Long term readers know that I both use body langue analysis and am highly critical of it. It is fascinating and if a baseline is established, it can be a useful, howbeit, minor tool.
It is not something that communicates well in teaching and success rates often mimic guess work. Those who are overly confident in it often fail, and as I say to professionals:
your first error may be your last. 80% isn't good enough.
If you declare an innocent person to be deceptive conclusively, you may lose the respect of your peers and damage the science we work hard at studying and employing. This is why students enrolled always have the support they need before submitting analysis in their work place.
Yet, I have covered some fascinating examples of it. Recall the video of Phoebe DiPietro who closed her eyes before each lie.
Here is another view of the eyes being closed and the samples are enough to establish a pattern which America will be watching.