False memories and fuzzy-trace theory: Misinforming gist versus verbatim memory

Item Type text; Dissertation-Reproduction (electronic)

Authors Titcomb, Allison Louise, 1963-

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 27/09/2021 08:18:04

Link to Item http://hdl.handle.net/10150/264412 INFORMATION TO USERS

This manuscript has been reproduced from the microfihn master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedtlirough, substandard margins, and improper alignment can adversely afreet reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drav^gs, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600

FALSE MEMORIES AND FUZZY-TRACE THEORY:

MISINFORMING GIST VERSUS VERBATIM MEMORY

by

Allison Louise Titcomb

copyright © Allison Louise Titcomb 1996

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF EDUCATIONAL PSYCHOLOGY

In Partial Fulfillment of the Requirements For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

19 9 6 UMI Number: 9713357

Copyright 1996 by Titconb, Allison Louise

All rights reserved.

UA-n Microform 9713357 Copyright 1997, by UMI Company. Ail rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 2

THE UNIVERSITY OF ARIZONA ® GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have read the dissertation prepared by Allison Louise Titcomb entitled False Memories and Fuzzy-trace Theory:

Misinforming Gist Versus Verbatim Memory

and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy

, jy ate^^

Da^e" W

Date

Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I hereby certify that 1 have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

?/ //A Dissertation Director ^ STATEMENT BY AUTHOR

This dissertation has been submitted in partial fiilfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the copyright holder.

(illJAhrn 4

ACKNOWLEDGMENTS

Many fine people supported me in this endeavor. I thank them all. In particular, I wish to acknowledge the contributions made by faculty, colleagues and fnends. Valerie Reyna was the key that unlocked my interest in false memories and misinformation effects. Her enthusiasm and insight as dissertation director, mentor, and fiiend inspired me throughout my program of study. Other members of my dissertation committee, Charles Brainerd, John Bergan, Lawrence Aleamoni, and John Kihlstrom, fine-tuned my understanding of the project with their focused queries and supportive attitudes. I appreciated most the committee's responsiveness to my own questions. Robert Belli graciously supplied the slides used in this study. Barb BCieman, Maria Romo, and John Fulginiti allowed me to recruit participants fi"om their courses. The volunteers made the study possible. Carol Bender, Gail Burd, Ann Christensen, Barb Fishel, Genevieve Kenney, Sarah Palmer, Cindy Rankin and other colleagues fi'om the Undergraduate Biology Research Program, Molecular & Cellular Biology, and Science Connection offered moral support, flexible work schedules, and steadfast belief in my abilities. My fellow graduate students, including Lilian Stein, Ambrosio Mojardin, and Susan Ellis, were always available with suggestions, questions and encouragement. Finally, my family and fnends, Shirley Titcomb, Chuck Fellows, and Nancy Gedney, deserve more thanks than I can put into words. TABLE OF CONTENTS

List of Figures 7 List of Tables 8 ABSTRACT 9 L INTRODUCTION 11 n. SELECTIVE REVIEW OF THE LITERATURE 15 Misinformation Paradigm 15 Theoretical Explanations for Misinformation Effects 17 Storage Explanations 17 Retrieval Explanations 19 Fuzzy-trace Theory 20 Specific Predictions 24 m. METHOD 26 Subjects 26 Design 26 Materials 27 Types of Misleading Information 27 Procedure 29 Analyses 30 IV. RESULTS 32 Signed Confidence Analysis of Variance 32 Memory Dependency Tests 37 V. DISCUSSION 39 Signed Confidence Analysis of Variance 39 Memory Dependency Results 41 Implications for Research 43 APPENDIX A; Instructions to Subjects 55 APPENDIX B: Comprehension Test Sample 56 APPENDIX C: Recognition Test Sample 57 APPENDIX D: Lure and Original Items 58 APPENDIX E: Description of Slides 59 APPENDIX F; Human Subjects Approval 61 APPENDIX G; Lure, Miseld, Consistency, Gist, Item, Time Interaction % Yes ANOVA Means 62 APPENDIX H; Lure, Misled, Consistency, Gist, Judgment, Time Interaction % Yes ANOVA Means for Verbatim Judgments 63 APPENDIX I: Lure, Misled, Consistency, Gist, Judgment, Time Interaction % True ANOVA Means for Truth Judgments 64 APPENDIX J; Signed Confidence ANOVA Interactions with Version and Item 65 APPENDIX K: Contingency Table % Responses 66 6

TABLE OF CONTENTS — continued

APPENDIX L; Dependency Results by Time; Verbatim (Yes-No) Judgments 67 APPENDIX M; Dependency Results by Time; Truth (True-False) Judgments 68 APPENDIX N; Mean Signed Confidence Ratings for Misled vs.Control Subjects Yes-No (Verbatim) Judgments 69 REFERENCES 70 7

LIST OF HGURES

Figure I, Lure-Original Ratings for Different Types of Misleading Information .... 45

Figure 2, Judgment Differences for Types of Misleading Information 46

Figure 3, Yes-No (Verbatim) Judgments of Lure-Original Items for Different Types of Misleading Information 47

Figure 4, Truth Judgments of Lure-Original Items for Different Types of NCsleading Information 48

Figure 5, Interaction Between Lure-Original Ratings and Time 49

Figure 6, Interaction Between Gist-Detail Ratings and Time 50

Figure 7, Effects of Timing on Original Ratings for Different Types of Misleading Information 51

Figure 8, Effects of Timing on Lure Ratings for Different Types of Misleading Information 52

Figure 9, Lure-Original Differences for Misinformed and Not Misinformed Consistent-Inconsistent Information 53 LIST OF TABLES

Table I, Summary of Memory Dependency Predictions

Table 2, Memory Dependency Results for Yes-No Recognition Test (Verbatim Judgments)

Table 3, Memory Dependency Results for True-False Recognition Test (Truth Judgments) 9

ABSTRACT Most misinformation studies have tested memory for altered details of an event (e.g., stop sign, wrench, swing set). This study varied the type of misleading information to include consistent gist, inconsistent gist and consistent details as well as inconsistent (or altered) details. Misinformation studies follow a three-stage procedure; an initial event, misleading information, and a final memory test. Here, too, adult participants initially observed a slide sequence. Misinformation was embedded in a comprehension test and memory was tested with a recognition test. The timing of the misleading information and the timing of the final memory test resuked in three between-subjects levels; Immediate misinformation and Immediate recognition test. Immediate misinformation and Delayed recogntion test, Delayed misinformation and Delayed recognition test. The delay was one week after the intial slide observation. The final recognition test queried memory for what subjects saw (a Yes-No "verbatim" task) and what they believed to be true (a True-False gist-based judgment). All subjects judged misleading items (lures) and original items for each kind of misleading information. Subjects also rated confidence for each answer.

Descriptions and examples of all stimuli are provided in the Appendices.

An analysis of variance conducted on the signed confidence data (judgment combined with confidence) revealed significant differences among the types of misleading information (gist consistent had highest rates of recognition, inconsistent items were rejected but less so when subjects were misled), differences between Yes-No "verbatim" and True-False judgments (truth ratings were greater than the verbatim ratings but were equally high for consistent gist), different effects of forgetting (decreased recognition of originals, increased recognition of lures), and significant misinformation effects.

Memory dependency analyses revealed that the relationship between memory for an event and memory for misleading information depends on the type of information. whether gist or detail in nature. Results are discussed in terms of storage explanations (e.g., discrepancy detection, misinformation acceptance), retrieval accounts (e.g., coexistence, blocking, discrimination) and fiizzy-trace theory. The conclusions support, general, predictions from fuzzy-trace theory. Implications for eyewitness testimony include the importance of early, neutral questioning. 11

CHAPTER 1

INTRODUCTION

False memories have gained recent attention from both the general public and

memory researchers. Their concern is prompted in part because of the recent spate of

highly publicized and poignant legal battles among family members accused of heinous

crimes based on supposedly "recovered" memories (Loftus & Ketcham, 1994; Smolowe,

1994). One of the most basic problems in memory research is reflected in the current arguments about repressed memories. After an extensive survey, Loftus and Loftus

(1980) reported that many lay persons and quite a few psychologists commonly assume that once an event or piece of information is stored in memory, it is there permanently, until an appropriate cue calls it forth. A more recent book states that, "Our capacity for storage of memories is almost limitless" (Terr, 1994, p. 51). This view is contrary to Loftus's own work and interpretation, which seeks to explain the malleability of memory through carefully controlled experimental studies. Loftus (1993) notes that, "This growing body of research shows that new, postevent information often becomes incorporated into memory, supplementing and altering a person's recollection" (p. 530). Does vivid memory for details necessarily mean that an event actually happened? Or is it possible that some people's recollections are tainted by recent questioning? These are crucial issues that misinformation studies attempt to answer.

Most misinformation manipulations involve altering a detail of the original event, e.g., traffic signs, tools, playground equipment (Loftus, Miller & Bums, 1978;

McCloskey & Zaragoza, 1985; Belli, Windschitl, McCarthy & Winfrey, 1992). However, it is possible to misinform memory for gist, as well as for verbatim details. Therefore, the experiment reported here incorporates three additional stimulus types (consistent gist. 12

inconsistent gist, and consistent details). Loftus and colleagues (e.g., Tousignant, Hall &

Lofhis, 1986) have predicted that better memory for an initial event leads to less

suggestibility (i.e., less susceptibility to misinformation). However, they have not tested

this assumption called "discrepancy detv. -tion." An alternative interpretation resides in fuzzy-trace theory (e.g., Reyna & Brainerd, 1995; Brainerd & Reyna, 1993) which posits that this relationship between original memory and suggestibility depends on the type of

misleading information (gist vs. verbatim details).

This study will not specifically address the recovered memory issue but will use a

less emotionally complex memory task to look at the effects of different types of

misleading information and the formation of false memories. Recent studies (Titcomb &

Reyna, 1992; 1994) were prompted by varied results in, for example, developmental reasoning research that turned on the gist/verbatim distinction (e.g., Reyna & Kieman,

1994). The research design in the Titcomb and Reyna work included tests of memory for both the original and the misleading information, results of which addressed issues from misinformation research, such as the untested discrepancy detection explanation (e.g.,

Titcomb & Reyna, 1993; 1995). The design elements also included different types of misleading information (gist consistent, gist inconsistent, inconsistent details and consistent details) and varied when subjects received the misleading information (either immediately or after a one-week delay). In these studies, long-term retention of written materials, adapted from Harris, Lee, Hensley and Schoen (1986), was measured with recognition tests given one week after the stories were read. Furthermore, different memory tasks were used. Verbatim tasks demanded precision in specific memory for details and gist-based tasks required more global, less-specific judgments. For example, Yes-No tests that asked what subjects specifically remembered from the original information required exact memory for details. Conversely, True-False tests queried 13

memory for belief about the truth of an item, regardless of explicit memory for its occurrence. The purpose of this experiment is to extend the results of previous studies

(Titcomb & Reyna, 1992; 1993; 1994; 1995) by including a third category for testing

immediate memory and to replicate previous findings with visual materials. Earlier

experiments (Titcomb & Reyna, 1993; 1995b) revealed a negatively dependent

relationship between memory for inconsistent details and original information but a positively dependent relation for consistent gist items and original information, a result not predicted by discrepancy detection. Such basic experimental work can provide much needed understanding about the underlying mechanisms of memory. Furthermore, the

theoretical insight can, in turn, guide future studies that delve into the complex realm of false memories in the courtroom.

The aim of this study is to investigate the effects of gist versus verbatim misleading information and the impact of forgetting on these effects. Recognition judgments, both gist and verbatim in nature, provide the data and fiizzy-trace theory (Reyna & Brainerd,

1995), a theory of reasoning and remembering, provides the theoretical perspective for interpretation of the results.

In particular, I investigate whether similar results can be obtained with visual stimuli that are comparable to those obtained with written stories (e.g., Titcomb & Reyna, 1993; 1995b). I expect the smallest misinformation effects when verbatim memory is best

(i.e., in the immediate condition) because better verbatim memory should allow subjects to reject misinformation. Similarly, the largest misinformation effects should occur with inconsistent details and will likely occur when misleading information is presented after a delay, or when verbatim memory is poorest. With regard to memory dependency predictions (see Chapter 2), and based on previous research, I predict that fuzzy-trace 14

theory will better match the results than the other theories (e.g., Reyna & Brainerd, 1995a).

The remaining sections of this paper include a Literature Review that briefly describes the classic misinformation paradigm and fundamental theoretical perspectives including discrepancy detection, coexistence theories, fiizzy-trace theory and others. That chapter ends with a description of the most recent work in this area and the specific predictions derived from those studies. The Methods chapter includes definitions of the different kinds of misleading information as well as descriptions of the design, subjects, procedure, memory tests and analyses. More detailed descriptions of the stimuli are included in the Appendices. The Results section summarizes, first, the results from the

"signed confidence" (which combines dichotomous choices with confidence) analysis of variance regarding differences between gist and verbatim information and judgments, the effects of forgetting, as well as significant misinformation effects. To sum up, significant differences in each of these sections revealed high ratings for consistent gist by all subjects, rejection of inconsistent lures particularly in the immediate condition, less rejection of lures after misleading information, higher truth ratings compared with Yes-No

(verbatim) responses, and significant forgetting over time which differentially impacted originals compared with lure items. Finally, the results of memory dependency analyses will show that fuzzy-trace theory best predicts the relationships between memory for original information and memory for misleading information. The last chapter will discuss the results in comparison to other studies and will elaborate on the predictions shown in Table 1. 15

CHAPTER 2

SELECTIVE REVIEW OF THE LITERATURE

Misinformation Paradigm

How do we study false memories? A classic interference design adapted by Loftus

(e.g., Loftus, 1979; Loftus, Miller & Bums, 1978) has been used by many researchers (for

a review see Titcomb & Reyna, 1995a). The basic procedure involves three stages: -I- an event, typically slides or video presentation; -2- exposure to misleading information, usually embedded in a questionnaire or written narrative that summarizes the original event; and -3- a memory test. For example, participants might watch a video of a car

accident in which a vehicle glides past a stop sign. Afterward, they might be asked, "How fast was the car going when it passed the yield sign?" Later, usually within 24 hours, subjects complete a memory test that queries their recollection of the original event. This memory test can be recall or recognition and, if the latter, can pit the original item (e.g., stop sign) against the misleading item (e.g., yield sign) or a novel item (e.g., no parking sign). (Again, for a fiill description of such measurement variations, see Titcomb & Reyna, 1995a). Note that this example involves memory for a specific detail (stop sign).

Most misinformation studies have focused on such altered or added details (but see

Titcomb & Reyna, 1992; 1994). Furthermore, few studies have included a substantial delay manipulation and those that have have failed to fully cross delay of the final memory test with delay of the misleading information (e.g., Belli et al., 1992). However, the importance of timing factors has been noted (Loftus, 1992; Loftus, Levidow, & Duensing,

1992), although these discussions have been limited to general points about forgetting, as opposed to specific explanations that include gist-verbatim distinctions (e.g. Reyna, 1995). 16

Although the magnitude of the results in these studies has varied depending on such factors as the type of test used, the timing of the memory test and the misleading information, the order of questions, and the modalities of the original and misleading events, the typical result is a performance deficit by misled subjects compared to control subjects. The bottom line explanation for these results fi"om Titcomb Reyna's (1995a) review is that factors that increase forgetting also increase misinformation effects. Of particular relevance to the current study, a delayed memory test or misleading information presented closer in time to a delayed memory test could produce larger misinformation effects. Similarly, Reyna and Titcomb (1994; in press) pointed out that factors that enhance verbatim memory for original events (e.g., early memory testing) decrease suggestibility even when memory is tested after a delay.

One of the measurement variations in misinformation studies turns on the nature of the memory test. Loftus's "standard" test (forced-choice between original and misleading items) was questioned by McCloskey and Zaragoza (1985) who proposed a "modified" test (forced-choice between original and novel items) to prevent responses based on social conformity. Unfortunately, the modified test has been criticized for lack of measurement sensitivity (Belli, 1989; Tversky & Tuchin, 1989) and for allowing subjects to select the correct answer merely by rejecting the novel distracter (Brainerd & Reyna, 1988). In fact.

Belli (1989) and Tversky and Tuchin (1989) found that control and misled subjects were equally good at rejecting these novel items. Another problem with these forced-choice tests is that subjects cannot reject or accept both responses. As shall be discussed below, theoretical assumptions about memory cannot be explicitly tested unless data include individual responses to both original and misleading information.

In summary, empirical results fi-om other misinformation research signal the relevance of several design features necessary for other studies. First, the timing of 17

misleading information and timing of the final memory test are clearly important. Immediate misinformation and memory tests, delayed misinformation and memory tests, and a third category with immediate misinformation followed by a delayed memory test should be included. Second, a memory test that measures responses to both original and misleading information gives subjects the opportunity to accept or reject both items, an option not available in multiple alternative forced-choice tests. Third, most misinformation studies typically have tested only memory for details and not tested memory for meaning. Therefore, the effects of different types of misleading information needs to be explored further (see Titcomb & Reyna, 1992; 1994, for preliminary studies).

The theoretical rationale for these and other design features will be described in the next section.

Theoretical Explanations for Misinformation Effects

Storage Explanations

Discrepancy Detection. As mentioned in the introduction, Loflus and colleagues have offered "discrepancy detection" as a likely explanation for misinformation effects.

That is, as memory for the original information increases, suggestibility to misleading information decreases. As early as 1978, Loflus et al. noted that "...the weaker the original trace, the easier it is to alter" (p. 25). Loftus and Hofl&nan (1989) restated this assumption: "...when memory for an original item is virtually perfea, people are unaffected by misinformation" (p. 103). More recently, Toland, HoflEman and Loftus

(1991) concluded that "...the stronger the original event memory, the more resistant to post-event contamination" (p. 237). Others have also noted this relationship. For example, Greene (1992) summarized Loftus's work by saying that the "... is only found when the misinformation subtly contradicts one's memory for the original event. More blatant and direct contradictions cause little interference" (p. 162). 18

The main evidence cited for this explanation revolves around better performance by subjects (i.e., they are harder to mislead) when the critical event is a "blatant" detail. Presumably, these details are more likely to be encoded and remembered, causing rejection

of misleading information when it is initially presented (Loftus, 1979). Tousignant, Hall

and Loftus (1986) measured and manipulated reading speed and inferred that better

attention to detail leads to less suggestibility. However, none of these experiments directly tested the "discrepancy detection" assumption that better memory for original information necessarily supports resistance to misleading information.

Misinformation Acceptance. Several researchers have discussed the possibility

that misinformation effects stem from a "gap filling" function (Belli, 1989; McCloskey &

Zaragoza, 1985). For example, Zaragoza, Dahlgren, and Muench (1992) pointed out that low levels of control subject performance should be associated with a greater misinformation effect. In other words, subjects who fail to encode event information are

more likely to be misled. Pirolli and Mitterer (1984) reported that false alarm rates following inconsistent presuppositions were higher among subjects who previously failed to recall target information than for subjects recalling targets. For example, they used a

free recall test in between the initial event (slides) and the final recognition test. They found that subjects who did not recall original items, such as stop sign, on the intermediate test were more likely to "recognize" misleading suggestions, such as yield sign, at the final test. Note that these examples parallel the same assumption about memory as discrepancy detection. However, Belli (1989) distinguished "misinformation acceptance" from other explanations for misinformation effects. When a subject fails to encode or store the original information, the "gap" might be filled with the misleading postevent item.

Misinformation acceptance could be characterized as an extreme end of a discrepancy detection continuum— that is, lack of any memory for the original event leads to the most 19

suggestibility. However, Belli described misinformation acceptance as being different from alteration hypotheses because, he argued, if there is nothing in memory in the first place, it cannot be changed or altered by subsequent information.

Lofhis and HofiOnan (1989) concluded that misinformation acceptance could be an important concept in understanding the creation of new memories. With regard to predictions about memory performance, Toland et al. (1991) predicted that

"Misinformation Acceptance would result in subjects responding 'no' when questioned about original and novel information" (p. 246). Therefore, misinformation acceptance can be distinguished from discrepancy detection in that it predicts more frequent rejection of original information and acceptance of misleading information for the misled subjects than responding "yes" to both (a pattern explained by various retrieval explanations described below) or "no" to the lure and "yes" to the original if memory is good for the initial event.

The control subjects, some of whom may also not remember item, would be more likely to be at chance, or guessing, levels when tested with the original and lure items because they had not been presented with the misleading information prior to the memory test. Retrieval Explanations

Coexistence. Storage explanations for misinformation effects have been questioned by researchers who claim that subjects actually remember both the original event and the misleading information. For example, Bekerian and Bowers (1983) stated that "...forgetting will depend on whether the retrieval conditions bias one to use an appropriate strategy for accessing the old memory" (p. 140). In their study, misleading information was presented as a set of questions that were presented randomly or in a sequential order matching that of the initial event. Memory was then tested, either randomly or sequentially. Because subjects performed better when memory was tested sequentially, Bekerian and Bowers assumed that memory for the original event was not 20

erased or altered by misleading information. Presumably, sequential order acts as a good

retrieval cue, unearthing the original memories that were there all along. When subjects identify original information under enhanced retrieval conditions (e.g. reinstating order at the time of test), many theorists reason that both pieces of information "coexist" in memory. Furthermore, these efifects are presumed to occur at the time of retrieval, or

testing, when both items "compete," and "access" to one item may "block" or interfere

with the retrieval of the other item. In other words, "Multiple trace hypotheses assume

that the original trace coexists with subsequently formed traces and that the original trace is unaffected by subsequent learning" (Chandler 1991, p. 115). Thus coexistence explanations would predict that, if offered the original item as a recognition stimulus, misled subjects should accept that item at the same level as control subjects.

Discrimination. Blocking and Cue-Chanee. More recently. Chandler and Gargano

(1995) used a word-list paradigm to investigate interference effects in general, rather than misinformation effects in particular, and tested several "retrieval" explanations relevant to the current study; (a) Discrimination, where more interference should occur under conditions of more difficult discrimination, e.g., when misinformation is presented immediately and when it is "similar" in meaning to the original information; (b) Blocking, where more interference is expected when misinformation is presented closer to the time of test; and (c) Cue-change, where interference is greater after delayed misinformation and with items that are different in meaning from the original information. Their results, greater interference with "delayed" and "different" distracters, favored the cue-change explanation.

Fuzzy-trace Theorv

Fuzzy-trace theory is a comprehensive theory of reasoning, remembering, and their relationship. Fuzzy-trace theory posits that memory representations can be described as 21

two, independent types: Gist, the essence or general meaning, and Verbatim, the surface details. These representations are assumed to be encoded in parallel and stored independently. Furthermore, each representation fades at different rates, with memory for gist as the most robust representation and memory for detail as most fragile and short­ lived. All of the subtleties of the theory will not be reviewed here (see, for example, Reyna & Brainerd, 1995a). However, tenets of fiizzy-trace theory that are most relevant to the interpretation of misinformation effects include the specificity of gist versus verbatim memory representations and differential forgetting rates of gist and verbatim details (Reyna & Titcomb, in press). After each of these is described below, recent research applying them to misleading information will be summarized. As mentioned above, fiizzy-trace theory describes two types of memory traces that are encoded in parallel and are independent of one another. These two types of memory differ in their specificity regarding actual events and information. Gist encompasses semantic, more global representations of experienced events. On the other hand, verbatim representations are more veridical, exact memories of the events and associated details.

Because gist, by definition, is more diffuse in nature, subjects cannot rely on it to reject details that do not contradict the gist. For example, using the above mentioned car accident scenario, if subjects retrieve the gist that the car passed some sort of traflSc sign, they might erroneously accept that it was a yield sign because it does not contradict their vague memory for the category. Other researchers have also noted such an important distinction. For instance, Bekerian and Bowers (1983) pointed out that "...when people view the original slide sequence, they extract general thematic information in addition to the specific details" (p. 143). However, they depicted the thematic information as an

"appropriate strategy" for a cueing retrieval of the original details and did not denote underlying differences in the memorial representation of the original information (p. 140). 22

It is well known that verbatim memory becomes inaccessible more rapidly than

memory for gist. For example, List (1986) claimed that, although inconsistent items were better remembered in the immediate condition, schema consistent items (i.e.,

gist consistent) were better remembered the next day. Ceci and Bruck (1993) predicted

that "...verbatim representations should not survive as long as gist representations because they are more susceptible to disintegration" (p. 413). Therefore, after a delay, subjects are

more likely to retrieve gist. As related earlier, delay is an important factor in obtaining

misinformation eflfects. There are factors which can reduce the disintegration of verbatim memory. For example, Brainerd and Omstein (1991) point out that early memory tests

enhance memory performance.

The relative benefits of robust gist memory over time seem, in general, to

outweigh the disadvantages. It is not uncommon for gist memory to assist a person in

reasoning out the correct answer to a problem (Reyna & Brainerd, 1992). Because

evidence for strong gist memory has been noted in research studies (Clark &. Clark, 1977),

theories of memory have arisen that predict varying memory relationships tested in this

study. For example, in a current text on memory and reasoning, Schwartz and Reisberg

(1992, p. 355) state that, "The presence of a schema improves memory for materials that

fit the schema, and works against memory for materials that do not fit. This means we will

be less likely to remember elements that were present but incongruent with the schema, and more likely to add elements that were absent but congruent with the schema."

Schema theory predicts that original details will not be as memorable as items consistent with the schema, i.e. consistent gist.

Fuzzy-trace theory has already been used in the interpretation of misinformation effects. Titcomb and Reyna (1992; 1993; 1994; 1995b) adapted written materials fi'om

Harris, Lee Hensley and Schoen (1988) and used Loftus's paradigm to look at the effects 23

different kinds of misleading information. Titcomb and Reyna (1992) found that inconsistent details were the hardest to misinform whereas misleading inconsistent gist produced the largest misinformation effects. Although both misled and control subjects showed high acceptance rates for consistent gist, the relationship between memory for original items and memory for consistent gist was positively dependent only for control subjects. This "autosuggestiblity effect" (Brainerd & Reyna, 1995), combined with the lack of any relationship for misled subjects' memory for original and consistent gist, indicates that external sources of misleading information may result in distinct memory representations. A fourth category of misleading information, consistent detail, was added in

Titcomb and Reyna (1994; 1995b). Note that "consistent" is not used in the same way as other misinformation studies (e.g., Bekerian & Bowers, 1983; Loftus et al., 1978; Pirolli

& Mitterer, 1984) where "consistent" information is equivalent to original events. Here, consistent details are new pieces of information that do not contradict existing details or meaning. The results of these two studies generally replicated the earlier results and consistent details were misrecognized by misled subjects more frequently than inconsistent details or inconsistent gist. Consistent gist again received the highest acceptance rates.

In summary, there is both empirical and theoretical support for differences between gist and verbatim types of misleading information. Furthermore, by varying the time at which misleading information is presented and at which memory is tested, an experiment can obtain data relevant to the assumption that gist and verbatim traces show differential forgetting rates. Also, different types of memory tests may cue different types of memory representations. In addition to the above mentioned yes-no (verbatim) test, a second judgment based on gist can be added. Last, obtaining individual judgments about both 24

misleading information (lures) and original information is the only avenue for correctly determining memory dependency.

Specific Predictions

The point of Table 1 is that most storage and retrieval explanations for misinformation effects make the same basic predictions about verbatim (inconsistent or altered) details with verbatim memory tasks (i.e., Yes-No verbatim test results). The storage explanations predict that recognition of misleading information will be necessarily greater when memory for original information is poor. Retrieval explanations predict that access to original information is blocked, or out competed, by the more recently presented misleading information. Both of these predictions are represented by negative dependency. However, these predictions are limited to the Yes-No (verbatim) test. They do not explicitly address the question of memory dependency with a True-False test.

Furthermore, the negative dependency predicted by the theories concerns only the inconsistent detail category, since past studies have included only altered (inconsistent) details as stimuli. Thus, the theories are almost indistinguishable on this level. In contrast, fiizzy-trace theory, with the relatively few assumptions mentioned above, makes many more specific predictions about the relationship between memory for original information and memory for misleading information.

In addition to the memory dependency relationships in Table 1, and based on assumptions fi-om Titcomb and Reyna (1995) and Reyna and Titcomb (in press), I predict greater misinformation effects when misleading information is presented after a delay and memory is tested after a delay. Memory for the specific details of the event are more likely to be forgotten in one week and the more robust gist traces are not likely to contradict the inconsistent details. However, strong memory for gist is likely to allow 25

rejection of inconsistent gist items by control subjects. In contrast, inconsistent gist will also show the greatest differences in misleading information. That is, misled subjects, once they have been exposed to inconsistent gist items, will incorporate this information into their memory for the event and come to accept the incorrect gist more than the control subjects. The specific nature of the relationship between memory for the original information and memory for the inconsistent gist will depend on the degree to which subjects forget the original event and how well they rely on their memory for gist.

Consistent gist, on the other hand, will be misrecognized by both misled subjects and control subjects, although not as much in the immediate condition when verbatim memory is at its best. Furthermore, when verbatim memory is best, the ability of subjects to reject inconsistent details should increase. In other words, subjects are least likely to be misled with altered (inconsistent) details in the immediate condition. 26

CHAPTERS

METHOD

After a description of the participants, this chapter will summarize the design, materials, procedure, and analyses. The main manipulation, type of misleading information, is included under the measurement subsection of materials.

Subjects The study participants were adults recruited from institutions of higher learning in southern Arizona. Most of these were college students in an upper division speech and hearing science course and a sophomore level developmental psychology course. Others were enrolled in a freshman level psychology course All of the previously mentioned subjects received course credit for their participation. The remaining subjects were volunteers from the university community. The average age was 26 years (min. 18, max.

66) and 86% of the participants were female. A total of 99 subjects completed the study, with 33 in the immediate tests condition, 34 in the immediate misinformation/delayed recognition test condition and 33 in the delayed tests condition. An additional two subjects did not complete the task and one additional subject wrote on his packet that he did not understand the recognition test instructions.

Design

The study involved a3X2X2X2X2X2X2 factorial design. Time ("Immed" with immediate misinformation and immediate recognition test, "Inmied/Delay" with immediate misinformation and a delayed recognition test, "Delay" with both delayed misinformation and a delayed recognition test) and Version (set A, set B) comprise the between-subjects factors. Assignment to a Version set determined the actual items on 27

which subjects would be misled. Within-subjects factors included Gist (Gist-Detail items) and Consistency (Consistent-Inconsistent information) which comprised the four types of misleading information (consistent gist, consistent detail, inconsistent gist, inconsistent detail). Item (two replications of each type of misinformation), and Judgment (Yes-No verbatim, True-False) were also within-subjects factors. Finally, all subjects rated Lure and Original items on which they were Misled or Not Misled (i.e. controls).

Materials A slide sequence (44 total slides) depicting a story about a woman and child was presented in fixed order at a fixed rate of eight seconds per slide. These slides were used in a previous misinformation study (Belli, et al., 1992). See Appendix E for a description of the slide contents. After viewing the slide sequence, participants completed a comprehension test in which was embedded misleading information. For example, the slides showed a woman hanging up a plaid shirt on the clothesline. The misleading question on the comprehension test then asked, "As the woman was hanging a pair of jeans out to dry, did the boy leave the house?" Participants selected "Yes" or "No" and indicated their confidence on a 7- point Likert-type scale. (See below and Appendix D for examples of misleading information and Appendix B for an example of a comprehension question and the complete list of comprehension items). Each subject was exposed to eight misleading questions (two of each type), eight questions containing original information fi-om the slides, and eight filler questions containing original information unrelated to the design.

All test questions were presented in random order.

Tvpes of Misleading Information A consistent gist item is a pragmatic inference that was not directly stated or shown. For example, the slides implied that a park was fairly close to the house but never 28

revealed the exact distance. The consistent gist statement was "The park was close to the house." The examples were drawn from a pilot study where nineteen participants responded to specific questions about information that was not explicit in the slides. In other words, when asked, "What was the distance fi"om the house to the park?" most people responded that it was "a couple of blocks" or "a few minutes" away. In other words, they thought the park was fairly close. Inconsistent gist was something that violated or was in opposition to a pragmatic inference drawn from the slides. For example, "The boy was hiding the pennies" goes against the implication in the slides that he was merely playing or counting them. The examples were created to oppose the "gist" represented by answers from the pilot study.

Inconsistent detail was a category that paralleled traditional "misinformation"—i.e., a detail that was directly contradictory to a detail in the slides. For example, the slides showed the boy on a swing set. The statement "The boy was found on a slide" was inconsistent with the original detail. This type of misleading information is also called

"inconsistent detail" in this paper. The consistent detail stimuli involved new details that neither contradicted existing details nor altered the meaning of the story. For example, the slides showed a couple of gold colored cushions on the sofa. The comprehension test asked a misleading question about "The velvet cushions on the sofa." The examples were generated by identifying an existing item in the slides where additional information (such as "velvet") could be true but could not be ascertained precisely by looking at the slides. Note that this definition of consistent differs from that of Loftus, Miller & Bums (1978) and Pirolli & Mitterer

(1984) where consistent meant something that was in the scene, i.e. repeated details that were actually seen in the original event. 29

After completing the comprehension test, all participants completed the same recognition test containing both misleading and original information. Two judgments were made about each item; "verbatim judgment" (i.e., Yes-No questions) and "truth

judgment" (i.e., True-False questions). The former asked whether they had seen the

information in the slides and the latter asked subjects to judge whether the information

was true whether or not they had seen the information. Subjects also rated their

confidence for each judgment on a 1-7 Likert-type scale. (See Appendix C for an example of the recognition test). The recognition test consisted of 16 "lures", eight of which were presented to the subject in the comprehension test as misleading information and eight

which were "controls," and 16 "originals," (i.e., accurate events and information from the

slides). Eight filler recognition items rounded out the recognition test. (e.g. total correct

responses included 18 Yes, 22 No; 22 True, 18 False; 26 Consistent; and 14 Inconsistent).

All items were presented in random order. Appendix D lists all the recognition items. Procedure Subjects participated in groups ranging in size from 5 to 45 persons. The

experiment occurred in classrooms and seminar rooms. After completing the human

subjects consent form, participants were instructed to pay close attention to the slides

because they would be asked questions later (See Appendix A for examples of written

instructions). After showing dummy slides aligned vertically and horizontally, subjects

moved if they wanted a better view of the slides. Once everyone was comfortable, the slides were presented at a fixed rate of 8s.

The procedure consisted of three stages: Participants first viewed the slide sequence, then completed the comprehension test after which they responded to the

recognition test. The exact procedure varied in the timing of the comprehension test and the recognition test. One-third of the subjects experienced all three stages in the first 30

session (Immed). One-third received the misleading information (embedded in the

comprehension test) immediately after the slides but took the recognition test at a second

session one week later (Immed/Delay). Finally, one-third of the subjects saw the slides in session one, but completed both tests in session two one week after slides (Delay). Subjects were randomly assigned to one of the three timing conditions and one of

two versions with comparable but different stimuli. After the slides, appropriate packets

were handed out and subjects were orally reminded about the written instructions (e.g..

Do not skip anything, proceed through the packet one page at a time, ask questions if you

need fiirther clarification, do not discuss the study with others until everyone is done). After completion of the packet, participants were fi^ee to leave the room. Analyses

The results of the recognition test were analyzed in two ways. The first analysis

involved a repeated measures analysis of variance (ANOVA) procedure. The main analysis for the recognition items was a 3 (Time: Immed, Immed/Delay, Delay) X 2 (Version: Set A, Set B) X 2 (Misinformation: Misled, Not Misled) X 2 (Lure-Original) X

2 (Judgment Type: Yes-No, True-False) X 2 (Consistent-Inconsistent) X 2 (Gist-Detail) X 2 (Item: 1, 2) design. Differences between means were confirmed by Student-Newman-

Keuls tests for multiple comparisons. The dependent measure, signed confidence, was computed by assigning a positive or negative value to the subject's confidence rating for each judgment. Confidence ratings for "yes" (verbatim) or "true" judgments were assigned positive signs and negative signs were attached to confidence ratings for "no" or "false" judgments. Thus, a +7 for the verbatim judgment would indicate maximum confidence that the information had been presented in the slides and a -7 would indicate maximum confidence that it had not been presented. A second ANOVA was also completed for the Yes-No, True-False responses. 31

The last set of analyses tested contingency tables of yes-no (verbatim) and true- false judgments of all "lures" and their paired original statements. In other words, the tables contained the frequency of responses in each of four cell categories; Yes to both original and lure. Yes to original and No to lure. No to original and Yes to lure, and No to both items. The True-False judgments were analyzed separately. The contingency tables were tested for significance with likelihood ratio analyses. Independent events were those for which the product of the marginals (or unconditional probability) did not differ statistically from the proportion of Yes responses to both lure and original items.

Dependent events (those with significant likelihood ratio chi square results) were labeled positive if the proportion of Yes to both responses exceeded the product of the marginals. Conversely, negative dependency occurred when the proportion of Yes to both responses was lower than the produce of the marginals. On the one hand, if subjects were more likely to say Yes to one item if they said Yes to the other, then the relationship was considered positively dependent. On the other hand, if subjects were more likely to say

No to one and Yes to the other item, and these judgments were contingent, then the relationship would be negatively dependent. An important point to keep in mind is that there must be some variance in the data in order to get a significant relationship. Ceiling effects and floor effects will result in statistically insignificant likelihood ratio analyses. 32

CHAPTER 4

RESULTS

Subjects completed recognition tests eitlier the same day after seeing the slides and misleading information (Immediate) or one week later (Immediate/Delay, Delay). The recognition test responses were analyzed with a repeated measures ANOVA. Reported differences in signed confidence data were confirmed with Newman-Keuls analyses and were significant at the .05 level. Where appropriate, the mean signed confidence is reported in the text. The remaining means are listed in the Appendices as are results fi'om the Yes-No, True-False (% Response) ANOVA. The results in this section will be described with an emphasis on gist-verbatim differences in types of stimuli and judgment, and the effects of forgetting. After these effects have been described, results fi'om the memory dependency tests will be introduced in the last section. Signed Confidence Analysis of Variance

Gist Versus Verbatim Details and Consistent Versus Inconsistent Stimuli. Gist items in general received higher ratings (mean 3.4) than did detail items (mean 1.0), F (1, 92) = 255.74, MSE = 36, g < 0.001. All Gist-Detail interactions occurred in tandem with the Consistent- Inconsistent or Judgment type factors and will be described below. The significant main effect for consistency, F (1,92) = 139.01, MSE =27.52 , £ < 0.0001, reveals that, on average, consistent items (mean 3.0) were rated more highly than inconsistent items (mean 1.5).

When types of misleading information were factored with Lure/Original statements, a significant three-way interaction, F (1,92) =216.47, MSE = 26.42, £ <

0.0023, showed that the consistent detail originals were much lower than the originals for 33

the other three types of misinfonnation (see Figure 1). Consequently, the low lure ratings for consistent details are diflBcult to interpret. The important results to note from Figure 1 are that, overall, subjects misrecognized consistent gist as often as they correctly recognized original items. In contrast, subjects were less likely to misrecognize inconsistent lures, whether they were gist or detail lures.

Diflferences Between Verbatim and Gist Judgments. All subjects made two types of judgments about each recognition item. The "gist" judgment asked whether the statement was true according to the slides. The "verbatim" judgment asked whether subjects had seen the information in the slides. The main effect of Judgment, F (1,92) = 74.37, MSE =25.74, g < 0.0001, showed that, in general, gist judgments (True-False) were made with greater confidence than verbatim (Yes-No) judgments. Although this pattern was consistent across all results, the Judgment factor interacted with Lure-Original ratings, Gist-Detail items, and Consistency. These interactions are described below and in

Figures 2, 3, and 4.

When subjects made verbatim judgments, the highest ratings (mean 5.1) were made for consistent gist (see Figure 2). Inconsistent details and inconsistent gist items received moderately low ratings (1.0 and 0.8 respectively) and consistent details received the lowest ratings (-0.1). This echoed the previously mentioned result of lower consistent details. For truth judgments, the pattern varied only in that inconsistent gist was rated slightly higher (2.3) than inconsistent details (1.8). The anchors of the continuum were still occupied by consistent gist (5.6) at the highest end and consistent details (1.5) at the lowest end. The other elements of the interaction showed consistent patterns with previous results, F (1,92) =34.19, MSE = 9.14, g < O.OOOI.

For verbatim judgments, consistent gist items were greater than inconsistent details which were equal to inconsistent gist items which, in turn, were greater than consistent 34

details. This was similar to the earlier pattern that showed consistent details at the bottom. For truth judgments, the pattern was slightly different. Consistent gist items were higher than inconsistent gist which were greater than either of the detail items. The interaction between Lure-Original, Consistency, Gist-Detail and Judgment, F (1,92) =14.92, MSE = 6.48, g < 0.0003, serves as a sununary for this section. The

important point about the verbatim, or Yes-No, judgments in Figure 3 is that most

originals were well-recognized with the exception of those for consistent details. Again,

this repeated the pattern mentioned above. For the originals, subjects rated consistent gist most highly. However, the lowest ratings were Yes-No (verbatim) judgments about inconsistent gist and inconsistent details. In other words, the participants had fairly good

memory for what they saw in the slides. Even though they did not recognize consistent

details items to the same extent that they recognized consistent gist items, these two types

were more accepted than either of the inconsistent types of stimuli. They were able to

reject "obviously" inconsistent information and accept correct information. Subjects also showed the familiar effect of misrecognizing consistent gist as something that actually

occurred. In all cases, truth judgments (Figure 4) were higher than verbatim ratings

although they reflected the same basic pattern of relationships.

Effects of Forgetting (Time). Although there was no main effect for Time, F (1,

92) = 0.85, MSE = 83.80, £ < 0.4302, interactions with other factors revealed that a one- week delay in taking a recognition test tended to result in smaller differences between ratings. For example, original ratings were greater than lure ratings across the three experimental groups (Immed, Immed/Delay, Delay) but original ratings decreased over time and lure ratings tended to increase over time, F (2,92) = 16.06, MSE = 34, g <

0.0001 (see Figure 5). Gist ratings tended to maintain high levels over time (Figure 6) while detail items, low to begin with, tended to increase over time, F (2,92) = 4.35, MSE 35

= 36.00 , B < 0.0157. The significant interaction of Time X Judgment, F (2,92) = 4.14,

MSE = 25.74 , £ < 0.0191, showed that verbatim judgments, across all items, tended to stay low across time (means of 1.6, 1.8, 1.6). In contrast, the truth judgments were higher in the Delay condition (mean 3.2) than in the other two conditions (means 2.6 and 2.5). This supports the notion that gist memory remains robust over time.

The four-way interaction among Lure, Gist, Consistency and Time factors, F (2,92) = 10.92, MSE = 26.42, p < 0.0002, summarizes the effects for this section (See

Figures 7 and 8). Except for consistent gist, where originals and lures were equally high, all other originals had higher ratings than their corresponding lures. The lowest, although still high, ratings for consistent gist originals occurred when both misleading information and the recognition test were completed one week after the slides (Delay). Inconsistent

gist originals also showed lowest ratings in the Delay condition. The largest effect of

Time on forgetting of originals is seen in the inconsistent detail condition. Inmiediate

testing appeared best for verbatim memory because this category showed the highest

ratings. These ratings tapered off in both of the following time conditions. Therefore, any delay in the recognition test was associated with decreased recognition of original items. The main point to note about consistent detail originals is that they were much lower than

any of the others. They also exhibited a crossover in that the highest recognition rates for

the category occurred with the Delay condition. This suggests that subjects did not

remember the originals very well, and, when compared to the lure ratings in Figure 8,

revealed a pattern more similar to lures than to original items.

Figure 8 shows that consistent gist lures maintained their high recognition rates over time (all three Time categories were equal). Inconsistent gist lures were equal for

Immediate and Immediate/Delay conditions (when misleading information was conveyed through an immediate comprehension test), but were lowest in the Delay condition (when 36

both tests were presented after a week). Inconsistent detail lures were easily rejected (-

3.9) in the Immediate condition, when verbatim memory was best. However, they came

to be more accepted over time (all three ratings are different). Consistent detail lures were more recognized than the inconsistent items, but are not as frequently misrecognized as consistent gist. Consistent detail lure ratings went up very slightly between the Immediate

(-0.3) and Delay (0.8) conditions, but were at zero (which was not different from -0.3) in

the middle condition. Misinformation Effects. Although the Misinformation main effect was not

significant, F (1,92) = 2.58, MSE = 24, £ < 0.12, there were significant interactions with

Lure-Original statements, Gist-Detail types of misleading information, Consistent-

Inconsistent statements, and the combination of all four elements. These interactions will

be described below. Figure 9 shows that Consistency and Lure-Original ratings varied depending on whether subjects were misled on the items, F (1,92) =406.13, MSE = 30.11, g = 0.0004.

Original ratings reflected the same pattern as the main effect with originals recognized more often and with greater confidence than lures with the exception of Misled consistent items which fit with the above mentioned Lure-Original X Consistency X Gist-Detail interactions. The important point to note about Figure 9 is that the differences between lure and original ratings were smallest for misled items. In other words, even though misled subjects, on average, rejected inconsistent lures and accepted original items, they did so with less frequency and lower confidence than those who were not misled. This, in essence, is the misinformation effect. Consistent Detail. Item and Version. As mentioned above, the consistent detail set of stimuli received unexpectedly low ratings, especially the original items as rated by control subjects. Consequently, the interactions with Consistency and Gist-Detail will be 37

reported in the Appendix J but will not be discussed here. Furthermore, these low ratings explain many of the Item and Version interactions as well. It seems a set of the consistent detail items were rather diflBcult to remember. The specific means for each item, with misled and control subjects in separate columns, are also reported in Appendbc N. Memory Dependency Tests In order to address questions about the underlying memory representations that might explain the preceding misinformation effects, lures were paired with their corresponding originals in order to measure the degree to which acceptances of lures were dependent on acceptances of originals. The following tables describe the results of likelihood ratio analyses and memory dependency judgments designed to test these dependencies. Tables that report results broken down by Time can be found in Appendices L and M.

Yes-No (Verbatim) Results. Table 2 reports the results of the likelihood ratio chi square analysis of Yes-No (yerbatim) responses to original items and their corresponding lure items. The most important thing to note is that different kinds of misleading information show different dependency results. Another important note is that the positive dependency for consistent gist is not the same kind of positive dependency as for consistent details. Inspection of the proportion of subjects responding Yes to both the original and the lure reveals the difference (see Appendbc K). For example, 80% of the misled subjects responded Yes to both the original and consistent gist lure items.

Conversely, only 30% or the misled subjects claimed to see both the original and consistent detail lure.

True/False Results. As can be seen in Table 3, the True-False judgments show roughly the same pattern of dependencies. It must be pointed out that the relationship between truth judgments for original information and consistent gist did not show any 38

variation and so the likelihood ratio chi square resulted in no statistical dependency. The pattern of responses, however, is the same as that reported in the verbatim dependency pattern. Most people responded True to both the original and consistent gist item.

Because of these ceiling effects, the positive dependency was masked by lack of variation in the data. 39

CHAPTERS

DISCUSSION

In general, the results from this study support the predictions described in the

Introduction and in the Literature Review. The first part of this chapter will focus on the results of the signed confidence analysis of variance and how they compare to previous misinformation studies. The second part will address the memory dependency results and how they relate to the theoretical predictions about memory representation. Finally, implications for applied aspects of memory and suggestions for further research will be discussed. Si|gned Confidence Analysis of Variance Gist-Details and Consistent-Inconsistent Differences and Effects of Time. With the exception of consistent details, this study generally replicates earlier experiments using similar design elements (Titcomb & Reyna, 1992, 1993, 1994, 1995b). Specifically, the types of misleading information (consistent gist, inconsistent gist, and inconsistent details) differed in their ratings when compared with each other and with original information.

Consistent gist was frequently misrecognized by all subjects. On the other hand, inconsistent information of both types was rejected by control subjects but was accepted more often by misled subjects. Furthermore, these patterns of acceptance and rejection varied over time, with original ratings lower and lure ratings higher when recognition tests were delayed after a week. Consistent gist recognition rates maintained the highest ratings over time, consistent with fiizzy-trace theory's prediction of robust memory for gist.

Likewise, subjects rejected inconsistent gist at fairly high levels, even after a week. This suggests that subjects did rely on memory for gist when they completed the final 40

recognition tests. Similarly, the highest rate of rejection for inconsistent details occurred in the immediate testing condition when verbatim memory is best. This, too, was as prediaed.

Differences between Gist and Verbatim Judgments. Zaragoza & Koshmider

(1989) predicted that "...although subjects will...claim to recognize inferences they have drawn as previously present, they do not make these errors when asked to distinguish between what they believe to be true and what they specifically remember seeing" (p. 248). The results in this study indicate that, to the contrary, subjects make the same mistakes but show higher confidence when asked to make True-False judgments about the items.

Misinformation Effects. Because there were misinformation effects even in the immediate condition, these results differ somewhat fi-om other studies (e.g.. Belli et al.,

1992). It should also be noted that the encoding conditions (8s per slide) were longer than in other studies. For example. Belli et al (1992) showed the same slides at 5s and .3 s rates. Tousignant, et al. (1986) showed a different set of slides at a 4s rate. Loftus et al. (1978) showed their slides at a 3s rate. Additionally, most of the items tested here were

"central" both perceptually and thematically. In other words, the context of this study provided opportunity for better than average memory for details. This can be seen in the relatively high recognition levels for original items across all three timing conditions. That misinformation effects resulted at all illustrates that memory is subject to change even under the best of conditions.

Consistent Detail. Item and Version Differences. The exception, mentioned several times, lies in the consistent detail items. These items were not central in the sense of the action/story of the slides. The low ratings for even the original version of these items echoes previous work (Cassel & Bjorklund, in press) which found that central items 41

were better remembered (who stole the bike) than peripheral details (the color of the bike). I assume, based on the results, that most of the consistent details were not encoded. Anecdotal remarks made by participants also support this assumption. For example, in response to the question about the "flowered" coffee mug (the information

"flowered" was the added, or "consistent" detail), one subject asked, "What coffee mug?"

However, this result does provide an interesting test of the assumptions of misinformation acceptance. In this case, lack of encoding did not lead to more susceptibility to misinformation. Although there were significant misinformation effects, the magnitude was less than the other types of misleading information rather than more. Furthermore, because the details were subtle, subjects could not use gist memory to reject them. Memory Dependencv Results

All theoretical accounts correctly predicted the overall negatively dependent relationship between memory for inconsistent details and memory for original items. That is, as memory for one increased, rejection of the other necessarily increased. However, some of the more subtle predictions did not hold. For example. Misinformation Acceptance predicts more frequent responses in the pattern No to originalA^es to lure. In the immediate condition for inconsistent details, the percentage of subjects with this response was very low (3%). I infer from this that Misinformation Acceptance does not explain the effects.

Since none of these theories makes predictions about the remaining conditions (i.e., truth judgments, differences among types of misleading information, differences between misled and control), we must turn to fiizzy-trace theory for insight. As with the storage and retrieval theories, fuzzy-trace theory also predicted negative dependency for inconsistent details and original information, albeit for different theoretical reasons.

Furthermore, it accurately predicted the positively dependent relationship between 42

memory for gist and memory for original information. What it did not predict were the dependency outcomes for inconsistent gist and consistent details. As mentioned previously, consistent details were not well remembered. This lack of memory was shown in the low ratings made by control subjects for original items. Inspection of the contingency tables showed that subjects tended to reject both the original and lure items, a diflferent pattern from that seen with consistent gist where subjects tended to accept both.

The high rates of acceptance in all gist consistent conditions (misled versus control) may reflect the tightly woven nature of the inferences tested. For example, the slides showed the woman fixing eggs and toast. The immediate inference which leaps to mind is that she is fixing breakfast. When I spoke with subjects after the experiment, they agreed that they did not really know for sure if it was breakfast (it could have been lunch), but this was after I asked them to recall what they saw. Remembering either of these would cause the subject to say yes to the other. This is another example of how gist memory can help or hinder, depending on the type of question being asked. The inconsistent gist items, on the other hand, were not as unrelated to the original items as the consistent gist was closely related to the original items. For example, one of the questions asked if the woman was still angry when they returned home. Pilot study responses indicated that she probably was not still angry because subjects remembered her hugging the boy. However, she could have been angry and hugged the boy, although this was unlikely given the slides. Perhaps some subjects responded in the afiBrmative to these items because they figured it could have happened. In other words, the original and lure did not necessarily contradict one another even if they intuitively contradicted one another. This, then, could result in a greater variation of responses which would lead to no significant difference in the contingency tables. The inconsistent gist result is different than those previously reported possibly because the consistent and inconsistent items in 43

earlier studies (Titcomb & Reyna, 1993; 1995b) involved the same original information.

Here, they represented comparable, but different, elements of the slides. In short, an attempt was made to ensure good encoding of all original materials. However, the items tended to vary in their memorability. Implications for Eyewitness Memory Research

Fuzzy-trace theory has several implications for eyewitness testimony (see Reyna & Titcomb, in press, for a review). The two most important suggestions supported by this study include early, neutral questioning and questions that focus on remembered details.

Based on the results that showed best rejection of inconsistent information in the immediate condition, interviews should be conducted as close to the witnessed event as possible to ensure the highest levels of verbatim memory. This conclusion is supported in other studies (e.g., Poole & White, 1991; 1993; in press) but is given further credence by the results reported here. Furthermore, these questions should focus on the specific details rather than on global judgments about the truth of experienced events or guilt of persons involved. In other words, asking eyewitnesses to focus on what they actually saw rather than what they believe to be true forces them, especially if questioned immediately, to access available memory for the details. This suggestion is based on the difference in specificity between memory for gist and memory for detail. As Reyna and Titcomb (in press) point out, "...misleading details might be accepted because they are compatible with how events are understood, rather than how they are remembered" (p. 17). For example a subject said (and I paraphrase), "I remember she found him at the park on some kind of playground equipment, but I don't remember what it was." If the outcome of a trial is determined by the accuracy of an eyewitness, and if that accuracy can be altered by misleading information, a clear distinction needs to be made between memory of gist and memory for detail. 44

Table 1 Summary of Memory Dependency Predictions

THEORY Relationship Between Memory Relationship Between Memory for Original Item and Memory for Original Item and Memor}- for Misleading Item: for Misleading Item: Verbatim Judgments Truth Judgments (Yes-No Test) (True- False Test)

Unitary Memory Theories: Storage Explanations; •Discrepancy Detection Negative —

•Misinformation Acceptance Negative — (No to Orig, Yes to Lure) Retrieval Explanations: •Coexistence (General) Negative — •Competition Negative — •Blocking Negative — •Discrimination Negative —

•Fuzzy-trace Theory: Misled Consistent Gist Positive Positive Inconsistent Gist Negative Negative, Independent, Positive (depending on forgetting and reliance on gist memory) Consistent Detail Independent Independent Inconsistent Detail Negative Independent

Not Misled Consistent Gist Positive Positive Inconsistent Gist Negative Negative Consistent Detail Independent Independent Inconsistent Detail Independent Independent Figure 1: Lure-Original Ratings for Different Types of Misleading Information

Original

Consistent Gist Inconsistent Gist Inconsistent Detail Consistent Detail c ra

-3

-5

-7

Stimulus Type Figure 2: Judgment Differences for Types of Misleading Information

Yes-No

Consistent Gist Inconsistent Gist Inconsistent Detail Consistent Detail • True-False

Stimulus Type

as Figure 3: Yes-No (Verbatim) Judgments of Lure-Original Items for Different Types of Misleading Information

7 T

B Original

Consistent Gist Inconsis ;ent Gist Inconsist mt Detail Consistent Detail D Lure

-5

-7

Stimulus Type Figure 4: Truth Judgments of Lure-Original Items for Different Types of Misleading Information

Original Consistent Gist Inconsistent Gist 1lnconsistent-0«i Consistent Detail

Stimulus Type Figure 5; Interaction Between Lure-Original Ratings and Time

3 - o c«> s 1 mmed c o u • Imm/Del •O0) -1' Lure Original cO) • Delay (0 -3

-7

Stimulus Type Figure 6: Interaction Between Gist-Detail Ratings and Time

H Immed

CD Imm/Oelay Gist Detail D Delay

Stimulus Type Figure 7: Effects of Timing on Original Ratings for Different Types of Misleading Information

7 T

• Immed

D Imm/Oel Gist Consistent Gist Inconsistent Detail Inconsistent Detail Consistent O Delay

-5

-7

Stimulus Type Figure 8: Effects of Timing on Lure Ratings for Different Types of Misleading Information

H M Immed D Imm/Del Gist Consistent nconsi J Incons istent Detail Consistent L D Delay

Stimulus Type Figure 9: Lure-Original Differences for Misinformed and Not Misinformed Consistent-Inconsistent Information

7 T

H Original

Consistent Consistent Inconsistent Incon! istent D Lure (Mis) (Not Mis) (MisI (IMot Mis)

Stimulus Type (collapsed across Gist-Detail)

Ui Table 2 Memory Dependency Results for Yes-No Recognition Test Verbatim Judgments about Originals and Lures

Type of Misleading Dependency Proportion Product of the LRCm Information Responding Marginals Probability Yes to Both Misled Consistent Gist PCS 0.80 0.78 0.0066 Inconsistent Gist IND 0.26 0.24 0.0714 Inconsistent Detail MEG 0.21 0.27 0.0000 Consistent Detail PCS 0.30 0.20 0.0000

Not Misled Consistent Gist POS 0.78 0.75 0.0001 Inconsistent Gist IND 0.14 0.16 0.1092 Inconsistent Detail NEG 0.16 0.18 0.0245 Consistent Detail POS 0.28 0.18 0.0000

Table 3 Memory Dependency Results for True-False Test Truth Judgments about Originals and Lures

Type of Misleading Dependency Proportion Product of the LRCm Information Responding Marginals Probability True to Both Misled Consistent Gist POS 0.87 0.85 0.0009 Inconsistent Gist POS 0.46 0.43 0.0131 Inconsistent Detail NEG 0.32 0.36 0.0032 Consistent Detail POS 0.49 0.38 0.0000

Not Misled Consistent Gist IND 0.87 0.86 0.0847 Inconsistent Gist IND 0.36 0.37 0.5108 Inconsistent Detail NEG 0.27 0.30 0.0084 Consistent Detail POS 0.45 0.35 0.0000 55

APPENDIX A: Instructions to Subjects

Instructions:

DO NOT SKIP ANYTHING. Read each question carefiilly and respond to each by marking your answer and circling your confidence. Answer each question in the order in which it is asked. DO NOT turn back in the booklet.

You will be asked two types of questions:

"Yes/No" Questions; When asked "Did you SEE this information in the slides?" your answer should be based on what you ACTUALLY saw, NOT what you may have inferred. If you remember seeing the information, then mark "Yes". Otherwise, mark "No."

"True/False" Questions: When asked "Regardless of whether you saw this information, is it true?" your answer should be based on what you believe is the case. Again, this is what you think is true, whether or not you distinctly remember seeing the information. If you think it's true, mark "True." Otherwise, mark "False."

Each question will be followed by a rating scale for you to indicate your confidence in your answer. Think of the confidence scale as equal steps of sureness with a "1" representing Completely Unsure (Guessing) and a "7" representing Completely Sure. 56

APPENDIX B: Comprehension Test Sample

When the woman hanged the jeans out to dry, was there snow on the ground? (Check one) QYes QNo

Confidence in my answer (Circle one); 1 2 3 4 5 6 7 Completely Somewhat Slightly Equally Slightly Somewhat Completely Unsure Unsure Unsiu« Sure and Sure Sure Sure (Guessing) Unsure

CRITICAL FEEMS: Gist Consistent/Original The boy went to a park close to his home. Did he pass other people along the way? The boy went to a park on foot Did he pass other people along the way? The boy pouted when the woman ihiwned at him. Did she scold him in the kitchen at that time? The boy stuck his lower lip out when the woman ihiwned at him. Did she scol him in the kitchen at that time? At the beginning, the boy was in his bedroom. Was he sitting on the floor? At the beginning, the boy was in a bedroom with toys. Was he sitting on the floor? The woman prepared breakfast Tor the boy. Did she use a cast iron skillet? The woman prepared eggs for the boy. Did she use a cast iron skillet? Gist Inconsistent/Original AAer eating, the boy wanted to play inside. Had he finished all his meal? After eating, the boy pointed to the door. Had he finished all his meal? The woman was still angry when they returned to the house. Later, did she give the boy a truck? The woman hugged the boy when they returend to the house. Later did she give him a truck? When the boy was hiding the penm'es. was there a glass jar nearby? When the boy was looking at the pennies, was there a glass jar nearby? The events happened in the spring. Was it a windy day? They work jackets when they went outside. Was it a windy day? Consistent Detail/Original The pink teddy-bear was on the clown bedspread. Did you see any other stuffed animals? The pink teddy-bear was dn the colorful bedspread. Did you see any other stuffed animals? In addition to the velvet cushions, was there also an afghan or blanket on the sofa? In addition to the gold cushions, was there also an afghan or blanket on the sofa? There was a flowered mug and a bottle of juice on the kitchen table. Was there also a milk carton? There was a coffee mug and a bottle of juice on the kitchen table. Was there also a milk canon? There were children's drawings of animals on the refrgerator. Were there photos there as well? There were children's drawings on the refiigerator. Were there photos there as well? Inconsistent Detail/Original As she was hanging a pair of jeans out to dry, did the boy leave the house? As she was hanging a shin out to dry, did the boy leave the house? The woman used a coffeepot while fixing the meal. Did you see a recipe book on the counter at that time? The woman used a toaster while fixing the meal. Did you see a recipe book on the counter at thai time? When the woman found the boy on the slide, was she upset with him for njnning away? When the woman found the boy on the swing set, was she upset with him for running away? Afier they read a cluldren's book together, did she give him a ball to play with? After they read a comic strip together, did she give him a ball to play with? FILLER ITEMS; The woman sat on a sofa at the beginning of the slides. Was it a brown plaid sofa? When the boy walked down the halL you could see his shirt Did it have a bird on it? The woman carried the laundry out in a basket Was the basket made of plastic? Some of the slides showed green grass. Was there also snow on the ground? Al the beginning, the woman wore red slippers. Later, were the tennis shoes she wore blue? After she discovered the boy was gone, did the woman drive to the park? At the end, was the boy alone when he played next to the drainpipe? In the slides that showed the woman and boy walking down the hallway, were the floors carpeted? 57

APPENDIX C: Recognition Test Sample The boy went to a park.

A. Did you SEE this information in the slides? (Check one) QYes QNo

Confidence in my answer (Circle one):

1 2 3 4 5 6 7 Completely Somewhat Slightly Equally Slightly Somewhat Completely Unsure Unsure Unsure Sure and Sure Sure Sure (Guessing) Unsure

B. Whether or not you saw this information in the slides, is the information true? (Check one); •True QFalse

Confidence in my answer (Circle one):

1 2 3 4 5 6 7 Completely Somewhat Slightly Equally Slightly Somewhat Completely Unsure Unsure Unsure Sure and Sure Sure Sure (Guessing) Unsure 58

APPENDIX D: Lure and Original Items

"Lures" a.k.a. Misleading Information Original Information

Gist Consistent: The boy went to a parlc close to his home. The boy went to a park on foot. The boy was in his bedroom. The boy was in a bedroom with toys. The woman fixed breakfast for the boy. The woman fixed eggs for the boy. The boy pouted when the woman frowned at him. The boy stuck his lower lip out when the woman frowned at him.

Gist Inconsistent; The events happend in spring. They wore jackets when they went outside. The woman was still angry when they returned to The woman hugged the boy when they returned to the house. the house. After eating, the boy wanted to piay inside. Alter eating, the boy pointed to the door. The boy was hiding the pennies. The boy was looking at the pennies.

Consistent Detail; The velvet cushions were on the living room The gold cushions were on the living room sofa. sofa. The flowered mug was on the kitchen table. The coffee mug was on the kitchen table. There were children's drawings of animals on There were children's drawings on the the refrigerator. refrigerator. The pink teddy bear was on the clown bedspread. The pink teddy bear was on the colorful bedspread.

Inconsistent Detail: The woman used a coffee pot while fixing The woman used a toaster while fixing breakfast. breakfast. As she was hanging a pair of jeans out to dry, the As she was hanging a shirt out to dry, the boy left boy left the house. the house. The boy was found on a slide at the park. The boy was found on a s\ving set at the park. They read a children's book together on the sofa. They read a comic strip together on the sofa.

Filler Items; (Same for everyone) The woman had long hair with bangs. (True Detail) The woman wore a dark vest. (True Detail) Boy wore mittens because it was cold. Gnconsistent Detail- Rainbow colored knit gloves) The boy was about twelve years old. (Inconcistent Gist/Detail— he was five years old) The woman was the boy's baby-sitter. (Inconsistent Gist- assume mother) The boy rode his bike. (Inconsistent Detail— he was never shown on a bike) There were signs of abuse. Gnconsistent Gist and Detail- No clear signs of abuse) LAST QUESTION ON RECOGNITION TEST; WITHOUT LOOKING BACK, would you change any of your answers now if you could? • Yes QNo If yes, which ones? 59

APPENDIX E: Description of Slides

Descriptive Contents of Slide Sequence from Belli et ai, 1992

1. Woman sitting on couch 2. Woman walks down hallway (backshot) 3. Woman walks down hallway (frontshot) 4. Young boy in bedroom with jar and pennies strewn across floor 5. 3/4 backshot of woman 6. Woman walks back down hallway 7. Woman goes in kitchen (refrigerator w/ child's drawings) 8. Woman pours glass of juice (milk carton on table) 9a. Woman pours water in coffeepot 9b. Woman puts bread in toaster 9c. Woman pours something in blender 10. Woman cooks eggs on gas stove 11. Woman serves eggs on plate at table 12. Woman "calls out" (mouth open) 13. Boy leaves his room 14. Boy smiles as he walks down the hall 15. Boy walks into kitchen 16. Boy eats eggs (slightly smiling) 17. Boy points to door, looks at woman 18. Woman frowns 19. Boy sits, pouting (Up out, head down) 20. Woman and boy return to bedroom via hallway— holding hands— woman leading 21. In room, both looking at floor, woman pointing with 1 hand, holding boy's hand with the other 22. Woman points to peimies on ground 23. Woman w/ straw laundry basket (wearing jacket) 24. Woman w/ laundry basket in doorway of house (going outside) 25a. Woman hanging up dress on line (leaves on ground, jacket on woman, no leaves on trees) 25b. Woman hanging up pants on line (leaves on ground, jacket on woman, no leaves on trees) 25c. Woman hanging up shirt on line (leaves on ground, jacket on woman, no leaves on trees) 26. Boy puts on jacket in bedroom 27. Boy at doorway of house (going outside) 28. Woman inside w/ empty laundry basket, "calling" (hand to mouth, mouth open) 29. Woman in bedroom, hands up, mouth open (as if in surprise), peiuiies scattered on floor 30. Boy going down street (leg raised as if numing) 31. Woman looking outside window 32. Boy further down street (snow on ground) 33. Woman goes outside 34. Boy at playground (next to open gate, slide, swinp in background) 35. Woman "runs" down street 36. Woman goes through gate (2 signs— Park Closed 7 pm- 7 am) 37a.Boy on slide- woman standing with raised hand and index finger ("shaking" her finger at him) 37b.Boy on swing- woman standing with raised hand and index finger ("shaking" her finger at him) 37c.Boy on monkey bars- woman standing with raised hand and index finger ("shaking" her finger at him) 38. Woman and boy go through gate (woman has her hand on the back of his neck— holding his collar?) 39. Woman and boy hugging- on steps to house Appendix E (continued) 40. Woman and boy taking off jackets in hallway 41. Woman and boy sitting on sofa— children's book, magazine and comics on cofiee table) 42a. Woman holding book Stars in the Night, boy sitting next to her (both mouths closed) 42b. Woman holding Ranger Rick magazine, boy sitting next to her (both mouths closed) 42c. Woman holding comics, boy sitting next to her (both mouths closed) 43. Boy putting jacket on. woman holding Tonka toy truck 44. Boy in yard of house playing in dirt with "truck" 61

Human Subiects Commutes ln22 E. \Ucel St Tucson. Aruona a 1520) oZ6-o7Zl

13 March 1996

Allison L. Titconb, Ph.D. Candidate Advisor: Valerie Reyna, Ph.D. Department of Educational Psychology Department of Surgery Life Sciences South, Room 250 PO BOX 210106

RX: rJHiSB MEMORIES AMD rUZZY-TRACE THEORY: MI8IM70RMIMG QIST VERSUS VERBATIM MEMORY

Dear Ms. Titcomb:

We have received documents concerning your above cited project. Regulations published by the U.S. Department of Health and Human Services [45 CFR Part 46.101(b) (2)] exempt this type of research from review by our Committee.

Thank you for informing us of your work. If you have any questions concerning the above, please contact this office.

Sincerely yours,

Hillieun F Denny, M.D. Chairman Human Subjects Committee

WFD:js cc: Departmental/College Review Committee Appendix G: Lure, Misled, Consistency, Gist, Item, Time Interaction % Yes Anova Means F (2, 93) = 4.35, MSE = 0.18, g < 0.016 Means collapsed across judgment and version

TIME

MISINFORMED IMMED IMMED/DELAY DELAY Item 1/Item 2 Item 1/Item 2 Item 1/ Item 2

Original Items GC 92/83 98/90 88/77 GI 78/89 78/88 74/91 DI 84/94 90/54 56/71 DC 56/33 63/59 66/53

Lure Items GC 94/94 93/91 89/95 GI 48/47 25/26 31/60 DI 23/23 21/57 65/62 DC 44/42 60/43 62/56

NOT MISINFORMED

Original Items GC 97/83 97/85 94/85 GI 88/97 85/93 91/80 DI 97/98 85/80 93/74 DC 73/44 66/38 76/50

Lure Items GC 84/89 87/100 93/80 GI 25/20 41/28 38/23 DI 13/16 26/38 23/47 DC 53/30 47/37 60/36 Appendix Hi Lure, Misled, Consistency, Gist, Judgment, Time Interaction % Yes Anova Means for Verbatim Judgments L(2,93) = 3.14, MSE = 0.05152,^ < 0.048 Means collapsed across item and version

TIME Verbatim Judgments

MISLED Immed Immed/Delav Delay

Original Items GC 88 93 86 GI 83 82 75 DC 34 57 48 DI 91 68 57

Lure Items GC 92 90 89 GI 39 21 28 DC 33 48 42 DI 17 37 59

NOT MISLED

Original Items GC 88 89 86 GI 91 90 85 DC 55 48 48 DI 97 81 79

Lure Items GC 81 91 83 GI 12 24 17 DC 30 40 34 DI II 29 22 Appendix I: Lure, Misled, Consistency, Gist, Judgment, Time Interaction % True ANOVA Means for Truth Judgments F (2,93) = 3.14, MSE = 0.05152, g < 0.048 Means collapsed across item and version

TIME Truth Judgments

MISLED Immed Immed/Delav Delay

Original Items GC 88 96 86 GI 84 84 89 DC 55 65 70 DI 88 76 70

Lure Items GC 95 94 95 GI 56 31 64 DC 53 44 62 DI 30 41 68

NOT MISLED

Original Items GC 92 94 92 GI 94 88 87 DC 63 56 78 DI 98 84 88

Lure Items GC 92 96 89 GI 33 46 44 DC 53 44 62 DI 17 36 48 APPENDIX J: signed Confidence ANOVA Interactions with Version and Item

Interaction F df MSE E-<

MGV 50.30 1,92 39.71 0.0000 MGIV 20.18 1,92 37.72 0.0000 MGV 6.40 1,92 40.38 0.0113 MCIV 13.90 1,92 29.45 0.0004 MTV 5.20 2,92 24.45 0.0074 MCGTV 4.21 2,92 27.39 0.0178 MCGIV 14.70 1,92 28.34 0.0003 MCGITV 6.20 2,92 28.34 0.0031 LMCV 12.50 1,92 30.11 0.0008 LMCGV 12.64 1,92 25.26 0.0007 LMGTV 4.61 2,92 36.50 0.0124 LMcrv 15.07 1,92 26.28 0.0003 CTV 4.72 2,92 27.52 0.0113 MGJV 8.03 1.92 9.33 0.0058 LMCJV 10.44 1,92 4.81 0.0018 LCIJTV 4.75 2,92 5.56 0.0110 LMCUV 8.20 1,92 5.31 0.0053 MCGUV 10.78 1,92 5.83 0.0016 LMCGIV 6.52 1,92 22.41 0.0124

Item 12.25 1,92 37.30 0.0008 LI 12.02 1,92 27.40 0.0009 MI 5.98 1,92 35.80 0.0165 CI 41.32 1,92 31.06 0.0000 CGI 6.17 1,92 31.81 0.0149 LCGI 14.60 1,92 30.24 0.0003 MCGI 9.93 1,92 28.34 0.0023 LCIT 6.44 2,92 20.19 0.0025 LMcrr 8.72 2,92 26.28 0.0004 UT 4.73 2,92 4.42 0.0112 LU 14.84 1,92 4.64 0.0003 LMGU 5.46 2,92 21.65 0.0059 LMCGIT 4.62 2,92 22.41 0.0124 APPENDIX K: Contingency Table % Responses

Yes-No Yes Orig Yes Orig No Orig No Orig Judements Yes Lure No Lure Yes Lure No Lure

MISLED

GC 80.3 6.1 10.1 3.5 GI 25.8 54.5 3.5 16.2 DC 29.8 17.2 11.6 41.4 DI 20.7 51.0 17.2 11.1

NOT MISLED

GC 78.3 9.1 7.1 5.6 GI 14.1 74.2 3.5 8.1 DC 28.3 22.2 6.6 42.9 DI 15.7 69.7 5.6 9.1

True-False True Orig True Orig False Orig False Orig Judgments True Lure False Lure True Lure False Lure

MISLED

GC 87.4 2.5 7.6 2.5 GI 46.0 39.9 4.0 10.1 DC 48.5 14.6 12.6 24.2 DI 31.8 46.0 14.6 7.6

NOT MISLED

GC 86.9 6.1 5.6 1.5 GI 35.9 53.5 5.1 5.6 DC 44.9 20.2 8.1 26.8 DI 26.8 63.1 6.1 4.0 67

APPENDIX L: Dependency Results By Time: Verbatim (Yes-No) Judgments

Stimulus Condition Dependency Proportion of Product IYEC Yes to Both of the Marginals MISLED GC Immed HMD .83 .80 0.10 Imm/Delay IND .84 .82 0.10 Delay IND .73 .71 0.15

GI Immed IND .36 .35 0.66 Imm/Delay IND .17 .16 0.64 Delay POS .25 .20 0.03

DC Immed POS .19 .11 0.01 Imm/Delay POS .40 .31 0.00 Delay POS .30 .18 0.00

DI Immed IND .14 .15 0.43 Imm/Delay NEG .17 .26 0.00 Delay IND .31 .34 0.32 NOT MISLED GC Immed IND .75 .72 0.14 Imm/Delay IND .80 .80 0.81 Delay POS .80 .71 0.00

GI Immed IND .09 .11 0.16 Imm/Delay IND .20 .22 0.15 Delay IND .13 .13 0.80

DC Immed POS .22 .16 0.04 Imm/Delay POS .34 .21 0.00 Delay POS .28 .15 0.00

DI Immed NEG .11 .14 0.00 Imm/Delay IND .19 .22 0.16 Delay IND .17 .17 0.91 68

APPENDIX Mr Dependency Results By Time: Truth (True-False) Judgments

Stimulus Condition Dependency Proportion of Product E-< Type True of the to Both Marginals MISLED GC Immed POS .88 .83 0.00 Imm/Delay IND .89 .84 0.49 Delay POS .86 .83 0.02

GI Immed IND .50 .49 0.59 Inun/Delay IND .30 .28 0.24 Delay POS .59 .54 0.01

DC Immed POS .35 .27 0.01 Imm/Delay POS .50 .38 0.00 Delay POS .59 .52 0.00

DI Immed IND .27 .26 0.75 Imm/Delay NEC .23 .33 0.00 Delay IND .47 .47 0.91 NOT MISLED GC Immed IND .86 .85 0.29 Imm/Delay END .89 .89 0.49 Delay POS .86 .84 0.05

GI Immed IND .30 .31 0.46 Imm/Delay IND .41 .42 0.58 Delay IND .36 .36 0.88

DC Inuned POS .41 .32 0.01 Imm/Delay POS .41 .27 0.00 Delay POS .53 .47 0.01

DI Immed IND .19 .20 0.07 Imm/Delay IND .26 .28 0.22 Delay IND .36 .39 0.12 69

APPENDIX N; Mean Signed Confidence Ratings Misled vs. Control Subjects Yes-No (Verbatim) Judgments

Stimulus Type Lure Items Original Items Misled Control Misled Control

GC park close 4.0 3.2 on foot 5.5 5.9 his bedroom 5.3 5.5 bedroom 3.4 4.2 w/toys breakfast 6.8 6.0 egp 5.9 6.6 pouted 5.7 4.9 lower lip 4.6 4.0

GI play inside -3.6 -3.3 pointed to 1.4 4.3 door hiding pennies -2.9 -5.9 looking 4.6 5.5 spring -2.5 -3.5 jackets 5.7 6.3 still angry -0.3 -2.9 hugged 5.2 4.8

DC drawings of 1.1 1.8 dra\vings 3.6 4.6 animals velvet - 1.0 -2.8 gold -0.7 -0.8 flowered mug -0.7 - 1.7 cofiFee mug - 1.1 0.0 clown -0.9 - 1.5 colorful -0.7 - 1.4

DI jeans - I.l -2.4 shirt 2.8 5.3 book 0.8 - 1.5 comic strip 2.5 4.7 slide -2.5 -5.8 swing set 5.0 5.8 coflFeepot -2.3 -3.6 toaster 3.3 4.1

Note: GC = Gist Consistent, GI = Gist Inconsistent, DC = Detail Consistent, DI = Detail Inconsistent. 70

REFERENCES

Bekerian, D. A., & Bowers, J. A. (1983). Eyewitness testimony: Were we misled? Journal of Experimental Psychology: Learning. Memory, and Cognition. 9, 139-145.

Belli, R. F. (1989). Influences of misleading postevent information; Misinformation interference and acceptance. Journal of Experimental Psychology; General. 118. 72-85.

BelU, R. F., Windschitl, P. D., McCarthy, T. T., & Winfrey, S. E. (1992). Detecting memory impairment with a modified test procedure; Manipulating retention interyal with centrally presented event items. Journal of Experimental Psychology: Learning. Memory and Cognition. 18. 356-367.

Bowers, J. M., & Bekerian, D. A. (1984). When will postevent information distort eyewitness testimony? Journal of Applied Psychology. 69. 466-472.

Brainerd, C. J., & Omstein, P. A. (1991). Children's memory for witnessed events; The developmental backdrop. In J. Doris (Eds.), The suggestibility of children's recollections (pp. 10-20). Washington, D.C.: American Psychological Association.

Brainerd, C. J., & Reyna, V. F. (1988). Memory loci of suggestibility development: Comment on Ceci, Ross, and Toglia (1987). Journal of Experimental Psychology; General. 117. 197-200.

Brainerd, C. J., & Reyna, V. F. (1993). Memory independence and memory interference in cognitive development. Psychological Review. 100. 42-67.

Brainerd, C. J., & Reyna, V. F. (1995). Autosuggestibility in memory development. . 2. 331-380.

Cassel, W. S., & Bjorklund, D. F. (in press). Developmental patterns of eyewitness memory and suggestibility: An ecologically based short-term longitudinal study. Law and Human Behavior.

Ceci, S. J., & Bruck, M. (1993). Suggestibility of the child witness; A historical review and synthesis. Psychological Bulletin. 113. 403-439. 71

Chandler, C. C. (1989). Specific retroactive interference in modified recognition tests: Evidence for unknown cause of interference. Journal of Experimental Psychology: Learning. Memory and Cognition. 15. 256-265.

Chandler, C. C. (1991). How memory for an event is influenced by related events: Interference in modified recognition tests. Journal of Experimental Psychology: Learning. Memory and Cognition. 17. 115-125.

Chandler, C. C., & Gargano, G. J. (1995). Item-specific interference caused by cue-dependent forgetting. Memory & Cognition. 23. 701-708.

Clark, H. H., & Clark, E. V. (1977). Psychology and Language. New York: Harcourt Brace Jovanovich.

Harris, R. J., Lee, D. J., Hensley, D. L., & Schoen. L. M. (1988). The effect of cultural script knowledge on memory for stories over time. Discourse Processes. 11. 413- 431.

List, J. A. (1986). Age and schematic differences in the reliability of eyewitness testimony. Developmental Psychology. 22. 50-57.

Loftus, E. F. (1979). Eyewitness testimony. Cambridge: Harvard University Press.

Loftus, E. F. (1991). Made in memory: Distortions in recollection after misleading information. In G. H. Bower (Ed.), The Psychology of learning and motivation: Advances in research and theory (pp. 187-2151 San Diego; Academic Press.

Loftus, E. F. (1993). The reality of repressed memories. American Psychologist. 48,518-537.

Loftus, E. F., Donders, K., Hoffman, H. G., & Schooler, J. W.(1989). Creating new memories that are quickly accessed and confidently held. Memory & Cognition. 17. 607-616.

Loftus, E. F., & Hoffman, H. G. (1989). Misinformation and memory: The creation of new memories. Journal of Experimental Psychology: General. 118. 100-104.

Loftus, E. F., & Ketcham, K. (1994). The mvths of repressed memory. New York: St. Martin's GriflBn. 72

Loftus, E. F., Levidow, B., Duensing, S. (1992). Who remembers best? Individual differences in memory for events that occurred in a science museum. Applied Cognitive Psvcholoev. 6. 93-107.

Loftus, E. F., & Loftus, G. R. (1980). On the permanence of stored information in the human brain. American Psychologist. 35. 409-420.

Loftus, E. F., Miller, D. G., & Bums, H. J. (1978). Semantic integration of verbal information into visual memory. Journal of Experimental Psvcholoev. 4. 19-31.

McCloskey, M., & Zaragoza, M. S. (1985). Misleading postevent information and memory for events: Arguments and evidence against memory impairment hypotheses. Journal of Experimental Psychology: General. 114. 1-16.

Pirolli, P. L., & Mitterer, J. 0. (1984). The effect of leading questions on prior memory: evidence for the coexistence of inconsistent memory traces. Canadian Journal ofPsvchologv. 38. 135-141.

Poole, C. A., & White, L. T. (1991). Effects of question repetition on the eyewitness testimony of children and adults. Developmental Psychology. 27. 975-986.

Poole, C. A., & White, L. T. (1993). Two years later: Effects of question repetition and retention interval on the eyewitness testimony of children and adults. Developmental Psychology. 29. 844-853.

Poole, C. A., & White, L. T. (in press). Tell me again and again: Stability and change in the repeated testimonies of children and adults. In M. Zaragoza (Ed.), Memory, suggestibility, and eyewitness testimony in children and adults. New York: Harper & Row.

Reyna, V. F. (1992). Reasoning, remembering, and their relationship: Social, cognitive, and developmental issues. In M. Howe, C. J. Brainerd, & V. F. Reyna (Eds.) Development of Long-term Retention (pp. 103-132^ New York: Springer-Verlag.

Reyna, V. F., & Brainerd, C. J. (1992). A flizzy-trace theory of reasoning and remembering: Patterns, paradoxes, and parallelism. In A. F. Healy, S. A. Kosslyn, & R. M. Shiffrin (Eds.), From learning processes to cognitive processes: Essays in honor of William K. Estes (DP. 235-260V Hillsdale, NJ: Erlbaum.

Reyna, V. F., & Brainerd, C. J. (1995). Fuzzy-trace theory: An interim synthesis. Learning and Individual Differences. 7. 1-75. 73

Reyna, V. F., & Kieman, B. (1994). The development of gist versus verbatim memory in sentence recognition: Effects of lexical familiarity, semantic content, encoding instructions, and retention interval. Developmental Psychology. 30. 178-191.

Reyna, V. F., & Titcomb, A. L. (in press). Constraints on the suggestibility of eyewitness memory: A fuzzy-trace theory analysis. In D. Payne and F. Conrad (Eds.), A synthesis of basic and applied approaches to memory research. Hillsdale, NJ: Lawrence Erlbaum.

Schwartz, R., & Reisberg, D. (1991). Learning and memory. New York: W. W. Norton.

Smolowe, J. (1994, May 23). Dubious memories. Time. 143. 51.

Terr, L. (1994). Lfnchained memories: True stories of traumatic memories, lost and found. New York: Basic Books.

Titcomb, A. L., & Reyna, V. F. (1992, November). Gist, verbatim memory, and suggestibility: Effects of different types of misinformation. Poster presented at meeting of the Society for Judgment and Decision Making, St. Louis, Mo.

Titcomb, A. L., & Reyna, V. F. (1993, November). Gist and verbatim misinformation: A fuzzv-trace theory analysis. Poster presented at meeting of the Society for Judgment and Decision Making, Washington, D.C.

Titcomb, A. L., & Reyna, V. F. (1994, November). Misinforming |gist versus verbatim memory: A fuzzv-trace theory analysis. Poster presented at meeting of the Society for Judgment and Decision Making, St. Louis, MO.

Titcomb, A. L., & Reyna, V. F. (1995a). Memory interference and misinformation. In F. N. Dempster and C. J. Brainerd (Eds.), New perspectives on interference and inhibition in cognition (pp. 263-294). San Diego, CA: Academic Press.

Titcomb, A. L. & Reyna, V. F. (1995b, November). Memory representation, contradiction and false memory: A fuzzv-trace theory analysis. Poster presented at meeting of the Society for Judgment and Decision Making, Los Angeles, CA.

Toland, K., Hoffman, H., & Lofhis, E. (1991). How suggestion plays tricks with memory. In J. F. Schumaker (Ed.), Human suggestibility (pp. 235-252). New York: Routledge.

Tousignant, J. P., Hall, D., & Loftus, E. F. (1986). Discrepancy detection and vulnerability to misleading postevent information. Memory & Cognition. 14. 329-338. 74

Tversky, B., & Tuchin, M. (1989). A reconciliation of the evidence on eyewitness testimony; Comments on McCloskey and Zaragoza. Journal of Experimental Psychology: General. 118. 86-91.

Zaragoza, M. S., Dahlgren, D., & Muench, J. (1992). The role of memory impairment in children's suggestibility. In M. L. Howe, C. J. Brainerd, & V. F. Reyna (Eds.y Deyelopment of long-term retention (pp. 184-216). New York: Springer-Verlag.

Zaragoza, M. S., & Koshmider, J. W. (1989). Misled subjects may know more than their performance implies. Journal of Experimental Psycholo|gv: Learning. Memory, and Cognition. 15. 246-255.

Zaragoza, M. S., McCloskey, M., & Jamis, M. (1987). Misleading postevent information and recall of the original eyent: Further evidence against the memory impairment hypothesis. Journal of Experimental Psychology: Creneral. 13. 36-44.