Selective encoding involves sifting out relevant information from irrelevant information. When new information is presented in natural contexts, relevant information for one's given purpose is embedded in the midst of large amounts of purpose-irrelevant information. A critical task for the learner is to sift the "wheat from the chaff," recognizing just what among the pieces of information is relevant for one's purposes (see Schank, 1990).

Selective combination involves combining selectively encoded information in such a way as to form an integrated, plausible whole. Simply sifting out relevant from irrelevant information is not enough to generate a new knowledge structure. One must know how to combine the pieces of information into an internally connected whole (see Mayer and Greeno, 1972).

Selective comparison involves relating new information to old information already stored in memory. It is not enough to encode and combine new information; the information has to be tied to some preexisting knowledge base. A good selective comparer recognizes how existing knowledge can be brought to bear on the present situation. A poor selective comparer does not readily see the relations between existing and new information. For example, a competent lawyer looks for past precedents, a competent doctor for old cases that shed light on new ones.

The various components of cognition work together. Metacomponents activate performance and knowledge-acquisition components. These latter kinds of components in turn provide feedback to the metacomponents. Although one can isolate various kinds of information-processing components from task performance using experimental means, in practice, the components function together in highly interactive ways, and are not readily isolated. Thus, diagnosis as well as instructional interventions needs to consider all three types of components in interaction rather than any one kind of component in isolation. But understanding the nature of the components of cognition is not, in itself, sufficient to understand the nature of cognition because there is more to cognition than a set of information-processing components. One could scarcely understand all of what it is that makes one person more intelligent than another by understanding the components of processing on, say, a cognition test. The other aspects of the triarchic theory address some of the other aspects of cognition that contribute to individual differences in observed performance, outside testing situations as well as within them.

The experiential subtheory. Components of information processing always are applied to tasks and situations with which one has some level of prior experience (including the null level). Hence, these internal mechanisms are closely tied to one's experience. According to the experiential subtheory, the components are not equally good measures of cognition at all levels of experience. Assessing cognition requires one to consider not only components but also the level of experience at which they are applied.

According the experiential subtheory, cognition is best measured at those regions of the experiential continuum that involve tasks or situations that are either relatively novel, on the one hand, or in the process of becoming automatized, on the other.

Several sources of evidence converge on the notion that skill to deal with relative novelty is a good way of measuring cognition. Davidson and Sternberg (1984) found that gifted children had greater insight to deal with novel problems than did nongifted children. Research on fluid cognition, which is a kind of cognition involved in dealing with novelty (see Cattell, 1971), suggests that tests that measure the skill to deal with novelty fall relatively close to the so-called general factor of cognition (Snow and Lohman, 1984).