Neuropsychologia 44 (2006) 18661877

When true memory availability promotes false memory: Evidence from confabulating patients
Elisa Ciaramelli a,b, , Simona Ghetti c,d , Massimo Frattarelli e , Elisabetta L` adavas a,b
Dipartimento di Psicologia, Universit` a di Bologna, Bologna, Italy Centro Studi e Ricerche di Neuroscienze Cognitive, Cesena, Italy c Department of Psychology, University of California, Davis, United States d Center for Mind and Brain, University of California, Davis, United States e Ospedale Bufalini, Cesena, Italy
b a

Received 22 February 2005; received in revised form 21 February 2006; accepted 22 February 2006 Available online 31 March 2006

Abstract We explored the extent to which confabulators are susceptible to false recall and false recognition, and whether false recognition is reduced when memory for studied items is experimentally enhanced. Five confabulating patients, nine non-confabulating amnesics including patients with (F amnesics) and without frontal-lobe dysfunction (NF amnesics) and 14 control subjects underwent the DRM paradigm [Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory and Cognition, 21, 803814.] in two experimental conditions. In both conditions participants studied eight lists of semantic associates, and free recall was tested after the presentation of each list. In the Standard condition recognition was tested after the presentation of all the lists, whereas in the Proximal condition patients were administered a six-item recognition task after the presentation of each list. Participants also provided remember or know judgements, and described the content of their recollections. All groups of patients recalled a lower proportion of targets and critical lures than did control subjects, but confabulators recalled more words unrelated to the studied lists than did NF amnesics and controls. All groups of participants improved true recognition across conditions. However, whereas normal controls suppressed false recognition to critical lures in the Proximal compared to the Standard condition, and non-confabulating amnesics showed comparable gist-based false recognition, confabulators showed increased levels of false recognition to critical lures across conditions. Furthermore, NF amnesics signicantly reduced false recognition to unrelated lures in the Proximal compared to the Standard condition, whereas confabulators were unable to suppress false recognition to unrelated lures across conditions. Analysis of the phenomenological experience showed that, unlike non-confabulating amnesics, confabulators characterized true and false memories with irrelevant information related to test items. Results are interpreted in light of confabulators monitoring decits. 2006 Elsevier Ltd. All rights reserved.
Keywords: False memory; Confabulation; Frontal-lobe; Strategic retrieval

1. Introduction Confabulation is a symptom (Burgess & Shallice, 1996; Dalla Barba, 1993a,b, 1995; DeLuca, 1993; Johnson & Raye, 1998; Korsakoff, 1889; Moscovitch, 1989, 1995; Schnider, 2003; Talland, 1961) that involves the production of erroneous memories, either false in themselves or resulting from true
Corresponding author at: Dipartimento di Psicologia, Universit` a di Bologna, Viale Berti Pichat 5, 40127 Bologna, Italy. Tel.: +39 051 2091347; fax: +39 051 243086. E-mail address: ciaramelli@psice.unibo.it (E. Ciaramelli).

memories misplaced in context and inappropriately retrieved or interpreted (Kopelman, Guinan, & Lewis, 1995). This symptom is commonly found in amnesic patients who survived anterior communicating artery (AcoA) aneurysm (Alexander & Freedman, 1984; DeLuca, 1993; Vilkki, 1985), often in association with damage to the orbitofrontal (Schnider, 2003) and medial frontal regions (Gilboa & Moscovitch, 2002). Some researchers found it useful to distinguish between provoked confabulations, which can be elicited by questions, and spontaneous confabulations, which patients produce without a recognizable motive (Berlyne, 1972; Dalla Barba, 1993b; Kopelman, 1987). Confabulators may occasionally act upon

0028-3932/$ see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.neuropsychologia.2006.02.008

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

1867

their confabulation, reecting their genuine belief in the false memory (Schnider, Gutbrod, Hess, & Schroth, 1996). Some authors attribute confabulation to a failure of memory control processes (Burgess & Shallice, 1996; Schacter, Norman, & Koutstaal, 1998). For example, Moscovitch argued that confabulation results from a failure in the strategic retrieval processes involved in focusing the memory search and in monitoring the veracity of retrieved memories. According to Moscovitch, a monitoring decit is necessary for confabulation to occur (Moscovitch & Melo, 1997). Other authors have emphasized that confabulation reects the inability to distinguish a memory derived from direct experience from one generated by imaginative acts (i.e. reality monitoring; Johnson, Hashtroudi, & Lindsay, 1993; Johnson & Raye, 1998). Reality monitoring decits are seen as a factor contributing to confabulation, rather than its cause, since they have been found in confabulating (Johnson, OConnor, & Cantor, 1997; Schnider, von Daniken, & Gutbrod, 1996), but also in non-confabulating patients (Janowsky, Shimamura, & Squire, 1989) and normal controls (Reyna & Brainerd, 1995; Roediger & McDermott, 1995; Payne, Elie, Blackwell, & Neuschatz, 1996). It is worth noting, however, that a failure in reality monitoring might result from several factors, including defective encoding operations or defective monitoring processes at retrieval (Johnson, 1991). Thus, there is the possibility that group-specic mechanisms lead different populations of patients to poor reality monitoring abilities. The aim of the present study was to explore the extent to which confabulators are susceptible to false recall and false recognition in two classes of lures, that is, lures that are semantically related and lures that are semantically unrelated to the studied material. Moreover, we investigated whether this susceptibility decreases when memory for studied words is experimentally enhanced. To this aim, we selected a paradigm originally developed by Deese (1959), and modied by Roediger and McDermott (1995). In the Deese, Roediger and McDermott (DRM) paradigm, individuals study lists of semantic associates (e.g. mug, handle, coffee, etc.) converging on a non-studied but semantically associated word (i.e. critical lure; e.g. cup), and are then required to recall the studied words and to recognize them among new words. The DRM recognition test includes both new words similar to the studied words in meaning (i.e. critical lures; e.g. cup) and new words unrelated to the studied words (i.e. unrelated lures; e.g. lion). Previous research has shown that, in the DRM paradigm, normal subjects are as likely to recall and recognize the critical lures as words that actually appeared on the studied list (Roediger & McDermott, 1995). False recognition of critical lures has been attributed to subjects binding together studied items at study, thereby encoding the gist of the lists (i.e. critical lure) along with the studied items (Reyna & Brainerd, 1995). Several authors have demonstrated that amnesic patients with medial temporal lobe (MTL) lesions exhibit lower levels of true recognition and false recognition to critical lures, but higher levels of false recognition to unrelated lures compared to normal controls (Melo, Winocur, & Moscovitch, 1999; Schacter, Verfaellie, &

Pradere, 1996; Schacter, Verfaellie, Anes, & Racine, 1998). In contrast, non-amnesic patients with frontal-lobe damage show higher levels of false recognition to critical lures than do controls (Budson et al., 2002; Melo et al., 1999). Of relevance for the present study, Melo et al. (1999) administered the DRM paradigm to four patients with frontal-lobe pathology in addition to amnesia. Three of these patients were confabulators. Similar to the MTL amnesics, frontal amnesic patients evinced extremely low levels of true recognition and false recognition to critical lures, but high levels of false recognition to unrelated lures. It has been proposed that relatively high levels of false recognition to unrelated lures and low levels of false recognition to critical lures might be due to problems in understanding the semantic theme of the studied lists at study (i.e. gist extraction), or to poor gist memory at the time of retrieval (Melo et al., 1999). In contrast, high levels of false recognition to critical lures occur when individuals are able to extract and remember the gist of the studied lists, but unable to distinguish studied words from gist-consistent, but unstudied, lures. This failure may be due to poor memory for the studied words (i.e. item-specic memory; Schacter et al., 1998) or to defective monitoring processes (Melo et al., 1999). Indeed, previous research showed that normal subjects signicantly suppress false recognition to critical lures when item-specic memory is experimentally enhanced (Budson et al., 2002; Schacter et al., 1998). In the present study, we introduced a manipulation designed to enhance memory for studied words and, as a consequence, for the gist of the lists, assuming that individuals are able to extract it. This manipulation would allow us to verify whether confabulators may benet from increased true memory to suppress false recognition to critical and unrelated lures. At variance with the typical DRM procedure (Standard condition), in which recognition is tested after the presentation of all the lists, in the Proximal condition patients were administered a six-item recognition test after the presentation of each list. Each recognition test included studied items, the critical lure of that list and unrelated lures. We administered the DRM to confabulators, nonconfabulating amnesics without frontal-lobe dysfunction (NF amnesics), non-confabulating amnesics with frontal-lobe dysfunction (F amnesics), and normal controls, both in the Standard and in the Proximal condition. Moreover, to begin to explore the subjective experience of true and false memories in confabulators, patients were required to determine whether endorsed items were remembered or known (Tulving, 1985) and to describe their recollections by reporting exactly what they remembered about the learning episode (see also Norman & Schacter, 1997). Based on previous evidence (Melo et al., 1999), we expected that in the Standard condition both confabulating and non-confabulating amnesics would show lower levels of true recognition and false recognition to critical lures, but higher levels of false recognition to unrelated lures, compared to normal controls. Compared to the Standard condition, in the Proximal condition true recognition was expected to increase in all groups. In contrast, the effect of the Proximal condition

1868

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

on false recognition was expected to vary across groups, due to different abilities to use increased item and gist memory in order to reject critical lures and unrelated lures. In line with other studies (Budson et al., 2002; Schacter et al., 1998), we expected normal subjects would suppress false recognition of critical lures in the Proximal compared to the Standard condition. In contrast, we predicted that monitoring decits would prevent confabulators from using increased true memory to counteract gist-based false memory, thereby showing increased false recognition of critical lures across conditions (see also Moscovitch & Melo, 1997). On the other hand, we expected confabulators would be unable to benet from enhanced gist-memory to reduce false recognition of unrelated lures across conditions, consistent with the evidence that these patients have difculties in suppressing material irrelevant to the task (Schnider, 2003). Accordingly, we predicted that confabulators recollections would mainly involve irrelevant information about test items. Different from confabulators, NF amnesics were expected to effectively use increased item and gist memory to reduce false recognition of both critical and unrelated lures across conditions. However, the impoverished quality of true memories in these patients might prevent them from signicantly suppressing false recognition of critical lures as controls do. Accordingly, we predicted that in these patients true memories would be extremely poor of contextual details. Finally, F amnesics were expected to be as able as NF amnesics to use increased gist memory to reduce false recognition to unrelated lures across conditions, due to preserved abilities to suppress information irrelevant to the task (Schnider, 2003). However, as frontal-lobe pathology may increase the susceptibility to gist-based false recognition (Budson et al., 2002; Melo et al., 1999), F amnesics were not expected to be as able as

NF amnesics to reduce false recognition of critical lures across conditions.

2. Method 2.1. Subjects
Table 1 shows, for each subject, the aetiology and the site of brain damage as determined by MRI or CT scans, and demographic data (see also Fig. 1). Participants in this study included 14 individuals with brain damage and 14 individuals with no history of brain insult. Of the patients, ve were patients who suffered from AcoA aneurysm, and exhibited confabulation. Patients were classied as confabulators depending on (1) their behaviour in real life and (2) their performance on the Confabulation Battery (Dalla Barba, 1993b; see below). With regard to (1), all patients in the present study showed spontaneous confabulations (e.g. in the absence of any verbal probe by the examiner, patient M1 narrated that the examiner had participated in his daughters wedding; patient GB claimed that an orthopaedic doctor was waiting for him at the second oor of the hospital). Unstructured interviews with patients relatives conrmed the presence of spontaneous confabulations (e.g. patient GV narrated to his wife that in the morning he had an argument with a work associate, even though he has been unemployed for several months). Patients typically appeared very convinced about the truthfulness of their claims and gave the impression of vividly remembering their false memories. Sporadically, patients also acted according to their confabulations (e.g. patient ML asked the examiner to interrupt the session because she needed to go home to prepare dinner for some friends; patient GV often searched for his work uniform during the session). Confabulators had a mean age of 55 years (range 4565) and a mean of 9 years of education. Mean time since lesion was 34.4 months. The remaining nine patients were non-confabulating amnesics with various aetiologies. Five of them had damage to medial temporal, diencephalic, or posterior regions (NF amnesics). These patients exhibited amnesia but preserved executive functions (see Table 2). NF amnesics had a mean age of 54.6 years (range 3475) and a mean of 9.8 years of education. Mean time since lesion was 20.2 months. The other four amnesics had lesions in frontal-lobe regions (F amnesics). These patients were comparable to confabulators with respect to memory and executive decits (see Table 2). F amnesics had a mean age of 52 years (range 4059) and a mean of 9.7 years of education. Mean time since lesion

Table 1 Patients demographic data and lesion location Sex Confabulators GV M1 GB LG ML Frontal amnesics M2 M3 DP AT Age Edu Aetiology Time since lesion (months) 15 13 26 96 22 60 24 23 24 5 48 24 12 12 Lesion site

M M M M F M F M M

45 65 48 56 61 53 56 59 40 61 45 34 75 58 53

8 8 8 13 8 8 13 5 13 5 8 13 10 13 9.8

ACoA Aneurysm clipped ACoA Aneurysm clipped ACoA Aneurysm clipped ACoA Aneurysm clipped ACoA Aneurysm clipped Anoxia ACoA Aneurysm clipped Brain tumour Traumatic Brain Injury Anoxia Brain tumour PCA Aneurysm clipped Left thalamic stroke Left internal capsule stroke

Right orbitofrontal and ventromedial frontal lesion Left ventromedial frontal lesion Left orbitofrontal and ventromedial frontal lesion Right ventromedial lesion. A smaller lesion is also present on the left medial prefrontal wall. Left ventromedial lesion Right dorsolateral frontal, and temporoparietal Bilateral orbitofrontal Left frontal Left dorsolateral frontal and temporal Bilateral hippocampal and parahippocampal Left parietal and occipital Left parietal + mesial surface of left occipital Left parainsular, internal capsule and thalamus Left internal capsule and omolateral midbrain

Non-frontal amnesics BL F SM F AD M AP M RB M Controls mean

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

1869

Fig. 1. Location of brain damage in confabulators, according to the standard template provided by Damasio and Damasio (1989). was 32.7 months. Absence of confabulation in NF amnesics and F amnesics was determined through clinical assessment and unstructured interviews with the patients relatives. For F amnesics, scores on the Confabulation Battery were also available (see Fig. 2), but not for NF amnesics. Patients gave their informed consent to participate in the study according to the Declaration of Helsinki (BMJ1991;302:1194), that was approved by the local Ethical Committee. The control group consisted of 14 healthy individuals, matched to patients on the basis of age and education. Exclusion criteria were clinically signicant depression, alcohol or drug abuse, epilepsy, or other known neurological condition. Two ANOVAs showed that confabulators, NF amnesics, F amnesics and normal controls were not signicantly different with respect to age (p = 0.88) and level of education (p = 0.98). Moreover an ANOVA showed no signicant differences among the three groups of patients with respect to time since lesion (p = 0.64).

2.2. Neuropsychological evaluation

Table 2 shows performance on the psychometric tests for confabulators, NF amnesics, and F amnesics. All patients showed well-preserved intellectual skills, as indicated by the scores obtained on the Mini Mental State Examination (Folstein, Folstein, & McHugh, 1975), the Verbal Judgement Task (see Spinnler & Tognoni, 1987 for normative data) and the Raven Standard Progressive Matrices. In contrast, all patients exhibited long-term memory decits. On the Wechsler Memory Scale (Wechsler, 1987), all groups of patients obtained a mean general memory index (MQ) that was one standard deviation below average performance. In the BuschkeFuld Test (Buschke & Fuld, 1974), that is a standardized selective-reminding list learning task involving free recall, all groups of patients exhibited a highly pathological Consistent Long Term Retrieval score

1870

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

Table 2 Neuropsychological test scores for patients (in percentile value) Confabulators GV MMSE Verbal judgement task Standard raven matrices Wechsler memory scale BuschkeFuld test CLTR Prose recall test Tower of London test Total move score Rule violation score Wisconsin card sorting test Number of categories (raw score) Perseverative responses Verbal uency (Letters) Digit span Missing data are indicated with a (). 26 50 50 94 0 1 1 1 5 30 1 35 M1 23 50 50 92 8 2 1 1 1 18 10 25 GB 26 35 5 93 4 2 16 1 2 1 0 18 LG 24 35 19 82 0 0 1 1 0 1 8 35 ML 24 50 50 84 1 1 1 2 4 9 20 50 Mean 24.6 44 34 89 2.6 1.2 4 1.2 2.4 11.8 7.8 32.6 Frontal amnesics M2 27 2 2 97 2 3 1 1 1 1 0 25 M3 24 34 50 80 2 0 1 1 2 30 1 38 DP 24 35 50 88 0 0 35 35 5 10 0 50 AT 30 50 45 82 1 25 1 2 5 10 3 25 Mean 26.2 30.5 36.7 87 1.25 7 9.5 9.7 3.2 10.25 1 34.5 Non-frontal amnesics AD 30 50 35 92 0 18 5 35 50 50 SM 30 50 90 1 10 50 50 6 50 50 50 AP 26 50 50 87 5 0 35 35 RB 28 35 50 92 0 1 35 50 5 50 35 19 BL 26 19 82 0 5 10 50 4 10 34 50 Mean 28 46 38.5 89 1.5 6.8 25 46.5 5 36 41 41

(CLTR, i.e. the number of recalled words without further reminding until the last trial), with a mean percentile value below the 5th percentile (Spinnler & Tognoni, 1987). Patients were also administered a prose-passage recall task that has been standardized for the Italian population by Spinnler and Tognoni (1987). Specically, patients were read a short story and required to immediately recall as many details about it as they can. Immediately after the recall task, the passage was newly read to patients and, after 10 min of distracting activities, patients were asked to recall the passage again. A recall score accounting for both the immediate and the delayed performance was calculated based on Spinnler and Tognoni (1987). All groups of patients showed below-average performance. In contrast, short-term memory, as assessed by digit span, was normal. As far as executive functioning is concerned, in the Tower of London Test (Culbertson & Zillmer, 2001) confabulators and F amnesics presented a pathological Total Move Score (TMS; i.e. the number of moves executed by the subject minus the minimum number of solution moves), and also demonstrated many violations of the experimental rules, whereas NF amnesics showed normal performance. In the Wisconsin Card Sorting Test (Lezak, 1995), confabulators and F amnesics obtained a below-average number of categories, and also produced a high number of perseverative responses. In contrast, NF amnesics

obtained a normal performance. On the Letter Fluency Task (Lezak, 1995), confabulators and F amnesics obtained a below average performance, whereas NF amnesics obtained normal scores. 2.2.1. Assessment of confabulations To assess the nature of patients confabulations, the Confabulation Battery by Dalla Barba (1993b) was administered to confabulators and F amnesics. This test consists of six subsections of questions concerning personal semantic memory (PS), personal episodic memory (PE), orientation in time and space (O), general semantic memory (GS), I dont know semantic (DKS) and I dont know episodic (DKE) memory. The DKE and DKS sections contain questions for which I dont know is an appropriate response. When necessary, the questions were adjusted to the patients cultural background (see also Box, Laing, & Kopelman, 1999), e.g. Who is Enrico Berlinguer? (a popular leader, now deceased, of an Italian political party) replaced Who is Valery Giscard DEstaing?. In order to corroborate patients memory statements, interviews were conducted with their relatives, and Dalla Barbas criteria (1993b) for classifying answers as confabulations were applied. Fig. 2 presents the percentage of confabulatory and correct statements made by confabulators and F amnesics across all sections. The two groups gave a similar percentage of correct responses in all the sections investigated (p > 0.10 in all comparisons). However, compared to F amnesics, confabulators produced a higher number of confabulations in response to questions about personal episodes (50% versus 0%, p < 0.05), spatio-temporal orientation (30% versus 3%, p < 0.05), and general semantic (13% versus 3%, p < 0.05). In contrast, confabulators only minimally confabulated in response to personal semantic questions (2%). Notably, confabulators did not produce signicantly more confabulatory statements than F amnesics in the I dont know sections (DKS: 10% versus 2%; DKE: 4% versus 3%; p > 0.11 in both comparisons). This performance has not been observed in all confabulating patients, as some freely confabulate in these sections (Dalla Barba, 1995; Fotopoulou, Solms, & Turnbull, 2004; Kopelman, Ng, & Van der Broke, 1997), but resembles that of other confabulating patients, for example those described by Dalla Barba (1993a,b) and Box et al. (1999).

Fig. 2. Mean percentage of confabulatory and correct responses produced by confabulators (C) and frontal amnesics (FA) on the six sections (PE = personal episodic; O = orientation in time and space; PS = personal semantic; GS = general semantic; DKS = I dont know semantic; DKE = I dont know episodic) of the Confabulation Battery (Dalla Barba, 1993b).

2.3. Materials
The material consisted of 24 lists of 15 Italian words. These lists were the Italian translation of the lists of semantic associates by Stadler, Roediger, and McDermott (1999). Each list converged on a critical lure representing the gist

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877 of the list. In order to counterbalance the list order, the 24 lists were subdivided into three sets of eight lists each. A set of eight lists was studied in the Standard condition, another set of eight lists was studied in the Proximal condition, and the remaining eight lists were not studied. Four of the latter appeared on the recognition test that accompanied the Standard condition and the other four were used in the recognition tests for the Proximal condition. Studied and nonstudied lists were counterbalanced so that they were used equally often in the Standard condition and in the Proximal condition. In the Standard condition, the recognition test included 48 words, 24 studied and 24 non-studied words. The studied words (targets) were obtained by selecting the items in serial position 1, 8 and 10 for each of the study lists, whereas the non-studied words consisted of the eight critical lures of the studied lists and 16 unrelated lures. Unrelated lures included the critical lures corresponding to four non-studied lists (lure-controls) and the items in serial positions 1, 8, 10 of these non-studied lists (target-controls). In the Proximal condition, recognition memory was also tested on a total of 48 words, but the recognition test was split into eight recognition tests corresponding to each of the eight studied lists and included three studied and three non-studied words. The studied words (targets) were selected from items in serial position 1, 8 and 10 of the just presented list, whereas the non-studied words consisted of the critical lure of that list and two words selected among the critical lures or the items in serial positions 1, 8, 10 of the four non-studied lists. In both conditions, the order of test words was randomized, with the constraints that no more than two words of a particular type (targets, critical lures, target-controls, lure-controls) appeared consecutively. Lures appeared equally often in each half of the test.

1871

Table 3 Mean proportion of study list words, critical lures and non-critical intrusions (RJS = related to the just studied list; RPS = related to previously studied lists; N = unrelated to the studied list) recalled by confabulators (C), frontal amnesics (FA), non-frontal amnesics (NFA) and normal controls (NC), in the Standard and in the Proximal condition Proportion of recalled items List words Standard condition C 0.28 FA 0.34 NFA 0.34 NC 0.50 Proximal condition C 0.26 FA 0.34 NFA 0.32 NC 0.49 Critical lures Non-critical intrusions (RJS + RPS + N) 0.20 (0.02 + 0.05 + 0.13) 0.19 (0.06 + 0.06 + 0.07) 0.20 (0.15 + 0.04 + 0.01) 0.15 (0.09 + 0.05 + 0.01) 0.25 (0.05 + 0.10 + 0.10) 0.22 (0.13 + 0.06 + 0.03) 0.15 (0.11 + 0.03 + 0.01) 0.19 (0.11 + 0.06 + 0.02)

0.28 0.28 0.22 0.39 0.23 0.28 0.25 0.38

3. Results 3.1. Free recall Table 3 presents the proportion of recalled list words (i.e. number of recalled list words divided by the number of studied words), critical lure intrusions (i.e. number of recalled critical lures divided by the number of lists) and non-critical lure intrusions (i.e. number of recalled non-critical lures divided by the number of lists) produced by confabulators (C), F amnesics (FA), NF amnesics (NFA), and normal controls (NC) in the Standard and in the Proximal condition. We classied non-critical lure intrusions into three groups: (1) intrusions related to the just studied list (RJS), (2) intrusions related to previously studied lists (RPS) and (3) intrusions unrelated to the studied lists (N). Results in the Standard and in the Proximal condition were virtually identical, and were therefore collapsed for the purpose of data analysis. An ANOVA on the proportion of recalled words with Group (C, FA, NFA, NC) as a between-subject factor and Item Type (target, critical lure) as within-subject factor yielded a signicant effect of Item Type [F(1,24) = 5.2; p < 0.05], such that participants recalled more targets (0.35) than critical lures (0.29). The main effect of Group was also signicant [F(3,24) = 9.9; p < 0.01]: confabulators (0.26), F amnesics (0.31), and NF amnesics (0.29) recalled fewer targets and critical lures than did controls (0.45; p < 0.05 in all comparisons). No signicant difference emerged among groups of patients. To analyse noncritical lure intrusions, an ANOVA was performed with Group as between-subject factor and Intrusion Type (RJS, RPS, N) as within-subject factor. The main effect of Group was not signicant (p = 0.92), but we found a signicant main effect of Intrusion Type [F(2,48) = 5.1; p < 0.01] which was qualied by a signicant Group Intrusion Type interaction effect [F(6,48) = 5.3; p < 0.01]. Post hoc comparisons, performed with the NewmanKeuls test, showed increased production of N intrusions in confabulators compared to NF amnesics (0.11

2.4. Procedure
All participants were tested in two conditions, which were administered in two sessions separated by about 1 week. Half of the subjects received the Standard condition during the rst session and the Proximal condition during the second one. This order was reversed for the other half. In both conditions, all the lists were auditorily presented at a rate of one word every 1.5 s. Immediately after the presentation of each list, participants were required to recall as many words as they could remember. In the Standard condition the eight study lists were presented during a single session, which took about 20 min, and the recognition test was administered approximately 2 min after the completion of the recall task that followed the nal study list. In the Proximal condition subjects received a six-item recognition test immediately after the completion of the recall task for each list. In both conditions all test items were presented in columns on a sheet of paper and participants were required to indicate whether or not they heard each word at study by stating old or new. In addition, for each word they judged old, they were asked to indicate whether they remembered or rather knew the word. Instructions for the remember/know procedure (Tulving, 1985) were similar to those used by Melo et al. (1999). Participants were told to make a remember (R) response if they could travel back in their minds to the precise moment of hearing the word and remember something about that event. Otherwise, if subjects merely believed that a word was presented earlier, but they could not recollect anything specic about the moment of its occurrence, they had to give a know (K) response. Participants were also required to characterize R responses (Norman & Schacter, 1997), by reporting what they remembered about the items presentation at study. To characterize the qualitative content of subjects explanations, we had two raters evaluate the explanations. Raters were instructed to classify participants explanations as (1) associative (A responses), if subjects referred to remember information that was semantically related to the item (e.g. I remember this word came in the same list as cake and chocolate); (2) contextual (C responses), if participants reported specic details regarding the presentation of the word, such as the target words list context (e.g. This word was the rst of the list/This word came immediately before chocolate) or some sensory characteristics of the target word (e.g. I remember the sound of the word candy); or (3) thoughts (T responses), if subjects reported having a specic thought relating to the target word (e.g. I thought about the music I heard this morning/I remember hill very well because I live on a hill). The raters were provided with a random ordering of the set of explanations for this experiment; the raters were blind to participants identity and item type (i.e. whether a given explanation was provided for targets, critical lures or unrelated distracters). Inter-rater agreement on the classication of the explanations was 96%.

1872

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

versus 0.01; p < 0.05) and normal controls (0.11 versus 0.01; p < 0.05), and reduced RJS intrusions compared to NF amnesics (0.03 versus 0.13; p < 0.05). 3.2. Recognition Table 4 presents the proportion of old responses to targets (i.e. studied items), target-controls (i.e. words belonging to non-studied lists), critical lures and lure-controls (i.e. critical lures of non-studied lists) produced by confabulators (C), F amnesics (FA), NF amnesics (NFA) and normal controls (NC) in the Standard and in the Proximal condition. These proportions were used to derive corrected true- and false-recognition scores (see Table 4). Based on standard high-threshold procedures (e.g. Melo et al., 1999; Schacter et al., 1998), we calculated corrected true recognition scores by subtracting the proportion of old responses to target-controls from the proportion of old responses to targets, and corrected false recognition scores by subtracting the proportion of old responses to lure-controls from the proportion of old responses to critical lures. An ANOVA on old responses that included Group as a between-subject factor and Item Type (targets, lures, target-controls, and lure-controls) and Condition (Standard, Proximal) as within-subject factors showed a significant main effect of Item Type [F(3,72) = 236; p < 0.01], and a signicant Group Condition Item Type interaction [F(9,72) = 5.4; p < 0.01]. Post hoc comparisons, performed with the NewmanKeuls test, showed that in the Standard condition all groups of patients exhibited lower levels of hits (C: 0.52; FA: 0.49; NFA: 0.59; NC: 0.75; p < 0.05 in all comparisons) and false alarms to critical lures than did normal controls (C: 0.60; FA: 0.59; NFA: 0.63; NC: 0.79; p < 0.05 in all comparisons). As for the unrelated lures, all groups of patients presented higher levels of false alarms to target-controls (C: 0.27; FA: 0.17; NFA: 0.20; NC: 0.01; p < 0.05 in all comparisons) and lure-controls (C: 0.35; FA: 0.19; NFA: 0.25; NC: 0.02; p < 0.05 in all comparisons) compared to normal controls. Differences among confabulators

and non-confabulating amnesics in hits and false alarms to critical lures and unrelated lures were not signicant (p > 0.10 in all comparisons). Compared to the Standard condition, in the Proximal condition hits signicantly increased in confabulators and F amnesics (CA: 0.72 versus 0.52; p < 0.01; FA: 0.76 versus 0.49; p < 0.01) and did so in a non-signicant fashion, but in a similar direction, in NF amnesics and normal controls (NFA: 0.77 versus 0.59; p = 0.07; NC: 0.85 versus 0.75; p = 0.3). However, whereas in normal controls false recognition of critical lures decreased in the Proximal compared to the Standard condition (0.79 versus 0.60; p < 0.05), confabulators showed increased levels of false recognition of critical lures across conditions (0.60 versus 0.80; p < 0.05). In contrast, NF amnesics (0.63 versus 0.52; p = 0.2) and F amnesics (0.59 versus 0.63; p = 0.8) showed comparable levels of false recognition of critical lures across conditions (see Fig. 3). As for the unrelated lures, only NF amnesics signicantly suppressed false recognition to target-controls (0.20 versus 0.03; p < 0.05) and lure-controls (0.25 versus 0.05; p < 0.05) across conditions. In the Proximal condition hits were similar among confabulators (0.72), F amnesics (0.76), NF amnesics (0.77) and normal controls (0.85; p > 0.3 in all comparisons). However, confabulators presented higher levels of false recognition to critical lures than did normal controls (0.80 versus 0.60; p < 0.05), NF amnesics (0.80 versus 0.52; p < 0.05), and, marginally, F amnesics (0.80 versus 0.63; p = 0.06). As for unrelated lures, confabulators showed levels of false recognition to targetcontrols signicantly higher than those of NF amnesics (0.20 versus 0.03; p < 0.05), but only marginally higher than those of F amnesics (0.20 versus 0.06; p = 0.11). One could argue that controls evinced lower false recognition for critical lures in the Proximal condition because the test on the earlier lists may have served as practice for the later lists. In other words, if individuals realized the characteristics of the task (e.g. memory for semantically related material is tested with items that resemble the studied items in meaning), they may have started to use this knowledge in later trials (e.g. Hill, but not

Table 4 Recognition memory data for targets, critical lures, target-controls and lurecontrols for confabulators (C), frontal amnesics (FA), non-frontal amnesics (NFA) and normal controls (NC), in the Standard condition and in the Proximal condition Proportion of OLD responses Standard C Targets Target-controls Corrected true recognition Critical lures Lure-controls Corrected false recognition (gist memory) 0.52 0.27 0.25 0.60 0.35 0.25 FA 0.49 0.17 0.32 0.59 0.19 0.40 NFA 0.59 0.20 0.39 0.63 0.25 0.38 NC 0.75 0.01 0.74 0.79 0.02 0.77 Proximal C 0.72 0.20 0.52 0.80 0.25 0.55 FA 0.76 0.06 0.70 0.63 0.13 0.50 NFA 0.77 0.03 0.74 0.52 0.05 0.47 NC 0.85 0.01 0.84 0.60 0.04 0.56 Fig. 3. Mean proportion of old responses to targets and critical lures for confabulators (C), frontal amnesics (FA), non-frontal amnesics (NFA) and normal controls (NC), in the Standard condition and in the Proximal condition.

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

1873

mountain, was on this list). Our ndings are not consistent with this interpretation, because false alarms to critical lures were as frequent in the rst four list as in the last four lists (2 = 0.31; p = 0.6). Since both confabulation and false recognition involve the misattribution to direct experience of events that never occurred, we investigated whether there was a correlation between the amount of confabulations patients produced on the Confabulation Battery and their tendency to make false alarms overall in recognition. In confabulators we found a positive correlation between false alarms (in the Standard condition) and the proportion of confabulatory responses in the DKS section of the Confabulation Battery (r = 0.81; p < 0.05). In contrast, in F amnesics no correlation was found between false alarms and amount of confabulatory reports in the Confabulation Battery (p > 0.3 in all analyses). 3.2.1. Corrected true and false recognition An ANOVA was performed on corrected recognition scores with Group as a between-subject factor, and Item Type (true recognition, false recognition) and Condition as withinsubject factors. Results showed signicant effects of Group [F(3,24) = 14; p < 0.01], Condition [F(1,24) = 46; p < 0.01], and Item Type [F(1,24) = 8; p < 0.01], which were qualied by a signicant Group Condition Item Type interaction [F(3,24) = 3.7; p < 0.05]. Post hoc comparisons, performed with the NewmanKeuls test, indicated that in the Standard condition all groups showed similar corrected true and false recognition scores (C: 0.25 versus 0.25; FA: 0.32 versus 0.40; NFA: 0.39 versus 0.38; NC: 0.74 versus 0.77; p > 0.5 in all comparisons), meaning they were not able to distinguish targets from critical lures (i.e. reality monitoring). In contrast, in the Proximal condition, normal controls (0.84 versus 0.56; p < 0.01), NF amnesics (0.74 versus 0.47; p < 0.01), and F amnesics (0.70 versus 0.50; p < 0.05) showed higher levels of corrected true than false recognition, whereas confabulators still presented comparable levels of corrected true and false recognition (0.52 versus 0.55).

3.3. Phenomenological characteristics of true and false memories Table 5 presents the proportion of R (A, T, C) responses following endorsement of targets, critical lures, and unrelated lures by participant Group in the Standard and in the Proximal condition, respectively. Our primary interest in this section was ascertaining whether the frequency of R responses, and more specically the type of R responses (i.e. T and C), discriminated targets from critical lures. An ANOVA on R responses that included Group as a between-subject factor, Item Type (targets, critical lures) and Condition as within-subject factors showed a signicant effect of Group [F(3,24) = 16; p < 0.01], indicating that F amnesics (0.51) and NF amnesics (0.39) gave fewer R responses compared to normal controls (0.64; p < 0.05 in both comparisons), whereas confabulators and normal controls gave a similar amount of R responses (0.66 versus 0.67; p = 0.9), and a signicant effect of condition [F(1,24) = 4.3; p < 0.05], which meant that more R responses were given in the Proximal (0.59) compared to the Standard condition (0.51). Item Type was not signicant (p = 0.77), indicating that targets and critical lures were not distinguishable in terms of the amount of R responses they received, neither in patients nor in controls, and this is consistent with previous evidence (Payne et al., 1996; Roediger & McDermott, 1995). We next investigated whether targets and critical lures received qualitatively different R responses. Specically, given that confabulators have been found to encounter difculty in suppressing material that is irrelevant to the current task (Schnider, 2003), we asked whether confabulators would be more likely to provide T responses than non-confabulating amnesics. In contrast, because normal controls are thought to use R responses when they can retrieve contextual details of the learning episode, this group was expected to report a higher proportion of C responses to targets than to critical lures. An ANOVA on R responses that included Group as a between-subject factor, and Item Type, Condition and R Type (C, T) as within-subject

Table 5 Mean proportion of R responses (A = associative; T = thoughts; C = contextual) for item called old by confabulators (C), frontal amnesics (FA), non-frontal amnesics (NFA) and normal controls (NC), in the Standard condition and in the Proximal condition Proportion of R responses C Targets Standard condition A 0.28 T 0.31 C 0.02 Total R 0.61 Critical 0.33 0.33 0.03 0.69 0.36 0.17 0.12 0.65 Unrelated 0.10 0.24 0.05 0.39 0.09 0.20 0.01 0.30 FA Targets 0.30 0.04 0.05 0.39 0.44 0.04 0.17 0.65 Critical 0.34 0.06 0.08 0.48 0.35 0.00 0.18 0.53 Unrelated 0.04 0.00 0.04 0.08 0.25 0.00 0.00 0.25 NFA Targets 0.18 0.04 0.06 0.28 0.29 0.04 0.14 0.47 Critical 0.28 0.04 0.01 0.33 0.41 0.06 0.01 0.48 Unrelated 0.08 0.02 0.01 0.11 NC Targets 0.36 0.06 0.24 0.66 0.35 0.04 0.38 0.77 Critical 0.44 0.09 0.12 0.65 0.36 0.05 0.08 0.49

Proximal condition A 0.36 T 0.20 C 0.16 Total R 0.72

R responses for unrelated lures in NF amnesics are not reported, due to oor levels of old responses to these items in NF amnesics.

1874

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

factors yielded a signicant effect of Group [F(3,24) = 612; p < 0.01], and Item type [F(1,24) = 9.7; p < 0.05]. The Group R Type interaction was signicant [F(3,24) = 39; p < 0.05]. Post hoc comparisons, performed with the NewmanKeuls test, showed that confabulating patients gave a higher proportion of T responses (0.25) compared to F amnesics (0.03), NF amnesics (0.04), and normal controls (0.08; p < 0.05 in all comparisons). Finally, the Group Item Type R Type interaction was significant [F(3,24) = 8; p < 0.05]. Post hoc comparisons showed that C responses discriminated targets from critical lures in controls (0.31 versus 0.1; p < 0.05) but not in patients (C: 0.09 versus 0.07; FA: 0.11 versus 0.13; NFA: 0.10 versus 0.02; p > 0.15 in all comparisons). 3.4. T responses and confabulation Because confabulators, unlike non-confabulating amnesics, gave mainly T responses to remembered items, we investigated whether these responses related to confabulatory behaviour. If T responses are a specic correlate of confabulation, we would expect to nd (1) a correlation between indices of confabulation in standard measures and T responses. Moreover, if T responses are related to strategic retrieval decits, we should nd (2) a higher proportion of T responses in the Standard compared to the Proximal condition, because retrieval is more demanding in the former compared to the latter. With respect to (1), we found that T responses to false alarms signicantly correlated with the scores obtained by confabulators in the DKE section of the Confabulation Battery (r = 0.92; p < 0.05). With respect to (2), an ANOVA on T responses with Condition and Item Type (targets, critical lures, unrelated lures) as within-subject factors yielded a marginal effect of Condition [F(1,12) = 3.12; p = 0.10], indicating that T responses were more numerous in the Standard (0.29) than in the Proximal condition (0.19). Item Type was not signicant (p = 0.89). 4. Discussion In this study we explored false recall and false recognition in confabulators, non-confabulating amnesics and normal controls. Moreover, we investigated whether false memories are reduced when studied information is made more memorable. Consistent with previous evidence, at free recall both confabulators and non-confabulating amnesics produced fewer studied words and fewer critical lures than did controls (Melo et al., 1999; Schacter et al., 1996). However, confabulators recalled more words unrelated to the studied lists than did NF amnesics and control subjects. As far as recognition is concerned, in the Standard condition all patient groups endorsed lower proportions of studied words and critical lures, but higher proportions of unrelated lures compared to control subjects (see also Melo et al., 1999). This pattern of results might be potentially due to a decit in understanding the semantic theme of the studied lists at study (i.e. gist extraction) or in remembering gist information from study to test. The evidence that in confabulators false memory for critical lures strongly increased when recognition memory was tested immediately after study (Proximal

condition) indicates that patients are able to extract the gist of the lists, but simply cannot retain it for long delays (Standard condition). It is worth noting that in the Standard condition all groups of patients showed similar levels of false recognition to critical and unrelated lures. In contrast, signicant differences in false recognition susceptibility among groups emerged in the Proximal condition, that is, when true memory increased. As far as critical lures are concerned, in the Proximal compared to the Standard condition normal subjects signicantly suppressed false recognition to critical lures, whereas confabulators showed increasing levels of false recognition across conditions. By investigating the conscious awareness associated to R responses, we found that normal subjects recalled more contextual details in conjunction with studied items than with critical lures, consistent with previous evidence (Norman & Schacter, 1997; Mather, Henkel, & Johnson, 1997; Johnson, Foley, Suengas, & Raye, 1988; Lampinen, Neuschatz, & Payne, 1999). We argue that the temporal proximity of study and test in the Proximal condition promoted normal subjects use of these qualitative differences between studied items and lures when making old/new judgements, which resulted in suppression of critical lures (Johnson, 1991; see also Brainerd, Reyna, & Kneer, 1995). Different from normal controls, the close presentation of study and test did not help confabulators distinguish studied items from critical lures. One could argue that in the Proximal condition confabulators might still have poor item-specic memory compared to normal controls to efciently counteract increasing gist-based false memory, resulting in increased levels of false recognition to critical lures. However, were it the case, we should have found similar results in NF amnesics, who exhibited comparable memory for targets than confabulators. In contrast, false recognition of critical lures tended to decrease in NF amnesics across conditions, suggesting these patients were partially able to benet from enhanced true memory to reduce false recognition to critical lures. Also, the difference between confabulators and NF amnesics cannot be attributable to the latter group enjoying higher levels of true-memory vividness. Indeed, both confabulating and NF amnesics gave fewer C responses to studied words than controls did, indicating an inability to store the contextual details of studied words, against which test items have to be matched (Parkin, Ward, Bindschaedler, Powell, & Squires, 1999; Schacter et al., 1998). Thus, impoverished item-specic memory may be the reason why NF amnesics did not signicantly suppress false recognition of critical lures, but cannot alone account for confabulators increase in false recognition across conditions. On the other hand, it does not seem that a general frontal-lobe dysfunction (Budson et al., 2002; Butler, McDaniel, Dornburg, Price, & Roediger, 2004; Schacter et al., 1998), in addition to amnesia, necessarily produces the increase in false recognition to critical lures observed in confabulators in the Proximal condition. Indeed, though to a lesser extent than were NF amnesics, F amnesics were able to resist the effects of increasing gist memory, resulting in comparable, rather than increased, false recognition to critical lures across conditions.

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877

1875

Additionally, we found that confabulators were unable to use increased gist memory in the Proximal condition to reject unrelated lures. That is, while an increase in gist memory allowed NF amnesics to suppress false recognition to gist-inconsistent material, confabulators still endorsed high proportion of unrelated lures in the Proximal condition, consistent with the evidence that these patients have problems in inhibiting material irrelevant to the task (Schnider, 2003). We note, however, that a decreased ability to suppress unrelated lures across conditions is also observed in F amnesics, indicating that frontal-lobe dysfunction in itself may interfere with the rejection of lures unrelated to the studied material (see also Curran, Schacter, Norman, & Galluccio, 1997; Shimamura, 1995). The results of the present study show that confabulators did not benet as much as non-confabulating amnesics from enhanced true memory promoted by the Proximal condition to suppress false recognition, although in the Standard condition confabulating and non-confabulating amnesics showed similar reality monitoring abilities (see also Schnider et al., 1996b). As a consequence, in the Proximal condition confabulators showed higher levels of false recognition to critical and unrelated lures than non-confabulating amnesics, with comparable levels of true recognition. The nding that reality monitoring in confabulators did not signicantly benet from an enhanced memory for studied material suggests that, in these patients, the inability to distinguish true from false memories depends only partially on memory decits (Johnson, 1991; Moscovitch & Melo, 1997). Thus, it seems that monitoring decits, possibly related to ventromedial frontal damage (Gilboa, 2004; Treyer, Buck, & Schnider, 2003), interfere with reality monitoring abilities in confabulators. It is worth noting that a striking difference between confabulating and non-confabulating amnesics concerns T responses. These uent thoughts related to test items were quite frequent in confabulators, but virtually absent in the control groups, and correlated with their tendency to confabulate in real life, as assessed by the Confabulation Battery (Dalla Barba, 1993b). Of importance, in confabulators T responses were as frequent for targets as for critical and unrelated lures, and tended to decrease in the Proximal condition, which is supposed to require less retrieval effort. This suggests that these responses, rather than genuine recollections of the study episode, may be free associations at the time of test. Specically, they might be the products of retrieval attempts (Koriat, 1993). It has been shown that the amount of partial information about a memory resulting from retrieval attempts is computed in order to estimate the accessibility of that memory (Koriat, 1993), thereby inuencing the probability that subjects accept it as true (Kelley & Lindsay, 1993; Koriat & Goldsmith, 1996). This being the case, the inability of confabulating patients to distinguish studied from unstudied words might be related to a failure to inhibit vivid, but irrelevant, thoughts triggered by test items (e.g. I remember butter, because I hate it). This is because these parasitic thoughts may contribute to make the subjective experience of false memories as vivid as that of true memories. Accordingly, Johnson described a confabulating patient (GS) who reported abnormally vivid

descriptions of memories for imagined events (Johnson et al., 1997). In line with this hypothesis, Schnider and colleagues have shown that confabulators fail in suppressing memory traces not relevant to the current task (Schnider, 2003; Schnider & Ptak, 1999). Moreover, neuroimaging evidence shows that some particular regions within the orbitofrontal cortex, and commonly damaged after ACoA aneurysm, play a crucial role in inhibitory (Treyer et al., 2003) and felt-rightness mechanisms (Moscovitch & Winocur, 2002), and also support accurate feeling-of-knowing judgements (Schnyer et al., 2004). To conclude, the results of the present study conrm that confabulating, like non-confabulating, amnesics show low levels of true recognition and false recognition of critical lures but high levels of false recognition to unrelated lures in the DRM paradigm (Standard, delay condition), due to poor gist memory (Budson et al., 2002; Melo et al., 1999; Schacter et al., 1998). Moreover, we demonstrated that in an experimental condition (Proximal, short delay condition) that enhances memory for studied items control subjects and non-confabulating amnesics improved their reality monitoring abilities whereas confabulators remained unable to distinguish true from false memories, possibly due to a decit in inhibiting irrelevant recollections during retrieval. Acknowledgements We thank Morris Moscovitch, Sandra Priselac, and the anonymous reviewers for helpful comments on the paper. We also thank Gianfranco Dalla Barba for making the Confabulation Battery available to us, Michela Coccia and Silvia G. Bonifazi for referring patients AP, RB, and BL to us, and Sebastiano Lupo and Giuseppe di Pellegrino for their assistance with the analysis of the lesion sites. References
Alexander, M. P., & Freedman, M. (1984). Amnesia after anterior communicating artery rupture. Neurology, 34, 752757. Balota, D. A., Cortese, M. J., Duchek, J. M., Adams, D., Roediger, H. L., McDermott, K. B., et al. (1999). Veridical and False memories in healthy older adults and in dementia of the Alzheimers type. Cognitive Neuropsychology, 16, 361384. Berlyne, N. (1972). Confabulation. British Journal of Psychiatry, 120, 3139. Box, O., Laing, H., & Kopelman, M. D. (1999). The evolution of spontaneous confabulation, delusion, misidentication and a related delusion in a case of severe head injury. Neurocase, 5, 251262. Brainerd, C. J., Reyna, V. F., & Kneer, R. (1995). False recognition reversal: When similarity is distinctive. Journal of Memory and Language, 34, 157185. Budson, A. E., Sullivan, A. L., Mayer, E., Daffner, K. R., Black, P. M., & Schacter, D. L. (2002). Suppression of false recognition in Alzheimers disease and in patients with frontal lobe lesions. Brain, 125, 2750 2765. Burgess, P. V., & Shallice, T. (1996). Confabulation and the control of recollection. Memory, 4, 359411. Buschke, H., & Fuld, P. A. (1974). Evaluating storage, retention and retrieval in disordered memory and learning. Neurology, 11, 1019 1025. Butler, K. M., McDaniel, M. A., Dornburg, C. C., Price, A. L., & Roediger, H. L., III. (2004). Age differences in veridical and false recall are not

1876

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877 Mather, M., Henkel, L. A., & Johnson, M. L. (1997). Evaluating characteristics of false memories: Remember/know judgements and memory characteristics questionnaire compared. Memory and Cognition, 25, 826 837. Melo, B., Winocur, G., & Moscovitch, M. (1999). False recall and false recognition: examination of the effects of selective and combined lesions to the medial temporal lobe/diencephalon and frontal lobe structures. Cognitive Neuropsychology, 16, 343359. Moscovitch, M. (1989). Confabulation and the frontal system: Strategic versus associative retrieval in neuropsychological theories of memory. In H. R. Roediger & F. I. Craik (Eds.), Varieties of memory and consciousness: Essays in the honour of Endel Tulving. Hillsdale: Lawrence Erlbaum Associates. Moscovitch, M. (1995). Confabulation. In D. L. Schacter (Ed.), Memory distortions. Cambridge: Harvard University Press. Moscovitch, M., & Melo, B. (1997). Strategic retrieval and the frontal lobes: Evidence from confabulation and amnesia. Neuropsychologia, 35, 10171034. Moscovitch, M., & Winocur, G. (2002). The frontal cortex and working with memory. In D. T. Stuss & R. Knight (Eds.), Principles of frontal lobe function (pp. 188209). New York: Oxford University Press. Norman, K. A., & Schacter, D. L. (1997). False recognition in younger and older adults: Exploring the characteristics of illusory memories. Memory and Cognition, 25, 838848. Parkin, A. J., Ward, J., Bindschaedler, C., Powell, G., & Squires, E. L. (1999). False recognition following frontal lobe damage: The role of encoding factors. Cognitive Neuropsychology, 16, 243265. Payne, D. G., Elie, C. L., Blackwell, J. M., & Neuschatz, J. S. (1996). Memory illusions: Recalling, recognizing, and recollecting events that never occurred. Journal of Memory and Language, 35, 261 285. Reyna, V. E., & Brainerd, C. J. (1995). Fuzzy-trace theory: An interim synthesis. Learning and Individual Differences, 7, 125. Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Esperimental Psychology: Learning, Memory and Cognition, 21, 803814. Schacter, D. L., Verfaellie, M., & Pradere, D. (1996). The neuropsychology of memory illusions: False recall and recognition in amnesics patients. Journal of Memory and Language, 35, 319334. Schacter, D. L., Norman, K. A., & Koutstaal, W. (1998). The cognitive neuroscience of constructive memory. Annual Review of Psychology, 49, 289318. Schacter, D. L., Verfaellie, M., Anes, M. D., & Racine, C. (1998). When true recognition suppresses false recognition: Evidence from amnesic patients. Journal of Cognitive Neuroscience, 10, 668679. Schnider, A. (2003). Spontaneous confabulation and the adaptation of thought to ongoing reality. Nature Reviews Neuroscience, 4, 662671. Schnider, A., & Ptak, R. (1999). Spontaneous confabulators fail to suppress currently irrelevant memory traces. Nature Neuroscience, 2, 677 681. Schnider, A., Gutbrod, K., Hess, C. W., & Schroth, G. (1996). Memory without context. Amnesia with confabulations following right capsular genu infarction. Journal of Neurology, Neurosurgery, and Psychiatry, 61, 186193. Schnider, A., von Daniken, C., & Gutbrod, K. (1996). Disorientation in amnesia. A confusion of memory traces. Brain, 119, 1627 1632. Schnyer, D. M., Verfaellie, M., Alexander, M. P., LaFleche, G., Nicholls, L., & Kaszniak, A. W. (2004). A role for right medial prefrontal cortex in accurate feeling-of-knowing judgements: Evidence from patients with lesions to frontal cortex. Neuropsychologia, 42, 957 966. Shimamura, A. P. (1995). Memory and the frontal lobe function. In M. S. Gazzaniga (Ed.), The cognitive neurosciences. Cambridge: MIT Press. Spinnler, H., & Tognoni, G. (1987). Standardizzazione e Taratura Italiana di Test Neuropsicologici. The Italian Journal of Neurological Science, 6(Suppl. 8) (Milano: Masson Italia).

inevitable: The role of frontal lobe function. Psychonomic Bulletin and Review, 11, 921925. Culbertson, W. C., & Zillmer, E. A. (2001). Tower of London. Toronto: Multi-Health System Press. Curran, T., Schacter, D. L., Norman, K. A., & Galluccio, L. (1997). False recognition after a right frontal lobe infarcition: Memory for general and specic information. Neuropsychologia, 35, 10351049. Dalla Barba, G. (1993a). Different patterns of confabulation. Cortex, 29, 567581. Dalla Barba, G. (1993b). Confabulation: Knowledge and recollective experience. Cognitive Neuropsychology, 10, 120. Dalla Barba, G. (1995). Confabulation: Remembering another past. In R. Cambell & M. A. Conway (Eds.), Broken memories. Oxford: Blackwell. Damasio, H., & Damasio, A. R. (1989). Lesion analysis in neuropsychology. New York: Oxford University Press. Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58, 1722. DeLuca, J. (1993). Predicting neurobehavioural patterns following anerior communicating artery aneurysm. Cortex, 29, 639647. Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). Mini mental state: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189198. Fotopoulou, A., Solms, M., & Turnbull, O. (2004). Wishful reality distortions in confabulation: A case report. Neuropsychologia, 42, 727744. Gilboa, A. (2004). Autobiographical and episodic memory-one and the same? Evidence from prefrontal activation in neuroimaging studies. Neuropsychologia, 42, 13361349. Gilboa, A., & Moscovitch, M. (2002). In A. D. Baddeley, M. D. Kopelman, & B. A. Wilson (Eds.), The cognitive neuroscience of confabulation: A review and a model. John Wiley and Sons. Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Source memory impairment in patients with frontal lobe lesions. Neuropsychologia, 27, 10431056. Johnson, M. L. (1991). Reality monitoring: Evidence from confabulation in organic brain disease patients. In G. P. Prigatano & D. L. Schacter (Eds.), Awareness of decit after brain damage. Oxford: Oxford University Press. Johnson, M. L., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). Phenomenological characteristics of memories for perceived and imagined events. Journal of Experimental Psychology: General, 117, 371376. Johnson, M. L., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 328. Johnson, M. L., OConnor, M., & Cantor, J. (1997). Confabulation, memory decits and frontal dysfunction. Brain and Cognition, 34, 189206. Johnson, M. L., & Raye, C. (1998). False memories and confabulation. Trends in Cognitive Science, 2, 137145. Kelley, C. M., & Lindsay, D. S. (1993). Remembering mistaken for knowing: ease of retrieval as a basis for condence in answers to general knowledge questions. Journal of Memory and Language, 32, 124. Kopelman, M. D. (1987). Two types of confabulation. Journal of Neurology, Neurosurgery, and Psychiatry, 50, 14821487. Kopelman, M. D., Guinan, E. M., & Lewis, F. D. R. (1995). Delusional memory, confabulation, and frontal lobe dysfunction: A case study in the Clerambaults syndrome. Neurocase, 1, 7177. Kopelman, M. D., Ng, N., & Van der Broke, O. (1997). Confabulation extending across episodic, personal and general semantic memory. Cognitive Neuropsychology, 14, 683712. Koriat, A. (1993). How do we know that we know? The accessibility model of the feeling of knowing. Psychological Review, 100, 609639. Koriat, A., & Goldsmith, M. (1996). Monitoring and control processes in the strategic regulation of memory accuracy. Psychological Review, 103, 490517. Korsakoff, S. S. (1889). Medico-psychological study of a memory disorder. Consciousness and Cognition, 5, 221. Lampinen, J. M., Neuschatz, J. S., & Payne, D. G. (1999). Source attributions and false memories: A test of the demand characteristics account. Psychonomic Bulletin and Review, 6, 130135. Lezak, M. D. (1995). Neuropsychological assessment. New York: Oxford University Press.

E. Ciaramelli et al. / Neuropsychologia 44 (2006) 18661877 Stadler, M. A., Roediger, H. L., & McDermott, K. B. (1999). Norms for word lists that create false memories. Memory and Cognition, 27, 494500. Talland, G. A. (1961). Confabulation in the WernickeKorsakoff syndrome. Journal of Nervous and Mental Disease, 132, 361381. Treyer, V., Buck, A., & Schnider, A. (2003). Subcortical loop activation during selection of currently relevant memories. Journal of Cognitive Neuroscience, 15, 610618.

1877

Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 112. Vilkki, J. (1985). Amnesic syndromes after surgery of anterior communicating artery aneurysms. Cortex, 21, 431444. Wechsler, D. (Ed.). (1987). The Wechsler memory scale. London: Psychological Corporation.