Concepts, propositions, and schemata: what are the cognitive units?

In this paper I have reviewed the data and theoretical considerations that bear on the issue of whether propositions and schemata should be considered cognitive units. Assuming a certain general framework about working memory and long-term memory, the evidence tended to point toward a positive conclusion-that both can be cognitive units. Evidence for this comes from observations about all-or-none recall, heightened recall of units, associative priming, diminution of interference effects, considerations of implementation efficiency, and considerations of cognitive naturalness. This being said, I do not find the current picture so convincing that I would want to make a never-say-die commitment to the positive conclusion for cognitive units at all levels of knowledge structures. The idea seems sufficiently promising that it is worthwhile to develop a more explicit theory of what a cognitive unit would be like. So I have proposed a general notion of a cognitive unit that spanned propositions and schemata as special cases and specified the encoding and retrieval properties that such a cognitive unit would have. To be succinct, the important ideas associated with cognitive units are the following: (1) They can occur at multiple levels and enter into hierarchies. (2) If a set of working-memory elements can be put in correspondence with an existing knowledge structure, the elements can be joined in a cognitive unit by a single encoding act. (3) The elements in a cognitive unit are brought into working memory in a single retrieval act. (4) It is possible to evaluate general properties of a cognitive unit without having to expand it into its elements and inspect these. Perhaps this is done by measuring level of activation.