Ausubel divides learning into three categories: inferior learning, superior learning and parallel learning: (1) inferior learning (new knowledge is not as abstract as the original knowledge). Inferior learning, also called generic learning, is a process of attributing new ideas to a certain part of the original ideas in the cognitive structure and making them interrelated. Dependent learning includes two forms: derived generic learning and related generic learning. The former means that the new concept is a special case or example of the original concept in the cognitive structure, and the new knowledge is only a derivative of the old knowledge. Derived classes can not only make new concepts or propositions meaningful, but also enrich or confirm the original concepts or propositions. For example, if students already have the concept of "mammal", they can now learn the animal "whale" by learning the derived genus. When the newly learned knowledge belongs to an idea in the original cognitive structure, but it is not completely contained in the original idea, nor can it be completely expressed by the original idea, and the two are only an interrelated subordinate relationship, related generic learning will occur. At this time, new knowledge needs to partially adjust or reorganize the original cognitive structure, which is the product of the expansion, deepening, restriction or refinement of the original concept. For example, students used to know that hanging the national flag was a patriotic act, but now it has become a patriotic act to protect energy. (2) Cognitive learning (the newly learned knowledge is more abstract than the original knowledge) Cognitive learning is also called blanket learning, that is, learning that obtains meaning through comprehensive induction. When some concepts with low generalization ability have been formed in the cognitive structure, and a concept or proposition with higher generalization ability and tolerance is learned on the basis of these original concepts, epistatic learning will occur. For example, after learning the volume calculation formula of cube, cuboid, barrage and other shapes, learning the volume calculation formula of general cylinder belongs to upper learning. (3) Parallel and combined learning Parallel and combined learning is produced when the new knowledge and original ideas in the cognitive structure are neither universal nor general. For example, the relationship between learning quality and energy, heat and volume, genetic structure and variation, demand and price all belong to parallel combinatorial learning. Generally speaking, it is difficult to study side by side, and we must carefully compare the connections and differences between old and new knowledge before we can master it.