Why do cops use a mind-reading tool?

Matthew Best

Penultimate Amazing
Matthew Best
Joined
Jun 12, 2006
Messages
11,161
Location
Leicester Square, London
Why Are Cops Around the World Using This Outlandish Mind-Reading Tool?

Scientific Content Analysis, or SCAN for short. I must admit I'd never heard of it before, but it is apparently widely used in law enforcement all over the world, despite seeming, to my untutored eyes, like so much hogwash.

SCAN ... encourages the asking of a simple, open question: What happened? After the person writes a statement, the SCAN investigator looks for signs of deception, analyzing, among other things, pronouns used, changes in vocabulary, what’s left out and how much of a statement is devoted to what happened before, during and after an event. Indications of truthfulness include use of the past tense, first-person singular (“I went to the store”); pronouns, such as “my,” which signal commitment; and direct denials, the best being: “I did not do it.” Signs of deception include lack of memory, spontaneous corrections and swapping one word in for another — for example, writing “kids” in one place and “children” in another.

But who am I to judge, when I have the combined minds of the International Skeptics Forum to do that?
 
Matthew Best said:
Why Are Cops Around the World Using This Outlandish Mind-Reading Tool?

Scientific Content Analysis, or SCAN for short. I must admit I'd never heard of it before, but it is apparently widely used in law enforcement all over the world, despite seeming, to my untutored eyes, like so much hogwash.
Click to expand...

Your link mentioned work by Vrij who is a leading researcher on detecting deception. This is a more recent (2015) publication where on page 27 he compared SCAN with two other content-based methods (criterion-based content analysis (CBCA) and reality monitoring (RL)). He also concludes that none of the methods are sufficiently reliable to be used as evidence in criminal cases, although SCAN comes out worst in terms of supporting evidence.
 
There is a good reason for using such tools knowing they do not work
Police - This device says you are lying
Suspect - It is right, I confess.
Or
Suspect - The machine is wrong. I am telling the truth.

They then tell the truth. It is often how lie detectors work.
 
Elaedith said:
He also concludes that none of the methods are sufficiently reliable to be used as evidence in criminal cases, although SCAN comes out worst in terms of supporting evidence.
Click to expand...


Linguists are developing theories of a linguistic fingerprint - that each person has a word usage specific to him/herself. These theories are absolutely not widely scientifically tested, let alone proven reliable. In any case, it takes a trained linguist to attempt such a thing. Police officers, no matter how dedicated, aren't going to be able to master the skills currently needed to attempt such analysis.


rjh01 said:
There is a good reason for using such tools knowing they do not work
Police - This device says you are lying
Suspect - It is right, I confess.
Or
Suspect - The machine is wrong. I am telling the truth.

They then tell the truth. It is often how lie detectors work.
Click to expand...


Yeah, it's a version of good cop/bad cop where the machine/fake-science-thing is the bad cop. It works. That's why it's a favorite. However, it's never going to be sufficient evidence for a conviction (except in the deep south where the standards for scientific evidence have taken a vacation). Even a signed confession isn't sufficient evidence without corroborating physical proofs.

The lesson is: if you're going to be a criminal, read up on interrogation tactics first. The prepared criminal is the successful criminal, I always say.
 
Loss Leader said:
Linguists are developing theories of a linguistic fingerprint - that each person has a word usage specific to him/herself. These theories are absolutely not widely scientifically tested, let alone proven reliable. In any case, it takes a trained linguist to attempt such a thing. Police officers, no matter how dedicated, aren't going to be able to master the skills currently needed to attempt such analysis.





Yeah, it's a version of good cop/bad cop where the machine/fake-science-thing is the bad cop. It works. That's why it's a favorite. However, it's never going to be sufficient evidence for a conviction (except in the deep south where the standards for scientific evidence have taken a vacation). Even a signed confession isn't sufficient evidence without corroborating physical proofs.

The lesson is: if you're going to be a criminal, read up on interrogation tactics first. The prepared criminal is the successful criminal, I always say.
Click to expand...


Just keep your mouth shut until you have an attorney, even if you are one ;)
 
Loss Leader said:
Linguists are developing theories of a linguistic fingerprint - that each person has a word usage specific to him/herself. These theories are absolutely not widely scientifically tested, let alone proven reliable. In any case, it takes a trained linguist to attempt such a thing. Police officers, no matter how dedicated, aren't going to be able to master the skills currently needed to attempt such analysis.
Click to expand...

I'm a bit confused about whether you mean the theories linguists are developing are not tested or reliable, or the theories behind existing methods of content analysis that police in several countries already use, (that don't depend on linguistic fingerprints), are not reliable (although one criticism is that there are no valid theories behind the methods).

Yeah, it's a version of good cop/bad cop where the machine/fake-science-thing is the bad cop. It works. That's why it's a favorite. However, it's never going to be sufficient evidence for a conviction (except in the deep south where the standards for scientific evidence have taken a vacation). Even a signed confession isn't sufficient evidence without corroborating physical proofs.
Click to expand...


False confessions seem to be implicated in as many as 25% of wrongful convictions, and it has been discussed before in this subforum that different forms of evidence are not independent and that confessions can contaminate the collection and interpretation of other forms of evidence. People have even been wrongly convicted based on confessions where other forms of evidence such as DNA actually excluded them.

The lesson is: if you're going to be a criminal, read up on interrogation tactics first. The prepared criminal is the successful criminal, I always say.
Click to expand...

I would say everyone should read up on interrogation tactics (especially in the US) since they seem to be a frequent factor in wrongful confessions.
 
I remember hearing about a questioning technique that asked the suspect to work backwards.

"You ran out of the bank and into your car and drove away?"

"Yes."

"What street did you drive away on?"

"First."

"Did you turn right or left onto First?"

"Uh..."

Apparently it's a lot easier to tell a lie beginning-to-end, the way you formulated it, than to tell it end-to-beginning. A real experience, the memories flow with similar accuracy and detail and consistency in either direction when prompted. Fictions, not so much.
 
Last edited:
rjh01 said:
There is a good reason for using such tools knowing they do not work
Police - This device says you are lying
Suspect - It is right, I confess.
Or
Suspect - The machine is wrong. I am telling the truth.

They then tell the truth. It is often how lie detectors work.
Click to expand...

I know someone who applied for an intelligence job. At one point in the hiring screening they were hooked up to a polygraph, then the interviewers asked increasingly invasive questions, getting ruder and angrier the while. They weren't actually using the polygraph for anything other than a prop, the actual test was how well the applicant handled a situation of being aggressively interrogated by unpleasant people. It was a test for attitude, not truth.

They failed by becoming flustered and upset. If they'd actually had information they'd likely have slipped it.
 
interrogations and false confessions; SCANning for truth

Elaedith said:
False confessions seem to be implicated in as many as 25% of wrongful convictions, and it has been discussed before in this subforum that different forms of evidence are not independent and that confessions can contaminate the collection and interpretation of other forms of evidence. People have even been wrongly convicted based on confessions where other forms of evidence such as DNA actually excluded them.
Click to expand...
Obviously I agree, and I would like to add a link to a 2012 paper by Saul Kassin that I found enlightening. Among the references in this paper is to a study by Hasel and Kassin on eyewitness identifications and false confessions from 2008.

More generally, I don't have confidence in methods that claim to to discern true versus false accounts. There was an extensive discussion of Statement AnalysisTM on another thread some years ago, and I came away unimpressed. This is only my first impression, but SCAN looks like old wine in a new bottle.
 
My judgement is that the headline is dishonest. It's a behavior modeling tool. Not only that, but the scope of use is appropriate to its predictive power.

My judgement is also that propublica are douchebags who aren't your friends, and who are mis-framing the issue on purpose to cause you distress instead of give you information.
 
Should we ban SCAN?

There was a 2016 study in the journal Frontiers in Psychology, which is the same one alluded to in the ProPublica story. Here is the abstract: "The Scientific Content Analysis (SCAN) is a verbal veracity assessment method that is currently used worldwide by investigative authorities. Yet, research investigating the accuracy of SCAN is scarce. The present study tested whether SCAN was able to accurately discriminate between true and fabricated statements. To this end, 117 participants were asked to write down one true and one fabricated statement about a recent negative event that happened in their lives. All statements were analyzed using 11 criteria derived from SCAN. Results indicated that SCAN was not able to correctly classify true and fabricated statements. Lacking empirical support, the application of SCAN in its current form should be discouraged."
EDT
ProPublica did a story on an Indiana conviction in which SCAN was used.
 
Last edited:
TragicMonkey said:
I know someone who applied for an intelligence job. At one point in the hiring screening they were hooked up to a polygraph, then the interviewers asked increasingly invasive questions, getting ruder and angrier the while. They weren't actually using the polygraph for anything other than a prop, the actual test was how well the applicant handled a situation of being aggressively interrogated by unpleasant people. It was a test for attitude, not truth.

They failed by becoming flustered and upset. If they'd actually had information they'd likely have slipped it.
Click to expand...

Interesting. I had an acquaintance who applied to the FBI, but failed the polygraph. They claimed it said he was lying about ever using drugs. I know he was livid about failing, but I don't know how he reacted at the time of the test. I suppose this could have happened to him. I completely believe that he'd never even had a contact high.
 
Elaedith said:
I'm a bit confused about whether you mean the theories linguists are developing are not tested or reliable, or the theories behind existing methods of content analysis that police in several countries already use, (that don't depend on linguistic fingerprints), are not reliable (although one criticism is that there are no valid theories behind the methods).
Click to expand...


Both. Any honest linguist will admit that there have not been reliable tests of a linguistic fingerprint, let alone of veracity of a statement.

The tools police are using have no generally accepted "theories" behind them.

Linguistic can do a good job of demographic analysis - the age, wealth, and location of the speaker. A 70 year-old New Yorker is not likely to say, "This can of pop is lit, yo." This only really works if the speaker is unaware of being monitored. Otherwise, it would be fairly easy to beat the test.
 
theprestige said:
My judgement is that the headline is dishonest. It's a behavior modeling tool. Not only that, but the scope of use is appropriate to its predictive power.

My judgement is also that propublica are douchebags who aren't your friends, and who are mis-framing the issue on purpose to cause you distress instead of give you information.
Click to expand...

As far as I can see, there is no reliable evidence that it has any predictive power. What do you see as the scope of its use? I am sure it is used in the US to induce confessions (i.e. a type of bogus pipeline), but it is also used in other countries that do not allow the use of deception in interrogation. If it is used deceptively to induce confessions there is the risk of inducing false confessions. If investigators believe it does work, it will create a guilt-presumptive bias where the rest of the investigation including interrogation is slanted towards confirming what is already concluded.
 
Chris_Halkides said:
Obviously I agree, and I would like to add a link to a 2012 paper by Saul Kassin that I found enlightening. Among the references in this paper is to a study by Hasel and Kassin on eyewitness identifications and false confessions from 2008.

More generally, I don't have confidence in methods that claim to to discern true versus false accounts. There was an extensive discussion of Statement AnalysisTM on another thread some years ago, and I came away unimpressed. This is only my first impression, but SCAN looks like old wine in a new bottle.
Click to expand...

I couldn't find the thread, but from what I have read, criterion-based content analysis (a step in statement validity analysis) and reality monitoring have accuracy rates averaging to around 70% in lab studies (for both detecting truthful or detecting false statements). But that may not be helpful for predicting accuracy in a field setting where the use might be qualitatively very different and much more variable. In any case, an error rate of 30% is high, yet apparently statement analysis can be used for evidence in court in some European countries. SCAN has even less evidence from what I can see, yet appears to be even more widely used in investigation.
 
Stephen Truscott?

Elaedith said:
I couldn't find the thread
Click to expand...
Statement Analysis came up the Knox/Sollecito threads, but it might be hard to find, because there are so many continuation threads. Most of the discussion concerned one individual (not someone in law enforcement) who had analyzed Amanda Knox's statements.

I went through a few of the cases in the link I provided, and the only case that I remember Mr. McClish finding a false confession was the West Memphis Three case. Although it is not possible for Mr. Truscott to prove his innocence, the Stephen Truscott case in Canada is generally seen as a miscarriage of justice, yet somehow Mr. McClish finds Mr. Truscott possibly guilty. He also thinks that Brendan Dassey is guilty.
 
theprestige said:
I remember hearing about a questioning technique that asked the suspect to work backwards.

"You ran out of the bank and into your car and drove away?"

"Yes."

"What street did you drive away on?"

"First."

"Did you turn right or left onto First?"

"Uh..."

Apparently it's a lot easier to tell a lie beginning-to-end, the way you formulated it, than to tell it end-to-beginning. A real experience, the memories flow with similar accuracy and detail and consistency in either direction when prompted. Fictions, not so much.
Click to expand...

The professional examiners I worked with in insolvency (the aim: to get dodgy directors to tell the truth about their assets] would simply throw questions out at random, not even in any particular chronology. This is believed to throw even the most clever crooks off their stride, trying to remember what they had said before and how it should tie in with what they say next, and if they say X next they'd better remember it because otherwise the lie about about A, B, C will become apparent if their next answer is Y.
 
Chris_Halkides said:
There was a 2016 study in the journal Frontiers in Psychology, which is the same one alluded to in the ProPublica story. Here is the abstract: "The Scientific Content Analysis (SCAN) is a verbal veracity assessment method that is currently used worldwide by investigative authorities. Yet, research investigating the accuracy of SCAN is scarce. The present study tested whether SCAN was able to accurately discriminate between true and fabricated statements. To this end, 117 participants were asked to write down one true and one fabricated statement about a recent negative event that happened in their lives. All statements were analyzed using 11 criteria derived from SCAN. Results indicated that SCAN was not able to correctly classify true and fabricated statements. Lacking empirical support, the application of SCAN in its current form should be discouraged."
EDT
ProPublica did a story on an Indiana conviction in which SCAN was used.
Click to expand...

It's not really like for like with criminal investigation is it? In a creative writing group I was in we each had to tell one truth and one lie about ourselves and the others had to guess which was which. Now anyone can make up a lie that has no consequence and doesn't matter if you are found out.

So no surprise the above poorly designed experiment couldn't differentiate the two.
 
Unless there is solid evidence one was there doing suspicious stuff at the right time frame a good liar can use reasonable doubt to dodge out. Video usually cannot be denied easily.

I have known folks that did stuff they probably shouldn't have and were not convicted.

Most were smart enough to never risk it again.

Where we were a more common method of investigation was to scare friends and neighbors into giving information, we were a damned ignorant lot if police were asking.
It had to be frustrating.
 
You must log in or register to reply here.

Back
Top Bottom