Topics

• All experimentalists face a common problem of "the best" interpretation of their findings. The procedures used, however, contain often quite a bit of prejudices and temptation to see what one wants to see. If we expect to see, for some reasons, a peak of e.g. charge density in a particular place within the unit cell, any indication of the existence of such peak is usually treated as confirmation of our expectations. Then we end up is saying something like "our experimental observations are in agreement with theoretical predictions". Such a biased approach could be avoided if we used strict, robotic rules of reasoning. In order to work out such rules one has to revise the notion of probability and adapt, in fact, the oldest its meaning (as used by Bernoulli and Laplace) which states that the probability is nothing else but quantitative description of the state of knowledge of a given subject (event) or correctness of a sentence. This leads to so-called Bayesian reasoning in science, and to the Maximum Entropy Method in particular.