In 1969 Paul Ekman and Wallace Friesen published a paper about a woman they called Mary. She was a psychiatric inpatient, forty years old, depressed, and asking for a weekend pass to visit her family. She told her doctor she was feeling much better. She was granted the pass, and later admitted she had intended to kill herself.
Ekman and Friesen had the interview on film, and they went through it frame by frame. Her face was composed and appropriately cheerful. Her hands were not. The paper became the foundation document for what they called nonverbal leakage.
The argument has a specific structure, and it is worth following carefully, because almost everything else in this part of the guide is either an application of it or a correction to it.
Ekman and Friesen proposed that the channels of the body differ along two dimensions. Sending capacity: how much information a channel can transmit, which is highest for the face. Internal feedback: how well you can tell what that part of your body is currently doing, which is also highest for the face, because you have spent your life monitoring it. And external feedback: how much other people attend to it and comment on it, which is again highest for the face.
Now consider what that means for a person trying to conceal something. Concealment is not a mental act. It is a motor task — you are running a suppression routine over a physiological response. And you will run it where you are looking. The face gets the attention, the monitoring and the correction. The hands, the feet, the trunk and the legs get almost none, because you cannot see them without looking down and nobody has ever told you what your feet do when you are anxious.
So affect that is being suppressed does not vanish. It escapes downward, into the channels with low feedback and low scrutiny. This is the leakage hierarchy.
Two things follow, and they run in opposite directions.
The useful one: attend to the channels the person is not defending. In Mary's film the composed face was the performance and the hands were the state. The professional version of this instruction appears repeatedly in later chapters — Slepian's poker study in chapter 45 found observers read hand quality better from players' arms than from their faces, which is exactly this prediction confirmed in a domain where faces are managed obsessively.
The dangerous one: leakage tells you that something is being suppressed. It does not tell you what, and it does not tell you why. Mary's hands indicated distress that her words denied. They did not say suicide. An interviewee's hands may indicate that a question is uncomfortable because they are lying, or because they are guilty of something unrelated, or because they are frightened of you, or because the question touched a bereavement.
Almost every failure in this part of the guide — the airport screening program, the demeanor evidence that convicted an innocent woman, the confident interrogator who was wrong — comes from taking a leakage signal, which is real, and reading a specific cause into it, which is not available.
The correct output of channel-by-channel observation is not a verdict. It is a location: this question, not that one. What you do with the location is ask better questions, and that is what Part 4 of this guide is about.
See Defense: channel-by-channel observation