The research attacks two problems: the locus of interference effects and the nature of the retrieval process in secondary memory. It compares reaction time (RT) performance on the same materials for James's primary memory (PM) and secondary memory (SM) conditions in the Sternberg paradigm. In PM, where the materials were "never cut off in consciousness," the scanning process is accomplished with information already present; in the SM condition, the to-be-compared materials "of which mean time we have not been thinking" must be retrieved from inactive memory. Performance in the first case may be described as RT = a + (bm), and in the second by RT = [a + (bm)] + R, where R identifies a retrieval demand. Identical materials were used in a PM and SM Sternberg situation, the SM process being produced by introducing a distractor task between the termination of the memory set and the probe, mimicking a Brown-Peterson recall task. As in a release from the proactive interference (PI) paradigm, three consecutive trials from one taxonomic category were given and a shift was made to another category for 24 consecutive categories. The first trial on a category was defined as low in interference and the third was defined as high. The negative probe always came from the current category, and the frequency of the positive and negative probes was equal. Memory set sizes were 2 and 4. All resulting curves were parallel, differing only in intercept, with a slope of 38 msec per item. Interference effects occurred only in SM but not in PM, thus ruling out a perceptual interpretation of PI. Data show that retrieval times (difference between SM and PM) were the same for both two-and four-item memory sets. This finding demonstrates that the memory set is retrieved qua set and not item by item, in both the high and low interference conditions, agreeing with the generalized response competition and response-set interpretation of interference. Interference increased RT in SM by 19 msec per prior memory set. Negative probes gave higher RTs than positive probes, markedly so in SM, suggesting differential retrieval cue value of the two probes in SM.Overall, the results strongly support a response-set, list-differentiation, and an interference at retrieval interpretation of PI, in contrast to an encoding (perceptual) one, and stress the view that the initial retrieval act is retrieval of the address of the set and not of individual items.

This research extends the investigation of Wickens, Moody, and Dow (1981) on retrieval time and its characteristics using an adoption of the Donders-Sternberg paradigm in primary and secondary memory. The two experiments were centered around the earlier finding that retrieval time (primary memory [PM] performance subtracted from secondary memory [SM] performance) was independent of memory set size. Experiment 1 repeated Wickens et al.'s previous research but added a negative probe of a taxonomic category different from that of the other negative probe and from the categorically homogeneous memory set itself. Although the out-of-category probe produced a much flatter slope than the other probes, retrieval time (SM-PM) and retrieval characteristics did not differ. As in the Wickens et al. (1981) experiment, interference effects were found only in secondary memory. Experiment 2 used memory sets of one, two, and four items with a consonant vocabulary and again found retrieval time to be independent of set size, retrieval time of the one-item set being approximately equal to that of the four-item set. This implies that a single-item set is retrieved like a pluralitem set-namely, by first retrieving a pointer to the list, rather than by direct access to the item itself.

The solution of relational size problems in the context of ordinal series was facilitated when children were directed to locate a terminal member (the smallest) of the series. In contrast, subjects who were instructed to locate a nonterminal member without prior designation of a terminal member were unable to achieve solution of the problem. Age-and sex-related differences in performance were observed as well as differences associated with stimulus arrangements and stimulus sizes. The outcomes were discussed in terms of composite task demands.

No abstract

No abstract