Paradoxical effects of testing: Repeated retrieval attempts enhance the likelihood of later accurate and false recall

The testing effect is the finding that taking an initial test enhances the likelihood of later recall. The present report examines the extent to which this benefit of testing comes with a cost: an enhanced likelihood of erroneously recalling incorrect information. Subjects were given short lists of semantic associates (e.g., hill, valley, climb); each list converged upon a related nonpresented word (e.g., mountain). After presentation of some lists, the subjects received no initial test; after others, one initial free recall test; and after others, three successive free recall tests. The probabilities of final free recall (and the probability of reporting vivid recollection of the moment of encoding) of both studied and related, nonstudied words (e.g., mountain) were highest when three initial tests had been taken, intermediate following one initial test, and lowest when no initial test had occurred. The beneficial effects of testing carry the cost of increases in erroneous memory for related information.     

The effects of state anxiety on the suggestibility and accuracy of child eyewitnesses

The negative effect of misleading information on memory is a well‐established fact in eyewitness testimony. However, individual differences have rarely been studied in this context, particularly in children. This paper is one of the first to explore whether objectively measured state anxiety levels have a moderating influence on suggestibility. A group of 83 9–10‐year‐old schoolchildren took part in the experiment. They were tested on their recall of details surrounding a minor car accident shown on video. No effects of state anxiety on accuracy were found. Clear misleading information effects were found, however (p < 0.001). Furthermore, both analysis of covariance and Pearson's correlation coefficients showed that higher anxiety levels were associated with a reduction in the number of misled responses given by the misinformed participants relative to low‐anxious participants (p < 0.05). The theoretical relevance of these findings are discussed in the light of processing efficiency theory and the ‘affect as information’ hypothesis. The practical implications for children as witnesses are also considered. Copyright © 2002 John Wiley & Sons, Ltd.     

Effects of suggestibility and hypnosis on accurate and distorted retrieval from memory

Research investigated the hypothesis inferred from the theorizing of Loftus that suggestibility is related to the tendency to incorporate incorrect information into memory when this information has been subtly introduced after the to-be-remembered events have occurred. Specifically, it was predicted that if level of suggestibility is theoretically relevant to subjects' acceptance of misleading information, then more subjects who are highly hypnotically suggestible than those with a low level of hypnotic suggestibility will incorporate the incorrect information into memory. Hypnotic as compared with waking instruction should enhance this distortion effect by providing a context of testing in which subjects are readily prone to respond positively to suggestions. Eight independent groups of 12 subjects were tested. Separate groups of subjects of high and low suggestibility were presented with misleading or neutral information about a wallet-snatching incident and tested for memory under either waking or hypnotic instruction. Analysis of subjects' memory distortions indicated that suggestibility plays a somewhat different role than has been argued previously. The magnitude of distortion that was observed varied according to the stimulus features that were studied, but hypnotic suggestibility was not associated with the distortion effect. Despite the fact that hypnosis did not enhance recall in any way, subjects were frequently confident that distorted memories recovered under hypnosis were accurate.     

The effects of the age of eyewitnesses on the accuracy and suggestibility of their testimony

Previous studies have compared the performance of young adult eyewitnesses with that of children or elderly eyewitnesses, but few studies have allowed direct comparison of the performance of all three age groups. The accuracy and suggestibility of accounts of a video recording of a kidnapping were investigated using an experimental eyewitness paradigm. Subjects were drawn from three age groups: children (aged 7–9 years); young adults (aged 16–18 years) and elderly subjects (aged 60–85 years). Subjects' accuracy in answering non-misleading questions and their susceptibility to misleading information was measured. Both the elderly and child subjects gave fewer correct answers and more incorrect answers to non-misleading questions than did young adults. The elderly subjects gave fewer correct responses but also fewer incorrect responses to non-misleading questions than did child subjects. Children were more suggestible than either elderly or young adults. No significant difference was found in the suggestibility of elderly and young adults. Contrary to the trace strength hypothesis no relationship was found between accuracy of recall and suggestibility. © 1997 John Wiley & Sons, Ltd.     

The dark side of testing memory: repeated retrieval can enhance eyewitness suggestibility

Eyewitnesses typically recount their experiences many times before trial. Such repeated retrieval can enhance memory retention of the witnessed event. However, recent studies (e.g., Chan, Thomas, & Bulevich, 2009) have found that initial retrieval can exacerbate eyewitness suggestibility to later misleading information--a finding termed retrieval-enhanced suggestibility (RES). Here we examined the influence of multiple retrieval attempts on eyewitness suggestibility to subsequent misinformation. In four experiments, we systematically varied the number of initial tests taken (between zero and six), the delay between initial testing and misinformation exposure (~30 min or 1 week), and whether initial testing was manipulated between- or within-subjects. University undergraduate students were used as participants. Overall, we found that eyewitness suggestibility increased as the number of initial tests increased, but this RES effect was qualified by the delay and by whether initial testing occurred in a within- or between-subjects manner. Specifically, the within-subjects RES effect was smaller than the between-subjects RES effect, possibly because of the influence of retrieval-induced forgetting/facilitation (Chan, 2009) when initial testing was manipulated within subjects. Moreover, consistent with the testing effect literature (Roediger & Karpicke, 2006), the benefits of repeated testing on later memory were stronger after a 1-week delay than after a 30-min delay, thus reducing the negative impact of RES in long-term situations. These findings suggest that conditions that are likely to occur in criminal investigations can either increase (repeated testing) or reduce (delay) the influence of RES, thus further demonstrating the complex relationship between eyewitness memory and repeated retrieval.     

Testing enhances both encoding and retrieval for both tested and untested items

In forward testing effects, taking a test enhances memory for subsequently studied material. These effects have been observed for previously studied and tested items, a potentially item-specific testing effect, and newly studied untested items, a purely generalized testing effect. We directly compared item-specific and generalized forward testing effects using procedures to separate testing benefits due to encoding versus retrieval. Participants studied two lists of Swahili–English word pairs, with the second study list containing “new” pairs intermixed with the previously studied “old” pairs. Participants completed a review phase in which they took a cued-recall test on only the “old” pairs or restudied them. In Experiments 1a, 1b, and 2, the review phase was given either before or after the second study list. Testing benefited memory to the same degree for both “new” and “old” pairs, suggesting that there were no pair-specific benefits of testing. The larger benefit from testing when review was given before rather than after the second study list suggests that the memory enhancement was due to both testing-enhanced encoding and testing-enhanced retrieval. To better equate generalized testing effects for “new” and “old” pairs, Experiment 3 intermixed them in the review phase. A statistically significant pair-specific testing effect for “old” items was now observed. Overall, these results show that forward testing effects are due to both testing-enhanced encoding and retrieval effects and that direct, pair-specific forward testing benefits are considerably smaller than indirect, generalized forward testing benefits.     

Simultaneous and sequential lineups: decision processes of accurate and inaccurate eyewitnesses

The cognitive processes and decision‐making strategies of eyewitnesses were tested for their predictive qualities in determining the accuracy of identifications from lineups. The sequential lineup presentation was compared with the traditionally employed simultaneous lineup under culprit (target) present and culprit absent conditions. Consistent with previous research the sequential presentation resulted in an equivalent number of correct identifications compared to the simultaneous lineup but reduced false identification rates. Although sequential lineups were found to be associated with the use of absolute strategies, those shown a simultaneous lineup reported the use of both relative and absolute strategies. Accurate identifications and rejections were found to be associated with the use of absolute strategies, irrespective of lineup presentation or presence of target. Also accurate identifications, at least with a sequential lineup, were generally made faster than inaccurate identifications. These results are compared to previous studies with respect to the effect that mode of processing (relative versus absolute judgements) has on a witness's decision making and identification accuracy. Copyright © 2001 John Wiley & Sons, Ltd.     

Enhancing retention through reconsolidation: negative emotional arousal following retrieval enhances later recall

When information is retrieved from memory, it enters a labile state rendering it amenable to change. This process of reconsolidation may explain, in part, the benefits that are observed in later retention following retrieval of information on an initial test. We examined whether the benefits of retrieval could be modulated by an emotional event occurring after retrieval. Participants studied Swahili-English vocabulary pairs. On a subsequent cued-recall test, each retrieval was followed by a blank screen, a neutral picture, or a picture inducing negative affect. Performance on a final cued-recall test was best for items whose initial retrieval was followed by negative pictures. This outcome occurred when a negative picture was presented immediately after (Experiment 1) or 2 s after (Experiment 2) successful retrieval, but not when it was presented after restudy of the vocabulary pair (Experiment 3). Postretrieval reconsolidation via emotional processing may enhance the usual positive effects of retrieval.     

Altering memory through recall: The effects of cue-guided retrieval processing

Three experiments were designed to investigate the influence of initial recall on memory by assessing delayed recall after different immediate cued-recall tests. In all experiments, subjects performed semantic and phonemic encoding tasks on a word list. The subjects then received a cued-recall test that cued the target using the same word as the context word in the encoding task, a test that cued the target with a word from the same level at which the target was encoded, a test that cued the target with a cue from a different level at which the target was encoded, or no immediate-recall test. One day later, the subjects performed a final cued-recall test in which the type of cue (semantic or phonemic) was varied. Consistently, delayed recall was facilitated primarily when the cue on the immediate test was from the same level as the cue on the delayed test. This pattern of facilitation suggests that immediate cued-recall produces an elaboration of an existing memory representation that is closely tied to the type of cue used on the immediate test.     

Confidence judgments in children's and adults' event recall and suggestibility

The present work investigated the role of children's and adults' metacognitive monitoring and control processes for unbiased event recall tasks and for suggestibility. Three studies were conducted in which children and adults indicated their degree of confidence that their answers were correct after (Study 1) and before (Study 2) answering either unbiased or misleading questions or (Study 3) forced-choice recognition questions. There was a strong tendency for overestimation of confidence regardless of age and question format. However, children did not lack the principal metacognitive competencies when these questions were asked in a neutral interview. Under misleading questioning, in contrast, children's monitoring skills were seriously impaired. Within each age group, better metacognitive differentiation was positively associated with recall accuracy in the suggestive interview.     

Event frequency and children's suggestibility: a study of cued recall responses

Research examining the effect of repeated experience on children's suggestibility for particular kinds of information has produced differing results. In one study, responses to recognition questions revealed heightened suggestibility for variable details in children who repeatedly experienced an event compared to children who experienced an event once. In other studies, no such effect was found with cued recall. In this study, 4–5‐year old children engaged in one or four play sessions. Children were later given a biasing interview wherein half of the details were incorrectly represented. Children were then given a final memory test using free and cued recall prompts that was preceded by one of three instruction types: no special instructions, moderate instructions, or opposition instructions. Children in the repeated‐event condition were more suggestible than those in the single‐event condition, regardless of instructions. No significant differences in suggestibility were found across instructions conditions. Copyright © 2004 John Wiley & Sons, Ltd.     

Guided visualization and suggestibility: effect of perceived authority on recall of autobiographical memories.

The purpose of this study was to investigate a possible component process in the formation of childhood pseudomemories in adults. Participants recounted a childhood event, the details of which came from hearing others tell it (a know event) rather than from their personal experience (a remember event). Then participants were placed in 1 of 4 possible conditions: They completed a guided visualization task led by an expert, a guided visualization task led by a nonexpert, a visual search task, or a verbal list-learning task. For the guided visualization task, participants listened to a middle-aged man on audiotape, who asked them to imagine details about their know event. Half believed the person on the tape was a well-known and esteemed psychologist (an expert), and half were led to believe that he was someone who had gone back to school to study communications (a nonexpert). As predicted, guided visualization led participants to rate their know event closer to a remember event. Planned comparisons demonstrated that the effect was significantly greater for the expert versus nonexpert conditions. Results were applied to the process of false memory formation and the use of visualization procedures in psychotherapy.     

Individual and developmental differences in eyewitness recall and suggestibility in children with intellectual disabilities

This study examined two key issues: (1) whether there were developmental improvements in eyewitness memory performance for children with intellectual disabilities (ID); and (2) whether standardised measures of cognitive ability and suggestibility would relate to eyewitness recall and suggestibility. Children with ID and age‐matched controls (ages 8/9 and 12 years) watched a video of a crime and were asked a range of open‐ended and specific questions about the event in a subsequent interview. Free recall increased between the two age levels for children with and without ID, but at a faster rate for those without ID. For other question types, differences in performance between children with and without ID were far more marked than age differences. Standardised measures of interrogative suggestibility (Gudjonsson Suggestibility Scale, GSS), verbal IQ, non‐verbal IQ, mental age and speed of information processing were related to eyewitness performance. In particular, higher eyewitness recall scores (free recall, non‐leading specific questions) were related to higher scores on the standardised GSS free recall measure; and higher eyewitness suggestibility scores were related to higher scores on the standardised GSS suggestibility measures. Mental age was a better predictor of performance on a range of eyewitness memory question types than verbal or non‐verbal IQ; and speed of information processing showed some relationships with eyewitness performance. Copyright © 2006 John Wiley & Sons, Ltd.     

Source-constrained recall: front-end and back-end control of retrieval quality

Research on the strategic regulation of memory accuracy has focused primarily on monitoring and control processes used to edit out incorrect information after it is retrieved (back-end control). Recent studies, however, suggest that rememberers also enhance accuracy by preventing the retrieval of incorrect information in the first place (front-end control). The present study put forward and examined a mechanism called source-constrained recall (cf. Jacoby, Shimizu, Velanova, & Rhodes, 2005) by which rememberers process and use recall cues in qualitatively different ways, depending on the manner of original encoding. Results of 2 experiments in which information about source encoding depth was made available at test showed that when possible, participants constrained recall to the solicited targets by reinstating the original encoding operations on the recall cues. This reinstatement improved the quality of the information that came to mind, which, together with improved postretrieval monitoring, enhanced actual recall performance.     

Long-term effects of testing on the recall of nontested materials

The isolation paradigm is a staple in the study of distinctiveness and memory. Isolated items are better remembered than non-isolated controls, and the standard interpretation of this effect is that subjective experience of salience recruits extraordinary processing to the isolated item. This interpretation is at odds with data showing an isolation effect when the isolate is not perceived as salient (e.g., von Restorff, 1933). All available research on the early isolation effect has tested memory after a relatively short retention interval. Perhaps the effect of salience on memory in the isolation paradigm would be revealed following longer retention intervals. The experiment reported here examines the effect of isolation following a 48-hour retention interval when the isolate evokes an experience of salience compared to when the isolate does not evoke that reaction. The isolation effect was substantial after the 48-hour delay but equally so for early and late isolation. Salience appears to have nothing to do with the memory processes even at the longer retention intervals.     

The effects of free recall testing on subsequent source memory

The testing effect is the finding that prior retrieval of information from memory will result in better subsequent memory for that material. One explanation for these effects is that initial free recall testing increases the recollective details for tested information, which then becomes more available during a subsequent test phase. In three experiments we explored this hypothesis using a source-monitoring test phase after the initial free recall tests. We discovered that memory is differentially enhanced for certain recollective details depending on the nature of the free recall task. Thus further research needs to be conducted to specify how different kinds of memorial details are enhanced by free recall testing.     

Distinctiveness effects in recall:

Unusual information is generally recalled better than common information (the distinctiveness effect). Differential processing accounts propose that the effect occurs because unusual material elicits encoding processes that are different from those elicited by common material, and strong versions of these accounts predict distinctiveness effects in between-list as well as within-list designs. Experiment 1 employed a between-list design and manipulated presentation rate. Contrary to differential processing predictions, no distinctiveness effect emerged, nor did recall patterns for atypical versus common sentences differ as a function of presentation rate. Experiment 2 further tested differential processing accounts as well as representation accounts via a within-list manipulation and conditions that included experimenter-provided elaborations. Distinctiveness effects emerged in all conditions and, contrary to differential processing predictions, the pattern of recall in the elaborated conditions did not differ from that in the unelaborated conditions. Taken together, the results of this study lend more support to a representation view that suggests mechanisms related to the representation and subsequent retrievability of elements in the memory record play a major role in the distinctiveness effect.     

Semantic relations between encoding and retrieval in cued recall

Two experiments were conducted to test the encoding-specificity hypothesis. In Experiment 1, subjects were presented pairs of coordinates to study, followed by a cued-recall test. Semantically weak and strong coordinates (defined by the degree of featural overlap shared with the pairs) served as encoding and retrieval cues. The semantic strength of the retrieval cue proved to be the most important factor in recall, whereas the re-presentation of an unrelated cue previously seen during encoding led to no facilitation of recall. In Experiment 2, three noun coordinates of varied semantic interrelatedness were presented for study and then were cued for recall by novel semantically strong or weak coordinates. Maximum recall was achieved when a strong encoding condition was matched with a strong retrieval cue. Implications of these findings for an encoding-specificity hypothesis were discussed.