A Theory of Information Structure I. General Principles*

Current cognitive theories are founded on the processing cornpurer analogy, i.e., that a cognitive system is analogous to a computer. In contrast to this view, the following paper proposes a content computer analogy which states that cognition is the attempt to model phenomena as computers. In fact, the content computer analogy is extended to a content machine analogy, because the term machine is understood as having two kinds of structure: (1) computational, or algebraic structure, and (2) stability. The paper develops a theory of the cognitive role of these two factors. Algebraic structure determines the generative or connective aspects of stimulus sets. However, stability is shown here to crucially constrain the allowable decomposition and ordering of that generative structure. It is argued that an understanding of this constraint is important to the study of cognition. The constraint results in (1) a cognized structure of nested environmental control, and (2) the imposition of a reference frame. It is thereby argued that the concept of reference frame is strongly related to that of control. A general theory of reference frames is offered. The constraint just described forces reference frames to be usymmetric-sequential; and it is argued that this explains a number of phenomena in perceptual organization, categorization, linguistic grammar, and planning structure.