In addition to utilizing short-term memory, problem solving also requires accessing relevant information - the solver's knowledge base about the problem - from storage in long-term memory (Ormrod, 2004). Critical factors in this retrieval of information include what has been stored and how it has been stored, and the cues or patterns present in the problem that help the individual perceive what information to access from memory. In order to be retrieved, knowledge from a content domain must be present in memory to begin with; it must also be organized in a way that facilitates its retrieval in an appropriate context.
With experience in a content domain, it is believed that problem solvers develop cognitive structures called “problem schemata” that allow them to recognize a problem as belonging to a particular category (Sweller, 1988). This mental classification of problem types can trigger particular actions for solving the problem, based on the perceived similarity of the presented problem to the same category of others stored in memory.
Marshall (1995) describes a schema as an “organized memory structure” (p. 31) or “basic storage device” (p. 46) that is centered around a general concept. This general concept develops as a person repeatedly encounters instances of that concept and abstracts common features from multiple situations. Schemas are believed to have a network structure, meaning that the subtopics of a schema are connected to each other and linked to other related schemas. These structures can vary in their size and complexity, and a schema is flexible in that its connections can change over time. A schema can be composed of both declarative knowledge (facts) and procedural knowledge of how to use or apply the information. The organization of an individual's schemas in memory and the strength of connections between concepts have a profound influence on the accessibility of information for problem solving, which will be described more in the section on Expert-Novice Research.