Thapar, A. & McDermott, K. B. (2001). False recall and false recognition induced by presentation of associated words: Effects of retention interval and level of processing. Memory and Cognition, 29, 424-432.

The DRM paradigm allows the study of false recall and false recognition by comparing the probability that subjects mistakenly recall/recognize words that are highly related to presented words, but that were not presented. Research utilizing the DRM has focused on determining how variables, known to effect accurate memory, effect false memory. This research focuses on retention interval and level of processing during study phase.

Retention interval is concerned with the length of time between study and test phases and the effect on the probability of false recall/recognition. Past research has used repeated within subjects designs, but a between subjects design is necessary to completely understand the effect of retention interval.

To what extent and in what direction does false recall/recognition change across time? Expect an increase in the ratio of false recall/recognition to accurate recall/recognition. This can be explained by fuzzy trace theory and source monitoring. Fuzzy trace theory accounts for this because, with time, subjects would rely more on gist as exact details fade. Source monitoring accounts for this because as perceptual details are lost, subjects would emphasize the memory for cognitive/emotional processes. Activation theories, such as semantic priming, would predict rapidly declining levels of false recall with time.

To date only one study has used a between-subjects design. Toglia, Neuschatz, and Goodwin found no decrease in false recall across time.

Level of processing is concerned with how deeply an item is processed at encoding. Is the item encoded shallowly, such as by color, or more deeply encoded, such as by the pleasantness of the word? Various experiments show that non-presented critical items are recalled/remembered with the same probability as presented items. McDermott (1997) has shown that non-presented items are even subject to perceptual priming. It can be assumed that non-presented items have level of processing effects and, therefore, might show a higher probability of false recognition/recall. This effect can be seen in that items presented in a blocked fashion tend to exhibit higher false recognition/recall.

The reverse can also be predicted, as deep processing should reduce false recognition/recall. In fact, McDermott and Watson (in press) show that slower presentation rates enhance veridical memory by allowing for deeper levels of processing. Subsequently, false recall/recognition would be reduced because improved memory for studied items should reduce the errors in commission.

Four studies have addressed this issue. Two report that semantically processed lists led to greater false recall than superficially processed items, while the others report the no effect. Thapar and McDermott site problems with these studies and indicate that they predict a level of processing effect.

Method:

Ninety-nine undergraduate students from an introductory psychology class received extra credit for participation. Subjects were randomly assigned to one of the retention intervals, 33 per group, and tested in groups of no more than 4 subjects each.

This study was a 3 X 3 design. There were 3 levels of encoding: color, vowels, pleasantness (shallow to deep processing); and three retention intervals: immediate, 2 day, and 7 day delays. Encoding was manipulated within subjects, while retention intervals were manipulated between subjects.

Thapar and McDermott utilized the first 12 items of the 24 lists used by Roediger and McDermott in 1995. The three lists corresponding to critical words king, man, and girl were replaced with trash, smoke, and smell. WHY? No explanation given. The 24 lists were randomly divided into 8 list, which were then assigned to the three encoding conditions so as to ensure that they are seen an equal number of times by all subjects. Word order was strongest to weakest in terms of their association to the critical non-presented item. Prior to hearing the lists, subjects were randomly told which encoding method to utilize.

Subjects were told that their memory for the presented words would be tested after presentation. The pleasantness, vowel counting, and color of word instructions were then presented. All subjects reported responses on rating sheets provided by the experimenter. All words were presented for 4 seconds. At test phase, which varied from immediate to a 7 day retention interval, subjects were given as much time as required to recall to write down all the words they freely recalled.

Results and Discussion:

• For items processed deeply, there was a decline in recall with longer retention intervals.
• For items processed shallowly, there was a decline in recall with longer retention intervals.
• For critical non-presented items, there was a decline in recall with longer retention intervals.
• Deep processing was associated with higher probabilities of recall than shallow processing, although both these probabilities declined with larger retention intervals.
• Recall of studied items declines faster over time than do critical items.
• Significant drop in recall from immediate to 2 day interval for studied items
• Marginal drop in recall from 2 day to 7 day interval for studied items
• Critical items only show a drop from immediate to 2 day interval, no drop in recall between 2 and 7 day interval

• Higher probability of recalling studied items than critical items at immediate interval, but this reverses after the 7 day retention interval
• Length of retention interval affected studied items more than non-presented items, particularly for deeply processed items

• Recall of deeply processed items declines faster over time than do shallowly processed items.
• Floor effects must be considered in shallow processing - ???
• Level of processing had a greater effect at immediate interval testing than it did on the longer intervals although the effect was still reliable at the 7 day interval

Summary:

1. Forgetting occurs for studied and non-presented items
2. Studied items are forgotten quicker than non-presented items
3. Level of processing effects occurred for critical items in much the same way as it occurs for studied items…deeper processing results in increased false recall over superficially processed items

Method:

Seventy-two undergraduate students from an introductory psychology class received extra credit for participation. Subjects were randomly assigned to one of the retention intervals, 24 per group, and tested in groups of no more than 4 subjects each.

This experiment was similar to Experiment 1 in all respects, except the test phase. Subjects were required to judge whether or not they remembered words on a 144 item test by indicating yes/no, rather than free recall. Twenty-four of these words were presented in study, 24 critical non-presented targets, 24 weakly related non-presented words (1 of each from the original 24 lists) and 72 distracters. The test was self-paced and administered at the same retention intervals as was the test phase in Experiment 1.

Results:

Results in experiment 2 were similar to those in experiment 1.

d' - used because baseline false alarm rates increased across retention interval

• Recognition of studied items declines faster over time than do critical items.
• Recognition decreases over time for both studied and critical non-presented items
• Recognition of deeply processed items declines faster over time than do shallowly processed items.
• Recognition was better for studied items than critical items at the immediate testing interval, however probabilities were the same at the 7 day interval
• Level of processing had a greater effect on studied items than on non-presented items, although both item types exhibit this effect (marginally significant)
• For both studied and non-presented items there was significant forgetting between immediate and the 2 day retention interval, but not between the 2 day and 7 day retention intervals

Summary:

Discussion:

False recall and recognition decline across time, and do so at a slower rate than do the probabilities of accurate recall and recognition. Deeper processing results in higher recall overall, despite forgetting over the retention interval. Results are consistent with the predictions of fuzzy trace theory and source monitoring. With time, details are lost and a greater emphasis is placed on gist/semantic/cognitive/emotional features.