can cause that response. In a polygraph test, somebody does not just walk into the room, get hooked up to a polygraph and then get asked questions. It is a fairly lengthy process. It takes at least an hour: typically two or three hours for a polygraph test. Most of that is spent in a pre-test interview where you go through information with the examinee with the aim of making sure that, if he is responding, he is responding because he is being deceptive and not for some other reason. That is where are lot of the training comes from and that is what differentiates a good polygraph examiner from a bad one: the way they have approached the interview and the test means that those responses are seen because of deception. It doesn’t always happen, which is why we get the one in five, one in 10 error rate. What you are looking for is physiological responses associated with deception. They can be associated with other things as well, but the aim of the polygraph test is to try to make sure it is because somebody is being deceptive.
Q Thank you for your evidence, Professor Grubin. I am interested in your view of how valuable the polygraph test is in assessing an offender’s intention in the long term, in comparison with the other tools available by way of standard psychological testing.
It is not valuable at all. You cannot use polygraph testing as a means of testing intentions. The polygraph is looking specifically at behaviours. Your colleague referred to concrete, very narrow questions of the type, “Have you done this?” They can be screening-type questions, or they can be very specific, such as, “Did you rob the bank?”, “Did you shoot the gun?” or whatever. It is not a tool for eliciting intentions or validating responses to those sorts of question.
Q You said that in normal circumstances the test is 80% to 90% accurate. My hon. Friend the Member for Stockton North raised questions about those with mental health problems and the issue of immaturity. I want to ask you about some other categories as to whether there are also concerns about the accuracy or appropriateness of tests: people with learning disabilities; people who are neurodiverse or have personality disorders; and non-English speakers. Are there concerns about the use of polygraph tests for those people and for any other vulnerable people?
For people with an intellectual disability, you are absolutely right that the accuracy of the test decreases once IQ drops below a certain level. In the sex offender testing, we will typically test down to 60, but we are much more cautious with the test outcome. It is still valuable, because of the disclosure aspect; you still get information and information gain—the point about information gain is the main one I want to leave you with—from the test, even with someone with an intellectual disability.
Again, examiners need to be trained; they need to address their questions in a different way, one that is much more concrete. The test has to be modified. It has to be shorter because of fatigue and issues such as that. So, you are absolutely right that accuracy decreases, but you must remember that nothing hinges on a test outcome alone. If it is a deceptive response and you have no other concerns, you would still look further. You might say, “We have to be more cautious because of IQ.”
There is no evidence to suggest it works any differently with people with personality disorders from how it works with anybody else. Again, because of misunder- standings about how polygraph works, people think, “It does not work with psychopathic individuals because they don’t feel anxiety.” First, the test is not based on anxiety. Sometimes when we do talks, and we will have an examiner, we can do demonstrations of polygraph testing. We used to like to get a volunteer from the audience who we can hook up. I try to pick somebody who is also a psychopath, so we can kill two birds with one stone. I know that here we would not be able to do that, but in the audiences I speak to there are often one or two psychiatrists who would fit the bill for a psychopath. There has been some testing of personality disorders and there is no evidence that the test itself is any less valuable. Again, part of the training of the examiner is that they need to know how to interview these individuals, because of the challenges that they may present.
I believe the third group you were thinking about was those with neurodevelopmental disorder or autistic spectrum disorder. Again, the evidence is that the test works just as well with them as with anybody else, but you have to make allowances in the interview, because of the concrete nature of a lot of their thinking, language difficulties and so on. You need to take that into account in terms of the interviewing, but there is no evidence to suggest that the test itself works any differently with them from how it works with anybody else.
We do test with interpreters, and they seem to work just as well. Again, it does take training for the examiner to know how to work with an interpreter, and the interpreter needs training as well. Certainly, security services in other countries use it with interpreters quite regularly.
Q Professor, you talked about a failure rate of up to 20%. What drives that failure rate? Is it the fact that people are tricking the test or that the examiner is making a mistake? What are the drivers behind that 20%?
There is a range of reasons that people can give either false positives or false negatives. Apologies for not looking at you while I am answering. Sometimes it is because the test hasn’t been set out properly, the examinee hasn’t been prepared properly in the pre-test interview or the questions haven’t been formulated well, and so on.
The examinee may have some other experience that is close enough to the way the question is being asked to cause that sort of response. For example, there was a very good study carried out in Israel. I won’t go through all the details of it, but they were able to debrief afterwards as we were with police officers who were applying for promotion. There were two false positives. The ground troops knew that these two people had been telling the truth, but they were said to be lying. One of them had said that he had previously made an insurance claim in Israel. At that time, the insurance companies in Israel would test people making insurance claims to see if they were honest or not. He said that he was being honest, but he was told that he was lying. He couldn’t get that out of his mind during the test. That causes the cognitive processing we were talking about, and it made him respond in that way.
The other person was more interesting. The experiment itself was about a test that the examinees could cheat on. You would know if they cheated or not. The second police officer said that he cheated when he took the test, but there was something wrong when he took it and the examiners had him do the test a second time. When he did it the second time, he said, “I don’t think I had better cheat again,” so he did it honestly. When he was asked if he had cheated on the test the second time, he said that he was thinking about having cheated the first time, which is why he reacted as he did.
There are other reasons as well, but it is hard to explain without going into the details about how polygraph testing works. Basically, you are comparing the relevant questions that you are interested in with so-called comparison questions. If those comparison questions are not evocative enough to elicit a response when a person is telling the truth to the relevant question, or vice versa, when they are too hot and the person is much more concerned about that question than about the relevant one, you can also get mistakes on the test.
The final reason is that sometimes we just don’t know; it just happens.
Q I have a quick follow up to that. In terms of the formulation of questions, with the sex offender work that has been done so far, how has that worked most effectively? What lessons can be learned from that when we think about applying it in terrorism offenses?
It is very similar. In sex offender testing, the majority of questions relate to their licence conditions and they are asked specifically about those conditions. You have to remember in a polygraph test and a screening test you get, at most, three relevant questions, so if they have 15 licence conditions you are only going to be able to test three of them. You can ask about all of them during the pre-test interview and, of course, the examinee won’t know which ones he will be asked on the test, which is why you get disclosures.
By and large, they are about licence conditions, and I would think that with this group that is what they would be. The things you would be interested in are undisclosed internet devices, have they been in contact with certain individuals, have they travelled to certain places and those sorts of question. The sex offenders are also asked about fantasies, but I am not sure that you would be particularly interested in that with this group.
Q My hon. Friend the Member for Brentford and Isleworth talked about people with disabilities in relation to polygraph testing. You said that the success rate goes down to about 60%. Is that a fair success rate to be used as part of the evidence for a recall to prison?
I think what I said was when IQ gets down to around 60; I did not say that the accuracy was around 60. I said that it becomes less accurate as the IQ lowers and that we typically would not test somebody with an IQ below 60.
Q That is extremely kind; I have two final points. We had some commentary from colleagues earlier, before you arrived, that there was no evidence that this would work with terrorist offenders. Given the work with sex offenders and the work on polygraphs around the world, can you comment on whether you believe this could be used to test terror offenders as part of their licence condition monitoring?
There are a couple of aspects to the answer to that. First, there is no reason to think it would not work similarly with terrorist offenders. They are people and they respond to polygraph testing like anybody else. It is used with terrorist offenders in other countries, but the problem is that that sort of work is not published. My understanding of it is anecdotal and what people have told me. They certainly find that its use is successful, and they get the same types of response that you would expect from the sex offender work. There is no real difference there. But none of that is published, so it is anecdotal.
One other thing to say from the sex offender work is that we looked at whether, after polygraph tests, there was an increase in actions taken by the probation officers managing those people. You get an increase by a factor of 10, sometimes higher, in actions taken. That does not necessarily mean recall to prison or charging with a new offence, but actions that mean you have an opportunity to reduce risk, which is really what you are looking for here. With any sort of offence where you have time to intervene, polygraph testing provides a good means to get that information to allow you to intervene and reduce risk.
Q That brings me on to the very last question. You have mentioned that one of the main benefits of polygraph testing is that it prompts or helps to persuade the offender to disclose information that they would not otherwise disclose. You described that earlier as “information gain”. Could you give us some examples of how that happens and the kind of information gain that you have seen occur as this has been used?
I will give you a couple of examples of that, but the first thing to say is that we do not know why it happens. There are various psychological attempts to explain it, but I know that I have been polygraph tested as part of our training and it was all I could do not to confess to the crime that I was meant to have committed. There is a real urge to disclose that I do not really understand, but there are various theories that I am happy to discuss later on.
To give you a couple of examples off the top of my head, one interesting case was a sex offender who was released from prison. Everything was thought to be going well with him. He disclosed that he had a new girlfriend, which was not known to the offender manager. That seems pretty mundane, but when they found this girlfriend it turned out that she was a single mother, that she was a vulnerable woman, and that this man was visiting her and helping her to paint her sitting room. He would do that in his underwear because he did not want to get his clothes painted. Her daughter was present at that time. A lot of that mirrored the way he had offended before, so that one disclosure about having a new girlfriend led to that man being recalled to prison —not based directly on the disclosure, but only indirectly, once the girlfriend was found and interviewed.
I am not sure just what you would mean. I can tell you, because I checked these figures before I came, that in the probation testing about 65% of tests resulted in new disclosures in the pre-test. That is information that was important to management but was not known. That might be small bits of information or it might be big bits. After someone fails a test, they are asked to explain why that might be, and about 60% of those tests result in further disclosures to try to explain that. What I cannot say is how many of those were in tests where there were no pre-test disclosures, so it is likely that about two thirds or 70% of tests result in new information.
That does not count something that I think is important but that is always overlooked: the truthful tests with no disclosures that provide reassurance, because decisions can be made on that. In the police world, they do voluntary testing of sex offenders on the register. Someone who is on the register for 15 years and wants to come off it may have been visited once a year for the past five years; there may be no intelligence on him, and an inspector is expected to sign off this person based on that information. If he passes a polygraph test and nothing of concern comes up, that gives them reassurance. Often, though, in those cases we find that bits of information do come up that they should have been aware of, and then they can move forward.
CTSB01 Dr Kyriakos N. Kotsoglou, Senior Lecturer in Law (Criminal Evidence), Northumbria University, and Marion Oswald, Vice Chancellor’s Senior Fellow in Law, Northumbria University
CTSB02 Dr Charlotte Heath-Kelly, Reader in International Security, Politics and International Studies, University of Warwick
CTSB03 Dr Rob Faure Walker (As part of SOAS COP for SOAS University of London)
CTSB04 Prison Reform Trust
CTSB05 Law Society of Scotland