NeuroImage 118 (2015) 146–153

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Structural brain correlates of associative memory in older adults Nina Becker a,b,⁎, Erika J. Laukka b, Grégoria Kalpouzos b, Moshe Naveh-Benjamin c, Lars Bäckman b, Yvonne Brehmer a,b a b c

Center for Lifespan Psychology, Max Planck Institute for Human Development, Berlin, Germany Aging Research Center, Karolinska Institutet and Stockholm University, Stockholm, Sweden Department of Psychological Sciences, University of MO, USA

a r t i c l e

i n f o

Article history: Received 16 January 2015 Accepted 2 June 2015 Available online 5 June 2015 Keywords: Associative memory Episodic memory Inter-individual differences VBM Gray matter Aging

a b s t r a c t Associative memory involves binding two or more items into a coherent memory episode. Relative to memory for single items, associative memory declines greatly in aging. However, older individuals vary substantially in their ability to memorize associative information. Although functional studies link associative memory to the medial temporal lobe (MTL) and prefrontal cortex (PFC), little is known about how volumetric differences in MTL and PFC might contribute to individual differences in associative memory. We investigated regional graymatter volumes related to individual differences in associative memory in a sample of healthy older adults (n = 54; age = 60 years). To differentiate item from associative memory, participants intentionally learned face–scene picture pairs before performing a recognition task that included single faces, scenes, and face–scene pairs. Gray-matter volumes were analyzed using voxel-based morphometry region-of-interest (ROI) analyses. To examine volumetric differences specifically for associative memory, item memory was controlled for in the analyses. Behavioral results revealed large variability in associative memory that mainly originated from differences in false-alarm rates. Moreover, associative memory was independent of individuals' ability to remember single items. Older adults with better associative memory showed larger gray-matter volumes primarily in regions of the left and right lateral PFC. These findings provide evidence for the importance of PFC in intentional learning of associations, likely because of its involvement in organizational and strategic processes that distinguish older adults with good from those with poor associative memory. © 2015 Elsevier Inc. All rights reserved.

Introduction Episodic memory (EM) is the conscious remembrance of events and relations between events situated in time and place (Tulving, 1972). Within EM, a distinction can be made between remembering single items (e.g., a name, an object, or a word) versus associated information (e.g., face–name, object–location, or word–sound pairs; Davachi, 2006). Although item memory remains relatively well preserved in aging, older adults' associative memory is markedly impaired (Chalfonte and Johnson, 1996; Naveh-Benjamin, 2000). The associative-deficit hypothesis, which attributes age-related EM deficits to problems in encoding and retrieving associated information (Naveh-Benjamin, 2000), has been supported by numerous studies using different materials (e.g., word–word, word–font, face–name, object–location, and object–object combinations; Old and Naveh-Benjamin, 2008). In addition to mean age-related alterations in performance, individual differences in EM increase in aging (Lindenberger, 2014; Lindenberger et al., 2013). Both cross-sectional and longitudinal studies ⁎ Corresponding author at: Max Planck Institute for Human Development, Lentzeallee 94, 14195 Berlin, Germany. E-mail address: [email protected] (N. Becker).

http://dx.doi.org/10.1016/j.neuroimage.2015.06.002 1053-8119/© 2015 Elsevier Inc. All rights reserved.

reveal greater inter-individual variability in EM functioning in older compared to younger adults (Christensen et al., 1999; Morse, 1993; Wilson et al., 2002). With regard to memory for associative information, greater variability has been reported for older compared to younger adults and for associative compared to item memory (Dennis et al., 2008; Glisky et al., 2001; Kilb and Naveh-Benjamin, 2011; Naveh-Benjamin, 2000; Naveh-Benjamin et al., 2009; Rajah et al., 2010a). Only few studies have systematically investigated the neural underpinnings of these individual differences. Magnetic resonance imaging (MRI) studies show that EM functioning in general draws on medial– temporal lobe (MTL) and lateral prefrontal (PFC) structures (Buckner et al., 1999; Cabeza, 2006; Dickerson and Eichenbaum, 2010; Mayes et al., 2007; Simons and Spiers, 2003; Van Petten, 2004). Findings from functional MRI studies have linked associative memory to hippocampal and item memory to parahippocampal activity, and reported greater PFC activity for associative than for item memory (Blumenfeld et al., 2011; Lepage et al., 2003; Staresina and Davachi, 2008; Westerberg et al., 2012). Structural MRI studies are less conclusive. Equivocal findings have been observed regarding the association between hippocampal volume and item memory (Hackert et al., 2002; Kalpouzos et al., 2009; Köhler et al., 1998; Rosen et al., 2003; for a review see Van Petten, 2004). Structural studies investigating associative

N. Becker et al. / NeuroImage 118 (2015) 146–153

memory are generally less common. To our knowledge, only one study examined both item and associative memory (using a word–pair recognition paradigm) from early to late adulthood with a ROI-based approach including the hippocampus (HC), entorhinal cortex, and lateral PFC (Rodrigue and Raz, 2004). Findings revealed a positive relation between hippocampal volume and associative memory across age. Shing et al. (2011) examined associative memory in relation to hippocampal subfield volumes in older adults and found a positive correlation between the CA3–4 and dentate gyrus subfields and memory performance. However, another study focused on differences between younger and older adults by comparing the relation of hippocampal volume to associative memory in each age group, and found a positive relation between gray-matter volume and associative memory in younger, but not older, adults. These data were interpreted to mean that older adults might depend less on HC, but rely more on frontal regions when intentionally learning associations (Rajah et al., 2010a; see Rajah et al., 2010b for supporting functional results). Although the relation of MTL and PFC gray-matter volumes to item and associative memory is still ambiguous, there is consensus on the functions these regions serve. The MTL, especially the HC, is important for binding single items together. Hence, we expect structural graymatter volume differences in the hippocampal region to be one underlying source of inter-individual differences in associative memory in older adults. The lateral PFC is known to enhance binding through control operations (Fletcher et al., 2000) and strategy use (Kirchhoff and Buckner, 2006; Kirchhoff et al., 2014), which facilitates item-item binding (Dunlosky and Hertzog, 1998; Naveh-Benjamin et al., 2007). In healthy aging, the PFC undergoes structural and functional changes accompanied by increased inter-individual variability in volume (Lindenberger, 2014; Lindenberger et al., 2013; Raz et al., 2005). Given that also differences in strategic processes might underlie individual differences in associative-memory performance, we hypothesize that individual differences in binding are, in part, explained by volumetric differences in PFC. However, to date no structural MRI study has investigated associative memory in a sample of older adults specifically in relation to PFC. Hence, the contribution of gray-matter volume in PFC to individual differences in associative memory remains unknown. We investigated whether inter-individual differences in PFC and MTL volumes account for inter-individual differences in associative memory in a sample of healthy older adults. To differentiate the relative contribution of brain structures to item and associative memory, a recognition task that captures both memory for single items and associated information was used. Methods Participants Data were collected within the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K), a population-based study targeting people 60 years and older living in the Kungsholmen district in central Stockholm. The current sample was taken from a cohort added in 2010–2013 (wave 4 in SNAC-K). All 678 participants in this new cohort were 60 years old and randomly selected from population registries. The examination in SNAC-K took about six hours and consisted of three parts: a nurse interview, a medical examination, and a neuropsychological testing session. In addition to the standard cognitive test battery of SNAC-K (Laukka et al., 2013), this cohort was also assessed with an item-associative-memory task. A subsample of 57 individuals participated in an MRI assessment. The sample for this study included cognitively healthy persons with data on MRI and the item-associative memory task. Participants were screened for microvascular lesions, cerebral infarcts, and motion artifacts by two independent and experiences raters. Three participants in the MRI sample were excluded due to missing data on the item-associative memory task; thus the final sample consisted of 54 participants (30 females; mean years of

147

education = 15.3; see Table 1). The selection bias of the MR subsample relative to the whole sample was positive but small (effect size of selectivity b .29 in a number of cognitive variables, including the itemassociative memory task, Mini-Mental State Examination (MMSE), episodic memory, perceptual speed, semantic memory, and educational background).

Material and procedure Item-associative-memory task (IAMT) The IAMT was administered at the end of the neuropsychological testing session. During encoding, participants saw 24 face–scene picture pairs on a computer screen. Each pair was presented for 4 s with a fixed inter-stimulus interval of 1 s. The face stimuli were male and female, old and young faces, exhibiting either a neutral or happy expression (Ebner et al., 2010). Scene stimuli were colored photographs of indoor and outdoor scenes (Chen and Naveh-Benjamin, 2012). Subjects were instructed to memorize both the single pictures and their combinations. The encoding phase was followed by a distractor task, in which subjects had to count backwards from 89 in steps of two for one minute. Immediately following the distractor task, three self-paced recognition tasks were administered: two item-memory and one associativememory task with their order counterbalanced across participants. In each item-memory task (one for the faces and one for the scenes), subjects saw 16 single pictures (i.e., 16 faces and 16 scene pictures), of which eight had been studied during encoding. The remaining eight new pictures served as lures. Participants were instructed to indicate whether they had seen a particular face or scene picture in the encoding phase by pressing the buttons “yes” or “no” on a computer keyboard. In the associative-memory task, subjects saw 16 face–scene pairs of which all had been previously presented. However, half of the pairs were intact (old), whereas the other half was recombined (composed of faces and scenes that appeared in the study phase, but not together). Again, subjects indicated whether or not they had seen a particular face–scene combination during encoding by pressing a “yes” or “no” button on the computer keyboard (see Fig. 1). Each face or scene appeared in only one of the three tests.

Free-recall task To control for differences in task difficulty between item and associative memory in the IAMT, a separate and more difficult item-memory task was assessed and used as regressor in the analyses. Here, participants studied 16 concrete Swedish nouns, presented in black on a 14.5 × 21 cm white paper. Each word was shown and read out aloud by the experimenter. Immediately after presentation of the last word in the series, participants were asked to free recall the words orally (Laukka et al., 2013).

Table 1 Demographic and cognitive measures of the study sample. Variable

Mean

SD

Age (years) Education (years) MMSE score Item memory (scenes)

60.47 15.30 29.20 6.87 0.04 6.83 6.85 1.81 5.04 6.50 3.31 3.19 8.56

.21 3.17 1.00 1.03 1.25 1.06 1.00 .19 1.48 1.23 2.05 2.56 1.62

Item memory (faces)

Associative memory

Free recall

H FA H-FA H FA H-FA H FA H-FA H

148

N. Becker et al. / NeuroImage 118 (2015) 146–153

Fig. 1. Experimental design and exemplar trials from the item-associative-memory task. During encoding, 24 face–scene picture-pairs were presented for 4 s each. Participants were instructed to memorize both the single pictures and the combinations. At retrieval, three self-paced recognition tasks were administered. In the item-memory tasks subjects saw 16 single pictures; half of the pictures had been studied and the other half served as novel lures. In the associative-memory task subjects saw 16 face–scene pairs. All stimuli had been studied, but half of the pairs were intact and the other half was recombined. Participants were told to indicate whether they had studied a particular item or item-pair by pressing the buttons “yes” or “no” on a computer keyboard.

MRI data acquisition and preprocessing Structural brain data were collected with a 3 T GE750 scanner equipped with a 32-channel head coil. T1-weighted MRI scans were collected using the SAG FSPGR BRAVO sequence with the following parameters: 172 adjacent sagittal slices with a thickness of 1 mm and an in-plane resolution of 0.94 mm × 0.94 mm, repetition time (TR) = 8.204 ms, echo time (TE) = 3.22 ms, field of view (FOV) = 24 cm × 24 cm, matrix = 256 × 256). We used voxel-based morphometry (VBM) to identify brain regions relevant for associative memory (Ashburner and Friston, 2000; Whitwell, 2009). Data were processed and analyzed in SPM12b (Statistical Parametric Mapping, Wellcome Department of Imaging Science, Functional Imaging Laboratory, http://www.fil.ion.ucl.ac.uk/spm/) in Matlab R2012b (Mathworks Inc, MA, US). The individual T1 images were segmented into gray matter, white matter, and cerebrospinal fluid (CSF) using the unified segmentation approach (Ashburner and Friston, 2005), and the “light cleanup” option to remove odd voxels from the segments. The resulting segments were visually verified. VBM was processed on the gray-matter images using DARTEL with customized templates, an algorithm for diffeomorphic image registration (Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra; Ashburner, 2007). First, the images generated in the segmentation process were imported into DARTEL space with a voxel resolution of 1.5 × 1.5 × 1.5 mm. Second, using the newly generated and aligned individual gray-matter images, optimized customized gray-matter templates were built. This included several iterative steps of improved template crisp and subject-specific flow fields

(i.e., deformation maps registered to the most representative template; Template 6). The native gray-matter images were finally spatially normalized using the final flow fields and further registered to Montreal Neurological Institute (MNI) space. Volumetric information that was lost during spatial normalization was re-integrated by modulating the images using Jacobian determinants from the deformations for the signal to be as close as possible to the signal of the native T1 images. Finally, gray-matter images were smoothed with a Gaussian FWHM (full width at half maximum) convolution kernel of 8 mm in the 3 directions. A binary gray-matter mask based on the normalized gray-matter images was created and used as an explicit mask in all analyses. Statistical analyses Behaviorally, we subdivided participants into three equally large groups based on their individual associative-memory performance (i.e., poor binders being in the lower tercile, middle binders in the middle tercile, and good binders in the upper tercile) to investigate whether (a) good and poor binders differed in item-memory performance; and (b) the response pattern of persons with good and poor associativememory performance, that is how good binders differ from poor binders regarding H and FA rates. To analyze gray-matter volume differences multiple regressions within the framework of the General Linear Model in SPM12b were applied. Separate multiple regressions were conducted with hits minus false alarms (H-FA), hits (H), and false alarms (FA) of the associativememory scores as variables of interest. Individuals' item-memory performance, i.e., H minus FA values from the face- and scene-recognition

N. Becker et al. / NeuroImage 118 (2015) 146–153

149

tasks, were added as additional regressor in the analyses to investigate which brain regions were related to associative-memory performance independent of item memory. To account for potential task-difficulty or retrieval type differences (familiarity vs. recollection) between item- and associative-memory recognition, a multiple regression with associative-memory performance (H-FA) as variable of interest was conducted, controlling not only for item recognition (faces and scenes) but also for item recall performance (i.e., number of freely recalled words from the free recall EM task). To investigate the relation of gray-matter volume and item memory, we performed separate multiple regressions with faces, scenes, and free recall as variables of interest. Analyses were conducted using a ROI approach. As we hypothesized to find gray-matter differences in PFC and hippocampal regions to be related to differences in associative memory, we conducted all analyses of specific ROIs by explicitly masking our gray-matter images. Two ROIs were created using the Brodmann anatomical template provided by MRIcro, one in dorsolateral PFC (including BA 8, 9, and 46) and one in ventrolateral PFC (including BA 45, and 47), as previous studies have linked these areas to episodic memory (Achim and Lepage, 2005; Bunge et al., 2004; Kirchhoff and Buckner, 2006; Rodrigue and Raz, 2004). Two additional MNI-based ROIs of the HC and parahippocampal gyrus (PHG) were created using the Automated Anatomic Labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002). A Family-Wise Error (FWE)-corrected cluster-extent threshold of p b .05, with a cluster-forming threshold of p b .001 (uncorrected for multiple comparisons) was applied to all analyses. All analyses were adjusted for sex, education, and total intracranial volume (TIV; calculated by adding the volumes of gray matter, white matter, and CSF). Given the novelty of the study, we repeated all analyses using a whole-brain approach to investigate if regions not captured by the ROI analyses accounted for individual differences in associative memory. Results from these analyses can be found in the supplementary materials.

Free recall task On average, participants remembered about 50% of the 16 words (mean = 8.56; min. = 5 words, max. = 11 words). Recall performance was uncorrelated with performance in both item and associative memory (rs b .14).

Results

Correlations between gray-matter volume and associative memory

Behavioral results

Controlling for item-memory performance, ROI analyses of graymatter volumes in dorsolateral and ventrolateral PFC in relation to HFA rates revealed one significant cluster located in left dorsolateral PFC (BA 8; Table 2). Larger gray-matter volume in this region was related to higher H-FA rates in the associative-memory task. To investigate the relative contribution of gray-matter volume differences for H and FA rates in associative memory, two additional multiple regressions

Item-associative-memory task Mean performance of both item-recognition tasks was generally high, reflecting a high number of H and a low number of FA. In fact, the mean number of FA in the scene-memory task did not significantly differ from zero. Performance in the associative-memory task was significantly lower compared to both item-memory tasks (ps b .001). Worse associative-memory performance mainly originated from higher FA rates than for item memory (ps b .001), whereas H rates were only marginally lower than H rates for item memory (ps N .067; see Table 1). Importantly, associative-memory performance (H-FA) was normally distributed with a skewness of −.179 (SE = .32) and a kurtosis of −.743 (SE = .64), leaving sufficient variance to investigate individual differences in underlying structural brain correlates (Fig. 2). To further investigate the response patterns in associative memory and their relation to item memory, the sample was split into terciles based on associative-memory performance (i.e., good binders (H-FAgood = 5 to 8), middle binders (H-FAmiddle = 2 to 4), and poor binders (H-FApoor = − 2 to 1)). Six one-way ANOVAs revealed no group differences in the two item-memory tasks for H, FA, or H-FA (p N .266). However, in associative memory, all three groups significantly differed in number of FA (F (2, 51) = 82.97, p b .001, η2p = .76), with poor binders having higher FA rates than middle binders (p b .001) and middle binders having higher FA rates than good binders (p b .001). By contrast, significant differences in H rates (F (2, 51) = 7.22, p b .005, η2p = .221) were reliable only between good and poor binders (p b .005; Fig. 3). An analysis of covariance (ANCOVA) revealed

Fig. 2. Distribution of H minus FA rates in the associative-memory task. Memory performance was normally distributed across individuals in the study sample.

significant differences in associative-memory performance (H-FA) between all three groups when controlling for performance in both item-memory measures (F (2, 285) = 159.93, p b .001). Thus, differences in associative memory mainly originated from differences in FA rates, and were independent of subjects' performance in both itemmemory tasks. The correlation between associative memory and both item-memory tasks was non-significant (rs b .23).

Fig. 3. H and FA rates in the associative-memory task across group (poor binders, middle binders, good binders). Error bars represent standard errors around the means.

150

N. Becker et al. / NeuroImage 118 (2015) 146–153

Table 2 Clusters related to associative memory in ROI analyses for the PFC. Region

L/R

BA

k

tvalue

MNI coordinates x

y

z

Regions with a positive relation to H-FA rates controlling for item recognition Dorsolateral PFC L 8 339 4.62 −3 23 52 3.90 −2 23 63 Regions with a positive relation to H rates controlling for item recognition Dorsolateral PFC R 46 663 4.92 40 41 4.17 48 45 4.15 30 47 Ventrolateral PFC R 45 393 5.17 40 41 4.15 46 47

23 16 24 22 18

Regions with a negative relation to FA rates controlling for item recognition Ventrolateral PFC R 47 110 4.88 32 21

−26

Regions with a positive relation to H-FA rates additionally controlling for free recall Dorsolateral PFC L 8 328 4.62 −3 23 52 Note. Main peaks in clusters of gray-matter volume related to associative memory. 1) Larger gray-matter volume in this cluster was related to better associative memory (H-FA). 2) Larger gray-matter volumes in these clusters were related to more H in the associative-memory task. 3) Smaller gray-matter volume in this cluster was related to more FA in the associative-memory task. 4) Larger gray-matter volume in this cluster was related to better associative memory (H-FA). PFC = Prefrontal cortex. k = cluster size (in voxels). BA = Brodmann area. L = left, R = right. ROI analyses for HC and PHG revealed no clusters significantly related to associative memory.

gray-matter volumes in this cluster made fewer FA (Table 2). ROI analyses for the HC and PHG revealed only two small clusters in right and left PHG that did not survive FWE correction. Recent findings have provided support for the assumption that vascular risk factors, such as high blood pressure, influence memory functioning among older adults, as it affects shrinkage of the HC (Bender et al., 2013; Shing et al., 2011). However, taking diastolic and systolic blood pressure values into our statistical model did not alter the results. Gray-matter volume differences related to associative memory controlled for free recall performance In an additional analysis, multiple regression with H-FA rates as the dependent variable was repeated, controlling for individuals' item memory (H-FA rates in face and scene recognition) and their freerecall performance. This was done to investigate whether gray-matter volume differences were uniquely related to differences in associative memory rather than to relative task difficulty. Controlling for recollection-based free-recall performance, ROI analyses for PFC, HC, and PHG resulted in one brain region located in left dorsolateral PFC (BA 8), overlapping with the cluster related to individual differences in associative memory from the prior analysis (BA 8; Table 2; Fig. 4). Gray-matter volume differences related to item memory

were conducted. ROI analyses on H rates revealed a cluster overlapping with both right dorsolateral (BA 46) and right ventrolateral PFC (BA 45). A negative relation between FA and gray-matter volume was observed in right ventrolateral PFC (BA 47) reflecting that individuals with larger

Additional ROI analyses for PFC, HC, and PHG were conducted to investigate potential gray-matter differences related to individual differences in the two item-recognition tasks (H-FA). However, no brain region showed a significant link to either of these two measures. Further

Fig. 4. Gray-matter volume correlates of associative memory. A. Left DLPFC (BA 8) positively related to H-FA rates. B. Right DLPFC and VLPFC (BA 45, BA 46) positively related to H rates. C. Right VLPFC (BA 47) negatively related to FA rates.

N. Becker et al. / NeuroImage 118 (2015) 146–153

analyses including H and FA rates for both types of items again did not reveal significant clusters linked to these scores. Similarly, a multiple regression with free recall, the recollection-based item-memory task, as covariate of interest revealed no cluster significantly linked to recall performance.

Discussion The aim of this study was to investigate the relation between regional gray-matter volumes and individual differences in associative memory after controlling for individual differences in item memory. Behaviorally, results indicated large variability in associative memory in a sample of healthy older adults that mainly originated from variability in FA rates. Further, individual differences in associative memory were independent of item memory. Larger gray-matter volumes in ventrolateral PFC (VLPFC), and dorsolateral PFC (DLPFC) were related to better associative memory (i.e., more H and fewer FA). These volumetric differences accounted for variability in the ability to bind together an across-domain item pair exceeding individuals' ability to remember the items separately. In contrast, individual differences in item memory (remembering a face or a scene separately) were unrelated to differences in regional gray-matter volume. The absent relationships between brain volumes and item memory is unlikely to stem from ceiling effects as H rates in the associative-memory task, which did not differ significantly from H rates in both item-memory tasks, were related to one prominent cluster in both VLPFC and DLPFC.

Inter-individual differences in associative memory are independent from inter-individual differences in item memory The behavioral findings are in line with previous research. Older adults' associative-memory performance was significantly lower than their memory for item information (e.g., Naveh-Benjamin, 2000) and lower associative-memory performance was primarily driven by higher FA rates (Jacoby and Rhodes, 2006; Shing et al., 2011; Shing et al., 2008). Interestingly, even though the study sample was age homogenous, similar with regard to educational level, and, because of their age, should be considered as “young-old adults”, large inter-individual differences in associative memory were observed. These individual differences were not accounted for by differences in item memory. Importantly, this statistical independence could not be explained by differences in task difficulty between the associative- and item-memory tasks as the inclusion of an additional, more difficult, item-memory task (free recall) revealed the same pattern (see Glisky et al., 2001, for similar findings). A possible explanation for the apparent difference between item and associative memory might be related to differences in the underlying retrieval processes. According to dual-process theory, recognition relies on two qualitatively different processes, familiarity and recollection (Yonelinas, 1994). Although single-item recognition can be based on familiarity, recognition of an item pair (original vs. recombined) primarily relies on recollection (Yonelinas, 1997). Previous work shows that aging has a larger detrimental effect on recollection than familiarity (Jennings and Jacoby, 1993, 1997). However, in the current study associative memory and the recollection-based item-memory task (free recall) were also unrelated, suggesting that poorer associative memory did not solely originate from a decrement in recollection. Hence, to bind two elements together may not only be more difficult than remembering a single item (see also Kilb and Naveh-Benjamin, 2007), but also to depend on processes that go beyond recollection. Such processes might be unique for the cognitive operation of binding two or more pieces of information and occur at different levels, including low-level perceptual binding, binding of items, and integration of events with their context.

151

Individual differences in associative memory are related to gray-matter volume in PFC This is the first study showing that individual differences in associative memory in older adults are related to between-person differences in gray-matter PFC volumes. Specifically, DLPFC (BA 8 and BA 46) and VLPFC (BA 45 and BA 47) accounted for individual differences in associative memory. Results remained unchanged after additionally controlling for a recollection-based free-recall task, which supports the suggestion that individual differences in the face–scene associative memory task might stem from processes other than recollection being implicated in visual associative memory. Such processes might include inter-item binding at encoding, as well as recall-to-reject (Rotello and Heit, 2000), and inhibitory processes at retrieval. DLPFC and VLPFC have been linked to information maintenance, binding, inhibition, monitoring, and control processes during encoding and retrieval of item pairs (Achim and Lepage, 2005; Bunge et al., 2004; Fletcher et al., 2000; Qin et al., 2007; Wagner et al., 2001; Wheeler and Buckner, 2003). In addition, these regions have been related to the self-initiated use of memory strategies (Kirchhoff and Buckner, 2006; Kirchhoff et al., 2014). In the present study, subjects studied item pairs under intentional learning conditions. Specifically, they were asked to study the items and their combinations for a later recognition test. Between-item binding can be enhanced by the use of strategies of verbal or visual nature (Kirchhoff et al., 2014). Research shows that older compared to younger adults have difficulties in initiating strategies (Hertzog et al., 2013; Naveh-Benjamin et al., 2009), most likely due to reduced PFC integrity. If larger gray-matter volume is related to better associative memory, individual differences in associative-memory performance may result from individual differences in strategic processes. If no strategy is used, or if the binding strategy fails, subjects face the combination of two highly familiar items in the re-arranged pair-recognition condition. Thus, to successfully recollect studied pairs, control processes are needed in order to overcome interference. In the current study, poor individual associative-memory performance originated largely from an increase in FA rates, suggesting familiarity-based interference due to deficient monitoring processes (Naveh-Benjamin et al., 2009). Hence, individual differences in PFC volume might lead to between-person variability in control processes. This interpretation is supported by a recent meta analysis, which yielded that greater PFC volume is related to better executive functions, including interference control (Yuan and Raz, 2014).

Structural differences in MTL do not account for individual differences in associative memory The current data do not suggest that structural differences in MTL account for individual differences in associative memory. At first glance, this stands in contrast to previous functional MRI studies, which have shown involvement of HC in between-item binding among younger adults (Chua et al., 2007; Qin et al., 2009; Rodrigue and Raz, 2004; Westerberg et al., 2012). However, the present results are in line with findings on older adults by Rajah et al. (2010a,b), who reported no relation between HC volume and context-memory performance. One study by Zamboni et al. (2013) found a link between HC volume, but not PFC volume, and performance in a visuospatial associative memory task in older adults. However, contrary to the present study, Zamboni and colleagues used incidental learning conditions (see also Kalpouzos et al., 2009). The HC is known for its pivotal role in rapidly forming associations between relational information in long-term memory (Eichenbaum, 2002; Olsen et al., 2012). Thus, HC volume may play a less important role and PFC volume a more important role when learning of associations is intentional and effortful (Atienza et al., 2011; Zamboni et al., 2013).

152

N. Becker et al. / NeuroImage 118 (2015) 146–153

Between-person differences in associative memory originated from greater variability in FA rates, whereas H rates showed less variance. Hence, participants were generally able to successfully form and retrieve associations as long as they could rely on familiarity-based response selection. Individual differences in performance seemed to emerge from difficulties to recollect, monitor, and verify associated information, for example through strategic or recall-to-reject processes (Rotello and Heit, 2000), resulting in higher FA rates. As frontal regions are involved in control processes (Miller and Cohen, 2001), especially when familiarity of the stimuli is high and individuals are unable to recollect associated information (Bunge et al., 2004), it is conceivable that differences in the PFC – and not MTL – accounted for differences in associative memory. Note that this finding does not imply that the hippocampal region is not involved in the formation and retrieval of associations; it rather suggests that volumetric differences in HC did not account for individual differences in associative memory, because these seem to be primarily related to differences in monitoring and strategic processes drawing on the PFC. This result complements on a structural brain level the suggestion from fMRI studies (Bunge et al., 2004; Fletcher et al., 2000) that between-item binding in an intentional learning context largely depends on PFC-related control and monitoring functions. Limitations of the study The current study was cross-sectional; thus, no conclusions can be drawn regarding the origin of individual differences in gray-matter volumes. Such differences might reflect variations in rate of age-related decline or between-person differences in brain volume already at younger ages. Thus, examining associations between structural brain changes and associative memory requires a longitudinal approach. Moreover, the findings of this study cannot be generalized to persons over the age of 60 years, where age-related brain atrophy is more pronounced, especially in HC (Raz et al., 2005). The present study investigated regional brain differences in relation to associative memory, but did not assess the interaction of regions in relation to memory performance. Although no relation between associative memory and HC was found, future studies should investigate the inter-regional interplay between PFC and HC in associative memory by using structural and functional connectivity analyses. The free-recall task used to control for task difficulty differs from the IAMT in terms of materials (words vs. pictures) and underlying cognitive processes. Therefore, this free-recall task might not have fully accounted for difficulty differences between item and associative memory in the current task. The current study investigated the relative contribution of graymatter volumes to H and FA rates. However, the findings need to be interpreted with caution, as FA rates showed greater variance than H rates, resulting in greater power to detect significant associations. Conclusion We observed that inter-individual differences in associative memory in 60-year-old adults are related to gray-matter differences in PFC. This finding underscores the importance of PFC in associative binding in an intentional learning context, most likely because of organizational and strategic processes that distinguish older adults with good from those with poor associative memory. Future studies should investigate the importance of distinct brain areas with regard to binding under both intentional and incidental learning conditions. Acknowledgements We thank all the participants and staff who were involved in the collection and management of the data for their contribution to the study. The Swedish National study on Aging and Care, SNAC, (www.snac.org)

is financially supported by the Ministry of Health and Social Affairs, Sweden, the participating County Councils and Municipalities, and the Swedish Research Council. In addition, specific grants were obtained from the Max Planck Society (YB), Stiftelsen för Gamla Tjänarinnor (YB) and the Swedish Research Council (EJL). LB was supported by grants from the Swedish Research Council, the Swedish Research Council for Health, Working Life, and Welfare, an Alexander von Humboldt Research Award, and a donation from the af Jochnick Foundation. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.neuroimage.2015.06.002. References Achim, A.M., Lepage, M., 2005. Dorsolateral prefrontal cortex involvement in memory post-retrieval monitoring revealed in both item and associative recognition tests. NeuroImage 24, 1113–1121. http://dx.doi.org/10.1016/j.neuroimage.2004.10.036. Ashburner, J., 2007. A fast diffeomorphic image registration algorithm. NeuroImage 38, 95–113. http://dx.doi.org/10.1016/j.neuroimage.2007.07.007. Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry—the methods. NeuroImage 11, 805–821. http://dx.doi.org/10.1006/nimg.2000.0582. Ashburner, J., Friston, K.J., 2005. Unified segmentation. NeuroImage 26, 839–851. http:// dx.doi.org/10.1016/j.neuroimage.2005.02.018. Atienza, M., Atalaia-Silva, K.C., Gonzalez-Escamilla, G., Gil-Neciga, E., Suarez-Gonzalez, A., Cantero, J.L., 2011. Associative memory deficits in mild cognitive impairment: the role of hippocampal formation. NeuroImage 57, 1331–1342. http://dx.doi.org/10. 1016/j.neuroimage.2011.05.047. Bender, A.R., Daugherty, A.M., Raz, N., 2013. Vascular risk moderates associations between hippocampal subfield volumes and memory. J. Cogn. Neurosci. 25, 1851–1862. http:// dx.doi.org/10.1162/jocn_a_00435. Blumenfeld, R.S., Parks, C.M., Yonelinas, A.P., Ranganath, C., 2011. Putting the pieces together: the role of dorsolateral prefrontal cortex in relational memory encoding. J. Cogn. Neurosci. 23, 257–265. http://dx.doi.org/10.1162/jocn.2010.21459. Buckner, R.L., Kelley, W.M., Petersen, S.E., 1999. Frontal cortex contributes to human memory formation. Nat. Neurosci. 2, 311–314. http://dx.doi.org/10.1038/7221. Bunge, S.A., Burrows, B., Wagner, A.D., 2004. Prefrontal and hippocampal contributions to visual associative recognition: interactions between cognitive control and episodic retrieval. Brain Cogn. 56, 141–152. http://dx.doi.org/10.1016/j.bandc.2003.08.001. Cabeza, R., 2006. Prefrontal and medial temporal lobe contributions to relational memory in young and older adults. In: Zimmer, D., Mecklinger, A., Lindenberger, U. (Eds.), Handbook of Binding and Memory: Perspectives from Cognitive Neuroscience. Oxford University Press, Oxford, pp. 595–626. http://dx.doi.org/10.1093/acprof:oso/ 9780198529675.003.0024. Chalfonte, B.L., Johnson, M.K., 1996. Feature memory and binding in young and older adults. Mem. Cognit. 24, 403–416. http://dx.doi.org/10.3758/BF03200930. Chen, T., Naveh-Benjamin, M., 2012. Assessing the associative deficit of older adults in long-term and short-term/working memory. Psychol. Aging 27, 666–682. http://dx. doi.org/10.1037/a0026943. Christensen, H., Mackinnon, A.J., Korten, A.E., Jorm, A.F., Henderson, A.S., Jacomb, P., Rodgers, B., 1999. An analysis of diversity in the cognitive performance of elderly community dwellers: individual differences in change scores as a function of age. Psychol. Aging 14, 365–379. http://dx.doi.org/10.1037/0882-7974.14.3.365. Chua, E.F., Schacter, D.L., Rand-Giovannetti, E., Sperling, R.A., 2007. Evidence for a specific role of the anterior hippocampal region in successful associative encoding. Hippocampus 17, 1071–1080. http://dx.doi.org/10.1002/hipo.20340. Davachi, L., 2006. Item, context and relational episodic encoding in humans. Curr. Opin. Neurobiol. 16, 693–700. http://dx.doi.org/10.1016/j.conb.2006.10.012. Dennis, N.A., Hayes, S.M., Prince, S.E., Madden, D.J., Huettel, S.A., Cabeza, R., 2008. Effects of aging on the neural correlates of successful item and source memory encoding. J. Exp. Psychol. Learn. Mem. Cogn. 34, 791–808. http://dx.doi.org/10.1037/0278-7393.34.4. 791. Dickerson, B.C., Eichenbaum, H., 2010. The episodic memory system: neurocircuitry and disorders. Neuropsychopharmacology 35, 86–104. http://dx.doi.org/10.1038/npp. 2009.126. Dunlosky, J., Hertzog, C., 1998. Aging and deficits in associative memory: what is the role of strategy production? Psychol. Aging 13, 597–607. http://dx.doi.org/10.1037/08827974.13.4.597. Ebner, N.C., Riediger, M., Lindenberger, U., 2010. FACES—a database of facial expressions in young, middle-aged, and older women and men: development and validation. Behav. Res. Methods 42, 351–362. http://dx.doi.org/10.3758/BRM.42.1.351. Eichenbaum, H., 2002. The Cognitive Neuroscience of Memory: An Introduction. Oxford University Press, New York http://dx.doi.org/10.1093/acprof:oso/9780195141740. 001.0001. Fletcher, P.C., Shallice, T., Dolan, R.J., 2000. “Sculpting the response space”—an account of left prefrontal activation at encoding. NeuroImage 12, 404–417. http://dx.doi.org/10. 1006/nimg.2000.0633. Glisky, E.L., Rubin, S.R., Davidson, P.S., 2001. Source memory in older adults: an encoding or retrieval problem? J. Exp. Psychol. Learn. Mem. Cogn. 27, 1131–1146. http://dx.doi. org/10.1037/0278-7393.27.5.1131.

N. Becker et al. / NeuroImage 118 (2015) 146–153 Hackert, V.H., den Heijer, T., Oudkerk, M., Koudstaal, P.J., Hofman, A., Breteler, M.M., 2002. Hippocampal head size associated with verbal memory performance in nondemented elderly. NeuroImage 17, 1365–1372. http://dx.doi.org/10.1006/nimg. 2002.1248. Hertzog, C., Fulton, E.K., Mandviwala, L., Dunlosky, J., 2013. Older adults show deficits in retrieving and decoding associative mediators generated at study. Dev. Psychol. 49, 1127–1131. http://dx.doi.org/10.1037/a0029414. Jacoby, L.L., Rhodes, M.G., 2006. False remembering in the aged. Curr. Dir. Psychol. Sci. 15, 49–53. http://dx.doi.org/10.1111/j.0963-7214.2006.00405.x. Jennings, J.M., Jacoby, L.L., 1993. Automatic versus intentional uses of memory: aging, attention, and control. Psychol. Aging 8, 283–293. http://dx.doi.org/10.1037/08827974.8.2.283. Jennings, J.M., Jacoby, L.L., 1997. An opposition procedure for detecting age-related deficits in recollection: telling effects of repetition. Psychol. Aging 12, 352–361. http:// dx.doi.org/10.1037/0882-7974.12.2.352. Kalpouzos, G., Chetelat, G., Landeau, B., Clochon, P., Viader, F., Eustache, F., Desgranges, B., 2009. Structural and metabolic correlates of episodic memory in relation to the depth of encoding in normal aging. J. Cogn. Neurosci. 21, 372–389. http://dx.doi.org/10. 1162/jocn.2008.21027. Kilb, A., Naveh-Benjamin, M., 2007. Paying attention to binding: further studies assessing the role of reduced attentional resources in the associative deficit of older adults. Mem. Cognit. 35, 1162–1174. http://dx.doi.org/10.3758/BF03193486. Kilb, A., Naveh-Benjamin, M., 2011. The effects of pure pair repetition on younger and older adults' associative memory. J. Exp. Psychol. Learn. Mem. Cogn. 37, 706–719. http://dx.doi.org/10.1037/a0022525. Kirchhoff, B.A., Buckner, R.L., 2006. Functional–anatomic correlates of individual differences in memory. Neuron 51, 263–274. http://dx.doi.org/10.1016/j.neuron.2006.06. 006. Kirchhoff, B.A., Gordon, B.A., Head, D., 2014. Prefrontal gray matter volume mediates age effects on memory strategies. NeuroImage 90, 326–334. http://dx.doi.org/10.1016/j. neuroimage.2013.12.052. Köhler, S., Black, S.E., Sinden, M., Szekely, C., Kidron, D., Parker, J.L., Foster, J.K., Moscovitch, M., Winocour, G., Szalai, J.P., Bronskill, M.J., 1998. Memory impairments associated with hippocampal versus parahippocampal-gyrus atrophy: an MR volumetry study in Alzheimer's disease. Neuropsychologia 36, 901–914. http://dx.doi.org/10.1016/ S0028-3932(98)00017-7. Laukka, E.J., Lövdén, M., Herlitz, A., Karlsson, S., Ferencz, B., Pantzar, A., Keller, L., Graff, C., Fratiglioni, L., Bäckman, L., 2013. Genetic effects on old-age cognitive functioning: a population-based study. Psychol. Aging 28, 262–274. http://dx.doi.org/10.1037/ a0030829. Lepage, M., Brodeur, M., Bourgouin, P., 2003. Prefrontal cortex contribution to associative recognition memory in humans: an event-related functional magnetic resonance imaging study. Neurosci. Lett. 346, 73–76. http://dx.doi.org/10.1016/S03043940(03)00578-0. Lindenberger, U., 2014. Human cognitive aging: corriger la fortune? Science 346, 572–578. http://dx.doi.org/10.1126/science.1254403. Lindenberger, U., Burzynska, A.Z., Nagel, I.E., 2013. Heterogeneity in frontal lobe aging. In: Stuss, D.T., Knight, R.T. (Eds.), Principles of Frontal Lobe Function, Second ed. Oxford University Press, New York, pp. 609–627 http://dx.doi.org/10.1093/med/ 9780199837755.003.0043. Mayes, A., Montaldi, D., Migo, E., 2007. Associative memory and the medial temporal lobes. Trends Cogn. Sci. 11, 126–135. http://dx.doi.org/10.1016/j.tics.2006.12.003. Miller, E.K., Cohen, J.D., 2001. An integrative theory of prefrontal cortex function. Annu. Rev. Neurosci. 24, 167–202. http://dx.doi.org/10.1146/annurev.neuro.24.1.167. Morse, C.K., 1993. Does variability increase with age? An archival stuy of cognitive measures. Psychol. Aging 8, 156–164. http://dx.doi.org/10.1037/0882-7974.8.2.156. Naveh-Benjamin, M., 2000. Adult age differences in memory performance: tests of an associative deficit hypothesis. J. Exp. Psychol. Learn. Mem. Cogn. 26, 1170–1187. http:// dx.doi.org/10.1037//0278-7393.26.5.1170. Naveh-Benjamin, M., Brav, T.K., Levy, O., 2007. The associative memory deficit of older adults: the role of strategy utilization. Psychol. Aging 22, 202–208. http://dx.doi. org/10.1037/0882-7974.22.1.202. Naveh-Benjamin, M., Shing, Y.L., Kilb, A., Werkle-Bergner, M., Lindenberger, U., Li, S.C., 2009. Adult age differences in memory for name–face associations: the effects of intentional and incidental learning. Memory 17, 220–232. http://dx.doi.org/10.1080/ 09658210802222183. Old, S.R., Naveh-Benjamin, M., 2008. Differential effects of age on item and associative measures of memory: a meta-analysis. Psychol. Aging 23, 104–118. http://dx.doi. org/10.1037/0882-7974.23.1.104. Olsen, R.K., Moses, S.N., Riggs, L., Ryan, J.D., 2012. The hippocampus supports multiple cognitive processes through relational binding and comparison. Front. Hum. Neurosci. 6, 146. http://dx.doi.org/10.3389/fnhum.2012.00146. Qin, S., Piekema, C., Petersson, K.M., Han, B., Luo, J., Fernandez, G., 2007. Probing the transformation of discontinuous associations into episodic memory: an event-related fMRI study. NeuroImage 38, 212–222. http://dx.doi.org/10.1016/j.neuroimage.2007.07. 020.

153

Qin, S., Rijpkema, M., Tendolkar, I., Piekema, C., Hermans, E.J., Binder, M., Petersson, K.M., Luo, J., Fernandez, G., 2009. Dissecting medial temporal lobe contributions to item and associative memory formation. NeuroImage 46, 874–881. http://dx.doi.org/10. 1016/j.neuroimage.2009.02.039. Rajah, M.N., Kromas, M., Han, J.E., Pruessner, J.C., 2010a. Group differences in anterior hippocampal volume and in the retrieval of spatial and temporal context memory in healthy young versus older adults. Neuropsychologia 48, 4020–4030. http://dx.doi. org/10.1016/j.neuropsychologia.2010.10.010. Rajah, M.N., Languay, R., Valiquette, L., 2010b. Age-related changes in prefrontal cortex activity are associated with behavioural deficits in both temporal and spatial context memory retrieval in older adults. Cortex 46, 535–549. http://dx.doi.org/10.1016/j. cortex.2009.07.006. Raz, N., Lindenberger, U., Rodrigue, K.M., Kennedy, K.M., Head, D., Williamson, A., Dahle, C., Gerstorf, D., Acker, J.D., 2005. Regional brain changes in aging healthy adults: general trends, individual differences and modifiers. Cereb. Cortex 15, 1676–1689. http:// dx.doi.org/10.1093/cercor/bhi044. Rodrigue, K.M., Raz, N., 2004. Shrinkage of the entorhinal cortex over five years predicts memory performance in healthy adults. J. Neurosci. 24, 956–963. http://dx.doi.org/ 10.1523/JNEUROSCI.4166-03.2004. Rosen, A.C., Prull, M.W., Gabrieli, J.D.E., Stoub, T., O'Hara, R., Friedman, L., Yesavage, J.A., deToledo-Morrell, L., 2003. Differential associations between entorhinal and hippocampal volumes and memory performance in older adults. Behav. Neurosci. 117, 1150–1160. http://dx.doi.org/10.1037/0735-7044.117.6.1150. Rotello, C.M., Heit, E., 2000. Associative recognition: a case of recall-to-reject processing. Mem. Cognit. 28, 907–922. http://dx.doi.org/10.3758/BF03209339. Shing, Y.L., Werkle-Bergner, M., Li, S.C., Lindenberger, U., 2008. Associative and strategic components of episodic memory: a life-span dissociation. J. Exp. Psychol. Gen. 137, 495–513. http://dx.doi.org/10.1037/0096-3445.137.3.495. Shing, Y.L., Rodrigue, K.M., Kennedy, K.M., Fandakova, Y., Bodammer, N., Werkle-Bergner, M., Lindenberger, U., Raz, N., 2011. Hippocampal subfield volumes: age, vascular risk, and correlation with associative memory. Front. Aging Neurosci. 3, 2. http://dx.doi. org/10.3389/fnagi.2011.00002. Simons, J.S., Spiers, H.J., 2003. Prefrontal and medial temporal lobe interactions in longterm memory. Nat. Rev. Neurosci. 4, 637–648. http://dx.doi.org/10.1038/nrn1178. Staresina, B.P., Davachi, L., 2008. Selective and shared contributions of the hippocampus and perirhinal cortex to episodic item and associative encoding. J. Cogn. Neurosci. 20, 1478–1489. http://dx.doi.org/10.1162/jocn.2008.20104. Tulving, E., 1972. Episodic and semantic memory. In: Tulving, E., Donaldson, W. (Eds.), Organization of Memory. Academic Press, New York, pp. 381–403. Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., Mazoyer, B., Joliot, M., 2002. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. NeuroImage 15, 273–289. http://dx.doi.org/10.1006/nimg.2001.0978. Van Petten, C., 2004. Relationship between hippocampal volume and memory ability in healthy individuals across the lifespan: review and meta-analysis. Neuropsychologia 42, 1394–1413. http://dx.doi.org/10.1016/j.neuropsychologia.2004.04.006. Wagner, A.D., Maril, A., Bjork, R.A., Schacter, D.L., 2001. Prefrontal contributions to executive control: fMRI evidence for functional distinctions within lateral Prefrontal cortex. NeuroImage 14, 1337–1347. http://dx.doi.org/10.1006/nimg.2001.0936. Westerberg, C.E., Voss, J.L., Reber, P.J., Paller, K.A., 2012. Medial temporal contributions to successful face–name learning. Hum. Brain Mapp. 33, 1717–1726. http://dx.doi.org/ 10.1002/hbm.21316. Wheeler, M.E., Buckner, R.L., 2003. Functional dissociation among components of remembering: control, perceived oldness, and content. J. Neurosci. 23, 3869–3880. Whitwell, J.L., 2009. Voxel-based morphometry: an automated technique for assessing structural changes in the brain. J. Neurosci. 29, 9661–9664. http://dx.doi.org/10. 1523/JNEUROSCI.2160-09.2009. Wilson, R.S., Beckett, L.A., Barnes, L.L., Schneider, J.A., Bach, J., Evans, D.A., Bennett, D.A., 2002. Individual differences in rates of change in cognitive abilities of older persons. Psychol. Aging 17, 179–193. http://dx.doi.org/10.1037/0882-7974.17.2.179. Yonelinas, A.P., 1994. Receiver-operating characteristics in recognition memory: evidence for a dual-process model. J. Exp. Psychol. Learn. Mem. Cogn. 20, 1341–1354. http://dx. doi.org/10.1037/0278-7393.20.6.1341. Yonelinas, A.P., 1997. Recognition memory ROCs for item and associative information: the contribution of recollection and familiarity. Mem. Cognit. 25, 747–763. http://dx.doi. org/10.3758/BF03211318. Yuan, P., Raz, N., 2014. Prefrontal cortex and executive functions in healthy adults: a meta-analysis of structural neuroimaging studies. Neurosci. Biobehav. Rev. 42, 180–192. http://dx.doi.org/10.1016/j.neubiorev.2014.02.005. Zamboni, G., de Jager, C.A., Drazich, E., Douaud, G., Jenkinson, M., Smith, A.D., Tracey, I., Wilcock, G.K., 2013. Structural and functional bases of visuospatial associative memory in older adults. Neurobiol. Aging 34, 961–972. http://dx.doi.org/10.1016/j. neurobiolaging.2012.07.011.