In most mediations, there is a divide between the parties as to what the actual facts are, and each party feels he or she is the right one. This leads to discussions about the basis for each sides' belief or position. Often everyone simply has to "agree to disagree", but It also happens that an email, memo or other document will surface which turns one sides' version of facts on its head. Their faces reflect bewilderment, "how can that be? All this time I was so sure I was right!" Wheels are turning in their heads as to how to refute the documentary evidence.
This article about Anchor/Journalist (NBC) Brian Williams and his recent mis-recollection, reveals through neuroscience how someone can believe a fact to be true, when it's not. http://www.washingtonpost.com/news/speaking-of-science/wp/2015/02/05/the-science-behind-brian-williams-mortifying-memory-flub/
So, how does this play out at mediation? For one, it helps in the process of collaboration and compromise for all sides in the dispute to understand that someone can honestly believe their version to be accurate, yet be wrong. As this article demonstrates, it does not automatically mean that someone is a liar, or intending to deceive. A common example used by lawyers and mediators is an auto accident at an intersection. Assuming 4 witnesses, it is assured that there will be 4 versions of what happened - what lane someone was in; who turned first; what color the traffic signal was; where traffic was. As neuroscience proves, our brains are wired to make snap perceptions which stay with us afterward, even if we misperceived. So, when you are in a mediation and hear the other side give statements you "know" to be false, perhaps it is better to give the benefit of the doubt that they really did believe their "truth", instead of assuming they are pathological liars. Not that the latter can't be the case, it certainly can. But, in my experience, I usually see decent people having different perceptions, each thinking the other is lying. It is an easy assumption to make, not always accurate.