Bayesian frame calculates conditioned probabilities using reversed conditions. That is calculating the probability of B after A with Bayes' formula allows to calculate the probability of A after B. This is probability, not simple inversion (like 'simple' division of primary operations. You need also all other issues of B and A independently. Opposing (by the ends) 2 trees of 2 x 2 one tree successively (left to right A then B) and the other but right to left and B then A) it is very easy to understand the mechanics of reversal contained in Bayes' formula..
Now, if you may think that complexity confuses the end products links or statements, so the frequencies of B respect to A (like evolution giving roles according which one appear first), and imagine that the end of such tree establish correspondance to an image not like in the mirror; by confusing definitions or complex matching, which are not as easy in the simple case. So Simple case does not crisscrosses. But out this simple case we tried to see if complex matches could show patterns of matching which is for olicognograph below; first observing 3 differents actors (as in replicator's case - in picture below basic red-green-blue) and 2 x 2 trees then matching their extremities not in a simple way with one actor to another, but all connected with the same number of connexions. Those links are shown in blue-green, purple and yellow in the drawing.
Now, with geometric intuition we see that central triangular grid which is the simple most 'rational' is not the lightest one. But around there are 3 patterns of links looking like inspiring complexity.
Of course logically, this is an abuse in the first logic formula design and a nightmare for such sort of logicians. But thinking it in terms of complexity, blurred definitions, matches since different perspectives, on abstract terms, not the same from one type of actors to another, with different perspectives and affinities, we may start to think that out exact primary logical definitions, can coexist with 'free' transformations in the same world (where there are only 4 forces and also few particles). Those shown are, not so but complex, more patterns that prevent too soon complex atomic structures. So as not to be perfectly reversed and immediatly burst by entropy. This does not mean that in complex nature do not exist also simple first order logic structures and simple forms of links that obey effectivelly to bayesian calculus as rigorously defined: the purpose of diversifying is also to create non confusing identities, even at the complex level of their definition but potentially needing to catch easilly, to try to incorporate or reject if non economically compatible.