Representation of order information: An analysis of grouping effects in short-term memory.

Exps I–IV tested the independence model, 2 hierarchical models, and the dependence model of matrix relationships with a probed recall procedure in which 91 undergraduates were presented with a grouped sequence of items and were then required to recall the position and group of 1 of the items. This technique provided information about how well Ss correctly recalled both the group and position, the group only, the position only, and neither the group nor the position of an item. Findings reveal that when the items in a group were letters, digits, or musical notes, the data conformed to a hierarchical structure. When the nonalphanumeric characters were used, a matrix structure emerged. Exps V–VII required 50 undergraduates to judge the dissimilarity of 2 sequences of grouped items, the 2nd of which (the variation) was a reordering of the 1st (the original). The variation was made by reordering the groups in the original, reordering positions within groups, or reordering both groups and positions. Results show that when the members of a group were able to be encoded as single verbal units, the data supported a hierarchical system. When this was not possible, a matrix system fitted the data best. It is concluded that there is no general code for representing the order of grouped sequences and that the results are more compatible with a theory that postulates a number of specific subsystems in short-term memory, each with its own format for preserving order, than one that assumes a generalized order code. (42 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)