share interactive transcript request transcript/captions live captions
download
|
MyPlaylist
JEFF HANCOCK: This study actually starts as a story almost a decade ago in Halifax, Nova Scotia, where Mike Woodworth, who will be dialing in soon, and I were PhD students in psychology. He was on a clinical track, which means he's somebody that works with people. In particular, he's a forensic clinical psychologist, so he works with people both that-- in an everyday clinic, but also in prisons.
And I was an experimental psychologist. I study how people use language. In particular, how people use language in online contexts on the internet. So I study, for example, deception, and how deception gets changed by digital communication environments.
And we were talking one day about what the two of us were doing. He was interviewing psychopathic and non-psychopathic homicidal killers in a maximum-security prison in New Brunswick. And it was this really amazing project. Very few people had ever been able to get access to these people actually interview them.
And I was really excited because he was generating all this language that were psychopaths telling the story, describing the day of the murder and non-psychopathic murders describing the day of their murder. And to both of us, this seemed like a really great place for us to collaborate, because Mike was getting this-- these transcripts, this text produced by these very-- you know, unique populations.
And I was just then exploring how to do automated-language analysis. And so we wanted to put those two things together, but it takes a long time to do something new like this. It's very unusual and novel. So that's why it's taken us so long to get it all done.
But since the last, maybe two years, it's sort of exploded into things we've been able to do. So let me start a little bit by describing some of the things about psychopaths that made us think that they might talk differently, especially when they're talking about a murder.
Then I'll describe a little bit about how I do language analysis, and we'll go over some of the results. And then at that point, we can open up for questions and Mike will be on. Mike is one of the people that-- he studies psychopaths quite deeply. And so if you have a lot more of the more subtle questions around psychopaths, we can hold on and ask him because that's his primary focus.
So there's a couple of things about psychopaths that are really interesting. One that you may all know about is the sort of emotional barrenness that they have. So they don't experience emotions in the same way that most people do. So they lack a conscience, for example. They have a difficult time understanding other people's emotions, though some of them can learn to use those to-- how to use emotions to manipulate others.
And so one very simple thing there was to look at the emotional angle of somebody trying to describe a murder, a highly-emotionally salient event. Well, there's a couple things. One is something that is called psychological distancing. It's also called temporal construal. The basic idea is that we put negative events in our lives further in the past.
So, for example, when people describe 9/11, right after 9/11 happened, some people found that they were using more past-tense verbs, for example, than they would when describing other things that happened equally long ago. Say three weeks ago was when 9/11 happened, and three weeks ago was when some other event happened. They would use more past tense, less present tense.
And so what we do in our minds gets reflected in language. This is one example. So if I had a horrible event, I can actually psychologically distance myself from it. The way this happens is same for all of the things I'm going to talk about. Some psychological dynamic happens, presumably in the mind. Somewhere-- you're experiencing some psychological dynamic. And then that is reflected in the words that we end up using.
For a long time, we've believed that we control very carefully our words, and that may be true for some words, like nouns and verbs. But the vast majority of the words we use on an everyday basis, we pay zero attention to, both as listeners and speakers. These are called function words.
They make up about half of all the words that we use. They're the tos, and ahs, and thes, and ons, all the little words that go between the big words, basically. And the beautiful thing about them is they are unconsciously produced. Same as when we think about words using tense. So we don't think about what is the past-tense verb I need to use here? It comes out automatically.
If you think about it, if you have an English-as-a-second-language-person friend, what are the mistakes they end up making? They screw up he, she. They screw up a, the. They get their tenses wrong. Because we, in English, it comes unconsciously to us automatically.
We can use that as a way to see into the psychological dynamic that's happening without the person necessary being aware of what's actually going on. So they're speaking in a very spontaneous, natural way, and we're able to look at some of the features of the language, like tense, for example, to get at that.
Another way emotionally that we should see some differences there very more direct is they should use less intense emotional words when describing the murder, and they should be less pleasant. So they tend to have sort of a darker world perspective, if you will. So we looked at also the emotional tone.
The last that's related to the emotions is something called disfluencies. I just did one right there. You probably didn't hear it because we, when we're speaking, don't process them. I said um. And that's a disfluency. Um. Repetitions-- the, the.
Turns out, when you look at videotape and you go back to conversations, they're are not nearly as smooth as we believe them to be. They're very, very messy, but we don't hear it. We hear just beautiful language being spoken all the time, with the exception, for example, of professional speakers like, radio announcers. So we believe that when describing a really emotional event, that they should have difficulty sort of conceptualizing, because this is a very emotional thing, that they would produce more disfluencies, more of these ums and ahs. OK.
One of the other areas that we knew a lot about psychopaths because of the work that Michael Woodworth and his advisor, Steve Porter, who's also an author on this paper-- both of them are at University of British Columbia, Okanagan-- is earlier, they had shown that psychopaths view the world in a more instrumental manner. OK?
So when they commit a crime it's usually to accomplish something. Whereas about half of the murders that a regular murderer will do, about half of them are what are called reactive. That is, you find the guy in bed with your wife, and you see red and you murder somebody. You're reacting to a situation.
Much rarer-- almost all of the psychopathic murders that Mike and Steve looked at were instrumental. Much higher proportion. Well, we should see that in their descriptions of the murders as well. We can look at things that are related to causality. So words like because, since, so, these are the words that begin subjunctive clauses. And we can just look at the language to see how many of those causal-related words were in the psychopathic speech compared to the non-psychopathic speech.
OK, the last one is sort of the most astonishing findings, I think, from the work. So psychopaths have been categorized as developmentally challenged in terms of the things that they look for in life, that they have needs for.
There's a very old psychological sort of system for this called Maslow's hierarchy of needs. Some of you may remember it from your own [INAUDIBLE]. It's not very well used in developmental psychology anymore, it's just been shown to be very difficult to operationalize. But the basic idea is sort of still accepted, and that's the idea that we have these lower-level needs, which are about what are we going to eat that day? What are we going to use-- hey, Erica. Come on in. Come on in. Erica is one of my former students, and now at CNN.
So we-- OK. So back to Maslow's hierarchy. We have these needs that are-- we can call basic, which are related to things that we eat and drink, they're related to shelter, money, which is something about how we get food, drink, shelter. These are our basic needs. And then there are higher-level needs which are more socially emotional. They're related to family, self-esteem development, and ultimately, in self actualization.
And the literature suggests that psychopaths are much more focused on this lower level. They are really unconcerned with self actualization, one reason being is because they sort of have this grandiose idea of who they. They're highly, highly narcissistic. So the need for self actualization isn't really there for them.
Instead, we can-- and partly related to this instrumental sort of approach to the world, they are very interested in material things. So that's what their need focus lays out. So in this, we looked at the kinds of things that they talked about. So we can actually classify transcripts according to how often they talk about lower-level things-- food, drink, shelter, money-- and how often they're related to love, family, religion, evil, god, OK? So those kinds of things.
So those are the three sort of general things we looked at. So emotional-related language, we looked at this instrumental kind of language, and we looked at where they focus their attention when they're telling these stories.
And so Mike interviewed-- Mike and his colleagues interviewed 18 psychopaths, clinically-identified psychopaths, using something called the psychopathy checklist revised, that were in a maximum-security prison that had committed murder. And they also interviewed 38 murderers in the same prison that had-- that were not identified as psychopaths, OK?
And Mike gave me those transcripts, and we looked over those transcripts. We used a couple of tools to analyze them automatically. So it wasn't like I was going in and hand coding things. We used a tool. One of them was called W Matrix, which allows you to identify each of those kinds of things, except emotion, really carefully. And another one was an emotion analysis tool called D.A.L, dictionary of affect and language.
And then what do we have? At the end of that, we have a pot of language over here that is psychopathic talking about the murders and a pot of language over here that is not psychopathic talking about the murders. And the psychopaths and the non-psychopaths were controlled for a number of things, like how old they were, how far back in time the murder happened, how long they'd been in jail for those sort of things.
And what we found for each of those predictions were supported. So the first, the instrumental one, for example, we found that psychopaths talked a lot more causally-- using because, so, these subjunctive-coordinating clauses-- than the non-psychopaths, suggesting that they-- when they described the murder, they were describing it because they were trying to accomplish something. The murder was a means to an end, compared to the non psychopaths who more often, we're reacting to an event, like finding their spouse in bed with somebody else.
For the emotional one, we found that even though the murders-- statistically, we could control for the same length of time had passed, they use more past tense to describe the murders and less present tense suggesting, that in their mind, they were construing them as taking place farther away from where they were in time. We also saw a less intense emotional language, as you'd expect from these emotionally-barren individuals.
And we found more disfluencies. They found it more difficult to describe the murders. And this is consistent with other work that suggests that even though psychopaths are very glib and very charming, they have very-- their speech is less cohesive than most.
OK, and lastly, sort of what I, again, think is the more fascinating stuff was what they talked about. And so psychopaths talked a lot about what they ate that day, which is one of the reasons that we named the paper, "Hungry Like the Wolf," because it is was one of the biggest effects was when we asked them to describe their murder, they talked a lot about what they ate that day. What they drank that day was also more frequent in the psychopathic language than the non-psychopathic.
They talked about money more often, which fits both with that instrumental thing and with the lower-level needs where they were focused, whereas, the non-psychopathic murderers talked a lot more about spirituality, religion, and family. And this, again, reflects sort of what most non-psychopathic people would think about and focus on when they just committed murder.
And one really important thing to point out, all these findings occurred despite being incarcerated for, on average, 10 years. So even though they'd been to training, and they'd been sort of in a lot of counseling to get over this idea that they weren't responsible for their murder, they still saw it as this instrumental thing that they were doing. They still saw it as less emotional. They still focused on these peripheral events, like eating and drinking, despite having been in jail for 10 years on average. So the effect holds up for a very long time.
So that's the story that began in Halifax--
[CHUCKLES]
--almost a decade ago now and ends with this paper, which looked at psychopaths and non-psychopaths. We're now moving into looking at stories that don't involve murder. So we have these people describe a positive and negative event in their life, and we've now analyzed that. We can talk a little bit about that, but it's not published yet. It's under review. And we're just now conducting studies looking at can we use social media that people produce-- use that to predict psychopathic tendencies. So we're going to move out of the prison space, because that's a pretty specialized population, and out into the social-media world.
It's going to be much more difficult. I mean, these psychopaths that were in jail were pretty serious bad dudes. And whether we'll be able to find it in the psychopathic tendencies that exist in people in the general population, as we talked earlier, there's estimates from 0.5% of the population to about-- to, on average, 1%. Whether we'll be able to find a strong enough signal in social media to connect to psychopathic self-reported levels of psychopathy. So that's the story.
So I thought I'd maybe open it up for questions, and clarifications, or anything else that--
AUDIENCE: How do you-- did you-- how do you distinguish psychopaths from sociopaths, and without any use in your study?
JEFF HANCOCK: Right. So there's been a number of different ways that that's been defined, and what I'll do is I'll leave it to Mike to distinguish between sociopath and psychopath. For my purposes, as the person that does the language analysis, they are essentially equivalent. So we focused on the psychopathy construct because it's well defined, right? We're using Hare's Psychopathy Checklist, which has, you know, been validated many, many times and is considered sort of the gold standard, if you will.
So in my view, I don't distinguish between the two. But Mike can talk about that, if he likes, when he gets on.
AUDIENCE: Is the psychopath more likely to act out than the sociopath?
JEFF HANCOCK: That's one we'll hold on for Mike. Yeah. He'll be the one to distinguish between those two. Yeah. Yeah, yes.
AUDIENCE: How are you studying the social media? What's the sort of methodology there?
JEFF HANCOCK: Right. So what we're doing is something we've developed over the last four years in the lab. We're taking advantage of the fact that everything we do digitally now leaves a record. So everything that I do, including looking at deception, which Erica actually has written a paper with me on, we now can use what people wrote. So any text messaging.
If I were to get you guys to pull out your cell phones, there'd be really nice digital records of real messages that you sent in the real world going back several days at least. So we get people to pull out their phones and look at text messages, and we get them to identify how many lies that took place in those messages. It's really lovely.
What we're doing here is we get people to pull out their last 20 texts messages, their last 10 e-mails, and then to give us any social media that they have, like Twitter, if they have a blog, even the last Facebook posts. And we take all of those and we then ask them to fill out something called the SRP, the Self-Report Psychopathy Scale. And what we'll be doing is correlating the kinds of language things I'm talking about here with scores on the self-reported psychopathy test.
And we'll be working with Cornell students to start. We already have some data in from University of British Columbia students. It was just them typing a positive and negative story. It wasn't from actual real produced social media. But even there, we're already starting to see some patterns that are emerging.
AUDIENCE: Don't people write differently than they speak? I mean, how--
JEFF HANCOCK: They do. They do. They do, but luckily for us, that notion of the unconscious still works when writing. In fact, many of us type almost as much as we speak now. Sadly-- I've just been doing some analysis of my own email-- I email about seven times-- I produce seven times more words in my emails than I do in my academic writing, which just astonishes and saddens me to no end.
But the same patterns, the same unconscious patterns, will take place in text as they do in speech, except for disfluencies. So we don't see the same kinds of disfluencies-- the ah's and the um's-- though we may be able to look at the number of times people backspace, for example, as an instance of disfluency.
AUDIENCE: And is this-- are you going to try and work with, you know, criminologists outside of psychopaths? Or are you going to try to identify-- use this as sort of a new, I guess, lie detector test--
JEFF HANCOCK: Yeah.
AUDIENCE: To sort of see--
JEFF HANCOCK: Yeah.
AUDIENCE: If their patterns fit into somebody that would commit a murder like this, or wouldn't?
JEFF HANCOCK: Right.
AUDIENCE: What is the usage going to be?
JEFF HANCOCK: Right. You can definitely see potential applications if there are solid findings in there. I could see it as one other tool that various types of professionals could use. So if you're a clinician, you could imagine using that to do research on your client to help them. If you're an investigator, you could imagine it being useful at different stages of the investigation. If you're looking for a suspect, then looking at any social media they've produced-- I'm thinking of the Long Island serial killer here using Craigslist. He left a very nice trail of digital messages.
If you've got a person that you're then interrogating, my understanding from Mike and Steve, is that you interrogate and interview psychopaths differently than you do non-psychopaths. So knowing before they've received the psychopathy checklist whether they have psychopathic tendencies could be useful there as well. John, yes?
AUDIENCE: Mike's a go.
JEFF HANCOCK: OK. So we'll dial him in. Does that answer your question roughly?
AUDIENCE: It did. I mean, I'm interested to see more of how you guys plan on-- if you are going to use this in the criminology field, and what clues would you-- it would be more of a training that you would do for investigators.
JEFF HANCOCK: Right.
AUDIENCE: Have you worked or approached the FBI and their behavioral analysts? Are you sort of--
JEFF HANCOCK: Right. We've-- and Mike can speak to this, because he was the first off on this-- we've just finished several articles for the FBI'S main journal, which you might know the name of but I'm forgetting. I think it's called The FBI Bulletin.
We have one article in it that describes this work, and then we also wrote a joint piece with some former FBI investigators. Marino Sullivan was the main person there. Yeah. So-- and Mike will know more, again, about that side of it, though I do think that some of these tools will be used by law enforcement, definitely.
AUDIENCE: What about Education, in terms of early child instances? For instance, dyslexics, kids that are on Ritalin, and so on and so forth, as far as affecting speech patterns and the kinds of things you're talking about?
JEFF HANCOCK: It's a really, really interesting question. And you know, we obviously will have to be cautious with how these sort of things are used. And this is the same from my deception research too. We've-- for example, we've been talking a lot about our algorithms for detecting fake hotel reviews. And it's the same thing where it can only be one piece of the evidence that you're using, right? So this is not sort of a stamp of like, obviously, this person is a liar or this person is a psychopath.
But when I work with law agencies, or more in particular, intelligence agencies, I sort of think of it as a-- the first set of triage. So if you're receiving thousands of messages a day and you have to analyze them for which one needs to be attended to more carefully, then some software that can do first pass type things on them and highlighter mark then, that becomes really useful. I mean, a lot of times, analysts tell me they feel like they're drinking from a fire hose.
And so we need-- as we produce more social media, we need tools to be able to analyze the social media. But that changes the way the drugs are being administered and things for kids, that's a very big question. Let me just dial Mike in here.
[DIAL TONE]
[BEEPING]
[RINGING]
And we'll get his secretary--
DONNA: Good morning. [INAUDIBLE] Professional Group. Donna speaking.
JEFF HANCOCK: Hi, Donna. It's Jeff calling for Mike.
DONNA: One moment please.
JEFF HANCOCK: Thanks. He finished his last session at 1:00. He has to clear his clients.
[HOLD MUSIC PLAYING]
All right. Good tunes.
[LAUGHTER]
Yeah.
AUDIENCE: Could you talk a little more about disfluencies?
JEFF HANCOCK: Yes. Sure. So disfluencies, again, these are a broad category of terms that relate to speech that literally is not fluent. So whenever I have to do a repetition of a word-- so the--
MIKE WOODWORTH: Uh, um.
JEFF HANCOCK: For example, uh and um. Thank you, Mike.
[LAUGHTER]
How's it going?
MIKE WOODWORTH: Good, good.
JEFF HANCOCK: Very good. We have six reporters in the room, and a videographer, and our guy that organized it all, John Carberry. And I just finished going over the study, and they've started to ask some questions. We held off the psychopathy-related ones--
MIKE WOODWORTH: OK.
JEFF HANCOCK: --For you.
MIKE WOODWORTH: Great. Hello, everybody.
AUDIENCE: Hi.
[LAUGHTER]
JEFF HANCOCK: Would you like to ask your psychopathy-sociopath question?
AUDIENCE: Yeah. Do you differentiate between psychopathy and sociopath-- between psychopaths and sociopaths in terms of their language and also in terms, I would think, of how far they go? The psychopath might end up murdering, whereas the sociopath might not murder, but would create trouble in other ways.
MIKE WOODWORTH: Well, the main differentiation between psychopaths and sociopaths, or what we would call from a clinical diagnosis of antisocial personality disorder, would be that sociopaths and people with APD, they show all the behavioral characteristics, a lot of really negative anti-social characteristics, but they don't often have that accompanying emotional issues and some of the interpersonal issues that psychopaths do.
So we may see that there's some differences in their language between the two groups, and it's an interesting question. We haven't compared against a group that had, say, anti-social personality disorder or sociopathy. We may see that some of these things we found with psychopaths are unique to them considering the different types of deficits that they have.
JEFF HANCOCK: Great.
AUDIENCE: You talked a little bit before about-- you mentioned the Craigslist killer on Long Island. Can you sort of speak a little bit how some research like this might be used when investigating a case like that? Like, what-- how would it have a practical application?
JEFF HANCOCK: Sure. And Mike, I'm not sure if you're familiar with that case or not.
MIKE WOODWORTH: Yeah. Yeah.
JEFF HANCOCK: Would you like to speak to it? Do you want me to give it the first pass?
MIKE WOODWORTH: You go ahead, then I can--
JEFF HANCOCK: OK.
MIKE WOODWORTH: I'm just getting warmed up. I've just come from therapy, so--
JEFF HANCOCK: Right.
MIKE WOODWORTH: My brain could use a minute or two.
JEFF HANCOCK: Change brain cells.
MIKE WOODWORTH: Yeah. Change content.
[LAUGHTER]
JEFF HANCOCK: Yeah, so in that case, there were a number of Craigslist posts, which was how he lured his victims. And the way something like this could be used is, as I mentioned before, anything you do digitally leaves a trace. And it's a really nice trace in that it's searchable and copyable which is what we need.
Once we have that, we can then use that to run the message against the database of psychopathic and non-psychopathic messages we have and try to classify whether that message looks like it was written more by a psychopath or more by a non-psychopath.
It's going to be difficult at this stage. We're not anywhere close to that at this point because right now we have the murder language, and we have the positive and negative story language to compare against. So what we would need to do there is we need to develop our database so that it encompasses things like online dating so we have Mark Twitchell, who is a murderer in Edmonton, that lured people using online dating sites, saying he was a woman bringing them there.
So we'll be able to build up these databases. And then we can have if a-- say an investigative body wants help, we can then compare it to the most relevant kinds of messages that we have for both psychopaths and non psychopaths. And as I said, and Mike can speak more to this, once an investigator knows whether somebody is more likely be psychopathic or not, it changes both the way they conduct the investigation. And if they have the suspect, how they go about interviewing the person as well.
MIKE WOODWORTH: Yeah. I think that one of the most key things that's come out of this research is that, previous to this research, literally nothing else has looked at the language of these individuals in a way that can get at what they're saying beyond their conscious control. Psychopaths have such an uncanny ability to micromanage their presentation. I mean, literally, it's sort of like almost a chameleon-like level. There's this vividness of how they present themselves to others, and it's very tough to assess them in a face-to-face context or when you're not able to sit back and consider their language. I mean, conning and manipulation is one of the key characteristics of these individuals.
So having an ability to look at their language and analyze it in ways that they aren't able to micromanage, at least not unless they're uncannily sophisticated, is a really important tool for, not only how to detect what some of their underlying motivations might be, but also as another strategy of how to interrogate them. We may even see a future where it's determined it's better to interrogate psychopaths not with a live human interrogator.
AUDIENCE: Do you ever think about the civil liberties aspect of this? Because this is a concern for the FBI. Because you can monitor and troll these social media sites, and you don't need any sort of court protection.
JEFF HANCOCK: Right.
AUDIENCE: Which is not the same, you know, if you're going to tap a wire.
JEFF HANCOCK: Right.
AUDIENCE: So what are you guys doing to protect yourselves, because this could become an issue? And do you think that you should, that you need to?
JEFF HANCOCK: Right. Some I don't know if you could hear that question?
MIKE WOODWORTH: About-- in terms of monitoring people online?
JEFF HANCOCK: Yeah, and the concern about civil liberties, that this kind of tool could be used without any kind of warrants because you'd be analyzing social media. So this is our first paper in this space, but you're right. It is something that we've started to think about. We haven't done anything specifically to protect ourselves because we've published everything that we've done.
So all the data, everything that we have, is published. Could agencies or other entities use this information to do social-media monitoring? Yes. And I think that companies-- marketing and branding companies, for example-- are already doing this, not necessarily with psychopathy, but they're tracking, for example, who is the thought influencer in this space. So they have these indexes for people that get followed the most.
AUDIENCE: [INAUDIBLE]
JEFF HANCOCK: Right, right. So in this case, it's-- rather than a consumer space, it's in a legal and investigative space. And so, yeah. I think it is important. I think, though, it's at this stage, at most, sort of like the polygraph, right, which cannot be admitted into court. So that's I think where this kind of science, which is sort of at the leading edge, is at. Mike, did you have any other thoughts on that?
MIKE WOODWORTH: Not on that aspect of it, just because the research right now is actually only-- we've only been concerned, ironically, with the liberties in terms of the offenders that we interviewed, and the psychopaths, and the non psychopaths in the prison to make sure that their confidentiality and anonymity was ensured by the way that we publish the data.
JEFF HANCOCK: Yes. Mike getting access in the prison, as I mentioned earlier, was a huge amount of work because it's a very protected population, so they had to do a lot of things to make sure it was anonymous. Yeah.
AUDIENCE: Could you [INAUDIBLE] how psychopaths are identified?
JEFF HANCOCK: Mike, do you want to talk about how psychopaths are identified? Maybe the PCOR?
MIKE WOODWORTH: Sure. The main tool that is commonly used around the world is called the Psychopathy Checklist. It's been revised a couple of times, but the newest version is from 2003. And it was published by a man, Dr. Robert Hare, who many would consider to still be the world's leading expert on psychopathy.
And there are 20 items that are used to assess an individual on psychopathy, and obviously, the more those items that the person seems to have at least some qualities of that trait, the higher they can score. It's from zero to 40. Anyone scoring above 30 would be considered to be a psychopath. As you get farther away from 30 and closer to 40, you're getting individuals that are highly psychopathic.
It's very difficult to score that highly on the tool as well. I should mention, when we're talking about these individuals that we assessed, those individuals that were psychopathic, they would encompass sort of all of the main qualities that you could potentially see across 20 different areas of functioning.
AUDIENCE: And is this done based on an interview?
MIKE WOODWORTH: It's based on an extensive file review, of course, considering how deceptive, and pathological lying, and conning and manipulation. You have to be very careful in terms of an extensive file review, as well as an extensive interview, which can last for hours and hours. When you get into the prisons and interview these guys, they are such a wily interview individuals that you have to be very, very diligent in your work when interviewing them.
JEFF HANCOCK: Yeah?
MIKE WOODWORTH: Yeah, I find the aspect of the manipulation very interesting because some-- just when I have nothing else to do, I do bunches of surveys on the web, and I was fascinated to find out how much I can manipulate them to the extent that, oh, now I'm getting a terribly expensive financial magazine delivered free at my door every Saturday because of a little lie I told. And when you're working with computers, whether it's social networking or what, there's lots of room for manipulation. Face to face, you've got body language, which is a whole other issue. But how can you adjust for manipulation ?
JEFF HANCOCK: Right. I mean, that's a really good question. We-- when it comes to, say, deception online, we found that people don't necessarily lie more online than off. We had that feeling that they should because there's no nonverbal cues, but-- it's because we've been interacting with a body for 60,000 years, ever since we started using language.
And when that body goes away, we get very suspicious. And this is natural. But we don't actually find that. In fact, Erica's honors thesis was to get two people to talk, to get to know each other, either in chat or face to face, as one example. And we had to look really hard to find differences in lying.
Overall, lying rates didn't change. People that were lying about face to face lied about things that were outside the conversation, whereas people that were lying in computer-media communication, they could lie about themselves more. So there was this-- that was the only change.
But we see, for example, LinkedIn-- we get more honest resumes produced on LinkedIn when we do an experiment than we do traditional paper printed resumes. Yes. Mike and I have also done a study that shows that-- OK, so maybe there's not that much more lying online, especially with people you know. But that if you are highly motivated to lie, you're actually going to be more successful online. And so we did that study-- that's that came out in 2010, didn't it, Mike?
MIKE WOODWORTH: Yeah. 2010.
JEFF HANCOCK: We found there that when we motivated one kind of liar and the other liar wasn't motivated, that they actually succeeded more than any of the other kinds of liars. And people are very bad at detecting deception. There's a lot of consensus on that. We perform, you know, typically around chance. But these guys who were highly motivated, their partners performed significantly below chance. So there is something special about the online space.
MIKE WOODWORTH: Yeah. Just to follow up on the part Jeff did mention is that in face to face, you actually see what we call a motivational impairment effect, where people do do worse as they become more motivated to lie. The good news is, they do do worse. They're not as effective at lying. We see the flip side.
So that was a-- that question was a really good one in that and in a computer-mediated domain, we actually see motivated individuals doing better. And so Dr. Hancock and I actually coined it. We called it a motivational enhancement effect, but you only see it in computer-mediated spaces. So when people are online, that motivation actually gives them a better chance of telling a lie, which is the exact opposite of what people have found in study after study in face to face.
MIKE WOODWORTH: So when you're interviewing psychopaths, for instance, do you check out the direction of the eyes, what they do with their hands, and basic facial expressions as well as all of these scientific aspects?
MIKE WOODWORTH: Well, you know what? They move so fast.
[LAUGHTER]
It's tough. They're shifty characters. You can watch a video of Paul Bernardo, who is sort of one of the more infamous Canadian serial killers, you can go on YouTube and watch him being interrogated a couple of years ago for another murder they suspect he committed. And it's a very apt example of these guys. He basically overtakes the interviewer and-- actually, two interviewers, two interrogators-- and the amount of body movement, and distraction, and manipulation, it's unbelievable.
We are hoping to start looking at some of that nonverbal behavior as well, but it is speculated that the amount of illustrators and body movement that they are doing is one of the ways that they're distracting the listener from the underlying message because they are so good at conning and manipulating people face to face. It's unbelievable. You can spend two or three hours with a psychopath and come out of there feeling like you've been hypnotized. And-- literally, all right? And you know, it's definitely time for a glass of wine and a cold shower.
And we're not quite sure what it is, because there's-- as we've talked about today and Jeff was talking about, I'm sure, before, their language is actually quite distinct. And in some ways, you don't think it would lend itself to being so conning and manipulative. So your question is great, and that people are now speculating-- a paper came out by Hart and colleagues a couple of years ago that maybe it's some of the nonverbal stuff that's going on that's really duping individuals.
JEFF HANCOCK: Whereas, the language actually provides the window.
MIKE WOODWORTH: Yeah. They flatter you by listening, for instance.
JEFF HANCOCK: Right. Right. Right. And we're find that with deception as well, that there's been this focus on the nonverbal so much that now we're seeing that language is the place where we see more action with deceptive, non-deceptive messages.
AUDIENCE: I'm curious if there's any approach to treating psychopathy?
JEFF HANCOCK: I'll let Mike answer that one. I think I know what he'll say, but-- Mike, is there-- what kind of approaches are there for when psychopaths are identified in prison and--
MIKE WOODWORTH: Yeah. The latest word on that is that for a period of time, people were very hopeful we could employ a treatment strategy, then it went to being absolutely no hope of treating these guys. And right now, I'd say that probably 20% of experts in the field feel that you could maybe at least manage them in some way.
Standard methods of treatment-- just standard cog-behave, cognitive behavioral treatment, things that would try and tap into their empathy, all of those type of treatments have been shown not only to be largely ineffective, they've actually shown to be detrimental in some studies, where the psychopaths learn some of those traits and actually use them to be more effective criminals.
So right now, there's a debate about how we can tap into treating these guys. For example, sex offenders. You used to not treat denying sex offenders. If they totally denied their offenses, it was very difficult to get them into a treatment group.
A few years ago, they started running groups specifically for these guys, saying, OK. Let's just say you didn't do it, but if you did-- at least you put yourself in a position where people think you did-- what kind of things can we work on? Some people suggested you try a similar strategy with psychopaths, saying, OK. You're not a psychopath, but we want to work on some things that are going to make your life easier to live and easier to function in terms of having your needs met, where you don't end up in prison or you don't end up having to harm other people.
Some people don't believe in that even. They feel that it's largely impossible to treat them, but others are working on that. Some have said maybe even just minimizing the amount of damage they do rather than thinking of it as treating them where they won't do damage, just the extent and breadth of damage maybe can be tamed, contained a bit, by treatment.
JEFF HANCOCK: John?
AUDIENCE: This is just because I'm curious. [INAUDIBLE] Is there an app out there to tell me not to go on my next-- is there-- in the future, are we looking at an app that says-- that can quickly tell me, don't go on the next Match.com date?
[LAUGHTER]
I mean, I'm thinking practical application. Law enforcement is phenomenal, but you know, some of these things, as you mentioned, are folks who used Craigslist or online things for ill ends.
JEFF HANCOCK: Yeah.
AUDIENCE: So you know, I'm thinking here thinking, boy, is there going to be a plugin for Match.com that, you know, that will give me an index on the side that says, you don't want to be alone with this person.
JEFF HANCOCK: Right. Well, Mike and I actually have a good friend that recently sold an online dating company in the UK, and so we got some insight into these questions. They are constantly having to look for bad guys. They have a small team of people, and this would be the case for any company. Match.com would be a lot more larger. Those people are trained to identify the people that are really problematic so when people report, oh, this guy was, you know, violent or whatever, then they will go on and do something to his file. Something, for instance, called ghosting, where the person still thinks that they're on the site, but no one else can see them, and so they think that they're just, you know, unattractive.
We've done-- we've looked at deception language patterns in online dating profiles, and we can see language aspects there. Our friend was really excited about this technology, and the fact that we could detect these things, because he thought there was something there. Again, that analyst drinking from a fire hose aspect, where there's now millions of these online dating profiles-- over 25 million Americans are using online dating. So to try and analyze all of those, you need some algorithmic power. Yeah. Did you have a question or a follow up?
AUDIENCE: No.
[LAUGHTER]
I-- this subject is fascinating. It's incredible. I can think of so many ways it could be used and misused because if you're stepping down from the experimental, academic level into use, everyday use, a lot of these things become distorted, simplified to an extent where they are meaningless. For instance, something like this could be used to screen employees by people who really don't understand the principles that Sarah said. I mean, aside from the civil liberties fact, I can see how it could get down to lower levels where it's misused.
JEFF HANCOCK: Absolutely. Mike, did you want to speak to that at all?
MIKE WOODWORTH: Well, certainly in terms of civil liberties or just in terms of individual's rights, there's no question that, at least from the part of claiming that you can assess psychopathy and claiming you may have additional tools to do that, yes. I mean, if you're diagnosed as a psychopath, especially in the United States, studies have shown that independent of any other factor, same crime, same everything, you are far more likely to get the death penalty, far more likely to get more serious offenses.
But the term, at this point, really takes on a negative characteristic for people, although that's starting to slightly change, in terms of people suggesting maybe individuals of psychopathy have a reduced culpability. But I think the positive thing is, here, there's such applied value from a law-enforcement perspective, at this point, in terms of both interrogation and prevention, and even a little bit, some could argue, from a treatment perspective, like somebody asked earlier. Where if we see changes in their self report-- that again, they are unable to control because we're using automated programs where they're not able to micromanage for all of that-- if we could see some changes in a positive direction from their language, maybe this is indicative, at least at a small level, that there's been some sort of cognitive shift or change that would make them slightly less-- slightly less dangerous.
AUDIENCE: I just think that the tough part is going to be to have all these major media companies report to you, to law enforcement. I mean, it's going to be a big nut to crack, because then you're going to have to-- law enforcement's not going to be able to have a team that's monitoring Facebook all the time. It's going to have to be an internal team from Facebook, and if they see something, to notify law enforcement.
JEFF HANCOCK: Right.
MIKE WOODWORTH: Is that the relationships you're thinking about setting up, or--
JEFF HANCOCK: No. I wouldn't have any sort of predictions about how that's going to work, but that idea of how law enforcement and these social media companies are going to interact is currently being worked out. I mean, we don't know how to do this yet. There's a huge privacy area here.
At the same time, Facebook doesn't want to be known as the place where, you know, predators hang out. That's right. And so, you know, New York has particularly tough laws in that space, and it was initially aimed at Facebook and Myspace for allowing underage people to get on, for allowing known sex offenders to get on.
So it's an area I know much less about-- and maybe Mike can speak more to it-- the legal side of it. But this is a huge space of turmoil right now, I would say. Yeah. Yeah. And I think all of the notions around the policy aspects, around how this is used in court, all of these things-- how this is implemented at the company level are all up for grabs right now.
AUDIENCE: Do your criteria work equally well when you have people-- say you're a psychopath, or you're not a psychopaths [INAUDIBLE], when you have people of different educational levels, different intelligence levels, different regionalities, et cetera, et cetera, et cetera, all of which affect their language, do your parameters sort of allow for these things and compensate for them?
JEFF HANCOCK: Well, I'll let Mike answer the question about where all of these guys came from. They're all Canadian, and most of them were Maritime, I think. Right, Mike?
MIKE WOODWORTH: Yup. Primarily in the Maritimes.
AUDIENCE: A level playing field.
JEFF HANCOCK: Right. That's right. So, no. We did not control for where they came from or anything like that. And those will have some impact. Thankfully, the level that we're looking at-- and most of the types of language we're looking, like these disfluencies, they're very, very, very common in language. And so I think they would work across these things.
The-- that isn't to say that there won't be issues here. So we have data from Mike's University, University of British Columbia students there, that told these positive and negative stories. We're now collecting that from Cornell. So we get a really nice set of different types of people, east coast, west coast, Canadian, US. So we'll see if it works there. But the answer to your question is no. It's-- we are looking just generally at that language.
AUDIENCE: This is another complication.
JEFF HANCOCK: Yes, absolutely.
AUDIENCE: [INAUDIBLE] a second class, I mean--
JEFF HANCOCK: That's. Right Yeah. But Mike, you can talk about how they came from different backgrounds, right?
MIKE WOODWORTH: Sure.
JEFF HANCOCK: Different education--
MIKE WOODWORTH: I mean, we do control-- like, we have controlled for some stuff where we could see if there was any differences in the corpus of the overall language, say based on their IQ or something. We just need to have access to that information.
And that's the great thing with the psychopathy research. If you can get access to these guys, they usually have really detailed background files. The majority of guys in this study were individuals we interviewed back east at a couple of medium and maximum-security prisons. And they were all primarily Caucasian, all male. And we also considered things like how long they'd been in jail for, how long it had been since their murder. We did consider things like that.
But you're right. There are other sort of little idiosyncratic aspects of maybe the way they spoke from where they grew up that could impact. But we just sort of hope that it all averages out in the mix. And for the most part, it was a fairly equal sample in this case. It will be interesting to compare across though, with our student sample and other samples. All these guys were in there though for at least one murder, and it was all maximum or medium-security prison.
AUDIENCE: Did you expect any gender differences? Do you think women might-- does it manifest differently for women?
MIKE WOODWORTH: Fantastic question. Right now there's a lot of speculation that female psychopathy may actually more truly encompass the core aspects of the disorder than even males and in terms of some of the real interpersonal and emotional aspects of it. And so how-- what you'd expect to see, I think, if you were to do this with a large sample of female psychopaths, you may actually see stronger results for some of the things that we found, in terms of how they would be conveying these incidents, or let's say other incidents.
There's so little research done overall on female psychopaths, just purely based on logistics and that there's just less of them, that it's tough to say, but that would truly be a fascinating study.
AUDIENCE: Why do you think it would manifest more?
MIKE WOODWORTH: I don't actually personally believe that, but some have suggested that some of the core characteristics of psychopathy, in terms of the emotional angle and the way that they might go about manipulating people outside of using real overt, anti-social behavioral means, it kind of gets back to that stereotypical idea that women, in some ways, are more likely to use covert types of ways to kind of control things, whereas men might, just with bravado and anti-social behavior, use more overt styles.
What you find is with some studies, females do score-- they definitely do score differently on the PCLR, and they often do show less of those more stereotypical, kind of criminal or antisocial type behaviors. And they do show more of what often people think of as core characteristics of psychopathy around the emotional deficits and they're ability to be very callous and very conning and manipulative. They do it in more subtle ways, ways that really make a great psychopath.
And I mean that in a negative sense, but if you're a really prototypical psychopath, most people hang their hat on the fact that it comes down to the emotional deficits and some of the interpersonal stuff. And some women psychopaths really seem to delineate this best. So how that would play out in their language, that's great. I've never really thought of that, in terms of a study. But that would be fascinating to compare a bunch of males and females. And I think we can maybe kind of do that from our student sample actually. maybe actually Dr. Hancock has even briefly looked at that. I'm not sure.
JEFF HANCOCK: You know, we didn't break it down by gender yet. So when I get back, I'll do that tonight.
MIKE WOODWORTH: Hey, well, I have to give you credit for that. That would actually be really interesting to see if there's certain aspects that we can start to look at differences in male and female psychopathy, but do that from the perspective of their language.
AUDIENCE: How many students are taking part in that social media?
JEFF HANCOCK: We've just begun collecting the data, so--
AUDIENCE: What's your hope for the sample size?
JEFF HANCOCK: About 200.
AUDIENCE: OK.
JEFF HANCOCK: And the UBC students, we have about 200-- 180, I think there.
MIKE WOODWORTH: Yeah, yeah. Closing in on 200.
JEFF HANCOCK: Yeah. Great. Well, thank you guys so much. It's 1:30, so I just want to say thanks for all the questions.
Using computerized text analysis, Cornell professor of communication Jeff Hancock and colleagues at the University of British Columbia found that psychopathic criminals tend to make identifiable word choices when talking about their crimes.
Hancock and UBC professor of psychology Michael Woodworth discussed the implications of their study at the October 17, 2011 Inside Cornell session at Cornell's ILR Conference Center in Midtown Manhattan.
