LEO's in suspect interrogation often use a method called the Reid Technique. It usually starts with several hours of questioning and rapport building to wear down a suspect (fatigue plays a HUGE factor in our ability to deceive). At some point the interrogator will begin moving to a "help us out here, we want to understand" kind of attitude.
One facet of the technique is to identify the individuals values and priorities (kids, job, etc) and offering up potential explanations of the crime that implies they are a bad father, husband, employee etc. If the person is sufficiently fatigued and has built some kind of rapport with the interrogator, the idea is that they will offer up a full confession as a means of explaining why what they did makes them a good father, husband, employee, etc.
Military interrogation is more about general information gathering. Like you describe, a lot of that experience comes out of WWII where we would collect simply vast amounts of information from POWS that individually is largely meaningless, but in aggregate is informative.
Current research with body language, eye tracking, etc indicates most of that is junk. An increase in activity can identify when an individual is nervous about something, but it doesn't necessarily indicate deception and is incredibly sensitive to gender, culture, and (interestingly) language background. The literature talks about these kinds of things as Pinnochio's Nose; some behavior that manifest only when the person is lying, and every time the person is lying. Unfortunately this singular diagnostic behavior doesn't exist.
Source: Worked for a couple of years as a deception researcher, exploring various methods of deception detection.