Creating false memories remembering words not presented in lists pdf

After a delay, subjects are asked to recall or recognize these words. In the recognition memory version of the task, subjects are asked whether they remember previously presented words, as well as related but never presented critical lure words 'doctor'.

Typically, the critical word is recognized with high probability and confidence. This false memory effect has been robustly demonstrated across short e.

The second critical dimension is the number of gist-inducing items on each list. Situation lists contained a greater number of gist-inducing items than did either the mid- or low-gist lists.

The strong and mid-gist lists did not differ significantly although they did differ on BAS. The mid- and low-gist lists differed significantly on gist-inducing items. The recognition test consisted of 36 studied items taken from Serial Positions 1, 8, and 10 on each of the 12 word lists. A further 36 items that had not been previously presented to participants were also included. These were the 12 nonpresented critical items along with 24 fillers randomly selected from other DRM word lists not used in the current study.

These 72 items were randomly ordered. After each list is presented you will be asked to complete 2 minutes of arithmetic problems and then you will have 1 minute to recall as many of the list words as possible. The participants then were instructed to work on arithmetic problems for 2 minutes. They then were given one minute to recall as many of the words that they could remember on a sheet provided. This procedure was repeated until all 12 lists had been presented and recalled.

Subsequently, the recognition test was administered. To make a remember judgement, participants were instructed that they should have a specific memory of that word being on the lists presented earlier. They were instructed to make a know judgement to an old word if they did not have a specific memory for the word being on one of the lists presented earlier, but that they were fairly certain that the word had been presented earlier.

The experiment took approximately one hour. The results are presented in five sections. Veridical recall is reported first, then false recall. Veridical and false recognition then are presented. False-memory results were followed up with pairwise planned comparisons because specific tests were planned for those data. Veridical and false-recall and recognition rates are presented in Table 7.

Veridical recall was measured as the proportion of correctly recalled studied items. Overall veridical recall collapsed across the four list types was quite good,.

In general, a greater number of gist-inducing knowledge-type items on a word list resulted in better veridical recall, with the best recall being for situation lists.

False recall was measured as the proportion of participants who recalled the nonpresented critical list word for each list.

Veridical recognition or hit rates was measured as the proportion of studied list words that were judged as old. The mid- and low-gist lists did not differ significantly. These analyses again show that a greater number of gist-inducing knowledge-type items on a word list results in better veridical recall. Overall, participants were more likely to falsely recognize a nonpresented critical word from lists that contained a greater number of gist-inducing semantic relations i.

These findings are consistent with explanations that attribute false memory to gist extraction. There were no other reliable differences. Overall, remember responses were more likely to be produced for falsely recognized nonpresented critical items for lists that contain a greater number of gist-inducing items, a result that is consistent with the phantom recollections reported by Brainerd et al.

As with the remember data, know responses are more likely to be produced for falsely recognized nonpresented critical items for lists containing a greater number of gist-inducing items. Experiment 4 adds further behavioural evidence concerning false-recall variability found in DRM lists and the related theoretical implications. On the other hand, situation lists produced a higher false-recall rate than did mid-gist lists. The false-recognition results are more problematic.

Furthermore, although the situation, mid-gist, and low-gist lists had similar mean BAS, situation lists produced a far higher rate of false recognition.

Finally, false-remember rates were similar for situation and strong lists and were much higher for both than for mid- or low-gist lists. Fuzzy-trace theory predicts that false memories result from processing the gist of an experienced event. In Experiment 4, gist was manipulated in two ways: using lists composed entirely of situation features, and by manipulating the number of gist-inducing items in the three groups of DRM lists.

The false-recognition results fall nicely in line with the fuzzy-trace prediction that the stronger the gist of a list of words, the more likely false memories will be produced although mid- and low-gist lists did not differ. The false-recall data are more potentially problematic, although successful recollection rejection may have decreased false recall of locations and events with the situation lists. That is, the two-minute distractor task between presentation and recall apparently did not sufficiently disrupt source monitoring or reduce the trace of the verbatim trace.

As a result, it may have been reasonably easy for participants to discount the nonpresented critical locations and events. In contrast, the recognition task was completed at the end of the study, after all 12 word lists had been studied and recalled. The longer delay plus the intervening lists appear to have decreased the verbatim trace sufficiently.

The observed pattern of remember responses made by participants when falsely recognizing a nonpresented critical list word also supports a monitoring explanation. A greater proportion of remember responses were made to nonpresented critical items on situation and strong lists than on mid- and low-gist lists. This indicates that participants had more phenomenological experiences associated with nonpresented critical words from word lists designed to have stronger gist.

One potential issue concerns whether false-recognition rates were inflated due to previous recall. Some studies have reported higher rates of false recognition for lists that were previously recalled than for lists that were not. However, although it intuitively seems that previous recall may increase false-recognition rates, in general, there is no or little evidence that it does.

For example, Brainerd et al. They were concerned about the potential of previous list recall to influence recognition. In Stadler et al. Brainerd, Watson, et al. The mean false-recognition rate for the 36 lists was actually slightly lower when recall was included. That is, without a recall phase, Brainerd et al.

Further evidence for the lack of an influence of prior recall on false recognition comes from Gallo Gallo looked at 14 studies that compared recognition rates when there was or was not previous recall. Therefore, including a previous recall task does not compromise Experiment 4. Another point regarding the false-recognition rates for the situation lists is that the design of Experiment 4 enables direct comparison among DRM and situation lists.

That is, the possibility that recall might inflate false-recognition rates would potentially be problematic if Experiment 4 had not included DRM lists as a direct comparison. Also, if falsely recalling a critical nonpresented word increases the likelihood of falsely recognizing that item, the false-recognition rate for the strong DRM lists should have been much more strongly influenced by the recall task. The false-recall rate was. A further aspect of the Experiment 4 data is that the difference between false-recognition and false-recall rates was much higher for the situation lists than for the DRM lists situation:.

These differences are consistent with our claims about the role of source monitoring with respect to the situation as compared to the DRM lists. When the situation lists are recalled following the presentation of each list, it is relatively easy for a participant to distinguish the name of the event or location from the types of people, animals, or objects that typically are involved in that event or typically are found at that location.

However, source monitoring is more difficult in the recognition task, particularly with respect to the situation lists, because participants are now dealing with 12 lists, 3 of which are situation lists, and 9 of which are standard DRM lists for which the critical nonpresented word is not qualitatively different from the list words, and the list words are a mixture of types of concepts.

Therefore, not only has time passed, but that time has been filled with lists of different types. This produced a context in which, during the recognition task, participants had a more difficult time deciding whether the name of the event or location had actually occurred on the list, and thus false recognition of those critical nonpresented words was quite high, and there was no difference in remember judgements for situation versus strong DRM lists.

In summary, because the recognition task makes it more difficult for participants to distinguish the event or location name from the list words, false recognition for the situation lists is significantly higher than for the DRM lists that were matched for BAS, and rates were similar to the rates for the strong DRM lists that contained numerous gist-inducing items and had a much higher BAS.

Finally, an experimenter oversight actually provides additional insight into gist formation. Although the low-gist list differed from the other types in mean number of gist-inducing items, whistle actually contained eight situation features. However, inspection of these items illustrates that they do not converge coherently on a gist because they come from multiple types of situation.

The whistle situation items were stop , dog , train , song , boy , blow , tune , and lips. Compare those items to those of the doctor strong list that contained the eight situation items sick , medicine , hospital , ill , patient , office , stethoscope , and cure as well as the synonyms physician and surgeon. It is obvious that the doctor list is much more semantically coherent than is the whistle list.

Of course, relational strength varied as well, with BAS for the doctor situation items being much higher than that for the whistle items. Apparently, producing high rates of false memories without high BAS demands list coherence, as in the situation lists that were coherent due to the method by which they were constructed. Although both can explain much of the same data, it has been argued that they have strengths and weaknesses.

Associative strength provides an operational index of item activation, but is theoretically mute on what underlies BAS. On the other hand, fuzzy-trace theory has a well-developed theoretical rationale but could be more clearly elaborated regarding identifying when gist is aroused.

The basic aim of the experiments reported here was to bridge these differences by identifying semantic relations that underlie BAS and by considering whether these relations can provide insight into gist extraction.

We found that BAS could be decomposed into various knowledge types and that situation features, synonyms, antonyms, and taxonomic relations predicted BAS. Further, these knowledge types, particularly situation features, synonyms, and taxonomic relations, predict the probability of false recall. As such, these findings provide a theoretical basis for considering associative strength in terms other than the mere probability of one word eliciting another word.

Finally, because the lists commonly used in the DRM paradigm vary in both BAS and in the features we take as an index of gist extraction, consideration of gist factors can explain variability in false recall not predicted to date by associative explanations. These conclusions are similar to those of Brainerd et al. In Experiment 2, a new type of word list was developed to understand the role of situation information in false memories.

The situational feature lists strongly converged on the nonpresented event or location. Moreover, DRM lists that are likely to elicit a false memory are also likely to produce a gist response, supporting further the contention that published DRM lists confound BAS and gist characteristics.

In Experiment 3, the situation lists produced false-recall rates that were more consistent with associative than gist extraction explanations. Experiment 4 tested whether the low false-recall rate in Experiment 3 was due to successful monitoring processes engendered by the situation lists in which the intended false memory an event or location differed substantially from the list items.

In false recognition however, the situation lists were as good as the strong DRM lists and much better than the mid- or low-gist lists. Remember responses also followed this pattern.

The results implicate both associative and gist extraction in false-memory production. Importantly, the findings give gist extraction a more central role than has been the case and thus support fuzzy-trace theory. It is clear that the situation lists enabled participants to form strong verbatim traces because these lists produced high rates of veridical recall. As such, the lack of finding high rates of false recall with the situation lists can be attributed in part to the strength of the verbatim memory trace.

Other results have discounted associative strength as the sole explanation in recent years. For instance, Meade, Watson, Balota, and Roediger reported semantic priming experiments in which the DRM lists were used in a lexical decision task. They concluded that this activation process does not persist long enough to account for the false-recognition effect in DRM studies.

One prediction from the present experiments is that our strongly gist-inducing situation lists should produce false memories that persist over time see Toglia et al.

To account for the high false-recognition rates observed in typical DRM studies, Meade et al. They further argued that items on the recognition task can serve as cues to reactivate an episodic associative network based on a previous episodic experience, which would be the encoding phase.

Therefore, false recognition is due to the reactivation of a previously encoded episodic experience. Importantly, Meade et al. Note that we agree with this conclusion to the extent that BAS indexes relation strength and thus the probability with which list words elicit a coherent gist.

Although the theoretical refinement outlined above is consistent with studies using DRM word lists, it fails to account for why the low-BAS situation lists elicited false-recognition rates that were similar to strong high-BAS DRM lists. Furthermore, reactivation of an episodic associative network fails to account for the false-recognition differences between the situation lists and the mid- and low-gist DRM lists that had similar mean BAS.

Clearly, the fact that the situation lists elicited greater false recognition is problematic for theories that emphasize associative strength as the primary theoretical construct. Another issue that deserves consideration relates to the associations among list items i. Interitem associative strength may explain why the situation lists produced very few critical-word intrusions on the recall tasks, but did produce high rates of false recognition.

McEvoy, Nelson, and Komatsu ; see also Deese, a have reported that word lists with stronger interitem associative strength result in reduced levels of false recall. McEvoy et al. On a recall task, the list items activate each other as a result of their strong interitem associations, but in doing so they prevent, or interfere with, the activation of the nonpresented critical word.

That is, strong interitem association leads to a stronger thematic representation, or gist, of the list items, and this influences recognition more strongly than recall. Although interitem associative strength was not controlled in the current experiments, the interitem connectivity among list items on the situation lists is fairly strong, ranging from a low of 1. Interitem connectivity is the number of items on a list that are associated to any degree. Future research with situation lists could be conducted in which this variable is manipulated between lists, while holding gist strength and backward associative strength constant, in order to see whether higher false-recall rates might be obtained.

The present experiments provide insight into at least some conditions that lead to gist extraction. Fuzzy-trace theory proposes that an encoding experience results in a verbatim trace and a gist trace. However, some have argued that the theory is unclear on how the semantic content of an episodic experience is constructed in the gist trace. We speculate that gist extraction might proceed in the following manner, at least in the case of situation lists.

The presentation of list words produces the encoding of a verbatim trace, which through conscious or nonconscious awareness activates simulations for each list item on the basis of situational knowledge and other semantic relations. That is, conceptual processing is situation based, and situational simulations are intrinsic to it.

Further, because gist processing occurs in parallel with verbatim processing, gist extraction results from integrating the individual simulations that are active at the time of encoding into a background situation, and this background situation represents the gist of the encoded episodic experience.

It also should be noted that although we have emphasized situational factors as well as featural overlap in the form of synonyms, and taxonomic relations , false memories in experiments using the DRM procedure can occur in the absence of semantic information.

For example, Zeelenberg, Boot, and Pecher found that high rates of false recognition result even when using nonwords that overlap orthographically and phonologically. Participants falsely recognized nonwords such as ploost after studying lists that contained items such as froost , floost , and stoost. Therefore, although semantic relations play a key role in false memories, other relations can matter as well. Co-occurrence probabilities influence processing at many levels, and, as such, they influence, for example, the probability of constructing a situated simulation.

Our point is that associative strength is not necessary nor by itself sufficient to explain false-memory effects in the DRM paradigm, nor in the real world. Rather, what is more important are certain types of semantic relations and the relations among them that are extracted from experiences with the world and then later used to reinstate that experience through simulation.

From this perspective, research using the DRM paradigm is applicable to understanding real-world false memories based on the gist overlap that produces these illusory memories. We thank Charles Brainerd, Diane Pecher, and an anonymous reviewer for their helpful comments on an earlier version of this article. External component: A three-dimensional component of an entity that, at least to some extent, normally resides on its surface. Entity made-of: A specification of the materials of which the entity is made.

Quantity: A numerosity, frequency, or intensity of an entity or its features. Associated abstract entity: An abstract entity associated with the target entity and external to it.

Systemic feature: A global systemic feature of an entity or its parts, including states, conditions, abilities, traits. Larger whole: A whole to which an entity belongs.

Spatial relation: A spatial relation between two or more properties within an entity, or between an entity and one of its properties. Function: A typical role that an entity serves for an agent. Action: An action that a participant performs in a situation. Location: A place where an entity can be found, or where people engage in an event or activity. Origin: How or where an entity originated.

Time: A time period associated with a situation or with one of its features. Manner: The manner in which action or behaviour is performed. Associated entity: An entity in a situation that contains the target concept. Spatial relation: A spatial relation between two or more things in a situation.

State of the world: State of a situation or any of its components except entities. Evaluation: A positive or negative evaluation of a situation or one of its components by either the subject or a participant. Contingency: A contingency between two or more aspects of a situation, including if, enable, cause, because, depends, requires. Representational state: Representational state in the mind of a situational participant, including beliefs, goals, ideas, etc.

Quantity: A numerosity, frequency, or intensity of an introspection or one of its properties. Negation: An explicit mention of the absence of something, with absence requiring a mental state that represents the opposite. National Center for Biotechnology Information , U. Search database Search term. Q J Exp Psychol Hove.

Author manuscript; available in PMC Nov David R. Katz 2. Albert N. Author information Copyright and License information Disclaimer. Correspondence should be addressed to David R. Copyright notice. See other articles in PMC that cite the published article.

Semantic relations and knowledge types One way to address the potential relations underlying BAS is to consider them as feature or knowledge types. Results and discussion Descriptive statistics for knowledge types We begin by providing descriptive statistics regarding the prevalence of the knowledge-type relations in the DRM lists see Appendix B. Open in a separate window. Table 1 Summary of multiple regression analysis using knowledge types to predict BAS in Experiment 1.

Predicting false recall with BAS and knowledge types To further understand the relation between BAS and its constituent components, a simultaneous multiple regression analysis was conducted in which BAS and the five knowledge types were used to predict the probability of false recall.

Table 2 Summary of multiple regression analysis using knowledge types and BAS to predict false recall in Experiment 1.

Table 3 Summary of multiple regression analysis using weighted knowledge types to predict false recall in Experiment 1. Table 4 Means and standard deviations for the weighted knowledge types as a function of intrusion rate in Experiment 1.

Table 6 Mean proportion of falsely recalling the nonpresented critical word and mean number of gist-consistent intrusions for the 12 situation lists in Experiment 3. Methods Citations. Results Citations. Figures and Tables from this paper. Citation Type. Has PDF. Publication Type. More Filters. False Memories Turned Against the Self. Memory construction is a phenomenon that has become intuitive to the experimental psychologist. In recent years, researchers have found that misleading postevent information can alter actual or … Expand.

On the prediction of occurrence of particular verbal intrusions in … Expand. View 4 excerpts, cites background. Study modality and false recall. Experimental psychology. False memories: Phenomena, theories, and implications. In recent years, there has been an explosion of research on false memories: the subjective experience of remembering something if that something did apparently not happen in reality. We review a … Expand. How are false memories distinguishable from true memories in the Deese—Roediger—McDermott paradigm?

A review of the findings. Psychological research. Highly Influenced. View 4 excerpts, cites background, results and methods. Incredible memories: How accurate are the reports of anomalous events.

Psychologists have studied the accuracy of eyewitness testimony for many decades and, more recently, there has been a great deal of re- search carried out on the topic of false memories.