In this study, four experiments are reported that examined circumstances that lead to episodic confusion errors and factors that affect one’s ability to identify and possibly remedy them. False memories have been defined as memories of events that did not occur. Episodic confusion errors (ECE’s) are memories of events that did occur, but they are remembered in the wrong episode. An eyewitness might falsely recall having seen a red pickup truck among the vehicles in a parking lot if the witness had actually seen the red truck in a different context. The witness correctly remembers seeing the truck but is mistaken or confused as to the appropriate episodic context of the event. Semantic Confusion Errors (SCE’s) are false memories that occur because the incorrect responses are meaningfully related to and confused with events that did occur. The experimental procedures that have been used for observing false memories are not ideally suited for observing episodic confusion errors or for separating out the effects of episodic confusion from the effect of semantic confusion.
The experiments reported in this article derive from a study by Smith, Ward, Tindell, Sifonis, and Wilkenfeld (2000), which showed that common nonpresented category members are often falsely recalled on category-cued-recall tests. This also showed that such intrusions were more likely to occur if the critical nonpresented word had been seen in the context of an incidental task. The ECE’s reported in the present study were always determined in relation to a baseline condition in which no critical items were incidentally presented.
An ECE is a type of source-monitoring failure. Source-monitoring procedures are those that correctly or incorrectly inform the rememberer as to the origin or source of a memory. An example of source-monitoring errors is false alarms in an old-new recognition memory task. In the present investigation, it is hypothesized that source information can be used to identify an episodic confusion error as a memory that actually came from a context different than the one in question. The encoding of source information as manipulated in the present study in terms of the level of processing of list items. McDermott and Roediger (1998) reasoned that according to the source-monitoring framework, people should be able to distinguish presented words from nonpresented ones on the basis of the perceptual information that can be retrieved in the service of recollection. ECE’s could be identified by the presence of specific source information rather than by the absence of source information.
Each of the current experiments examines various aspects of ECE’s. In all conditions, the most common category member omitted from each categorized list was expected to be falsely recalled. The level of false recall when the critical words were not presented in an incidental task was used as a baseline from which to calculate the ECE’s.
General Method
Participants – All were undergraduate volunteers participating for a course requirement. Each session was held in a group of approximately 5-15 participants at a time.
Materials – Eight lists were drawn from Smith et al.’s (2000) materials. Each list corresponded to a different category. Each list contained 15 items from a single category, with the critical item exempt from the list. Sixteen unrelated common English nouns were presented on the incidentally presented list of words. Stimuli were shown individually on a television screen using black letters on white background.
Procedure –
Experiment 1: either standard, inclusion, or exclusion
Experiment 2: standard
Experiment 3: standard
Experiment 4: stated that participants should write down not only the category members
category
Experiment 1
This tested whether participants were capable of avoiding episodic confusion errors by bringing source memory to bear during cued recall tests. It was predicted that participants who deeply processed incidental items would be able to reliably include and exclude such items on the cued-recall tests.
Results and Discussion – Carefully worded instructions that focused participants on distinctions between incidentally presented items and categorized list words were effective, in the deep processing conditions, for getting participant to bring source memory to bear during the category-cued-recall tests. Instructions had no effect on semantic confusion errors. Significantly more critical intrusions occurred in the include condition than in the standard instruction condition.
Experiment 2
The effect of level of input processing of critical items on ECE’s was examined. Also, the ability of participants to recognize post hoc which of their own falsely recalled critical items were actually presented on the incidental task was examined. It was hypothesized that deeper input processing of critical items would endow those memory traces with greater source memory potential, allowing for superior source identification of falsely recalled critical items.
Results and Discussion – A significant level of episodic confusion errors was found, and more critical words were falsely recalled when those items had been presented on an incidental task, as compared with the incidentally nonpresented condition. There was not a significant effect of level of processing on the amount of episodic confusion errors in recall. The results of the source-recognition test clearly indicate that deeper processing of the incidental list resulted in better source monitoring for falsely recalled items that were incidentally presented than for critical items that were not presented in the experiment.
Experiment 3
Participants were asked to indicate whether responses were "remember" or "know" judgments on the test in which they had tried to recognize the source of their own episodic confusion errors. This was done to test the predictions that deeper input processing increases the number of "remember" responses without influencing the number of "know" responses on a recognition test.
Results and Discussion – The results replicate and extend the results of Experiment 2. There was no significant effect of level of processing on episodic confusion errors. On the source-recognition test, remember source recognition judgments were affected by level of processing, whereas know judgments were not. Results from Experiments 1-3 are consistent with the idea that episodic confusion errors in recall occur when automatic retrieval of incidentally presented critical items is not successfully opposed by recollection of those items.
Experiment 4
Whether post hoc corrections of inappropriate memory attributions are accompanied by an escape from the illusion that ECE’s are correct responses from the categorized lists was tested. This was done by asking participants to indicate during the cued-recall tests whether each recalled word had been presented on the incidental list, on the categorized list, or on both lists. It was predicted that more accurate identification of episodic confusion errors would be accompanied by a reduction of the belief that incidentally presented items had actually been presented on the categorized lists. Also predicted was that deep input processing would better enable participants to identify which critical items had been presented on the incidental list.
Results and Discussion – Deeper processing appears to have endowed incidentally presented items with more source memory, allowing those items to be properly sourced at recall. Results from Experiments 1-4 indicate that the correct source of episodically inappropriate responses can sometimes be determined online, as the participant is recalling the list words. The memory illusion tested occurred 90% of the time when critical items had not been presented. Although episodic confusion errors can be remedied to some degree when participants use source memory, such errors nonetheless occur at a high rate.
General Discussion
The results of the four experiments indicate that false recall can be caused by episodic confusion and that such falsely recalled responses sometimes have the potential to be avoided or even corrected after their commission. They also showed that even when memory accuracy was emphasized, people often did not use available source memory during recall to edit out episodically inappropriate memories from the appropriate ones.
The ECE’s observed in these experiments may have been caused by automatic retrieval of episodically inappropriate items, unopposed by recollection of the appropriate episodic contexts of intruded items. They could also be due to increased familiarity caused by priming, whether incidentally primed items were shallowly or deeply processed. Although the present results offer some promise that people sometimes may be able to identify their own episodic confusion errors, they do not address false recall that is due to semantic factors such as strength of association. Errors made when episodic confusion compounds semantic confusion effects comprise an important class of false memories and should be investigated directly.
|
|
|
|
|
|
|
|