Reconstructability Analysis (RA) has been developed by G.J. Klir and others in order to get a method with which it is possible to obtain information about the structure of a system when only behavioral data are available. The elicitation of an adequate structure can be regarded as a gain of information, when its information content is compared to the information content of the behavioral data. Simulation experiments can be used to estimate the information distance between the reconstructed behavior and the original system behavior. The results are used to reinterpret the gain of information in terms of pragmatic information, i.e. in novelty and confirmation. It is demonstrated how the concept of pragmatic information might be able to explain the rationale of computer aided inductive reasoning.