The Wason selection task is designed as a refutation attitude and based on the K. Popper principle of falsificationism. The numerous studies have revealed that specific forms of the task are solved better than abstract ones. However, researchers explain this phenomenon differently. The aim of the study is to introduce and explain the better solution effect by varying the coherence level and the assumption form of selection task conditions. The notion of “coherence” is introduced in the theory-theory concept approach to denote the connection between two concepts, as well as between the attributes of one concept. We have introduced three levels of coherence: formal coherence means the absence of any meaningful connection, external coherence implies the presence of an external “context”, and internal coherence includes the causal connection between the assumption concepts as well. Based on the selection task modifications analysis, it was found that conditions are implicitly formulated in the form of rules or hypotheses, which implies different testing strategies. The following hypotheses were tested: 1) the selection task solving efficiency will increase with the growth in the assumption attributes coherence level; 2) tasks with rules will be solved better than tasks with hypotheses on the same assumption attributes coherence level. The sample includes 193 subjects (median age 23 years, 60% female, 51% with higher education). Selection task modifications were used as stimulus material, where the coherence levels and the assumption form were varied. As a result, both hypotheses found statistical support. The novelty of this paper lies in the methodological separation of selection task modifications both by the assumption form – rules or hypotheses, and by the coherence level of the tested assumption – formal, external and internal. The introduction of two forms and three levels of assumption coherence allows us to integrate the selection task solution factors revealed in different studies within a single explanation of the inclusion of the tested assumption in different kinds of theories.