The Australian Remote Memory Battery, 10 Years on
John O'Gorman, School of Applied Psychology, Griffith University, Mt Gravatt Campus M24, Brisbane, QLD 4111, Australia. email: email@example.com
The article reviews research conducted with the Australian Remote Memory Test (ARMB) since its development first reported in 2001. Data from case studies and experimental investigations that have used the ARMB are summarised, and the use made of these data in reviews of the literature is noted. The ARMB has been found sensitive to disease processes such as herpes simplex, stroke, and epilepsy and lesion effects such as temporal lobectomy. The pattern of findings is consistent with current understanding of memory function. The studies have, however, pointed to factors that need to be considered in using the test, including patient medication and education level. Directions for future research with the test are briefly noted.
Memory for events and people in the distant past, remote memory, involves autobiographical memory and public memory (Kopelman, 2002). Autobiographical remote memory is the recall or recognition of people or events from the past with personal meaning or reference for the individual, such as classmates at school or the setting of a childhood holiday. Public remote memory, on the other hand, relates to people or events that are part of the public store of knowledge for all those who have lived in a particular era, such as the opening of a landmark building or the image of a particular national leader. The accurate assessment of autobiographical and public memories of the remote past is important for clinical diagnosis and management in health care settings (Kopelman, 2002; Lezak, Howieson, & Loring, 2004). This is because identification of one or both of these two types of memory impairment can contribute to differential diagnosis of patients and assist clinicians to develop treatment/management plans. The assessment of each is, however, attended by its own set of challenges.
For autobiographical memory, there is the difficulty of verifying the accuracy of recall of personal events. Partners or family members can assist in this, but because memory is for the long past, even they may be of little help in some instances (Lezak et al., 2004). There is also the problem of rehearsal as important events or people form part of an ongoing narrative, which is repeated sufficiently frequently for the content to be part of the person's recent and not distant past. For public memory, there is the difficulty of selecting items that are not so widely known as to be available to most people irrespective of their age. For example, one would not have had to live through the world wars of the 20th century to be aware of them or to have lived in the 1950s to have heard of Elvis Presley. There is as well a potential confounding with education level and mental ability when items of public knowledge are tested.
The Australian Remote Memory Battery (ARMB; a title given to it by its users rather than its authors, but readily accepted by them) was developed for the assessment of public memory for the distant past for long-time residents of Australia (Shum & O'Gorman, 2001). It was modelled on similar tests developed in the UK and the USA, and involves a public events subtest and a famous faces subtest, which cover the period from the 1930s to the 1990s. There are nine items of each type for each of the six decades making for a total of 54 items on each subtest. Some of the items have an international reference (e.g., the face of Franklin D. Roosevelt; the question “Who was the first man to land on the moon?”), but the aim was to identify items that were sufficiently specific to Australia (e.g., the face of Lionel Rose; the question “What was the Ash Wednesday disaster?”) for knowledge of people and events current in Australia in the different decades to be tested. When scores are computed by decade, unselected samples of participants were shown to have better recall for more recent than remote events and faces. Both subtests employ recall and recognition formats, with recall always presented prior to recognition. For the recognition version, there are three options for each question. The purpose of the two formats is to allow examination of problems in storage and retrieval.
The items were carefully selected using a pilot sample and then tested using samples of males and females varying in age. As expected of a valid test of remote memory, scores for older participants compared with middle-aged and younger participants were better for events and faces in the earliest decades, whereas scores for the middle-aged participants were better than those for the younger in the middle decades, with scores for the three age groups not differing for the most recent decade. This was generally true for both recall and recognition. In a second study, performance on the subtests for two patient samples for whom retrograde amnesia is known to be a significant symptom was compared with that of normal controls. Both clinical groups, one comprising patients with Korsakoff's syndrome and one with Alzheimer's disease, differed significantly from their control groups on both recall and recognition forms of the famous faces and public events subtests.
Our development work on the ARMB pointed to it being a potentially useful test of remote memory that could be employed for two principal purposes. One was to contribute, with other memory tests, to the diagnosis of patients with pathology of the central nervous system arising from brain injury or disease. The other potential use was to assist in the testing of hypotheses derived from models of memory. Among the possibilities, the test allows examination of an interaction between side of temporal lobe lesion and type of stimulus material. For example, is performance poorer for both faces and events for both left and right temporal lesions, or is there some preservation of memory for faces in the case of right temporal lobe lesions? The use of material from a number of decades allows the test to be used to examine the slope and direction of any temporal gradient in memory. For example, is memory for remote events spared relative to recent events in certain neurological conditions? The inclusion of both recall and recognition formats allows examination of differences in storage and retrieval of material. For example, is recognition adequate in the face of impaired recall suggesting a retrieval and not a storage problem.
The test has now been used in a number of Australian studies, and the results have been included in international reviews of research on remote memory. The following sections examine these studies with a view to determining what we now know about the ARMB that can guide its further use and development. The studies were identified using Scopus to search for all citations of the ARMB up to 2011, and the results were checked using Google Scholar. In all, 10 articles were identified. Two were review articles, two reported case studies, and four involved comparisons of patient groups with normal controls. There were two other foreign language articles. One was a thesis in German and one was a publication in a Spanish language journal. As neither of these used Australian participants, they were not included in the present review. The empirical studies that were included are summarised in Table 1. The original validation study has been included in the table for completeness.
Table 1. Studies Using the ARMB
|Geffen et al. (2008)||HSVE||Recall and recognition faces and events; 1970s and 1980s||Recall faces||0|
|Recognition faces||Moderate impairment|
|Recognition events||Chance level|
|Hepner et al. (2007)||Stroke (MT), 11 NC||Recall and recognition faces and events; relevant decades||Recall faces||≈NC|
|Group comparison studies|
|Batchelor et al. (2008)||10 RHS, 10 LHS, 10 NC||Recall and recognition events only; decades appropriate to age||Recall events||RHS ≈ LHS ≈ NC; MT < NC.05|
|Recognition events||RHS ≈ LHS ≈ NC; MT < NC.05|
|Lah et al. (2004)||15 RTL, 15 LTL, 15 NC||Recall and recognition faces and events; three decades only||Recall faces||LTL < RTL.005 and NC.001; RTL ≈ NC|
|Recognition faces||LTL ≈ RTL ≈NC|
|Recall events||LTL and RTL < NC.005; LTL ≈ RTL|
|Recognition events||1970s: LTL ≈ RTL ≈ NC; 1980s: LTL < NC.001; 1990s: RTL < NC.005|
|Lah et al. (2006)||14 RTLE, 15 LTLE, 15 NC||Recall and recognition faces and events; three decades only||Recall faces||LTLE < NC.001|
|Recognition faces||LTLE ≈ RTLE ≈ NC|
|Recall events||LTLE < RTLE.04 and NC.001|
|Recognition events||LTLE < RTLE.02 and NC.001|
|Lah et al. (2008)||8 RTL, 7 LTL||Recall faces and events; three decades only||Recall faces||PreLTL ≈ PostLTL; PreRTL ≈ PostRTL|
|Recall events||PreLTL ≈ PostLTL; PreRTL ≈ PostRTL|
|Shum and O'Gorman (2001)||9 Alz, 9 Kkf, 9 + 9 NC||Recall and recognition faces and events; all decades||Recall faces||Alz < NC.01; Kkf < NC.01|
|Recognition faces||Alz < NC.01; Kkf < NC.01|
|Recall events||Alz < NC.01; Kkf < NC.01|
|Recognition faces||Alz < NC.01; Kkf < NC.01|
Hepner, Mohamed, Fulham, and Miller (2007) reported the case of SG, a stroke patient, who had sustained infarctions in the mesial temporal lobe and retroplenial region bilaterally. Site and extent of damage was established by magnetic resonance imaging (MRI) and fluorodeoxyglucose positron emission tomography. Retrograde autobiographical, public, and topographical memories were examined using a number of tests, including the ARMB, and performance compared with that of 11 normal controls matched for age and educational level. Three scores were calculated for childhood (the first full decade of life), recent life (the most recent full decade), and midlife (the average of intermediate decades). SG's performance on the famous faces subtest of the ARMB was within the range of scores for control subjects for both recall and recognition. On the public events subtest, however, SG's performance was poorer than that of control subjects for most time periods for recall and recognition, and in the case of the childhood period, recall and recognition scores were significantly different from controls. By contrast, SG's recall of autobiographical events from childhood, as assessed by interview and by response to cue words and verified by his partner, remained intact.
SG's performance on famous faces given his medial temporal lobe damage, Hepner et al. reasoned, contraindicated a posited role for this brain region in the identification of famous faces but was consistent with two other case studies they reviewed and with the role of the fusiform gyrus in this function. Damage to this region had been spared in SG's case. Thus, performance on the ARMB was useful in checking current accounts of aspects of memory functioning as well as illuminating the nature of the deficits suffered by SG as a result of stroke.
Geffen, Isles, Preece, and Geffen (2008) reported the case of JL who had become densely amnesic following herpes simplex viral encephalitis (HSVE). MRI confirmed swelling of the temporal and frontal lobe bilaterally, although more extensively on the left side. Her immediate memory, perceptual priming, and cognitive problem solving were relatively intact, but she showed little memory of personal information or events from her early life as tested by interview. When JL was administered the 18 items of the ARMB from the 1970s and 1980s she recorded scores of 0 on recall of both famous faces and public events, a chance level of performance on the recognition form of the public events test, but only moderate impairment on the recognition form of the famous faces subtest. Geffen et al. did not comment specifically on the latter observation but it might suggest a retrieval problem for memories for faces, whereas storage problems seem more likely for memories of public events. The use of both recall and recognition formats in the ARMB makes such comparison possible. Although subsequent follow-up indicated that JL had learned or relearned a number of skills such as programming a video player/recorder, swimming, playing the flute, performing multiplication and division, and although she retained some of the procedural skills of her former profession, physiotherapy, the deficits in her memory and limitations in her social skills had meant that rehabilitation to her profession was not possible.
Group Comparison Studies
Batchelor, Thompson, and Miller (2008) reported a study of retrograde memory in stroke patients with unilateral lesions involving the frontal or temporal lobe. Autobiographical memory was tested by interview, memory for public events using the public events subtest of the ARMB, and memory for words entering the language at different time points was tested using an adaptation of a North American test developed for this purpose. A normal control group matched on age and educational level with the patient group was used for comparison. Items from the public events subtest were administered to subjects from the first decade following the decade of their birth, which was described as “childhood.” The items from the most recent decade were described as “recent life” and those for the intervening years were described as “midlife.” Statistical analyses of performance on all tests involved comparison of patients, grouped by site of lesion and side of lesion, with controls for the three life periods. The site of lesion comparison involved those patients with damage to the hippocampus with those with damage outside that region. For the public events test, there were no group, life period, or life period by group interactions for recall or recognition tests, when group was formed in terms of side of lesion. When, however, group was constituted by lesion site, there was a significant main effect for group for both recall and recognition, with those with hippocampal involvement showing poorer memory than normal controls. For autobiographical memory, there were side and site of lesion effects with the right-sided group and the hippocampal group showing poorer performance than normal controls. There were no significant differences on the new vocabulary test.
As the authors noted, the poorer performance of the hippocampal group on memory for public events as tested by both recall and recognition points to problems in storage and retrieval, and the specificity of the side of lesion effect to autobiographical memory rather than being general to autobiographical and public events memory, suggests the right hemisphere is particularly important in memory for events from the past with personal reference. There were few significant interactions with the life period factor (none for scores on the public events subtest) indicating no evidence of the temporal gradient referred to in the literature, with older memories being preserved in comparison with newer with some forms of disease or injury. Although eight of the 20 patients showed a level of performance two or more standard deviations (SDs) below the mean for the control group on the public events test, none complained of difficulties remembering the past, which the authors suggested may indicate a poor appreciation of or poor insight into their own retrograde memory abilities.
An incidental finding of some importance for the ARMB is that educational level correlated significantly with score on the public events test for both recall (r = 0.44, df = 30, p = .007) and recognition (r = 0.55, df = 30, p < .001). We comment on this point later.
Lah, Grayson, Lee, and Miller (2004) used the ARMB in a study of memory in patients who had undergone unilateral temporal lobectomy (TL) for the relief of epilepsy. Right and left TL patients were compared with normal controls on a series of tests of anterograde and retrograde memory and of executive functions (EFs). There were differences between patient groups and controls on recall of famous faces and public events. For famous faces, the left TL group performed more poorly than the right TL group and the controls; whereas for public events, the left and right TL groups performed more poorly than controls but did not differ significantly between themselves. These differences remained when adjustments were made for correlation with some EF measures. For recognition as distinct from recall, there were no significant differences between patient groups or controls for the famous faces subtest. There was a group effect for public events recognition but this was qualified by an interaction with time (the right TL group showed a drop in performance for the most recent decade) and was nonsignificant when adjusted for correlations with EF tests. Thus, there were robust differences between patient groups and controls for recall, although the difference in the case of famous faces was specific to the left TL group.
There were no correlations between any of the subtest scores on the ARMB and anterograde memory, consistent with the view that these are different types of memory. There were correlations between the Benton naming test and recall and recognition scores on the ARMB, and these correlations were stronger for measures using the items from the more recent decades. Importantly, there was a significant medication effect on most of the ARMB measures, with those patients on medication for epilepsy showing poorer performance than those who were medication free.
Taken as a whole, the authors interpreted their results as consistent with a model in which the anterior temporal lobes play a role in memory for unique information from the past. The ARMB has thus proved useful in furthering theoretical understanding of memory and not only in diagnosis.
Lah, Lee, Grayson, and Miller (2006) reported on the effects of temporal lobe epilepsy (TLE) on retrograde memory, with 29 patients who were candidates for unilateral TL for symptom relief. Their performance was compared with that of the control group used in their earlier study with patients who had undergone lobectomy. The ARMB was once again one of the tests used. Those with left TLE differed significantly from those with right TLE on recall of famous faces and recall and recognition for public events, using items from the 1970s to the 1990s. Those with left TLE recalled significantly fewer faces than controls and recalled and recognised significantly fewer public events than controls or those with right TLE. Unlike their earlier finding, where both left and right TL groups had difficulty with the public events subtest, there were no significant differences on this subtest between right TLE group and controls. There was no evidence of a temporal gradient for any of the measures.
In other findings, Lah et al. (2006) reported that patients with a history of early seizures had lower scores on all memory tests and the differences were significant for famous events recall and recognition. There was also some indication of poorer performance for those on polytherapy as compared with those on monotherapy. Unlike their finding in the previous study, there were some significant correlations between the anterograde memory measures and the ARMB measures for the patient group as a whole. There were also significant correlations between the ARMB measures and the Boston Naming Test and the Controlled Oral Word Association Test.
This study provides evidence of the sensitivity of the ARMB to the effects of epilepsy, which broadens the domain of disease processes for which differences with the ARMB have been detected beyond stroke and HSVE (see earlier) and Alzheimer's and Parkinson's (see Shum & O'Gorman, 2001). The earlier study by Lah et al. (2004) showed sensitivity of the ARMB to brain lesions, but in this study, participants had yet to undergo lobectomy.
In a third study, Lah, Lee, Grayson, and Miller (2008) compared patients with unilateral TLE before and after undergoing lobectomy, again using a range of tests of anterograde and retrograde memory. Sample size was smaller in this study (n = 15) than previously, and as the researchers comment, this would have reduced the power of the significance tests. For the ARMB, there were no statistically significant pre–post differences on any of the subtests, although there were significant differences on some of the other retrograde memory tests employed (e.g., autobiographical fluency for friends' names and personal events). Mean scores for the sample on recall or famous faces and public events on the ARMB were low, more so for faces than events, leading Lah et al. to comment on the difficulty level for this group of patients. It is difficult to draw definite conclusions about the ARMB from this study because of the power problem with it.
Gainotti (2007) reported a review of case studies of patients with damage to the anterior right or left temporal lobe who reported disorders of famous faces recognition and a review of patients with right or left TLE in which impairment of famous faces recognition had been investigated. The study of Lah et al. (2004) that used the famous faces subtest of the ARMB was included in Gainotti's review as one of the six studies that had investigated the question in TLE. Gainotti concluded that, despite a number of methodological problems, there was evidence that damage to the right temporal lobe results in loss of feelings of familiarity and of personal information in face recognition and that damage to the left temporal lobe damage results in impairment of naming of faces.
Lah and Miller (2008) provided a thorough review of research on retrograde memory in patients with lesions of the temporal lobe, including findings and range of methodological and patient variables that need to be taken into account in interpreting these. They included reference to the ARMB, noting the need for a culturally specific test of remote memory for use in Australia. They noted as well the likelihood of obsolescence with tests of this sort as material that is initially well known fades into insignificance and item content and norming need updating. A shortcoming of the ARMB, they noted, was the time it took to administer, but an advantage was that it allowed assessment of a temporal gradient for memory for the past. In their review, studies that reported impairment and that were restricted to patients with damage to mesial temporal lobe structures all reported a temporal gradient. Studies where the lesion extended beyond the mesial temporal region showed impairment to be temporarily pervasive.
In developing the ARMB, we sought to provide an additional clinical tool for the investigation of remote memory for use with long term residents of Australia and an experimental tool for examining hypotheses advanced in the neuropsychological literature. Although the pool of potential users in Australia is small and by design the ARMB is for use only with those who have lived there for a considerable period, the test has been successful in meeting its purpose. Both case studies and group comparisons have shown the value of the test for examining subtle differences in memory performance of which participants are not necessarily aware. The research as well adds to our understanding of the construct validity of the test.
As one would expect of a valid test of memory, damage to brain structures known to be involved in recall and recognition, such as the temporal lobe, affects performance on the test in predictable ways. For example, damage to the hippocampus resulting from stroke when compared with damage outside that region leads to impairment of recall and recognition on the public events subtest of the ARMB (Batchelor et al., 2008). Likewise, TL results in poorer performance on this subtest compared with that of controls (Lah et al., 2004).
The discriminant validity of the test is shown by its generally low correlations with tests of anterograde memory such as those on the Wechsler Memory Scale – Revised (Lah et al., 2004; Lah et al., 2008; but see Lah et al., 2006 where some correlations were significant) and by differences in lesion effects for measures of autobiographical memory and public events memory. These dissociations are broadly supportive of the validity of the test because they are consistent with the complexity inherent in the emerging understanding of memory and indicate that the ARMB subtests provide additional information in investigations of memory functioning. Important in this regard is the capacity of the subtests, because of their sampling of a number of decades, to demonstrate a temporal gradient in memory.
A potential limitation on validity is the finding by Batchelor et al. of a moderate correlation (0.44–0.55) between ARMB scores and educational level. In our original validation study, we had examined correlations between education level and recall and recognition scores for each of our three groups formed in terms of age. Only one of the six correlations proved to be statistically significant and we saw that as evidence for the substantial independence of the test from differences in education. However, education would be expected to increase knowledge of public events either directly or through encouraging awareness of world events. Kapur, Thompson, Kartsounis, and Abbott (1999) reported that exposure to media may compromise tests of knowledge of people and events. In their study, score on a test of knowledge of news events correlated 0.76 (p < .001) with score on the Dead-or-Alive test in which subjects must decide whether a name is that of a person who is alive or dead. Further study of the potentially confounding effects of differences in educational level and exposure to mass media on the ARMB is warranted. The finding of Batchelor et al., however, means that educational level must be a consideration when using the ARMB until more is known about this effect. Where groups are compared, educational level can of course be used as a matching variable, but in the study of individual cases, without control groups, the possible effect of educational level needs to be considered.
One other consideration that has emerged from the research to date with the ARMB is that there may be floor effects for some patients because of the difficulty level of some of the items. Lah et al. (2008) reported these in their sample of epilepsy patients, where mean scores for recall of famous faces were as low as 5 for those in the left TLE group, with the maximum being 27 (nine items for each of three decades). A mean score at this level was also reported by Lah et al. (2006) for their left TLE group (again maximum 27). At such low levels, any further decline in performance is difficult to demonstrate. However, for non-patient groups, no such floor effects have been reported. Batchelor et al. reported mean scores for recall and recognition on the public events subtest of ARMB of 23.4 (SD = 2) and 26.1 (SD = 0.08), respectively, for their sample of 10 normal controls (mean age = 57). Maximum scores were 27 in both cases. For the 15 control subjects (mean age = 36.3 years) in the study of Lah et al. (2004), recall and recognition scores were at a similar level (22 and 25, respectively), with maximum score again being 27. For the famous faces subtest (which was not used by Batchelor et al.), recall and recognition scores were 14 and 24, respectively, out of a possible 27. These values are close to those found in our original study. The possibility of floor effects with patient groups should, however, be noted.
Also of note in using the test is the medication effect reported by Lah et al. (2006), with scores being lower for patients on medication, although not significantly so. There was also some suggestion in their later study (Lah et al., 2006) that patients on polymedication performed more poorly than those on single medication. These results are not surprising but point to the need to consider medication level as a possibly confounding effect in using the test.
The ARMB is now 10 years old and, as Lah and Miller (2008) noted, tests of this kind can age as the store of public knowledge changes. The work to date encourages further revision of the test so that its usefulness for practitioners and researchers is maintained. Such research should assist in developing norms for the test, which are currently rudimentary at best. One useful direction for future research is development of a computer version, which would improve standardisation of procedures for administering and scoring the test and increase its availability.
Memory for the distant past is an important component of the pattern of behaviours and cognitions that fall under the rubric of memory. It is one, however, less studied than memory for recent information because of a number of instrument issues. The ARMB was developed to assist diagnosticians and researchers in Australia to probe systematically one aspect of remote memory to do with famous people and events. The present review indicates that the test has been reasonably successful in accomplishing this aim, with data now reported from clinical case studies and experimental investigations and the results included in secondary sources.