Jeffrey N. Rouder

William H. Batchelder

University of California, Irvine

This chapter defines and analyzes a new family of multinomial processing
tree models for an experimental paradigm frequently used in memory
research. The paradigm involves a paired-associate learning phase, where
subjects study a list of A-B items and subsequently they receive two suc-
cessive memory tests. First, they receive a free-recall test, and subjects
attempt to recall all the A and B items. Then, some time after completing
the free-recall task, subjects receive a cued-recall task, in which they attempt
to recall each B item when given a suitable cue, such as the corresponding
A item or a category label for the B item. Hereafter we refer to the paradigm
as free-then-cued recall.

The free-then-cued-recall paradigm is derived from a between-subjects
paradigm originally employed by Tulving and Pearlstone ( 1966), where
one group of subjects performed the free-recall task, and the other group
of subjects performed the cued-recall task. The cued-recall and free-recall
performances were considered proxy measures of item "accessibility" and
item "availability," respectively. The paradigm has been used by a number
of researchers, such as Drachman and Leavitt ( 1972), Hirshman ( 1988),
Hirshman, Welley, and Palij ( 1989), Hultsch ( 1975), Pra Baldi, de Beni,
Cornoldi, and Cavedon ( 1985), and Thomson and Tulving ( 1970), because
it provides information about the separate contributions of storage and
retrieval processes in paired-associate learning. The cued-recall test is pre-
sumably less demanding on retrieval processes than the free-recall test. In
fact, some researchers (e.g., Drachman & Leavitt, 1972) have operationally

-195-