Some effects of remembering on forgetting

Abstract

Three experiments investigated the effect of testing for recall on forgetting. One experiment demonstrated that the effect of testing is restricted to the items tested. A second experiment showed that the difference between tested and untested items increased with longer retention intervals. Increasing the study time had no effect on the retention of tested items. but increased the recallability of untested items. A third experiment ruled out rehearsal during the retention interval as an explanation. It was concluded that the critical aspects of a test are the retrieval operations themselves. Suggestions are made as to how these operations attenuate forgetting.Both common sense and countless experiments indicate that information in memory becomes less accessible when it is not activated or utilized for some time. This decrease in accessibility is commonly called "forgetting." It is also well established that retrieving information from memory renders that information more accessible upon some later occasion.The research reported in this paper was concerned with testing effects on the course of forgetting over relatively long periods of time. The experiments were motivated primarily by certain conceptual and methodological problems involved in the study of forgetting, but the results have important implications for the understanding of testing effects in general.The problem originates with the very definition of forgetting. Conceptually, the notion of forgetting implies that information that was once accessible can no longer be retrieved. Operationally, then, the measurement of forgetting requires two retrieval tests separated by a time interval. Working under the assumption that any event that is perceived is immediately available (e.g., Tulving, 1964), some researchers may not provide the initial test, but forgetting can only be demonstrated by determining the actual accessibility of that information at two different times. Normally, the first test is given as close to the end of input as is possible, but there is no logical necessity for this restriction, and forgetting may be validly measured between any two points in time.Although the general operations required to measure forgetting are straightforward, the realization of concrete procedures presents somewhat of a problem. Two kinds of design are available. One procedure is to give both tests to the same person at two different times.

Assessing the distinguishability of models and the informativeness of data

“…In this situation, we might safely conclude that this noise is the source of the unrepresentativeness. The cause of the outliers in Runquist (1983;middle panel) is less clear-cut, since the study was a little more controlled, and the data do not violate monotonicity quite so extensively. However, the violations are large enough to conclude that noise was probably a contributing factor to the unrepresentativeness of these data.…”

“…The experiments in our database vary enough along both dimensions to assess the effect of each fairly independently of the other. The top row of graphs in Figure 3 shows the effects of N on model Krueger (1929;left;N=280, |t|=6), and Runquist (1983;right;N=1728, |t|=6). Those in the middle row are from Wickelgren (1968;left;N=40, |t|=5) and Strong (1913;right;N=40, |t|=13).…”

“…distinguishability when |t|=6. The HY/SE landscape in the left graph is from a recall study by Krueger (1929;N = 280), and the one on the right is from a cued recall study by Runquist (1983;N = 1728). The effect of increasing N is dramatic.…”

“…Of the 8 most informative designs (bottom panel), it is noteworthy that the only one with N < 400 has 13 time intervals (#18), and that the three most distinguishable designs all have N >1000. These are the designs of Burtt and Dobell (1925), Rubin et al (1999), and Runquist (1983).…”

One hundred years of forgetting: A quantitative description of retention.

“…The autobiographical memory data sets reported in Table 9 are included here because they have been claimed to be reten- Nelson et al (1980) la Peterson & Peterson (1959) 3 Rubin(1989) 1 Runquist(1983) Ht lot Itu lou Schonfield(1969) . Note.…”