|
Roediger, H.L., McDermott, K. B. (1995) Creating False Memories: Remembering Words Not Presented in Lists. Journal of Experimental Psychology, 21, 4, 803-814. Purpose: In what is considered the first study of false memories, Bartlett (1932) had subjects read and recall a story. He found that subjects’ memory became distorted over repeated recall trials. Bartlett also illuminated the distinction between reproductive memory, which is an exact regeneration of material, and reconstructive memory, which involves potentially inaccurate filling in of missing information. Most researchers believe that remembering simple materials involves reproductive memory, so materials such as word lists have rarely been used to study false memory. One exception is Underwood (1965), who used a list learning paradigm to demonstrate that subjects falsely recognized words that were associates of the words that actually appeared in the lists. While Underwood found this effect using a recognition test, Deese (1959) found false memory using a free-recall paradigm. Studies that use tests of recall are much harder to find in the literature than those that use recognition tests. Deese found that subjects produced a nonrepresented word when asked to recall some lists of associates. In other words, extralist intrusions occurred. The authors sought to replicate Deese’s findings by examining false memory in a single-trial, free-recall paradigm and to broaden his paradigm to recognition tests. Experiment 1 tests for false recall and recognition of nonpresented words. Experiment 2 attempts to generalize the effects to lists not previously used by Deese. Experimental Work: In Experiment 1, the authors presented subjects with six lists of associates, which were chosen because they produced high false recall when used by Deese. The targets, or critical nonpresented words, were chair, mountain, needle, rough, sleep and sweet. The associates came from Russel and Jenkins’ (1954) list of word association norms. Subjects heard the lists, read by the experimenter, and were then asked to write the words they remembered starting with the last few items. After the recall test, the experimenters administered a 42-item recognition test, which included 12 studied and 30 nonstudied words. The nonstudied words consisted of 6 critical words, 12 unrelated words and 12 words weakly related to the lists. The words were presented on a sheet and the subjects were asked to provide a rating of their confidence in having heard the words using a 4-point scale. The critical nonpresented words were recalled almost as frequently as the studied items presented in the middle of the list. Intrusions of the critical word occurred on 40% of the lists. On the recognition test, subjects indicated remembering 86% of the studied words. The false recognition rate for the critical lure, at 84%, was much higher than that for weakly related words and was very close to the hit rate. Overall, subjects’ judgments of the critical items looked more like those of the studied items than of other lures. The rationale for Experiment 2 was to replicate the first experiment with a broader set of materials, to determine whether prior recall of words affects recognition, to look at the false-alarm rate for critical items whose associates were not presented and to explore the phenomenological experience of subjects by asking them to discriminate between remembering and simply knowing that the words occurred. To these ends, the authors constructed 24 new word lists containing 15 associates each. Subjects completed recall tests for 8 of the lists, studied another 8 without recalling them and did not study the last 8 lists at all. After hearing the words, subjects either recalled the items or completed math problems. A non-recall condition was added to examine false recognition for lists that were only heard, not recalled. A recognition test followed, including old and new judgments as well as remember-know judgments. This last measure was used to determine whether false recognition of the critical items would be attributed to mentally reliving the presentation of the words or to a feeling of familiarity. Recall of the critical lure occurred on 55% of the lists. Subjects recalled these words at a rate greater than that for the studied words presented in the middle of the lists. The hit rate for recalled lists was greater than that for lists that were followed by math problems and there were more remember responses in the recall condition, suggesting that the recall task improved recognition. Results for the critical items showed that subjects failed to distinguish them from presented words. Additionally, false recall improved false recognition and increased remember responses. The authors also compared recognition of items listed by subjects during recall tests with recognition of items not recalled. Recognition of correctly recalled studied words was very high (98%), whereas recognition of non-recalled words was 50%. For falsely recalled critical items, recognition results were very similar to the results for the studied words, with recognition for items not produced on recall tests actually exceeding the results for non-recalled studied items. For both the studied items and the critical items, a majority of subjects indicated remembering the words that were produced, while close to half gave know responses for words that were not produced. Conclusion: In general, these experiments produced high levels of false recall and recognition. It is possible that the results were due to schematic processing, as occurs with the use of prose materials, because the critical nonpresented word could have served as a prototype for the lists of associates. The results of these experiments differed form previous work in that the authors showed that false memory effects can occur with both recall and recognition, that subjects indicate remembering rather than just knowing false information, and that recall influenced subsequent recognition of both accurate and false information. The authors offer a potential explanation for the false memory effects obtained here. False recognition may result from an implicit associative response involving a spreading of activation during encoding. The strong effects seen in the these experiments may be due to the length of the word lists used, because false recognition has been shown to be related to list size. The fact that the false memories in these experiments were judged to remembered rather than known was an important result. Subjects may have experienced source monitoring errors, which would account for the finding that recall enhanced the remembering judgment. The experience of remembering may have increased because the process of recalling the words made it easier for the subjects to think of them, a phenomenon that may cause misattribution to memory. The authors conclude that the ease with which false memories were obtained may have implications for real-world memory of events. Points for Discussion: 1. The authors go to great lengths to ensure that the subjects understand the difference between remember and know judgments and that they do not confuse having written the word during the recall test for having remembered hearing it. However, they acknowledge that the act of recalling the words probably influenced the subjects’ remember judgments regardless of the instructions. Is it a problem that this judgment is so difficult for subjects to make accurately or is the important result the fact that subjects do think that they remember events that have not actually occurred?
2. Does it matter that presenting the recognition task after the recall test affected the recognition results? Would examining recognition independently contribute anything to the results or is it just important to note this phenomenon and consider the effects of both tasks taken together? |
|
This page was last updated: |