Before the 1950s, the influential behaviorist school in psychology only recognized the behavior that people reacted in some form after being stimulated by some external environment, but denied the conscious process between stimulation and reaction, and rejected the introspection method of studying consciousness. When cognitive psychology studies human information processing mechanism with computer, it shows that there is a real information processing process between the input and output of computer, which consists of input classification, symbol operation, term storage and extraction, pattern recognition and so on. Although there is a substantial difference between computer information processing and human information processing, we can get inspiration from it and realize that there must be corresponding information processing between stimulus and response, which can only be attributed to consciousness. The rise of cognitive psychology has restored the position of consciousness research in psychology, and its leading factor is information processing. The similarity between computer information processing and human information processing in symbol processing is the origin of the name of artificial intelligence and the basis of its realization and development. Information processing is also the link between cognitive psychology and artificial intelligence. Cognitive psychology and artificial intelligence permeate and promote each other.

Due to the complexity and diversity of human cognitive activities, it is difficult to establish an all-encompassing cognitive model. Generally speaking, according to the modular hypothesis, each cognitive function has its corresponding structural principles, and each cognitive model generally reflects only one or several aspects of cognitive characteristics. The famous models that lay the foundation of cognitive psychology include EPAM, semantic network model of memory, HAM, ACT, MEMOD, etc. In addition to memory model, general problem-solving program GPS and opportunistic problem-solving model are cognitive models for researchers to solve problems. The former puts forward the commonly used means and purpose analysis technology in artificial intelligence, while the latter is based on the idea that people often do not necessarily consider specific opportunities in strict order when solving problems. For complex process control, enterprise and business management, socio-economic field and human thinking process simulation, it is more and more difficult to use traditional mathematical modeling technology for quantitative calculation, but it is often necessary to use artificial intelligence technology with the help of appropriate qualitative cognitive models, or to use a model combining qualitative analysis with quantitative calculation (such as fault control of nuclear reactors).