The cognitive effect of processing a piece of information is to allow fixation or revision of beliefs. Bayesian Belief Systems and software for manipulating Belief Networks deal with uncertainty management. Varying degrees of certainty giving a better match to real-world systems than logic requiring certainty. Fuzzy Logic has been applied specifically to deal with concepts that are vague; other approaches to problem solving include evolutionary techniques, Genetic Algorithms, Genetic Programming or Neural Networks that simulate the effect of neurons and synapses in the brain.
In bottom-up models of Pattern Recognition based on template matching, Prototype and Feature Comparison Theories (distinguishes between detecting and integration), processing starts with part of the pattern and through manipulation yields a more richly specified output. The system works in one direction starting from the sensory input and proceeding to final interpretation, uninfluenced by expectations or previous learning. Other models include Structural Description Theory and top down processes that focus on high-level cognitive processes, existing knowledge and expectations. The pattern of sensory input alone cannot explain the relatively stable and rich experience we have of our surroundings. The immediate perception of a specific interpretation clearly indicates that it is based on more than the sensory input or the information falling on our retina. The highly accurate guesses and inferences that are made rapidly and unconsciously are based on a wealth of knowledge of the world and our expectations for the particular moment. The influences of these sources beyond sensory input are collectively known as top-down influences.
Expert behavior involves highly specific Pattern Recognition employed in sensation and perception. Research on problem solving provides experimental support for a pattern-based knowledge acquisition approach for expert systems and development is increasingly based on patterns rather than linear hierarchies of rules. The pattern-based approach to knowledge acquisition is centered on recognition memory rather than the more error-prone recall memory used to build general rules. A great deal of human expertise seems to result from extensive experience in recognizing and reacting to specific patterns rather than the application of general rules to specific situations.