An object location memory paradigm for older adults with and without mild cognitive impairment

Journal of Neuroscience Methods 237 (2014) 16–25

Contents lists available at ScienceDirect

Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth

Clinical Neuroscience

An object location memory paradigm for older adults with and without mild cognitive impairment Nadine Külzow a,b,∗ , Lucia Kerti a , Veronica A. Witte a,b , Ute Kopp a , Caterina Breitenstein c , Agnes Flöel a,b,d,∗∗ a

Department of Neurology, Charité Universitätsmedizin Berlin, Berlin, Germany NeuroCure Cluster of Excellence, Neurocure Clinical Research Center, Charité Universitätsmedizin Berlin, Berlin, Germany c Department of Neurology, University of Münster, Münster, Germany d Center for Stroke Research Berlin, Charité Universitätsmedizin Berlin, Berlin, Germany b

h i g h l i g h t s • • • • •

Development of a simple standardized tool (LOCATO) for object-location memory (OLM). The parallel sets are equivalent within the groups of healthy older adults and MCI. Only small re-tests effects after 6 month without ceiling were revealed. LOCATO detects differences in formation of OLM between healthy older adults and MCI. Systematically assessment of OLM formation and its modulation by adjuvant therapies.

a r t i c l e

i n f o

Article history: Received 18 June 2014 Received in revised form 20 August 2014 Accepted 21 August 2014 Available online 28 August 2014 Keywords: Cognitive mild impairment Ageing Visuospatial memory Associative learning Transcranial direct current stimulation Non-invasive brain stimulation

a b s t r a c t Background: Object-location memory is critical in every-day life and known to deteriorate early in the course of neurodegenerative disease. New method: We adapted the previously established learning paradigm “LOCATO” for use in healthy older adults and patients with mild cognitive impairment (MCI). Pictures of real-life buildings were associated with positions on a two-dimensional street map by repetitions of “correct” object-location pairings over the course of five training blocks, followed by a recall task. Correct/incorrect associations were indicated by button presses. The original two 45-item sets were reduced to 15 item-sets, and tested in healthy older adults and MCI for learning curve, recall, and re-test effects. Results: The two 15-item versions showed comparable learning curves and recall scores within each group. While learning curves increased linearly in both groups, MCI patients performed significantly worse on learning and recall compared to healthy controls. Re-testing after 6 month showed small practice effects only. Comparison with existing methods: LOCATO is a simple standardized task that overcomes several limitation of previously employed visuospatial task by using real-life stimuli, minimizing verbal encoding, avoiding fine motor responses, combining explicit and implicit statistical learning, and allowing to assess learning curve in addition to recall. Conclusions: Results show that the shortened version of LOCATO meets the requirements for a robust and ecologically meaningful assessment of object-location memory in older adults with and without MCI. It can now be used to systematically assess acquisition of object-location memory and its modulation through adjuvant therapies like pharmacological or non-invasive brain stimulation. © 2014 Elsevier B.V. All rights reserved.

Abbreviations: PC, percent correct; NIBS, non-invasive brain stimulation; FR, free recall; CR, cued recall. ∗ Corresponding author at: Charitè-Universitätsmedizin, Department of Neurology, and Neurocure Clinical Research Center, Charitéplatz 1/Bonhoefferweg 3, 10117 Berlin, Germany. Tel.: +49 30 450560365. ∗∗ Corresponding author at: Charitè-Universitätsmedizin, Department of Neurology, and Neurocure Clinical Research Center, Charitéplatz 1/Bonhoefferweg 3, 10117 Berlin, Germany. Tel.: +49 30 450560284. E-mail addresses: [email protected] (N. Külzow), agnes.fl[email protected] (A. Flöel). http://dx.doi.org/10.1016/j.jneumeth.2014.08.020 0165-0270/© 2014 Elsevier B.V. All rights reserved.

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

1. Introduction Remembering the place of an object (object location memory) is crucial for adapting to changing environments in everyday life, an ability known to decline during ageing (e.g., Hedden and Gabrieli, 2004; Kessels et al., 2007) and with an acceleration in pathological conditions like mild cognitive impairment (MCI) or Alzheimer’s disease (AD) (Bucks and Willison, 1997; Petrella et al., 2007; Troyer et al., 2008; Vacante et al., 2013). Due to the continuous increase of the elderly population world-wide, the incidence of age-associated memory impairment may increase dramatically (Bishop et al., 2010; Plassman et al., 2011). Current research focuses on prodromal AD stages like MCI to allow for early intervention with the ultimate goal to delay progression of the disease (Langbaum et al., 2013). In the face of lacking effective pharmacological treatments (Bond et al., 2012; Tricco et al., 2013; Yue et al., 2012), non-pharmacological enhancement including cognitive training (Reijnders et al., 2013; Simon et al., 2012), physical activity (Ruscheweyh et al., 2011), dietary modifications or nutrition supplements (Janssen et al., 2010) as well as techniques like non-invasive brain stimulation (NIBS, Floel, 2014) have gained increasing attention in the treatment of MCI and AD. Importantly, cognitive training may be combined with and boosted by any of the other interventions. A well-defined cognitive training paradigm is therefore paramount not only to enhance cognitive function by training per se, but also to assess the impact of training-adjuvant therapies like NIBS. Moreover, this paradigm should be applicable in both older individuals and in MCI patients in order to test effectiveness of interventions under controlled conditions and to study pathological mechanism. The methodological quality of interventional cognitive training studies differed widely (see also, Papp et al., 2009; Reijnders et al., 2013; Simon et al., 2012). Participation-orientated interventions (generally including not only memory training but additional cognitive domains like attention and executive functions, or acquisition of memory strategies) and their outcome parameters were heterogeneous, rendering critical comparisons between interventions difficult and impeding assessment of training effects within a specific cognitive domain (Lovden et al., 2012). Furthermore, despite its obvious validity for activities of daily living, object location memory has rarely been used in larger training studies in MCI patients. In the current study, our main goal was therefore to develop an ecologically valid paradigm for assessing object-location learning and recall without floor or ceiling effects, which can be applied over several learning sessions in older adults with and without MCI. To achieve this goal, we adapted a previously established objectlocation learning paradigm (LOCATO) developed for healthy older adults (Floel et al., 2012). A shorter (15 instead of 45 items) and thus cognitively less demanding version of LOCATO was designed in two parallel versions. These versions were tested for equivalence in each group, as well as practice effect after 6 months. Moreover, we assessed if the short LOCATO version would be sufficiently sensitive to detect memory differences between healthy older adults and patients with MCI.

17

without non-invasive brain stimulation. Here we will provide only a short overview about the development of long LOCATO (stage I, for more details, see the original publication by Floel et al., 2012) and will then focus on describing specific modifications for a shorter and thus simpler version of the task and its validation in healthy older adults (stage II) and MCI patients (stage III). Data reported here were taken from three interventional studies conducted in our laboratory (two with healthy older adults, one with MCI patients). From these studies, baseline assessments as well as follow-up assessments in the respective “placebo conditions” were employed. All subjects were native German speakers and underwent a medical examination prior to baseline testing. The latter included structural magnetic resonance imaging (MRI) of the brain, several serum-based parameters, and a comprehensive neuropsychological test battery (comprising general intellectual functioning, attention, executive functions, and verbal memory; see e.g., Witte et al., 2013) for a complete description; and see below for details on memory testing). Memory impaired subjects were additionally tested with the cognitive subscale of the Alzheimer’s Disease Assessment Scale (Rosen et al., 1984). Depression was monitored using the Beck’s Depression Inventory (BDI; Hautzinger et al., 2001). The affective state at the time of the testing was assessed with the Positive and Negative Affect Schedule (PANAS, Watson et al., 1988), and handedness was determined by the Edinburgh Handedness Inventory (Oldfield, 1971). Healthy older adults were recruited via advertisements in the internet, local newspapers and the Charité University Hospital intranet in Berlin, Germany. They had to fulfil the following inclusion criteria: (1) no current intake of medication that affects the central nervous system (e.g., antipsychotics or antidepressants); (2) normal routine medical and neurological examinations; (3) no recreational drug use; and (4) no signs of dementia (Mini Mental State Examination of minimal 26 points (Folstein et al., 1975)); (5) no subjective memory complaints; (6) all neuropsychological test results within 1 SD of age/education norms. MCI patients were referred to the study from the local memory clinic of the Charité University Hospital. They fulfilled core clinical criteria for the diagnosis of MCI outlined by Petersen and others (Petersen, 2004; Petersen et al., 2001; Winblad et al., 2004) which did not comprise novel biomarkers as suggested in more recent MCI criteria (Albert et al., 2011). Patients reported subjective memory complaints, which were confirmed by standardized neuropsychological testing using the Consortium to Establish a Registry for Alzheimer’s Disease test battery (CERAD; Memory Clinic Basel, www.memoryclinic.ch) and the Verbal Learning and Memory Test (VLMT, Helmstaedter et al., 2001). All MCI patients had maintained independence and reported minimal if any impairment of function in daily life. A clinical interview, neurological examination, and structural MRI revealed no systemic or brain diseases accounting for declined cognition. Patients diagnosed with amnestic or amnestic plus MCI (in the following referred to as MCI) were included. All subjects received a small reimbursement and gave written informed consent prior to the study. Each part of the study was approved by the Ethics Committee of the Charité University Hospital Berlin, Germany, and was conducted in accordance with the declaration of Helsinki.

2. Materials and methods

2.2. Stages of development

2.1. Overview

2.2.1. Stage I The original version of the LOCATO task comprised the consecutive presentation of a series of 45 buildings (objects) on a street map (location). Associations between an object and a certain location had to be acquired based on the frequency of associations, that is, correct object-location pairings were shown more often than

In an earlier study, we used a standardized computer-based object-location learning task called “LOCATO” (learning of 45 correct positions of buildings (objects) on a street map (locations)) in healthy older adults in a study that assessed learning with and

18

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

incorrect ones over the course of several training blocks. For more details, see the original publication by Floel et al. (2012). 2.2.2. Selection of objects The internet (google picture database) was screened for coloured pictures of real-world buildings. Famous or highly salient objects were excluded. All of the 144 selected pictures were edited to achieve homogenous size and background. From the initial list, 90 buildings were selected based on the results of a rating study (N = 15, 7 woman; unpublished pilot data), in which subjects had to rate the degree of verbalizability, recognizability, complexity, salience, pleasantness and how likely it is that this building come across in a large German city. Thus, the final set of 90 stimuli was pictures of buildings common in larger German cities, rated with average pleasantness, medium saliency and complexity, and which were highly recognizable. To construct two parallel versions, the 90-item list was further divided into two sets with 45 pictures each, and items were matched on every rating scale. 2.2.3. Object-location-assignment We designed a two-dimensional, simplified street map in which streets where represented as grey lines on a white background (see Fig. 1A). Street corners were used as landmarks. The map was portioned into four quadrants to assure an almost uniformly, but random “correct” assignment of each building to one of the 45 street corners. For each building, ten additional positions were selected, which will be referred to as the ‘incorrect’ associations. The following constraints applied to the selection of “incorrect” locations: (i) each position was only used once for the same building, and (ii) the positions are roughly mapped equally across the four quadrants. For the parallel version (B) the same 45 correct positions and their ten alternates were assigned to the second set of buildings and rotated by 180◦ . 2.2.4. Learning Participant’s task was to learn the correct position of a building on the street map by presenting the “correct” position more frequently (for details see Fig. 1A: lower part, left) compared with “incorrect” positions (see Fig. 1A: lower part, right) over the course of five training blocks. Within one block 180 trials (one building on a streetmap at a time) were presented in random sequence on a computer screen for a duration of 3000 ms (see Fig. 1B). Within this time frame, subjects had to press one of two keys on a response pad to indicate if – in their opinion – objects and position matched, or did not match. Subjects were not informed about the underlying statistical principle (increasing frequency of correct object-location) and no online feedback was provided. Dependent variables were accuracy and response speed on a given trial. 2.2.5. Recall Recall was tested using a free-recall (FR) and a forced-choice memory test (cued recall, CR). For the FR-task the empty street map was presented together with one building above the map. Subjects had to move as precise as possible each item to its assumed correct position (Fig. 1B: lower part, left). Only 1 attempt was allowed and the response was scored as correct or incorrect. There were 45 trials (one for each object-location), there was no time limit for a subject’s response and again no feedback was provided. In the CR-task the street map and 3 positions (marked with 1,2,3) for a particular building was shown. Subjects had to indicate by button press which position – in their opinion – is the correct one for that building (Fig. 1B: lower part, right). A total of 90 trials were presented in randomized order. Each correct position was shown twice, distractor positions appeared only once. Distractor positions were counterbalanced across the quadrants as well as the labels (1, 2 or 3) used for correct and incorrect positions. Especially, a

correct position got never the same label for it’s first and second presentation. Again, no time limit for subject’s response was set and no online feedback was provided. 2.3. Stage II: Modification and validation of a shortened version for healthy older adults 2.3.1. Shortened LOCATO paradigm “short LOCATO” The shortened LOCATO task included two main modifications from the originally developed “long version” (Floel et al., 2012): First, we reduced the number of correct building-positionassociations from 45 to 15 items. Therefore, from the original 90 buildings (two lists of 45-items) a subset of 30 buildings was selected and divided into two lists of 15-items each (version A and B). Subsets were generated with respect to the rating results of the 90 buildings. Second, an additional explicit learning instruction (“try to find out and remember the right place of each building”) was given and the underlying statistical principal was explained. This modification was included because (i) everyday memory formation is a hybrid between explicit and implicit memory processes, (ii) this instruction may be helpful to form novel memories, particularly in patients with MCI, see below, and (iii) to avoid uncontrollable influences of some healthy individuals or MCI becoming aware of the inherent statistical principal, as seen during previous pilot testing of repeated sessions of LOCATO (unpublished observations). In detail, a total of 300 training trials (five blocks of 60 trials, thereof 30 correct and 30 incorrect presentations within each block) were presented in randomized order. Each trial comprised a picture of a schematized street map with one building presented for a duration of 3000 ms. An interstimulus interval of 1000 ms was added to capture slow responses as well. Within this time frame (that is, during the 4000 ms), subjects had to press one of two keys on a response pad to indicate as accurate as possible if – in their opinion – object and position matched, or did not match (button “correct” or “incorrect”). No online feedback on performance was provided. Recall performance was also measured by free (FR) and cued recall tests (CR; no time limit to give an answer). In the modified version, participants were asked to point to the assumed position of a presented building in the FR task, instead of dragging the building to its correct position as in the original version. The position was then documented by the investigator. 2.3.2. Cross-sectional dataset (healthy older adults; testing comparability between version A and B) Data were taken from a large cohort study comprising 141 healthy older adults performing LOCATO (Kerti et al., 2013). Sixty percent performed version A, 38 percent version B (2% drop outs unable to perform any version) at baseline. To obtain reliable data for task validation and to derive groups of identical sample sizes for versions A and B the sample was reduced stepwise according to the following criteria: First, only subjects with less than 5% of non-responses in the LOCATO-task were chosen (n = 111), and second, the remaining subjects were matched (across versions) for sex, age and years of education. Thus, data of 76 healthy older adults between 50 and 75 years (mean 62 ± 6.4 years; 42 women) and formal years of education between 10 and 25 years (mean 16.9 ± 3.6 years) were used in Stage II. No significant differences between the groups performing versions A and B were found with regard to other relevant neuropsychological test scores or self-reported affect or depression. 2.3.3. Longitudinal dataset (healthy older adults; testing practice effects after repeated application) Data of the placebo arms of two interventional studies of our lab involving healthy older adults were used for analysis. Participants of both studies had undergone an identical neuropsychological

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

19

Fig. 1. Schematic of the associative object-location learning paradigm. (A) Two parallel versions of the object-location learning paradigm were used, each with a different set of 45 buildings, and with the street map rotated for 180◦ for version B. For each building one correct and 10 incorrect positions for this particular building were assigned. (B) Within one block a particular building was shown four times on the street map (twice on the correct and twice on two different incorrect positions). Each trial comprised a picture of a schematized street map with 1 building. After each stimulus presentation subjects had to indicate by button press whether a building was in a “correct” location. For each subject there were a total of 900 training trials (five blocks of 180 trials with 90 correct and 90 incorrect presentations, respectively). Over the course of five training blocks the “correct” pairing position of a building (total of 45 buildings) occurred 10 times more frequently compared with “incorrect” positions (shown only once, respectively). Immediately after the training blocks, learning success was tested with a free recall (FR) task. Here, an interactive presentation program was used, where the street map was shown in the lower part of the picture and the building in the upper part, allowing subjects to drag each item to its assumed correct position on the map by using the computer mouse. The FR task was followed by a further cued recall test (CR). Here, the street map was shown in the lower part of the picture and the building in the upper part. Three possible locations for this particular building (marked with “1”, “2” or “3”) were shown on the street map and subjects had to indicate which number corresponded to the building’s “correct” position.

screening, and LOCATO was tested at baseline and 6 months later (N = 30; 21 subjects from the cross-sectional dataset described above). Healthy older subjects were between 52 and 75 years old (mean 62.2 ± 6 years; 18 women). Years of education varied between 9 and 25 years (mean 16.2 ± 3.1 years). Subjects performing versions A and B did not differ significantly on neuropsychological test scores except for scores on the “Trail making test Part A” (Tombaugh, 2004), P = 0.05). 2.3.4. Stage III: Application of short LOCATO to patients with MCI 34 MCI patients performed the same LOCATO version with 15 correct-building-positions as described above. From the original

cohort, 4 subjects had to be excluded because of a high number of non-responses (n = 2) or an extreme response bias (n = 2) in the LOCATO learning task. Thus, 30 MCI patients were included (age range 50 – 79 years, mean 67.97 ± 7.65, formal years of education between 8 and 20 years, mean 14.82 ± 3.17; 11 women). Fourteen MCI patients performed version A (5 women) and 16 MCI patients version B (6 women). Patients performing versions A and B did not differ significantly with regard to age, education, neuropsychological test scores, self-reported affect or depression. To compare performance of impaired to the performance of healthy older adults, samples were matched with regard to age, sex, education and LOCATO version A and B (N = 60; for matching results

20

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

Table 1 Means of demographic data and neuropsychological performance of selected domains in MCI (n = 30) and matched healthy older adults (n = 30) (SD in parentheses unless stated otherwise). Matched healthy older adults

MCI

n (male)

30 (19)

30 (19)

Demographics Age (years) Education (years)

65.40 (6.95) 15.35 (3.39)

67.97 (7.65) 14.82 (3.17)

General abilities MMSE

28.80 (1.38)

28.97 (0.93)

Long-term episodic memory (VMLT) Sum score (list 1–5) Delayed retrieval Recognition (hits) Recognition (adjusted for false positive)

49.97 (9.95) 9.47 (3.58) 13.42 (1.63) 10.77 (3.79)

43.50 (8.50) 7.13 (3.22) 12.97 (1.75) 9.30 (5.87)

Other cognitive domains Digit span forward Digit span backwards Trail making test Part A (s) Trail making Test Part B (s)a Verbal Fluency words (GR) Verbal Fluency category (sport-fruits) MWT

7.60 (1.99) 6.03 (1.87) 37.55 (12.46) 92.36 (48.55) 13.87 (3.41) 14.87 (3.22) 31.97 (2.59)

Mood scales PANAS positivea PANAS negative BDIa

34.14 (6.69) 12.48 (3.86) 6.69 (6.71)

t(58)

2.71** 2.65*

7.87 (1.36) 6.33 (1.49) 42.13 (15.68) 104.80 (40.16) 12.37 (4.05) 14.77 (3.86) 31.73 (3.29) 31.77 (5.74) 13.04 (3.32) 10.74 (6.71)

−2.26*

*

P < 0.05. P < 0.01. Independent t-test between healthy older adults and MCI: only significant t-values are reported; MMSE: Mini Mental State Examination scores (Folstein et al., 1975), VLMT: verbal learning and memory test (Helmstaedter et al., 2001), Digit span (Härting et al., 2000) Trail Making Test (TMT; Army Individual Test Battery, 1944; normative data e.g. in Tombaugh, 2004), Verbal Fluency: Regensburger Verbal Fluency Test (Aschenbrenner et al., 2000), MWT: Vocabulary Test (Lehrl, 2005), PANAS: positive and negative affective scale (Watson et al., 1988), BDI: Becks depression inventory (Hautzinger et al., 2001). a N reduced because of missing data in PANAS (MCI: n = 26; healthy controls: n = 29) and BDI (MCI: n = 27; healthy controls: n = 29). **

see details in Table 1). As expected MCI patients demonstrated selective deficits in several cognitive domains as compared to controls, including long-term memory as assessed by a verbal memory task (sum score of word list 1–5, and delayed memory score of the VLMT). Further, the MCI group scored significantly higher in the depression scale (self-reported) compared to the healthy control group. The groups did not differ with respect to reported positive and negative affects. MCI patients subsequently participated in an interventional study in our lab. Cognitive tests were repeated after 6 month. Follow up testings are still ongoing and therefore not presented here. 2.4. Statistical analysis 2.4.1. Data aggregation 2.4.1.1. Learning. Learning performance for each training block was assessed on the basis of hits and correct rejections. The index (percent correct, PC) was computed as follows: PC = ((number of hits + number of correct rejections)*100)/total number of buildings presented within a training block). Moreover, an index for response bias was calculated on the basis of “yes” responses as follows: response bias = (number of hits + number of false alarms)/total number of buildings presented within a training block). This score varied between 0 (tendency to respond with “no” to every stimulus) and 1 (tendency to respond with “yes” to every stimulus), thus a value of 0.5 reflects an “unbiased” response. Mean reaction times (RT) to correct responses were determined by averaging the individual medians for hits and correct rejections per training block. 2.4.1.2. Recall. On the FR task (15 runs), the number of correctly placed buildings, as a percentage of the total number of buildings that could have been placed correctly, was taken (correctly buildings*100/total number of buildings that had to be placed). Minor

deviations in the assignments to the precise position were tolerated and were counted as correct. In addition, buildings not clearly attributable to a street corner were rated as missing data. Analyses were always conducted by trained rater. On the CR task (same task structure as for the long version, with a total of 30 (2 × 15) trials), the percentage of correctly selected positions was calculated. 2.4.2. Data analysis 2.4.2.1. Stage II: Evaluation of 15-item version in healthy older adults (comparability and practice effects). Learning performance (PC, RT), and response bias were analyzed by using repeated measures analyses of variance (ANOVA) on PC to determine the learning slope of the learning curve (temporal aspects) in more detail. Comparability between versions A and B was tested by a version (A, B) as between- and block (1–5) as within-subject factors ANOVA. Performance in free and cued recall tasks was compared by independent t-tests. To test for practice effects, data of the placebo arm group were analyzed in a block (1–5) × time (baseline, follow up) × version (A, B) ANOVA along with post hoc t-tests. Recall (CR only) was compared by a time (baseline, follow-up) × version (A, B) ANOVA. 2.4.2.2. Stage III: Application of the 15-item version in MCI patients. Before data of version A and B were pooled comparability between these two versions were tested in MCI patients by a version (A, B) × block (1–5) ANOVA. Learning curves were analyzed by an ANOVA with the factor block. The factor, group (MCI vs. healthy older adults) was then incorporated as a between-subject factor to compare cognitively impaired vs. healthy older adults. Group differences in recall performance were analyzed by independent t-test. Sex was entered as an additional between-subject factor in control ANOVAs.

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

21

Fig. 3. Performance at baseline and follow-up visit in 15-item LOCATO. Percent correct scores (circles, solid lines) and reaction times (triangle, dashed lines) in LOCATO learning task (version with 15 items) at baseline and at six month follow-up testing over the five training blocks. At both test sessions performance increased and reaction times decreased across blocks. Note that no significant differences emerged between baseline and follow-up testing at the beginning and at the end of five training blocks. Error bars represent standard errors of the mean. **Significance level P < 0.01, (*) P < 0.10. For readability, significant increases across training blocks are not indicated by asterix in this figure.

Fig. 2. Performance in 15-item LOCATO. Learning: version A vs. B. Accuracy (2A) and reaction times (2B) in LOCATO learning task A and B (version with 15 items) in healthy older adults. Fig. 2A represent percent correct scores, Fig. 2B reaction times over the five training blocks. Both versions showed a significant linear slope of learning from block 1 to 5 with significant improvements between blocks (except from block 3 to 4) in accuracy (2A) and significant decrements in reaction times (2B) between blocks. Fig. 2C represents the course of response bias over the five training blocks. In Fig. 2D recall performance of version A vs. B is shown: Cued and free recall performance in percentage after learning for version A and B of the LOCATO task (15item version) in healthy older adults. Note that no significant differences emerged between versions A and B for any of the indices. Error bars represent standard errors of the mean. ***Significance level P < 0.001, *significance level P < 0.05 (for within group comparisons).

Significance level ˛ was set to P < 0.05 in all analyses. Reported effect size is partial 2 (subsequently shortened to 2 ). Results of two-tailed t-tests are presented unless reported otherwise. All statistical analyses were conducted using SPSS (Statistical Package for the Social Sciences for Windows, SPSS Inc. USA). 3. Results

75 percent correct answers (mean 75.26 ± 9.99%) at the end of five training blocks. The learning slopes were best described by a linear trend (polynomial contrast analysis: F(1,74) = 381.08, P < 0.001, 2 = 0.84). RT significantly decreased over the five training blocks (block: F(4,296) = 76.25, P < 0.001, 2 = 0.51, see Fig. 2B), but there was no effect for the factor version. Response behaviour showed no extreme response tendencies (see Fig. 2C). The corresponding ANOVA revealed no significant effects (all F < 1). In the recall tests (CR and FR; see Fig. 2D) none of the t-tests revealed a significant difference between versions A and B (all P > 0.17). Sex was not counterbalanced across version. Control analyses revealed no significant sex differences with regard to learning curve (main and interaction effects: all P > 0.19), recall (all P > 0.21) or response behaviour (all P > 0.08). In sum, healthy older adults showed significant improvements over five training blocks with an average performance of 75 percent correct. The results were highly comparable to the learning curves of the original 45-item LOCATO paradigm (on average 70 percent correct responses on the last training block). No systematic differences were observed between versions A and B or between men and women (within each version) with regard to learning curves and recall performance.

3.1. Stage I Data of the long LOCATO version (analyses and results) have been presented elsewhere (see Floel et al., 2012). In short, healthy older subjects showed smooth learning curves over the five training blocks, achieving a performance of around 70 percent of correct responses on the last training block. 3.2. Stage II: Results of evaluation of 15-item version in healthy older adults 3.2.1. Comparability between version A and B The version × block ANOVA revealed no significant effects of the factor version for PC (interaction F < 1; main effect: F(1,74) = 2.49, P = 0.12, 2 = 0.03) or reaction time (all F < 1) indicating comparable learning curves in versions A and B across the five consecutive training blocks. The significant effect of block on PC (F(4,296) = 149.94, P < 0.001, 2 = 0.67) suggests performance differences between blocks independent of version (see Fig. 2A). Performance significantly increased from chance level (mean 53.42 ± 6.63%) to over

3.2.2. Practice effects (baseline/follow-up analysis) The block × time × version ANOVA on learning data revealed two significant main effects, while all other effects failed to reach statistical significance (all P ≥ 0.10). Significant effects of block (PC: F(4,112) = 118.94, P < 0.001, 2 = 0.81, RT: F(4,108) = 76.72, P < 0.001, 2 = 0.74) and time (PC: F(1,28) = 4.84, P = 0.04, 2 = 0.15, RT: F(1,27) = 4.68, P = 0.04, 2 = 0.15) indicated substantial learning across training blocks and improved performance over time. As can be seen in Fig. 3 (solid lines) time effect in performance (PC) is mainly driven by a significant difference in the learning slope at training block 4 (t(29) = −2.18, P = 0.04; all other P > 0.12). Importantly, the learning curve showed that performance started around chance level at the first training block and improved to a similar performance level at the fifth training block at baseline (block 1: mean 52.33 ± 7.15%, block 5: mean 73.5 ± 7.22%) as well as at the six month follow-up visit (mean 54 ± 5.95; block 5: mean 76.17 ± 8.27%). RT curves were also similar at baseline and followup testing as indicated in Fig. 3 (dashed lines), but overall subjects responded somewhat faster at the sixth-month follow-up session

22

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

compared to baseline testing. Further, no significant effects were found for response bias (all P > 0.12) and recall performance (all P ≥ 0.09). In sum, some changes of the performance (PC) slope, that is, within training blocks, were observed with repeated testing and RT decreased on average over time. However, performance on the fifth training block and recall scores were relative stable over time and comparable at the baseline and follow-up visit without reaching ceiling effects. 3.3. Stage III: Results of the 15-item version in MCI patients 3.3.1. MCI patients Analyses comparing versions A and B revealed comparable performance (no significant main or interaction effects of the factor version with regard to learning curve or response bias (all F < 1)). Also, recall performance did not significantly differ between versions (t-test: all P > 0.23). For subsequent analyses data of versions A and B were pooled. Analysis of learning performance revealed a main effect of block on PC (F(4,116) = 27.15, P < 0.001, 2 = 0.48) and RT (F(4,116) = 31.14, P < 0.001, 2 = 0.52) indicating a statistically significant increment in mean performance (see Fig. 2, dashed line) and a decrement in RT, respectively, over the five training blocks. Learning curves were again best described as linear trend (polynomial contrast analyses: F(1,29) = 29.03, P < 0.001, 2 = 0.77). In contrast to the healthy older subjects, a sex effect was observed, with men (mean 60.84 ± 9.39%) performing better than women (mean 53.73 ± 9.59%; F(1,28) = 6.72, P = 0.02, 2 = 0.19). This might have been due to the higher age of women (age range: 67–79 years) compared to men (age range: 50–79 years) in this sample. Response bias varied within a tolerable range of 0.4 and 0.48 and showed no significant response tendencies across the blocks (all P > 0.08). Recall performance yielded a mean of 36.78 ± 14.21 (N = 29) percent correct for free recall and a mean of 61.31 ± 11.84 (N = 28) for cued recall (both better than chance level) (Fig. 4). 3.3.2. MCI vs. healthy older adults The ANOVA comparing learning performance (PC and RT) between healthy older adults and MCI patients revealed significant main effects of block (PC: F(4,232) = 73.59, P < 0.001, 2 = 0.56; RT: F(4,232) = 46.69, P < 0.001, 2 = 0.45) and group (PC: F(1,58) = 14.33, P < 0.001, 2 = 0.20; RT: F(1,58) = 23.98, P < 0.001, 2 = 0.29) indicating linear learning slopes for either group (see Fig. 2). Compared to matched healthy older adults, MCI patients performed on average significantly worse and slower. A separate ANOVA was conducted for type of correct response (hits vs. correct rejections), showing that difficulties in MCI are mainly due to recognizing correct locations (hits; group: F(1,58) = 9.52, P < 0.01, 2 = 0.14) and not to rejecting incorrect locations (correct rejections; group: n.s). The respective ANOVA for RT yield for both types increased RT for MCI compared to healthy controls (all P < 0.001). Within a group faster responses to hits compared to correct rejections as commonly observed in other studies (e.g., Ecker et al., 2007; Rubin et al., 1999) were evident only for healthy older controls (hit: mean 1564 ± 246 ms, correct rejections: mean 1690 ± 212 ms; F(1,29) = 19.05, P < 0.001, 2 = 0.40), but not for MCI patients (hit: mean 1925 ± 293 ms, correct rejections: mean 1932 ± 281 ms; F < 1) as indicated by an ANOVA of factor type on each group. Further, no extreme response tendencies were observed. MCI patients showed a trend to respond slightly more conservative, that is to respond rather with “no” than with “yes” to stimuli, compared to healthy subjects (marginal significant main effect: F(1,58) = 3.41, P = 0.07, 2 = 0.06). Recall performance differed only with respect to the CR-task. Here, MCI patients performed significantly worse than healthy older subjects (MCI: mean 61.31 ± 11.84 percent correct; healthy: mean 71.44 ± 12.59 percent

Fig. 4. Performance differences in 15-item LOCATO. Percent correct scores of learning (4A) and reaction times (4B) in LOCATO task of healthy older adults (n = 30) and patients with MCI (n = 30). Over the five training blocks both groups showed significant linear learning slopes, but with significant lower mean performance (Fig. 4A) and slower reaction times (Fig. 4B) of MCI patients compared to healthy older adults. Fig. 4C represents the course of response bias over the five training blocks for healthy older subjects and MCI patients. In Fig. 4D recall performance is shown: Cued and free recall performance in percentage after learning for healthy older subjects and MCI patients in the LOCATO task (15-item version). Note that significant differences emerged only in the cued but not in the free recall task. Error bars represent standard errors of the mean. ***Significance level P < 0.001, **P < 0.01, (*) P < 0.10 (for betweengroup comparisons). Note that both groups showed a linear increase of learning from block 1 to 5 with significant improvements between blocks (except from block 3 to 4 in older adults); but for readability, significant increases (within group comparisons) across training blocks are not indicated by asterix in this figure.

correct, t(56) = 3.15, P = 0.003; FR: t(55) < 1). Since MCI scored significantly higher in depression than healthy older adults, additional control analyses of covariance (ANCOVA) were done with BDIscore as covariate. However, all reported effects above remained significant. In sum, MCI patients showed smooth learning curves over the five training blocks. However, the learning curve was more shallow compared to healthy older adults. Compared to healthy controls, MCI had more problems in correctly identifying the right place of an object (reduced number of hits) and did not show a typically observed reaction time advantage for hits compared to correct rejections. Also, recall performance (CR) discriminated well between healthy and cognitively impaired older subjects. 4. Discussion The present study demonstrates the validity of a shortened and cognitively less demanding version of LOCATO in both healthy older adults and patients with MCI. Overall, healthy older adults were able to learn the correct position of an object across five training blocks through simple statistical associations, without reaching ceiling effects. Versions A and B were comparable with regard to learning curve and recall performance. Re-testing after 6 month showed small practice effects without reaching a ceiling effect. Moreover, MCI patients were likewise able to acquire object-location memory during the task with no significant differences between versions A and B, and produced smooth learning

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

curves over five training blocks. However, compared to healthy older adults, learning ability was significant lower in MCI patients, as indicated by lower accuracy as well as slower reaction times as compared to healthy controls. Subsequent recall showed significantly impaired cued recall memory in MCI patients compared to healthy older adults. The longer version of LOCATO (comprising 45 items; see Floel et al., 2012) was developed for use in healthy subjects. With this 45-item version significant learning and recall could be demonstrated in healthy older adults. Here, learning after five training blocks reached on average 70% correctly remembered items. Thus, learning curves were well-suited to assess the effects of adjuvant interventions, but free recall performance was rather low (29% correct answers). Turning to patients with MCI, known to show deficits particularly in formation of object-location memory (Bucks and Willison, 1997; Petrella et al., 2007; Troyer et al., 2008; Vacante et al., 2013) and recall, the 45-item version of LOCATO was therefore deemed inadequate. Importantly, it is known that too challenging tests can induce frustration (Bühner, 2011; Suen and McClellan, 2003). Further the long test format might cause fatigue. Both frustration and fatigue can mask true memory deficits (Wesnes and Harrison, 2003). Thus, we set out to develop and validate a modified shorter and simpler version of LOCATO without producing floor (MCI) or ceiling (healthy older adults) effects. At present it was not intended to test psychometric properties of this simple version, but rather to evaluate the usefulness of LOCATO-15 in training and assessing associative visuospatial memory in MCI and healthy older adults, including any extensive practice effects for re-testing after 6 months. Our data in healthy older adults demonstrated consistence of performance (learning curve and recall) for both forms of short LOCATO, with significant gains over the course of five training blocks without ceiling effects. On 6-month followup, practice effects were found for learning slopes, but this did not significantly affect performance at the end of the training sessions (at the fifth training block) and subsequent recall. This might be due to experience-based task-learning strategies which may facilitate encoding within a training session but which will not affect overall performance at the end of the training or on subsequent recall. Note that test and re-test were spaced by several months and larger practice effects cannot be excluded after shorter delays. However, in general long-term compared to short-term effects of interventions are of higher clinical interests, rendering re-assessments after long delays most relevant. Further, a more comprehensive training/familiarization with the task before testing may dilute practice effects and may provide a more reliable measure of change. In sum, apart from small practice effects during re-test sessions after 6 month, the short LOCATO form with its two equivalent versions constitutes an ideal scenario for experimental manipulations (pharmacological or non-pharmacological) or for monitoring cognitive changes over time. Turning to MCI patients, we found preserved associative visuospatial learning and retention performance beyond chance level, but performance was significantly lower as compared to healthy older controls. This pattern is in line with neuroanatomical findings reporting the first volumetric loss in the hippocampus and adjacent structures in patients with MCI, regions which are also strongly involved in associative learning (Achim et al., 2007; Mayes et al., 2007; Qin et al., 2009; Squire et al., 2004; Troyer et al., 2008), including object-location binding (Kessels et al., 2007; Postma et al., 2008). Differences in recall compared to healthy controls were shown in cued, but not in free recall. This finding seems surprising at first glance, given previous reports showing that controlled effortful memory processes are more affected in MCI (e.g., Anderson et al., 2008) than less effortful memory processes (e.g., familiarity effects). Moreover, cued recall has also been shown to differentiate between MCI and healthy controls in earlier studies

23

(Bennett et al., 2006), and in fact involves controlled recollection in addition to familiarity processes (Westerberg et al., 2006). Thus, the short version of LOCATO seems to be well-adapted to characterize differences in associative visuospatial memory (learning curve and cued recall) between normal and pathological ageing. To date, visuospatial memory has been assessed by various tasks, including mental rotation, visual span, navigational, visual, or visuo-constructive tasks (Iachini et al., 2009). Compared to these tasks LOCATO offers several advantages. First, it does not require drawing abilities, fine motor responses, mental imagery, language, or literacy. For example, LOCATO uses irregular grids in the street map in order to minimize verbal encoding strategies that often help to solve other “visual tasks” (e.g., “Location Learning Test”; Bucks and Willison, 1997, “The Placing Test”; Vacante et al., 2013). Second, even though LOCATO does not use real-life scenarios, it does use real-life stimuli. Thus, it combines the advantage of real-life stimuli with a controlled paradigm that focuses on assessing the association of objects to locations, with known hippocampal involvement of such binding processes (e.g., Kessels et al., 2007; Postma et al., 2008). At the same time it avoids the use of real-life contexts that necessitate complex cognitive operations, like egocentric processes, orientation, mental imagery, planning and executing of movements which probably involves widely distributed cerebral circuits (Weniger et al., 2011) and which might be differentially affected by ageing (e.g., Iachini et al., 2009). Third, since learning in daily life often depends on both implicit and explicit processes (e.g., Kessels et al., 2005) LOCATO was designed to comprise both types of features. Implicit learning is built on a statistical learning procedure facilitating intuitive learning by a cumulative increase of correct associations across several training blocks (Breitenstein and Knecht, 2002; Floel et al., 2008). This principle was explained to participants before the task. Given that the most profound decline in memory function in healthy ageing (Schacter and Tulving, 1994) and more so in AD (Kessels et al., 2005; Kuzis et al., 1999; Lustig and Buckner, 2004) is seen in explicit learning, a combined task might convey several advantages: it is conceivable that statistical information might be implicitly used to establish a labile representation of object-location-associations, which might be then stabilized by explicit knowledge that helps guide attention and promote more elaborate encoding by the use of specific memory strategies (Belleville et al., 2008). Fourth, LOCATO provides not only information on subsequent and/or delayed recall as seen in previous tests (e.g., “Room task” by Caldwell and Masson (2001) or “The placing test”; Vacante et al., 2013), but allows to assess several aspects of learning including temporal information (learning curve) along with a more refined analysis of correct responses including hits and correct rejections. In fact, we found that MCI and healthy controls differed notably in hits compared to rejections, which could not be explained by differences in response tendencies. Since hits and correct rejections involve distinct cognitive processes and brain activity (e.g., Yonelinas, 2002), these analyses might provide further information on the process of memory breakdown in neurodegenerative disease. 5. Conclusion Here we demonstrated for the first time that learning and recall of object-location associations can be systematically assessed in both healthy older adults and MCI patients by a simplified, shortened version of a standardized paradigm previously developed in our group. Given that the two parallel forms of the shortened paradigm showed comparable results within groups, and allowed for re-testing after 6 months with only small practice effects and without ceiling effects, it provides an optimal basis to assess beneficial effects of training-adjuvant pharmacological and

24

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25

non-pharmacological interventions in controlled experimental settings. Additionally, LOCATO requires simple non-verbal responses, which allows the application in cross-cultural studies. Finally, it constitutes a low-cost intervention that may even be applied at patients’ home. Future research should also be directed towards detecting its sensitivity in monitoring progression to AD. Funding This work was supported by grants from the Deutsche Forschungsgemeinschaft (Fl 379-8/1; 379-10/1, 379-11/1, DFGExc-257 NeuroCure), the Bundesministerium für Bildung und Forschung (Grants FKZ0315673A, 01GY1144, and 01EO0801), the Else-Kröner Fresenius Stiftung (Grants 2009-141 and 2011-119). Acknowledgement We thank Henrike M. Herrmannstädter and Angela Winkler for help with data acquisition. References Achim AM, Bertrand MC, Montoya A, Malla AK, Lepage M. Medial temporal lobe activations during associative memory encoding for arbitrary and semantically related object pairs. Brain Res 2007;1161:46–55. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:270–9. Anderson ND, Ebert PL, Jennings JM, Grady CL, Cabeza R, Graham SJ. Recollectionand familiarity-based memory in healthy aging and amnestic mild cognitive impairment. Neuropsychology 2008;22:177–87. Aschenbrenner A, Tucha O, RWT KL. Regensburger Wortflüssigkeits-Test. Hogrefe: Göttingen; 2000. Belleville S, Sylvain-Roy S, de Boysson C, Menard MC. Characterizing the memory changes in persons with mild cognitive impairment. Prog Brain Res 2008;169:365–75. Bennett IJ, Golob EJ, Parker ES, Starr A. Memory evaluation in mild cognitive impairment using recall and recognition tests. J Clin Exp Neuropsychol 2006;28:1408–22. Bishop NA, Lu T, Yankner BA. Neural mechanisms of ageing and cognitive decline. Nature 2010;464:529–35. Bond M, Rogers G, Peters J, Anderson R, Hoyle M, Miners A, et al. The effectiveness and cost-effectiveness of donepezil, galantamine, rivastigmine and memantine for the treatment of Alzheimer’s disease (review of Technology Appraisal No. 111): a systematic review and economic model. Health Technol Assess 2012;16:1–470. Breitenstein C, Knecht S. Development and validation of a language learning model for behavioral and functional-imaging studies. J Neurosci Methods 2002;114:173–9. Bucks RS, Willison JR. Development and validation of the Location Learning Test (LLT): a test of visuo-spatial learning designed for use with older adults and in dementia. Clin Neuropsychol 1997;11:273–86. Bühner M. Einführung in die Test- und Fragebogenkonstruktion. Pearson Studium; 2011. Caldwell JI, Masson ME. Conscious and unconscious influences of memory for object location. Mem Cognit 2001;29:285–95. Ecker UKH, Zimmer HD, Groh-Bordin C. Color and context: an ERP study on intrinsic and extrinsic feature binding in episodic memory. Mem Cognit 2007;35:1483–501. Floel A. tDCS-enhanced motor and cognitive function in neurological diseases. NeuroImage 2014;85(3):934–47. Floel A, Rosser N, Michka O, Knecht S, Breitenstein C. Noninvasive brain stimulation improves language learning. J Cogn Neurosci 2008;20:1415–22. Floel A, Suttorp W, Kohl O, Kurten J, Lohmann H, Breitenstein C, et al. Non-invasive brain stimulation improves object-location learning in the elderly. Neurobiol Aging 2012;33:1682–9. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–98. Härting C, Markowitsch H, Neufeld H, Calabrese P, Dejerine J, Deisinger K, et al. Gedächtnistest – Revidierte Fassung; WMS-R. Deutsche Adaptation der revidierten Fassung der Wechsler memory Scale. Hogrefe: Göttingen; 2000. Hautzinger M, Bailer M, Worall H, Keller F. Beck-Depressions-Inventar (BDI). Testhandbuch. Hans Huber: Bern; 2001. Hedden T, Gabrieli JD. Insights into the ageing mind: a view from cognitive neuroscience. Nat Rev Neurosci 2004;5:87–96. Helmstaedter C, Lendt M, Lux S. Verbaler Lern- und Merkfähigkeitstest (VLMT). Beltz: Göttingen; 2001.

Iachini I, Iavarone A, Senese VP, Ruotolo F, Ruggiero G. Visuospatial memory in healthy elderly, AD and MCI: a review. Curr Aging Sci 2009;2:43–59. Janssen IM, Sturtz S, Skipka G, Zentner A, Velasco Garrido M, Busse R. Ginkgo biloba in Alzheimer’s disease: a systematic review. Wien Med Wochenschr 2010;160:539–46. Kerti L, Witte AV, Winkler A, Grittner U, Rujescu D, Floel A. Higher glucose levels associated with lower memory and reduced hippocampal microstructure. Neurology 2013;81:1746–52. Kessels RP, Feijen J, Postma A. Implicit and explicit memory for spatial information in Alzheimer’s disease. Dement Geriatr Cogn Disord 2005;20:184–91. Kessels RP, Hobbel D, Postma A. Aging, context memory and binding: a comparison of “what, where and when” in young and older adults. Int J Neurosci 2007;117:795–810. Kuzis G, Sabe L, Tiberti C, Merello M, Leiguarda R, Starkstein SE. Explicit and implicit learning in patients with Alzheimer disease and Parkinson disease with dementia. Neuropsychiatry Neuropsychol Behav Neurol 1999;12:265–9. Langbaum JB, Fleisher AS, Chen K, Ayutyanont N, Lopera F, Quiroz YT, et al. Ushering in the study and treatment of preclinical Alzheimer disease. Nat Rev Neurol 2013;9:371–81. Lehrl S. Mehrfachwahl-Wortschatz-Intelligenztest MWT-B. Balingen: Spitta Verlag; 2005. Lovden M, Schaefer S, Noack H, Bodammer NC, Kuhn S, Heinze HJ, et al. Spatial navigation training protects the hippocampus against age-related changes during early and late adulthood. Neurobiol Aging 2012;33:620.e9–22. Lustig C, Buckner RL. Preserved neural correlates of priming in old age and dementia. Neuron 2004;42:865–75. Mayes A, Montaldi D, Migo E. Associative memory and the medial temporal lobes. Trends Cogn Sci 2007;11:126–35. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97–113. Papp KV, Walsh SJ, Snyder PJ. Immediate and delayed effects of cognitive interventions in healthy elderly: a review of current literature and future directions. Alzheimers Dement 2009;5:50–60. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med 2004;256:183–94. Petersen RC, Doody R, Kurz A, Mohs RC, Morris JC, Rabins PV, et al. Current concepts in mild cognitive impairment. Arch Neurol 2001;58:1985–92. Petrella JR, Prince SE, Wang L, Hellegers C, Doraiswamy PM. Prognostic value of posteromedial cortex deactivation in mild cognitive impairment. PLoS ONE 2007;2:e1104. Plassman BL, Langa KM, McCammon RJ, Fisher GG, Potter GG, Burke JR, et al. Incidence of dementia and cognitive impairment, not dementia in the United States. Ann Neurol 2011;70:418–26. Postma A, Kessels RP, van Asselen M. How the brain remembers and forgets where things are: the neurocognition of object-location memory. Neurosci Biobehav Rev 2008;32:1339–45. Qin S, Rijpkema M, Tendolkar I, Piekema C, Hermans EJ, Binder M, et al. Dissecting medial temporal lobe contributions to item and associative memory formation. NeuroImage 2009;46:874–81. Reijnders J, van Heugten C, van Boxtel M. Cognitive interventions in healthy older adults and people with mild cognitive impairment: a systematic review. Ageing Res Rev 2013;12:263–75. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer’s disease. Am J Psychiatry 1984;141:1356–64. Rubin SR, Petten CV, Glisky EL, Newberg WM. Memory conjunction errors in younger and older adults: event-related potential and neuropsychological data. Cogn Neuropsychol 1999;16:459–88. Ruscheweyh R, Willemer C, Kruger K, Duning T, Warnecke T, Sommer J, et al. Physical activity and memory functions: an interventional study. Neurobiol Aging 2011;32:1304–19. Schacter DL, Tulving E. What are the memory systems of 1994? In: Schacter DL, Tulving E, editors. Memory Systems 1994. Cambridge, MA: MIT; 1994. p. 1–38. Simon SS, Yokomizo JE, Bottino CM. Cognitive intervention in amnestic Mild Cognitive Impairment: a systematic review. Neurosci Biobehav Rev 2012;36: 1163–78. Squire LR, Stark CE, Clark RE. The medial temporal lobe. Annu Rev Neurosci 2004;27:279–306. Suen HK, McClellan S. Item construction principles and techniques. In: Huang N, editor. Encyclopedia of vocational and technological education. Taipei: ROC Ministry of Education; 2003. p. 777–98. Tombaugh TN. Trail Making Test A and B: normative data stratified by age and education. Arch Clin Neuropsychol 2004;19:203–14. Tricco AC, Soobiah C, Berliner S, Ho JM, Ng CH, Ashoor HM, et al. Efficacy and safety of cognitive enhancers for patients with mild cognitive impairment: a systematic review and meta-analysis. CMAJ 2013;185:1393–401. Troyer AK, Murphy KJ, Anderson ND, Hayman-Abello BA, Craik FI, Moscovitch M. Item and associative memory in amnestic mild cognitive impairment: performance on standardized memory tests. Neuropsychology 2008;22: 10–6. Vacante M, Wilcock GK, de Jager CA. Computerized adaptation of The Placing Test for early detection of both mild cognitive impairment and Alzheimer’s disease. J Clin Exp Neuropsychol 2013;35:846–56. Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Pers Soc Psychol 1988;54: 1063–70.

N. Külzow et al. / Journal of Neuroscience Methods 237 (2014) 16–25 Weniger G, Ruhleder M, Lange C, Wolf S, Irle E. Egocentric and allocentric memory as assessed by virtual reality in individuals with amnestic mild cognitive impairment. Neuropsychologia 2011;49:518–27. Wesnes KA, Harrison JE. The evaluation of cognitive function in the dementias: methodological and regulatory considerations. Dialogues Clin Neurosci 2003;5:77–88. Westerberg CE, Paller KA, Weintraub S, Mesulam M-M, Holdstock JS, Mayes AR, et al. When memory does not fail: familiarity-based recognition in mild cognitive impairment and Alzheimer’s disease. Neuropsychology 2006;20:193–205. Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Wahlund LO, et al. Mild cognitive impairment – beyond controversies, towards a consensus: report of

25

the International Working Group on Mild Cognitive Impairment. J Intern Med 2004;256:240–6. Witte AV, Kerti L, Hermannstadter HM, Fiebach JB, Schreiber SJ, Schuchardt JP, et al. Long-chain omega-3 fatty acids improve brain function and structure in older adults. Cereb Cortex 2013, bht163. Yonelinas AP. The nature of recollection and familiarity: a review of 30 years of research. J Mem Lang 2002;46:441–517. Yue J, Dong BR, Lin X, Yang M, Wu HM, Wu T. Huperzine A for mild cognitive impairment. Cochrane Database Syst Rev 2012;12:CD008827.