Get better at detecting deception
See, ODB, follow up, confirm.
I have at least two PhD’s that read this blog. If either of y’all are looking for a new project, I think I might have a lead on it with this.
There are studies that academics cite claiming that the average cop and the average college student are equally likely to be able to detect lies, both hovering around the 50% success rate. I have read a few of these studies. They often have volunteers simulating deception and others analyzing them to see if they can guess truth or lie in a few minutes. If these studies are saying that the subjects of the study are 50/50 on detecting individual lies, then I could be ok with that. In the real world, though, we want to know if the person is a liar and what subject they are lying about. I think a well-trained investigator with time and good technique should be significantly more successful than 50/50 at detecting liars and subjects of concern.
Lies are discovered with attention, follow up, and amplifying questions. I believe that an interviewer can achieve a much higher rate of success in detecting lies by adhering to a certain thought pattern. When they Observe Deceptive Behavior ODB (RIP ODB), it should be like a flag alerting them. When they see the flag, they commit to follow up on that behavior, but they maintain the mindset: it’s probably nothing; when I ask them some questions it will probably resolve itself. For example, if they ask a subject where they were yesterday morning, and the subject crosses their arms before responding, that nonverbal behavior - which we know is associated with deceptive behavior - would tell the investigator to follow up, but it might mean nothing! When the investigator returns to the type of questions being asked when the ODB (crossed arms) occurred, they ask questions, expecting to resolve it. Maybe, “You seemed to have something more to say about yesterday morning. What is it?” If they resolve it or even if it is inconclusive, then the investigator moves on and doesn’t count it one way or another. If the suspect continues to display deceptive behavior, the investigator would be right to mark that down as probably deceptive.
James Clear wrote a book called Atomic Habits. In it, he talks about how no one single action defines you, but rather each action is a vote for the type of person you want to be. If you go to the gym 5 days in a week and veg out the other two, you are, on balance, a healthy person. Your 5 votes for healthy outweigh your two votes for lazy. I think the same is true for detecting deception. If you get 10 ODB that you follow up on and still appear deceptive and 4 that you follow up on appear to resolve themselves, then, on balance, then you are probably talking to a deceptive person.
Don Rabon’s writing first introduced me to the idea of presuming innocence in investigative settings. He writes that we should go into a suspect interview presuming their innocence. Fight with that however you need to. I fought with it too. When you’re ready, let’s talk about it. His contention is that if you go in presuming innocence, and they display deceptive behavior, then you’ll be able to articulate exactly what they appear to be lying about. This ability to articulate deception will make you much more confident about your belief in their deception. This same concept applies to deceptive behavior. Assume it is nothing, but follow up. If you follow up and it’s nothing, then leave it alone. If it is something, test your belief until you are certain.
Let me try to illustrate this with more than a picture of a Wu Tang legend. I have a wife and two kids. When we are all together, and I am talking to my wife, there is a 99% chance that one or both of the kids will interrupt us while we talk. Now, I know that the kids might have something very serious to tell me (I work in child abuse!), but I also know that it’s more likely they want cereal, lost a sock, or saw a puppy video on YouTube that they want to tell me about. Either way, they are alerting me that they need something. When they interrupt, I hold my hand out to tell them to wait. When I finish my sentence, I return to the interruptor, and I ask, “What’s up?” Usually it’s the thing about the lost sock. Someday, though, one might tell me, “A boy touched me in a funny way, and I didn’t like it.” If that day comes, I’ll ask, “Funny like how?” They may say, “He pinched my nose and barked like a Walrus.” If so, I’ll probably bark like a Walrus too, and we can all laugh. On the other hand, they might outcry that the boy pinched their privates in a way that wasn’t appropriate. At that point, I’ll take their interruption seriously, and follow up. If I went around believing that every time my kid demanded my attention that he had been abused, I’d be a total head case. If we go around thinking that everybody that has an ODB demanding our attention is a liar, we may be similarly pitiable.
We need to approach ODB the same way. Anybody who sees that gulp or wiping of sweat or picking of lint and immediately concludes that the person is lying is not going to be successful. There is not one single definitively deceptive behavior, but that doesn’t mean that behaviors aren’t observable in deceptive subjects. Notice the flagged behavior and follow up with amplifying questions.
I believe that if a research study were to follow investigators that used this mindset, and gave the investigator time to work, it would find that that they detected liars and subjects of concern at a significantly higher rate than the stated 50/50 odds.
What do you think?


Hi, Lewis - I am an author and a journalist, working on a book project on deception and deception detection. Your blog is fantastic. Maria Hartwig pointed me to it. I’d love to talk to you on background, if you’d be kind enough to trust me. My name is Yudhijit Bhattacharjee and I’m at yudhijit@gmail.com. Thanks.