Real Life "Lie To Me"

You may be familiar with the new-ish show on Fox called "Lie To Me" where the incomparable Tim Roth plays a psychologist who can detect when people are lying (and numerous other emotions) through revealing facial expressions he calls "micro-expressions" or, my favorite, "deception leakage".  The show is based on the work of psychologist Paul Ekman, and now, in the real world, so are some airport security screening techniques (or via the MindHacks blog).
Of course many experts question how reliable these techniques are, particularly in light of the fact that Ekman's research seems unreplicable, and since he has shied away from publishing in peer reviewed journals in recent decades.  (I'm always so disappointed when real life differs from Hollywood.)  Anyway, my own take goes something like this:  The principle is obviously intriguing.  After all, anyone who has played poker quickly learns that their facial expressions can betray what kinds of cards they are holding.  However, if you play cards a lot, you may also know that it can take some time to learn what each individual person's "tell" may be.  And experienced gamblers can obviously manipulate the situation by intentionally displaying their facial tick or other betraying behavior when they want you to think they are bluffing.  Like this scene in Casino Royale:

More Casino Royale (2006) (Three-Disc Collectors Edition) Videos

What this tells us is that any system of "deception detection" would have to rely on either an intimate knowledge of the person being interrogated (to know what their specific "tells" are) OR on a set of facial expressions and/or body movements that are common to EVERYONE when lying. Since different people have different emotional responses to telling a lie (or even to the type of lie they are telling) and since different people often have different "tells" regardless of the extent of guilt or shame they feel, it seems to me that coming up with a system that provides cues used by everyone would yield some (likely unacceptable) level of false positives (i.e. thinking that someone is lying when they aren't) and false negatives (i.e. thinking that someone is not lying even though they are).  Anyway, if you read the article over at Nature you will see that the Department of Homeland Security is promising a "rigorous review" of the scientific merit of the programs they have put in place, so maybe we will get some data to support Ekman's ideas, or maybe we will just get more that debunks them.