Research Like This Frightens Me

Bruce Schneier posted a worrisome piece about new research claiming psychopaths can be detected based on their speech:

The researchers interviewed 52 convicted murderers, 14 of them ranked as psychopaths according to the Psychopathy Checklist-Revised, a 20-item assessment, and asked them to describe their crimes in detail. Using computer programs to analyze what the men said, the researchers found that those with psychopathic scores showed a lack of emotion, spoke in terms of cause-and-effect when describing their crimes, and focused their attention on basic needs, such as food, drink and money.


To examine the emotional content of the murderers’ speech, Hancock and his colleagues looked at a number of factors, including how frequently they described their crimes using the past tense. The use of the past tense can be an indicator of psychological detachment, and the researchers found that the psychopaths used it more than the present tense when compared with the nonpsychopaths. They also found more dysfluencies — the “uhs” and “ums” that interrupt speech — among psychopaths. Nearly universal in speech, dysfluencies indicate that the speaker needs some time to think about what they are saying.

So psychopaths have a tendency to be accurate in their speech? I don’t know about you but I almost exclusively talk bout past events using the past tense. When I say I traveled to Las Vegas for Defcon I dont state it as, “I’m going to Las Vegas for Defcon.” or, “I’m at Las Vegas for Defcon.” To me talking about past events in the present or future tense is a little more concerning than using the correct past tense.

On top of that I know a great number of people who constantly use dysfluencies in their speech and they certainly do not fit the description of being psychopaths. Using criteria like this would label a huge number of people and myself as psychopaths.

I’m not sure what the goal of this research is but if it’s detecting psychopaths then I guarantee a great number of false positives are going to be generated. I would also like to point out the fact that this research was performed on a very small sample size (52) and nothing stated in the article leads me to believe any real control group was used for comparison. This has all the red flags of a research projects done to acquire government grants by delivering results the state desires.