Do you think it is actually possible to quantify all these behavioral cues and ascertain whether a person is lying or not based on statistics created through test observations? It is given in the article that the accuracy of lie detectors is only about 50%. Then we might as well go for a heads or tails coin toss since the probability over there is also 50%. Every time we find a new observation or a new "behavioral cue", a lot of those figures are going to change. What I am trying to say is aren't we just trying to justify things with numbers after we've seen some observations, rather than creating concrete frameworks which would hold true and be able to predict correctly? And this thing is not possible, which I am sure, even you would agree. Thus are all these attempts futile?

Regards

Shehzaad