Mainly the latter, which is based on Paul Ekman's work. He's directly involved in the SPOT program.

Sad thing is, he used to do really great research — in fact, it's what inspired me to go to grad school to study social neuroscience (and is in part the basis of my thesis).

But then he stopped doing peer-reviewed, published research and started contradicting his own prior research which clearly showed that involuntary displays of basic emotions or stress (which do exist) cannot be reliably attributed to any particular underlying reason, like deception.

This is intuitively obvious if you think about this: suppose a cop asks you, in an interrogation room, whether you raped your ex. You're pretty damn likely to show a spike of CNS stress combined with anger, disgust, fear, or the like. Think of how many potential reasons there are for that reaction (which is all that can be measured externally, even if you're good at it*) that are not "because you're guilty".

* And FWIW, as my thesis work shows (duplicating prior research; see its cites), there's basically zero correlation between whether you actually are good at detecting facial emotion and how confident you are in your assessments, etc.