Download FALSE-MEMORY CONSTRUCTION and more Slides Construction in PDF only on Docsity!
FALSE-MEMORY CONSTRUCTION: THE EFFECT OF MEMORY CONFIDENCE
Christiane N. Schneider
University of North Carolina at Wilmington in Partial FulfillmentA Thesis Submitted to the Of the Requirement for the Degree of Masters of Arts Department of Psychology University of North Carolina at Wilmington 2004 Approved by Advisory Committee
Chair Accepted by
Dean, Graduate School
ii
TABLE OF CONTENTS
ABSTRACT ......................................................................................................................v ACKNOWLEDGMENTS ................................................................................................. vi LIST OF TABLES............................................................................................................ vii LIST OF FIGURES .......................................................................................................... vii INTRODUCTION ............................................................................................................... Implicit Activation Theory .................................................................................... Source Monitoring Theory..................................................................................... Fuzzy Trace Theory ............................................................................................... Criterion Shift Theory............................................................................................ Factors that Influence Memory.............................................................................. Presentation................................................................................................ Modality..................................................................................................... Distinctiveness ........................................................................................... Delay .......................................................................................................... Repetition................................................................................................... Explanation ................................................................................................ List Length ................................................................................................. Encoding and Retrieval.............................................................................. Social Factors............................................................................................. Development .............................................................................................. Recall Confidence...................................................................................... Metamemory ..............................................................................................
- Memory Confidence and Eyewitness Testimony ......................................
- Present Study .........................................................................................................
- Hypothesis 1...........................................................................................................
- Hypothesis 2...........................................................................................................
- Hypothesis 3...........................................................................................................
- Hypothesis 4...........................................................................................................
- Hypothesis 5...........................................................................................................
- METHODS ........................................................................................................................
- Participants.............................................................................................................
- Materials ................................................................................................................
- Health Questionnaire .................................................................................
- Pre-Test Memory Questionnaire................................................................
- Word Lists..................................................................................................
- Rey-Osterrieth Complex Figure Test.........................................................
- Intense Concentration Test ........................................................................
- WAIS-R Digit Span ...................................................................................
- Digit Symbol Substitution Test..................................................................
- Recognition Test ........................................................................................
- Post-Test Memory Questionnaire ..............................................................
- Procedure ...............................................................................................................
- RESULTS ..........................................................................................................................
- Analysis for Hypothesis I ......................................................................................
- Analysis for Hypothesis II .....................................................................................
- Pre- and Post-Test Memory Questionnaire............................................................ iv
- Rey-Osterrieth, Intense Concentration, Digit Span, and Digit Symbol Sub..........
- SUMMARY and DISCUSSION........................................................................................
- SUMMARY...........................................................................................................
- Remember, Know and Guess Judgments...............................................................
- Spreading Activation Theory/ Implicit Activation Hypothesis .............................
- Source Monitoring Theory.....................................................................................
- Fuzzy Trace Theory ...............................................................................................
- Criterion Shift Theory............................................................................................
- Limitations and Future Research ...........................................................................
- LITERATURE CITED ......................................................................................................
- APPENDIX........................................................................................................................
- Appendix A............................................................................................................
- Appendix B ............................................................................................................
- Appendix C ............................................................................................................
- Appendix D............................................................................................................
- Appendix E ............................................................................................................
- Appendix F.............................................................................................................
- Appendix G..........................................................................................................
- Appendix H..........................................................................................................
v
ABSTRACT
According to Roediger and McDermott (1995), a false memory is a memory of an event that never occurred. A large body of research has explored the false memory effect and the factors that influence false memory production. The purpose of the present study was twofold. The first aim was to examine the extent to which a participant’s confidence in their overall memory ability influences the production of false memories. The second aim was to explore the extent to which fluctuations in memory confidence potentially influence fluctuations in the false memory phenomenon. To these aims, participants were randomly assigned to one of three groups: high-confidence group, low-confidence group, or neutral group. Participants received positive, negative, or neutral feedback about their memory performance on three different memory tests in an attempt to experimentally manipulate participants’ confidence in their memory. Using the Roediger and McDermott (1995) paradigm, participants were administered a total of 32 word lists and were administered a recall test after each list was presented. Each list contained 15 words associated with one non-presented word (critical lure). After all 32 lists were presented, participants completed a recognition test in which they were asked to identify the words presented on each list and to make remember, know, and guess judgments (Tulving, 1985). The analysis on the recall and the recognition test revealed a false memory effect: studied items were recalled and recognized at a higher rate than critical lures which in turn were recalled and recognized at a higher rate than non-critical intrusions or new words. No significant differences between the three memory manipulation conditions were observed, indicating that the memory manipulation did not affect false memory production.
vii
LIST OF TABLES
Table Page
- Means (Standard Errors) of Age and Education by Group....................................
- Means (Standard Errors) of Probability of Recognition ........................................
- Means (Standard Errors) of Probability of Remember, Know, and Guess Judgments ..............................................................................................................
- Means (Standard Errors) and F-values for the Pre and Post-Test Questionnaires .......................................................................................
viii
LIST OF FIGURES
Figure Page
- Example of spreading activation network for the word ‘soft.’ ................................
- Hypothesized results for the recall of studied items for the high confidence group (top), the low-confidence group (middle), andthe neutral group (bottom) .....................................................................................
- Hypothesized results for the recognition of studied items for the high confidence group (top), the low-confidence group (middle), and the neutral group (bottom) .....................................................................................
- Probabilities of Studied Items, Critical Lures, and Non-Critical Items.................
- Actual results for the recognition of studied items (top panel) and critical lures (bottom panel) ............................................................................
- Proportion of remember, know, and guess responses by list.................................
Steblay, Mehrkens & Bothwell, 1994). In some unfortunate circumstances, patients have falsely accused family members, friends, or acquaintances of crimes that were never committed (e.g. sexual abuse). In recent years, the accuracy and appropriate use of eyewitness testimony has also been called into question. The current consensus is that there are many factors that influence the accuracy of an eyewitness’s testimony (Cutler, Penrod and Martens, 1987b; Kassin, Ellsworth, & Smith, 1989; Shapiro & Penrod,1986), and that at least under certain circumstances, eyewitness accounts can be wholly inaccurate (Migueles & Garcia-Bajos, 1999). James Deese (Deese, 1957; Deese, 1959; Deese & Hardman, 1954) was the first to conduct laboratory investigations of false-memory construction. In his research, participants were administered 36 word lists that contained 12 items each. Each list contained 12 words all semantically related to one non-presented word. For example, one list included the words hard, light, pillow, plush, loud and cotton and the highly associated word that was not presented on the list (i.e., the critical lure) was soft. The lists were randomized in five different orders, and presented auditorily. Participants orally recalled the words that were presented on each list. Deese calculated the percentage the critical lure was recalled and found that some lists produced few critical lures (butterfly = 0%) while other lists produced the critical lure at a much higher rate (sleep = 44%). He hypothesized that participants recalled the critical lure because the words on the lists activated other highly associated words not presented on the lists. The participants then recalled these highly semantically related words thinking that they actually appeared on the lists.
In order to explore this hypothesis, he administered a free-association test consisting of all the words from the 36 lists. He instructed participants to report the first word that came to their mind for each word that was presented from the 36 memory lists. He calculated the association strength for each word by summing the number of times the critical lure was paired with each word. The mean association strength for the entire list was calculated by summing the strength of the each word’s association to the critical lure and dividing that sum by the number of words on the list. Using this information, Deese calculated the relationship between the frequency that the critical lure was recalled to the mean association strength of the lists to establish the probability that critical lure would be recalled for each list. He found a positive linear correlation between the mean association strength of a list, and the recall of the critical lure (P Recall of Critical Lure = 3.7 + 1.63*mean association strength of a list; r = .873). Using this relation, if the mean association strength of a list was 60%, the probability of the critical lure being recalled would be 1.0. On the basis of the strength of the relation between the recall of critical lures and the mean association strength, Deese concluded that the words on each memory list activated other semantically related words that were subsequently incorrectly reported as studied words at recall (Deese, 1959). Recently, Roediger and McDermott (1995) sought to replicate Deese’s (1959) findings using 6 lists that produced the largest number of critical lures in Deese's study in an attempt to extend their findings to recognition memory. In the first experiment, participants were read words one a time from a word list. At the end of each list, participants were told to write down as many words from the list that they could remember. After all the lists were administered, participants completed a recognition test
words in each list was highly semantically related to one word that was not presented on the list. For example, one list included the words hard, light, pillow, plush, loud and cotton and the highly associated word that was not presented on the list (i.e., the critical lure) was soft. After each list was presented, participants either performed a recall test or completed a math test. The inclusion of a math test after half the lists enabled Roediger and McDermott to determine whether recalling a list immediately after it had been presented would affect recognition performance. After all 16 lists were presented, participants completed a recognition test similar to that presented in their original experiment and indicated whether each word was “old” or “new”. In addition, for all items rated as 'old', participants indicated whether they actually remembered the word being presented (remember) or if they just knew the word was in the list (know). The remember/know procedure, introduced by Tulving (1985), was adapted to examine the nature of the memory for the critical lure in order to explicitly determine whether the information participants recalled or recognized was retrieved from specific representations in memory or simply from strong feelings of familiarity. Results revealed that studied words were recalled at a probability of .62 while the critical lure was recalled at a probability of .55, replicating the results obtained in Roediger and McDermott's (1995) first experiment. For those participants who recalled the studied items immediately following presentation, studied items were correctly recognized at a rate of .79 and these were identified as remembered at a rate of .57. The critical lures were correctly recognized at a rate of .81. For those participants who performed the arithmetic test immediately after each list was presented, participants
correctly recognized the studied items at a rate of .65 and identified them as remembered at a rate of .41. The critical lures from the lists that were followed by arithmetic problems were incorrectly recognized at a rate of .72 and identified as old at a rate of .38, and identified as being remembered at a rate of .38. Not surprisingly, results revealed that recalling information immediately after it is presented facilitates later recognition of that information as well as increasing the probability of remembering that word as having been presented on one of the word lists. On the basis of the results from both experiments, Roediger and McDermott (1995) concluded that the pattern of memory performance that they obtained could be explained using a more general theory of memory called spreading activation. In this theory, memory is represented as a framework of interconnected nodes. A node is essentially a concept or a piece of information that is linked to several other concepts. Consider the word soft. This word is highly interrelated to other words such as hard, cotton, and loud. As such, each of these words would be represented as a separate node in a network and the relationships between these words would be reflected by their connections. The links between nodes are created on the basis of experience with those words and the strength of the connection is based on various criteria that designate the importance of that relationship. For the present discussion, it is sufficient to assume that each of the connections between the words is of the same strength. One possible network that specifies these relationships is portrayed in Figure 1. Hard is connected with soft and cotton because they are semantically associated with one another (because connections in a network of this type can represent synonym or antonym associations). However,
Soft
Hard
Cotton
Loud
Figure 1. Example of a spreading activation network for the word ‘soft’.
The process of memory retrieval begins with an input, which activates one node. Activation then spreads from that node to other nodes that are linked to it. Nodes that are not linked to the nodes that become activated are unaffected. As other inputs activate other nodes in the network, activation spreads from those nodes to the other nodes that are connected to them. Each activation of a node leads that node to become more and more activated such that all the activations sum together. The most highly activated node(s) in a network will comprise what is recalled or recognized at any given time. If too much time passes between activation of information in a network, retrieval or recognition may not be possible because activation in a network decays exponentially (Bechtel & Abrahamson, 1991). Roediger and McDermott (1995) used the theory of semantic activation to explain their results. Specifically, they argued that each word in a list and any high-associates of those words (e.g., soft) would be nodes in an interconnected network. As each word in the list is activated, activation would also spread to other connected nodes, including the critical lure node, ‘soft’. Because all the words in the word lists administered were high associates of the word soft, that node would receive large amounts of activation over the course of a list. After all the words in a list are presented, the node for each presented word and the node for the critical lure would have some level of activation above zero. At recall or recognition, any node that is activated above some threshold amount will be recalled or recognized as having been on the list. Although Roediger and McDermott (1995) used the spreading activation theory to explain their data, several other theories have been proposed to explain why false memories occur. Specifically, the implicit activation hypothesis (Underwood, 1965), the
activation of the critical lure: near-threshold presentation rates and concurrent memory load. In the first experiment, two groups were tested: the concurrent memory load group, in which participants were given a digit sequence to memorize before the first list was given; and a control group, in which participants received no digit sequences to memorize. Participants in both groups saw words presented for 20 ms, 250 ms, and 2 s. After all the lists had been presented, participants were administered a recognition test. Results revealed that hit rates were lower for those in the concurrent memory group than those in the control group. Correct recognition was also lower at near threshold presentation rates (20 ms) than slow presentations (250 ms and 2 ms). With regard to false recognition of the critical lure, the concurrent memory group showed lower false recognition rates than in the control group when the presentation rate was 250 ms and 20 ms. This is an important finding because it suggests that the false memory phenomenon is, at least to some extent, the result of an implicit activation of critical lures. In a second experiment, Seamon et al. (1998) attempted to replicate these findings using a within-subjects design. In this experiment, participants were administered four lists under concurrent memory load conditions and four lists under normal conditions. Not surprisingly, hit rates were lower for lists under concurrent memory loads than lists under normal conditions. Correct recognition for studied items and false alarms for critical lures was lower at shorter presentation rates than at longer presentation rates, consistent with the findings from Experiment 1. However, critical lures were still falsely recognized at all presentation rates. On the basis of this research, Seamon et al. (1998) concluded that these results could only be obtained if the words on the lists activated the
critical lure unconsciously, as Underwood (1965) suggested. In other words, the activation of the both the words and the critical lure were done automatically, without any conscious thought from the participants. As the participants would not be able to consciously encode the words from the list, no differences between remember and know judgments would be predicted under the implicit activation hypothesis framework. Source Monitoring Theory The source monitoring theory (Johnson et al., 1993) is a framework that is based on a more general theoretical framework of memory functioning. From this perspective, each event is stored in memory in conjunction with certain perceptual (e.g., color, texture, or sound), contextual (e.g., spatial or temporal), and reflective or affective (e.g., emotional reactions or semantic associations) characteristics. For example, the experience of hearing the word ‘hard’ might be stored in memory with other characteristics such as it’s position in the list, the tone of voice of the experimenter, and any images that may have been spontaneously produced when the participant heard the word. Johnson et al., (1993) argued that the ability for a participant to determine whether a specific word was presented on a list involves the activation of that memory and the activation of any additional characteristics stored with that memory. When probed about an event, a little or a lot of additional information may be necessary in order to judge whether that event was real, imagined, in the recent past, or in the distant past. The more characteristics a memory has associated with it, the greater the likelihood that it will pass the source monitoring test and be considered a memory of a true experience.
