Constant vs Varied Serial Order in Paired-Associate Learning

Abstract

Typical varied serial-order conditions of paired-associate (PA) learning were compared with a constant serial order on all trials in 5 experiments involving systematic variations in method, kind of material, and other potentially relevant factors. The results showed a small but relatively consistent facilitation by constant serial order limited primarily to later stages of learning. Since a shift following the first correct response to each pair from constant to varied serial order produced as much facilitation as did completely constant-order conditions, it was concluded that complex associations involving serial position are developed during early stages of constant-order PA learning, but that these exercise a facilitating influence primarily through the reduction of inter-pair interference late in learning.

A Test of the Retention Interval Hypothesis With a Constant Item Presentation Order

“…In summary, taking all of the present findings together, one may conclude that the retention interval hypothesis expects either that (a) the study-test method be superior to the anticipation method, or that (b) both learning procedures may produce about the same levels of performance under a constant item presentation order, in a manner similar to the random one, from the same theoretical rationale based on critical items that enter STM at the end of the study event. The present nonsignificant differences between the two learning methods and those of Battig et al (1963, Experiment 1) support (b) above, and the overall superiority for the study-test method over the anticipation method by Battig et al (1963, Experiment 4) and Izawa (1979a) is in agreement with (a) above. Thus, the data seem to support both types of expectations from the retention interval hypothesis under a constant item presentation order.…”

“…This implies the possibility of similar performance levels for both constant and random item presentation orders within each method. There are indeed, data supporting this expectation, as seen in e. g., Battig, Brown and Nelson (1963). On the other hand, there are grounds to expect superior performance for a constant order vis-a-vis a random one within each method: (a) The constant groups enjoy facilitation stemming from serial order cues, which is absent in the random groups; (b) whereas the random order conditions suffer from considerable variations in the length of the retention interval, such variations are absent in the constant conditions.…”

Interference among memory traces

“…The authors argue that the data did not indicate serial learning, but rather that position represented a discriminative cue with which already learned pairs could be associated. In a subsequent experiment (Battig, Brown, & Nelson, 1963), it would found that the amount of facilitation obtained with position constancy contingent upon the appearance of a correct response was equivalent to that obtained with items in constant position from the outset. Serial position, like frequency, has effects only after associations have been acquired.…”

Constant order or pairs in the presentation and testing of paired associates.

“…Martin and Saltz (1963) showed significant transfer from the S(S) and S(R) conditions to a derived serial list of the response terms which maintained the constant order of the paired-associate study trials. Although these authors concluded that serial associations had therefore been learned during pairedassociate training, Battig, Brown, and Nelson (1963) failed to replicate such positive transfer among roughly comparable conditions. It is, however, with the effects of constant serial presentation upon paired-associate learning itself that the present study is concerned.…”