## Abstract

Numerous studies have documented that older adults (OAs) do not perform as well as young adults (YAs) when task demands require the establishment or retrieval of a novel link between previously unrelated information (relational memory: RM). Nonetheless, the source of this age-related RM deficit remains unspecified. One of the most widely investigated factors is an age-related reduction in attentional resources. To investigate this factor, previous researchers have tested whether dividing YAs' attention during encoding equated their RM performance to that of OAs. However, results from these studies failed to replicate the age-related RM impairment observed in aging. The current study investigated whether a reduction in attentional resources for processing of relational information (i.e., relational attention) underlies age-related RM deficits. Using fMRI, we examined whether the effect of reduced attentional resources for processing of relational information is similar to that observed in aging at both behavioral and neural levels. The behavioral results showed that reduced attentional resources for relational information during encoding equated YAs RM performance to that of OAs. Furthermore, the fMRI results demonstrated that both aging, as well as reductions in relational attention in YAs, significantly reduced activity in brain areas associated with successful RM formation, namely, the ventrolateral and dorsolateral PFC, superior and inferior parietal regions, and left hippocampus. Such converging evidence from behavioral and neuroimaging studies suggests that a reduction in attentional resources for relational information is a critical factor for the RM deficit observed in aging.

## INTRODUCTION

Decades of research have shown that episodic memory, the ability to remember personally experienced episodes, is one of the most vulnerable cognitive functions affected by aging (Hoyer & Verhaeghen, 2006). Episodic memory involves the encoding and retrieval of the individual aspects of an event (i.e., item memory [IM]), as well as the associations or relationships formed between elements (i.e., relational memory [RM]; Tulving, 1983). Recent research indicates that older adults (OAs) have a disproportionate deficit in RM relative to IM (for a review, Old & Naveh-Benjamin, 2008). Such disproportionate deficits in RM in aging have been observed using different types of paradigms, including the association of separate stimuli (e.g., unrelated word pairs; Naveh-Benjamin, Hussain, Guez, & Bar-On, 2003; Naveh-Benjamin, 2000), spatial relations (e.g., location of word or objects; Park, Puglisi, & Lutz, 1982), and temporal order (Kausler, Salthouse, & Saults, 1988).

Consistent with these behavioral findings, neuroimaging studies have demonstrated age-related decreases in neural activity associated with successful RM encoding and retrieval. For example, Cabeza and colleagues (1997) observed age-related decreases in left PFC activity during intentional learning of word pairs and suggested a direct relationship between PFC function and OAs' deficits in forming new semantic associations. Mitchell, Johnson, Raye, and D'Esposito (2000) also reported weaker PFC activity in OAs compared with young adults (YAs) during an RM task with object drawings and locations. Additionally, they observed reduced hippocampal activation in OAs in an associative encoding task that involved objects in an array. Such age-related reductions in hippocampal activation have also found in other RM studies using stimuli such as face–name pairs (Sperling et al., 2003) and face–scene pairs (Dennis et al., 2008).

Although an age-related RM deficit has been observed at both behavioral and neural levels, the source of this deficit remains unspecified. One popular behavioral approach for examining the source of RM deficits in aging has been to tax attentional resources at encoding by dividing attention and then to assess RM and IM in YAs. On the basis of the data suggesting that normal aging is often accompanied by a reduction in attentional resources (Madden, Spaniol, Bucur, & Whiting, 2005; Craik & Byrd, 1982) as well as the finding that reductions in attentional resources during episodic encoding greatly impairs subsequent memory for studied information (Anderson, Craik, & Naveh-Benjamin, 1998; Craik, Govoni, Naveh-Benjamin, & Anderson, 1996), prior studies have imposed a concurrent attentional task at encoding and then asked YAs to retrieve IM and RM. With this paradigm, prior studies have demonstrated that divided attention (DA) during encoding equally affects IM and RM in YAs, which is dissimilar to the disproportionate RM deficits observed in aging (e.g., Kilb & Naveh-Benjamin, 2007).

Recently, however, we demonstrated that reduced attentional resources at encoding equated YAs' performance on an RM test to that of OAs with full attention (FA) if the attentional task tagged a critical cognitive process required during RM formation (Kim & Giovanello, 2011). Specifically, using a simple item detection attention task, which we labeled as an item attention task, we replicated previous findings that DA at encoding proportionately impaired both RM and IM performance in YAs. In contrast, when the concurrent attention task required relational attention, which we defined as a cognitive process to allocate attention to relational aspects of different stimuli, the attention task disproportionately impaired RM performance in YAs, as in the case of OAs under FA. Critically, we found a strikingly similar pattern of RM impairment between YAs under the relational attention condition and OAs under FA: The disproportionate impairment in RM arose from a decreased hit rate for intact pairs and an increased false alarm rate for recombined pairs. Thus, these data suggest that a lack of attentional resources for relational information plays a critical role in age-related RM deficits.

Unlike studies focusing on the behavioral effects of attention on RM encoding, the neural effects of attention on age-related RM deficits have not been widely studied. In fact, only one PET study has investigated the effect of DA during RM encoding regarding its effect on age-related RM deficits (Anderson et al., 2000). Specifically, the researchers imposed an auditory detection task during encoding and retrieval of word pairs and demonstrated significantly reduced activity in left PFC in both YAs and OAs under DA. They also found reduced activity in the medial temporal lobe (MTL) under DA conditions for OAs. On the basis of the comparable reduction in prefrontal activation observed in OAs under FA and YAs under DA condition, Anderson et al. (2000) concluded that aging and DA affected RM encoding in a similar manner.

Using event-related fMRI, these prior studies provide important insight into the neural effects of reduced attention on encoding which later supports item recognition or recollection. Nonetheless, the neural correlates of IM encoding do not necessarily extend to those of RM encoding. In fact, previous studies have reported that encoding of RM involves distinct areas in PFC and MTL compared with that of IM (Blumenfeld & Ranganath, 2007; Davachi & Wagner, 2002). Furthermore, although recollection is associated with RM, it may be related more to intraitem RM (i.e., item–context association) than to interitem RM (i.e., item–item association). Moreover, as Park and Rugg (2008) have noted, there is evidence that the successful encoding of interitem RM is supported by distinct neural mechanisms from those that support either IM or item–context associations (Diana, Yonelinas, & Ranganath, 2007; Eichenbaum, Yonelinas, & Ranganath, 2007). Finally, by testing only YAs, the prior event-related studies (e.g., Uncapher & Rugg, 2005, 2008; Kensinger et al., 2003) do not offer insight into the effect of attention on age-related RM deficits. Thus, here, we tested both YAs and OAs during encoding of interitem RM to examine the neural effects of attention on successful RM encoding with regard to its role in age-related RM impairments.

In the current study, we used event-related fMRI to investigate the effect of different types of attentional tasks on the neural underpinnings of RM encoding in YAs and OAs. On the basis of the prior literature, we proposed the following hypotheses. First, we hypothesized that there would be significant activation in ventrolateral PFC (VLPFC), DLPFC, and the hippocampus in YAs during successful RM encoding. Second, in line with previous findings, we hypothesized that there would be a reduction in activity in PFC and hippocampus in OAs relative to YAs, during successful RM encoding. Finally, using qualitatively, rather than quantitatively, different attention tasks, we hypothesized that a lack of resources for relational attention plays a critical role in age-related RM deficits at both behavioral and neural levels. On the basis of our previous study (Kim & Giovanello, 2011), we predicted that YAs with reduced relational attention during RM encoding would show attenuated activation in PFC and hippocampus compared with YAs under FA conditions. In addition, we hypothesized that the neural effects of reduced relational attention would be similar to those of aging during successful RM encoding.

## METHODS

Thirteen healthy YAs (age = 18–31 years, mean = 20.9 years, SD = 3.26 years; 9 women; mean education = 14.5 years, SD = 1.56 years) and 12 healthy community-dwelling OAs (age = 66–89 years, mean = 74.8 years, SD = 8.36 years; 8 women; mean education = 15.8 years, SD = 2.22 years) with normal or corrected-to-normal vision participated in the experiment. Each participant provided informed written consent, and all were paid $20/hr for participation. Before participation, all participants received an MR safety screen, as well as a general health screen, and completed a battery of neuropsychological tests to assess memory, language, attention, visuospatial abilities, and general intellectual functioning. These tests included the Mini Mental State Examination, the American National Adult Reading Test, the Trail Making Test Parts A and B, the Vocabulary from the WAIS-III, and the Morningness–Eveningness questionnaire (Table 1). All participants were right-handed, native English speakers with no history of neurological or psychiatric conditions and were not taking psychotropic medications or medications known to affect MR signal. All participants completed RM encoding blocks under FA, under DA with an item attention task (DA-I), and under DA with a relational attention task (DA-R) conditions. Table 1. Mean Scores on Battery of Neuropsychological Tests MMSE (/30) ANART (/50) Trail Making A (sec) Trail Making A Errors (%) Trail Making B (sec) Trail Making B Errors (%) Vocab (/66) Morn/Eve (/86) YA 29.08 37.08 25.46 0.00 44.23 0.00 52.46 42.38 OA 29.67 44.33 29.33 0.00 54.42 0.00 59.17 61.50 MMSE (/30) ANART (/50) Trail Making A (sec) Trail Making A Errors (%) Trail Making B (sec) Trail Making B Errors (%) Vocab (/66) Morn/Eve (/86) YA 29.08 37.08 25.46 0.00 44.23 0.00 52.46 42.38 OA 29.67 44.33 29.33 0.00 54.42 0.00 59.17 61.50 MMSE = Mini Mental State Examination; ANART = American National Adult Reading Test; Vocab = Vocabulary from the WAIS-III; Morn/Eve = Morningness–Eveningness. ### Materials and Tasks A total of 240 word pairs were used for the memory tests (one to four syllables long; mean frequency = 101 occurrence/million; Francis & Kucera, 1982). There were three attention conditions (described below), and participants performed two runs of each attention condition (i.e., a total of six runs). Forty word pairs were presented in each encoding and retrieval phases. To avoid possible floor effects (see Anderson et al., 1998), intentional encoding was used, in which participants were informed of the nature of the memory tasks. Numbers from 1 to 10 served as stimuli for the two DA conditions, with all numbers written in the English alphabet (e.g., “THREE”). For the FA condition, meaningless strings of Xs and Ys were used and located in the same position as the number stimuli to balance the amount of visual input across all conditions. The number and string pairs were presented below a fixation cross, and word pairs appeared above the fixation cross (Figure 1). In each encoding block, one third of the trials were control trials to assess baseline activity during memory encoding. Control trials, consisting of strings of dollar signs (“$”) and pound signs (“#”), were presented in the same locations as the word and number pairs in each experimental trial. Participants were asked to detect whether the strings of dollar signs appeared on the left or right. A randomized jittered ISI for control trials ranged from 2 to 4, 6, 8, or 10 sec to increase the power to detect hemodynamic response differences specific to individual trial events. OptSeq (surfer.nm.mgh.harvard.edu/optseq) was used to determine the list order for each experimental and control trial that would optimally assess event-related activation. During the test phase, participants received RM tests corresponding to their encoding blocks. The test phase consisted of two types of pairs: word pairs previously seen at study (intact pairs) and word pairs consisting of words presented during the encoding phase but not together (recombined pairs). In each block, participants saw 40 word pairs, and all word stimuli were counterbalanced across participants using a Latin square design. The experiment was programmed with MacStim (Darby, 2006) using an Apple iBookG4.

Figure 1.

Trial sequence for each encoding condition. In the FA condition, participants were told to memorize word pairs ignoring strings presented below the word pairs (A). In the DA with an item attention task condition (B; DA-I), participants were instructed to detect an odd number on each trial while memorizing word pairs. In the DA with a relational attention task condition (B; DA-R), participants were instructed to compare two numbers and detect the numerically bigger number on each trial while encoding of word pairs. One third of the total trials in each condition were control trials consisting of dollar and pound signs. Participants were instructed to detect the location of dollar signs for these trials. The ISI for control trials ranged from 2, to 4, 6, 8, or 10 sec.

Figure 1.

Trial sequence for each encoding condition. In the FA condition, participants were told to memorize word pairs ignoring strings presented below the word pairs (A). In the DA with an item attention task condition (B; DA-I), participants were instructed to detect an odd number on each trial while memorizing word pairs. In the DA with a relational attention task condition (B; DA-R), participants were instructed to compare two numbers and detect the numerically bigger number on each trial while encoding of word pairs. One third of the total trials in each condition were control trials consisting of dollar and pound signs. Participants were instructed to detect the location of dollar signs for these trials. The ISI for control trials ranged from 2, to 4, 6, 8, or 10 sec.

### Procedures

Before entering the scanner environment, all participants completed a practice session that consisted of each attention condition. During the scanned experimental session, participants received two study–test runs for each attention condition, along with an anatomical scan. Functional MRI data were obtained during the encoding phase only. Each encoding run included 40 unrelated word pairs and approximately 20 control trials, with word pairs presented for 6 sec and attention stimuli (i.e., numbers or strings) presented for 3 sec. To produce the equivalent visual changes between word pairs and attention stimuli, the word pairs switched their locations every 3 sec (e.g., A–B, then B–A) as the attention stimuli changed their identities. A central fixation cross was presented on the screen for 500 msec to separate the different word pairs in time. Figure 1 depicts the procedure in each encoding phase.

In the FA condition, participants were instructed to memorize all word pairs for a later memory test and were asked to detect strings of dollar signs. Participants were also instructed to press a button with either index (left) or middle finger (right) of their right hand when they saw the next word pair, as well as when the two words switched their location. This manipulation was used to equate response requirements for finger movement between FA and DA conditions. In the DA with item attention (DA-I) condition, participants saw both word and number pairs, and they were asked to detect an odd number on each trial while memorizing the word pairs. In the DA with relational attention (DA-R) condition, participants were asked to compare two numbers on each trial and instructed to detect the numerically larger number while memorizing the word pairs. Thus, only the attention task in the DA-R condition involved relational processing.

On the basis of prior pilot testing, we inserted a problem-solving task block between study and test runs only for YAs. There were two purposes for having the problem-solving task. First, we intentionally matched performance between OAs' FA condition and YAs' DA-R condition to replicate the pattern of our behavioral study. Second, the problem-solving task block was inserted to ensure enough miss responses for YAs, such that a subsequent memory analysis could be performed (Wagner et al., 1998). For the problem solving task, YAs saw 40 equations (e.g., 1832 + 29 = 1851) and were instructed to judge whether each equation was true or false. The importance of accuracy on the problem solving test was stressed. Each equation remained on the computer screen for 6 sec. After the problem-solving task for YAs and after each encoding block for OAs, all participants received an RM test corresponding to the previous encoding phase. Forty word pairs were presented in each test, and for each of word pair, participants were asked to decide whether they had seen two words together or not in the previous encoding phase. Half of the pairs were intact pairs (presented in the same form as the first presentation during encoding, e.g., A–B) and the other half of the pairs were recombination of encoded pairs (e.g., study C–D and test C–E). Responses were recorded by pressing a button on an MR-compatible response box. The order of the attention conditions was counterbalanced across all participants.

### Imaging Methods and Analyses

#### Image Acquisition

All imaging data were acquired at the University of North Carolina's Biomedical Research Imaging Center on a Siemens 3-T Allegra head-only imaging system equipped for EPI (Siemens Medical Systems, Iselin, NJ) using a three-axis gradient head coil. Visual stimuli were back-projected onto a screen and viewed on an MR-comparable mirror mounted above the participant's head. Responses were recorded via response box using each participant's right hand. Head motion was restricted with a pillow and foam inserts. For each participant, an anatomical scan was acquired using a high-resolution T1-weighted MPRAGE sequence (TR = 1700 msec, TE = 4.38 msec, flip angle = 8°, field of view = 280 × 320, 160 slices, matrix = 224 × 256, resolution = 1.25 × 1.25 × 1.25 mm, acquisition time = 382 sec). After the anatomical scan, six functional runs were acquired for each participant. For the functional runs, imaging was performed using a T2*-weighted EPI sequence designed to minimize susceptibility artifact in the anterior hippocampal regions and fully volume the long axis of the hippocampus (TR = 2000 msec, TE = 30 msec, flip angle = 80°). Each brain volume was composed of 34 transverse slices (field of view = 243 × 243, matrix = 64 × 64, resolution = 3 × 3 × 3 mm, with a 5-mm skip between slices; slices were oriented along the long axis of the hippocampus, collected interleaved, inferior to superior). For all functional runs, data from the first two volumes were discarded to allow for stabilization of magnetic fields.

#### Image Processing and Statistical Analyses

Imaging data were preprocessed using SPM 8 (Wellcome Department of Cognitive Neurology, London) run within Matlab (Matlab Mathwork, Inc., Natick, MA). For preprocessing, data were slice-time corrected for acquisition order (referenced to the first slice), realigned, and unwarped to correct for motion across runs. Next, the images were spatially normalized (with trilinear interpolation and preserving the intensities of the original images) to the SPM EPI template corresponding to the Montreal Neurological Institute (MNI)-defined standardized brain space and then spatially smoothed with a Gaussian kernel of 8-mm FWHM. The time series were high pass filtered at 128 sec.

Statistical analyses were performed using the general linear model for event-related designs in SPM 8. For each participant, a whole-brain voxel-wise analysis was conducted in which individual events were modeled as a canonical hemodynamic response. Each event type was first modeled for each participant using a fixed effects analysis. To identify the neural correlates of successful encoding, the subsequent memory procedure (Wagner et al., 1998) was used. In this procedure, neural activity elicited by a series of study items is back-sorted according to whether the items are subsequently remembered or forgotten on a later memory test. To investigate the neural correlates of successful RM under qualitatively different attentional loads, subsequently remembered trials (i.e., encoding trials leading to “together” recognition responses for intact pairs), subsequently forgotten trials (i.e., encoding trials leading to “separate” recognition responses for intact pairs), and control trials were modeled separately for each attention condition. Trials that were used to construct recombined test pairs were excluded from the subsequent memory analysis because test responses to rearranged pairs corresponded to multiple encoding trials. To ensure the statistical power, a minimum of six trials per event type were modeled (see Giovanello, Kensinger, Wong, & Schacter, 2010; Chua, Schacter, Rand-Giovannetti, & Sperling, 2006, for examples of modeling less than 10 instances per event type). For each attention condition, the neural correlates of successful memory encoding were identified by directly comparing subsequently remembered and forgotten trials. Furthermore, separate pairwise contrasts of interest were conducted to compare the neural correlates of successful memory encoding under FA condition to that of each DA condition. After completion of the fixed effects analysis for each participant, the resulting least squares parameter estimates of the height of the modeled hemodynamic response for each condition were entered into between-subjects random effects analyses. For all random effects analyses, we used combined intensity and cluster size thresholds of p < .005 with a minimum cluster size of 10 contiguous voxels (k ≥ 10) to balance between types I and II error rates (see Lieberman & Cunningham, 2009, for more information). On the basis of our prior hypothesis that the hippocampus would be involved in an episodic encoding task as well as the small size of this structure, we used a reduced cluster size of k ≥ 5 for activity in this region only.

To examine the effect of each attentional load on the neural basis of RM encoding, pairwise t tests were conducted to compare contrast maps of successful memory encoding under FA and those under each DA condition for each age group. To identify differences in the neural correlates of RM between YAs and OAs, two-sample t tests were conducted for contrast maps of successful memory encoding from YAs under FA and contrast maps from OAs under FA. The results from the two sample t test were subsequently used as an inclusive mask for identifying common effects of aging and relational attention load on RM encoding at p < .05 (k ≥ 10). Thus, the conjoint probability following inclusive masking approached p < .00025 (Fisher, 1950), but this estimate should be taken with caution given that the contrasts were not completely independent (see Dennis et al., 2008, for the same masking procedure).

## RESULTS

### Behavioral Results

#### Attention Performance

Accuracy and RTs for correct responses were averaged for each DA condition in each age group (Table 2A). A 2 (Age) × 2 (Attention) mixed ANOVA for the accuracy data revealed neither significant main effects nor an interaction between Attention and Age (all ps > .05). However, the same analysis for the RT data revealed a significant interaction between the two factors, F(1, 23) = 4.70, p < .05. Specifically, OAs showed significantly slower RTs to the attention task with relation processing (DA-R) than to the attention task with item processing (DA-I), t(11) = 2.23, p < .05, whereas RTs in these two attention tasks were not significantly different from each other in YAs. Moreover, independent sample t tests for the RT data between YAs and OAs revealed that OAs' responses were significantly slower in the relational attention task compared with that of YAs, t(23) = 2.17, p < .05. These results suggest that OAs exhibit more difficulty with an attention task involving relational attention than with an attention task requiring little or no relational attention. Such differences were not observed in the YA group, confirming that dual-task performance on the two attention tasks was equivalent.1

Table 2.

Behavioral Results for Each Attention and Memory Condition

A. Performance in Attention Tasks in YAs and OAs
Condition
YAs
OAs
Accuracy (%)
RTs (msec)
Accuracy (%)
RTs (msec)
DA-I 96 (4)  1403 (275) 97 (3)  1537 (357)
DA-R 95 (4)  1378 (265) 96 (3)  1633 (321)

B. Performance in RM Tasks in YAs and OAs
Condition YAs OAs
Hits (H) False Alarms (F) H − F Hits (H) False Alarms (F) H − F
FA 0.82 (0.12) 0.17 (0.12) 0.65 (0.18) 0.74 (0.13) 0.3 (0.14) 0.44 (0.22)
DA-I 0.80 (0.13) 0.26 (0.16) 0.54 (0.16) 0.72 (0.14) 0.28 (0.14) 0.44 (0.24)
DA-R 0.77 (0.08) 0.33 (0.14) 0.44 (0.16) 0.67 (0.13) 0.34 (0.12) 0.33 (0.22)
A. Performance in Attention Tasks in YAs and OAs
Condition
YAs
OAs
Accuracy (%)
RTs (msec)
Accuracy (%)
RTs (msec)
DA-I 96 (4)  1403 (275) 97 (3)  1537 (357)
DA-R 95 (4)  1378 (265) 96 (3)  1633 (321)

B. Performance in RM Tasks in YAs and OAs
Condition YAs OAs
Hits (H) False Alarms (F) H − F Hits (H) False Alarms (F) H − F
FA 0.82 (0.12) 0.17 (0.12) 0.65 (0.18) 0.74 (0.13) 0.3 (0.14) 0.44 (0.22)
DA-I 0.80 (0.13) 0.26 (0.16) 0.54 (0.16) 0.72 (0.14) 0.28 (0.14) 0.44 (0.24)
DA-R 0.77 (0.08) 0.33 (0.14) 0.44 (0.16) 0.67 (0.13) 0.34 (0.12) 0.33 (0.22)

Numbers in parentheses indicate standard deviation.

#### Memory Performance

For memory accuracy, the proportion of hits minus the proportion of false alarms was computed for each participant and then averaged over each attention condition for each age group (Table 2B). A 2 (Age) × 3 (Attention) mixed ANOVA revealed a significant main effect of Attention, F(2, 46) = 27.73, p < .05, and a significant interaction between Attention and Age, F(2, 46) = 4.29, p < .05. Paired t tests contrasting each attention condition in each age group revealed significant differences in RM performance between FA and DA-I condition, MD (mean difference) = .11, t(12) = 3.82, p < .05, between FA and DA-R condition, MD = .21, t(12) = 5.94, p < .05, and between DA-I and DA-R condition, MD = .10, t(12) = 2.68, p < .05, in YAs. These results indicate that a reduction in relational attention during RM encoding disproportionately impaired YAs' RM performance, although both item and relational attention tasks significantly worsened YAs' memory performance compared with the FA condition. The effect of Item Attention Task on RM was different in OAs; there was no effect of Item Attention Task on RM in OAs, MD = .01, t(11) = .32, p > .05. However, OAs showed significantly lower memory accuracy under the DA-R compared with both the FA, MD = .11, t(11) = 3.69, p < .05, and the DA-I conditions, MD = .10, t(11) = 3.72, p < .05. Finally, OAs revealed significantly lower RM accuracy than YAs under FA, MD = .20, t(23) = 2.61, p < .05, confirming the age-related RM deficits.2

For retrieval latency, a 2 × 2 mixed ANOVA with Age and Attention Condition as factors revealed neither main effects nor an interaction between two factors, all ps > .05. Separate t tests for each factor also revealed no significant differences in any of the comparisons, indicating that the findings from the memory accuracy data are not a consequence of speed–accuracy trade-off.

### Imaging Results

#### Neural Correlates of Successful RM Encoding in YAs and OAs under FA

Using the subsequent memory procedure (Wagner et al., 1998), neural regions for successful RM encoding were identified for both YAs and OAs under the FA condition. Consistent with prior studies examining the neural correlates of RM encoding, YAs showed significant activity in bilateral VLPFC (BA 11/45/47), bilateral DLPFC (BA 6/9/10/46), left inferior parietal lobule (IPL; BA 40), left superior parietal lobule (SPL; BA 7), and left parahippocampal gyrus (BA 36), as well as in the left anterior hippocampus (Table 3; Figure 2). The same analysis was conducted for OAs under FA. As listed in Table 3, the analysis revealed significant activation in left inferior and superior temporal gyrus (BA 20/22), right perirhinal cortex (BA 35), and right insula (BA 13). Notably, none of the brain regions activated by YAs during successful RM encoding under FA were significantly activated in OAs under the FA condition using the same threshold of p < .005, (k ≥ 10).

Table 3.

Neural Correlates of Successful RM Encoding in YAs and OAs

Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
Hit > Miss in YAs under FA Middle frontal gyrus 32 60 5.18 112
38 55 5.15
Superior frontal gyrus 10 13 56 4.84 31
Middle temporal gyrus 37 48 −60 4.82 36
Inferior frontal gyrus 11 −24 26 −20 4.71 32
Cuneus 19 −16 −92 27 4.63 46
Cuneus 30 −6 −70 4.56 128
Parahippocampal gyrus 36 −34 −24 −22 4.53 16
Precentral gyrus 34 37 4.5 51
Middle frontal gyrus 42 10 38 3.55
Cingulate gyrus 24 −12 −20 36 4.5 12
Inferior frontal gyrus 45 −34 28 4.45 17
Precentral gyrus −38 −11 48 4.4 35
Superior temporal gyrus 42 67 −28 18 4.29 14
Inferior frontal gyrus 45 44 22 15 4.25 47
Middle frontal gyrus 10 −30 38 15 4.24 21
Middle temporal gyrus 19 40 −80 22 4.18 18
Superior frontal gyrus −18 −12 61 4.13 24
Superior frontal gyrus 68 4.1 18
Inferior frontal gyrus 47 50 16 −1 4.09 35
Superior frontal gyrus −40 35 30 4.05 26
Hippocampus* n/a −20 −7 −22 3.83
Supramarginal gyrus 40 −59 −48 21 3.7 13
Superior temporal gyrus 13 53 −40 19 3.53 11
SPL 26 −56 51 3.46 13
Middle frontal gyrus 46 51 45 3.42 11
IPL 40 −38 −44 46 3.38 31
Hit > Miss in OAs under FA Inferior temporal gyrus 20 −38 −13 −33 4.64 11
Perirhinal cortex 35 22 −9 −21 4.41 18
Insula 13 46 −8 4.09 26
Superior temporal gyrus 22 −48 −6 −5 3.5 12
Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
Hit > Miss in YAs under FA Middle frontal gyrus 32 60 5.18 112
38 55 5.15
Superior frontal gyrus 10 13 56 4.84 31
Middle temporal gyrus 37 48 −60 4.82 36
Inferior frontal gyrus 11 −24 26 −20 4.71 32
Cuneus 19 −16 −92 27 4.63 46
Cuneus 30 −6 −70 4.56 128
Parahippocampal gyrus 36 −34 −24 −22 4.53 16
Precentral gyrus 34 37 4.5 51
Middle frontal gyrus 42 10 38 3.55
Cingulate gyrus 24 −12 −20 36 4.5 12
Inferior frontal gyrus 45 −34 28 4.45 17
Precentral gyrus −38 −11 48 4.4 35
Superior temporal gyrus 42 67 −28 18 4.29 14
Inferior frontal gyrus 45 44 22 15 4.25 47
Middle frontal gyrus 10 −30 38 15 4.24 21
Middle temporal gyrus 19 40 −80 22 4.18 18
Superior frontal gyrus −18 −12 61 4.13 24
Superior frontal gyrus 68 4.1 18
Inferior frontal gyrus 47 50 16 −1 4.09 35
Superior frontal gyrus −40 35 30 4.05 26
Hippocampus* n/a −20 −7 −22 3.83
Supramarginal gyrus 40 −59 −48 21 3.7 13
Superior temporal gyrus 13 53 −40 19 3.53 11
SPL 26 −56 51 3.46 13
Middle frontal gyrus 46 51 45 3.42 11
IPL 40 −38 −44 46 3.38 31
Hit > Miss in OAs under FA Inferior temporal gyrus 20 −38 −13 −33 4.64 11
Perirhinal cortex 35 22 −9 −21 4.41 18
Insula 13 46 −8 4.09 26
Superior temporal gyrus 22 −48 −6 −5 3.5 12

Regions significant at uncorrected p < .005 with a cluster extent threshold of k ≥ 10 (for *, k ≥ 5). R = right; L = left; BA = approximate Brodmann's area based on coordinates.

Figure 2.

Neural correlates of successful RM encoding in YAs. The regions were significantly activated for the contrast of “Hit > Miss” under the FA condition in YAs (p < .005 (uncorrected) with a cluster extent threshold of k ≥ 10). L = left; R = right.

Figure 2.

Neural correlates of successful RM encoding in YAs. The regions were significantly activated for the contrast of “Hit > Miss” under the FA condition in YAs (p < .005 (uncorrected) with a cluster extent threshold of k ≥ 10). L = left; R = right.

#### Neural Effects of Aging on Successful RM Encoding

To examine whether aging significantly attenuated activity in the brain regions associated with successful RM formation in YAs, two-sample t tests were conducted with YAs and OAs under the FA condition. The results showed that YAs engaged regions in lateral PFC and parietal cortex significantly greater than OAs during successful RM encoding. Specifically, brain activity in bilateral VLPFC (BA 11/45/47), bilateral DLPFC (BA 6/9/10/46), and left IPL (BA 40) were significantly attenuated in OAs compared with YAs (Table 4; Figure 3A). Relative to YAs, OAs exhibited greater activity in the right anterior cingulate (BA 24/32), right insula (BA 13), and cingulate gyrus (BA 24), possibly indicating that the RM task was more effortful for OAs compared with YAs (Allman, Hakeem, Erwin, Nimchinsky, & Hof, 2001). Finally, OAs showed greater activity in right superior temporal gyrus (BA 22) and left inferior frontal gyrus (BA 47) than YAs.

Table 4.

Neural Effects of Aging on Successful RM Encoding

Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
YA > OA for FA (Hit > Miss) Middle frontal gyrus 30 60 5.11 60
Precentral gyrus 34 36 4.34 133
Superior frontal gyrus 66 4.12 46
SPL −32 −56 54 3.92 58
Subgyral 18 57 3.87 16
Middle frontal gyrus 46 44 28 17 3.83 54
Superior frontal gyrus −22 68 3.79 21
Cuneus 17 −10 −83 13 3.73 14
Middle temporal gyrus 19 42 −80 24 3.65 14
Middle frontal gyrus 51 10 46 3.5 50
Middle frontal gyrus 10 48 47 14 3.49 11
Middle frontal gyrus 34 52 3.48 25
Middle occipital gyrus 19 32 −85 12 3.47 28
IPL 40 −42 −44 48 3.47 67
−32 −48 45 3.19
Inferior frontal gyrus 45 55 18 14 3.41 63
Inferior frontal gyrus 47 50 16 −1 3.2
Middle frontal gyrus 10 −28 36 17 3.41 16
Precuneus 26 −52 50 3.37 11
Middle frontal gyrus 11 −32 50 −11 3.25 23
Middle frontal gyrus −38 33 35 3.21 15
IPL 40 −63 −33 37 3.18 11
−65 −28 31 3.12
OA > YA for FA (Hit > Miss) Insula 13 36 −10 24 3.85 13
Cingulate gyrus 24 16 −7 45 3.82 17
Anterior cingulate 32 10 34 −10 3.56 30
Superior temporal gyrus 22 34 −50 14 3.38 10
Anterior cingulate 24 27 −1 3.35 44
Inferior frontal gyrus 47 −28 −14 3.27 19
Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
YA > OA for FA (Hit > Miss) Middle frontal gyrus 30 60 5.11 60
Precentral gyrus 34 36 4.34 133
Superior frontal gyrus 66 4.12 46
SPL −32 −56 54 3.92 58
Subgyral 18 57 3.87 16
Middle frontal gyrus 46 44 28 17 3.83 54
Superior frontal gyrus −22 68 3.79 21
Cuneus 17 −10 −83 13 3.73 14
Middle temporal gyrus 19 42 −80 24 3.65 14
Middle frontal gyrus 51 10 46 3.5 50
Middle frontal gyrus 10 48 47 14 3.49 11
Middle frontal gyrus 34 52 3.48 25
Middle occipital gyrus 19 32 −85 12 3.47 28
IPL 40 −42 −44 48 3.47 67
−32 −48 45 3.19
Inferior frontal gyrus 45 55 18 14 3.41 63
Inferior frontal gyrus 47 50 16 −1 3.2
Middle frontal gyrus 10 −28 36 17 3.41 16
Precuneus 26 −52 50 3.37 11
Middle frontal gyrus 11 −32 50 −11 3.25 23
Middle frontal gyrus −38 33 35 3.21 15
IPL 40 −63 −33 37 3.18 11
−65 −28 31 3.12
OA > YA for FA (Hit > Miss) Insula 13 36 −10 24 3.85 13
Cingulate gyrus 24 16 −7 45 3.82 17
Anterior cingulate 32 10 34 −10 3.56 30
Superior temporal gyrus 22 34 −50 14 3.38 10
Anterior cingulate 24 27 −1 3.35 44
Inferior frontal gyrus 47 −28 −14 3.27 19

Regions significant at uncorrected p < .005 with a cluster extent threshold of k ≥ 10. R = right; L = left; BA = approximate Brodmann's area based on coordinates.

Figure 3.

(A) Neural effects of aging on successful RM encoding. The figure shows neural regions that were significantly activated with the contrast of “YA(Hit > Miss) > OA(Hit > Miss)” under the FA condition (p < .005 uncorrected, with a cluster extent threshold of k ≥ 10). (B) Neural effects of reduced relational attention on successful RM encoding in YAs. The figure shows neural regions that were significantly activated with the contrast of “(FA > DA-R) for (Hit > Miss)” in YAs (p < .005 uncorrected, with a cluster extent threshold of k ≥ 10). L = left; R = right.

Figure 3.

(A) Neural effects of aging on successful RM encoding. The figure shows neural regions that were significantly activated with the contrast of “YA(Hit > Miss) > OA(Hit > Miss)” under the FA condition (p < .005 uncorrected, with a cluster extent threshold of k ≥ 10). (B) Neural effects of reduced relational attention on successful RM encoding in YAs. The figure shows neural regions that were significantly activated with the contrast of “(FA > DA-R) for (Hit > Miss)” in YAs (p < .005 uncorrected, with a cluster extent threshold of k ≥ 10). L = left; R = right.

#### Neural Effects of Different Attention Tasks on Successful RM Encoding in YAs

To test whether each attention task significantly attenuated activity in the brain regions associated with successful RM encoding in YAs, separate analyses between the FA and each DA condition were conducted. As shown in Table 5, only the right posterior cingulate (BA 29) and the right medial frontal gyrus (BA 9) showed significant attenuation by item attention task load in YAs compared with the FA condition. However, when YAs' attention was divided with the relational attention task during encoding, YAs exhibited significant reduction in activity in most of the neural regions associated with successful RM formation. As shown in Figure 3, there was a reduction in activity for YAs under the DA-R condition compared with the FA condition, in bilateral VLPFC (BA 47), bilateral DLPFC (BA 6/9/10), left IPL (BA 40), and left hippocampus.

Table 5.

Neural Effects of Attention on Successful RM Encoding in YAs

Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
FA > DA in YA: FA > DA-I for Hit > Miss Posterior cingulate 29 14 −42 18 6.17 38
Medial frontal gyrus −20 29 25 4.16 20
FA > DA-R for Hit > Miss Cuneus 18 −14 −75 17 6.62 216
Superior temporal gyrus 22 −61 −50 17 5.52 106
Middle temporal gyrus 39 −57 −64 11 3.74
Superior frontal gyrus −4 36 50 4.97 34
Middle frontal gyrus 10 −30 38 17 4.89 33
Middle frontal gyrus −42 33 32 4.81 149
IPL 40 −61 −41 39 4.8 82
−57 −33 48 3.5
Inferior temporal gyrus 20 −51 −55 −12 4.57 23
Inferior temporal gyrus 20 51 −11 −33 4.47 38
Middle frontal gyrus 11 −28 46 −14 4.43 77
Middle frontal gyrus 10 −24 54 −9 3.55
Precentral gyrus −38 −11 52 4.43 40
Superior temporal gyrus 22 −38 −52 17 4.38 16
Superior frontal gyrus −12 49 45 4.32 38
Precentral gyrus 38 37 4.28 29
Middle frontal gyrus −24 20 56 4.21 25
Superior frontal gyrus −20 16 51 3.87
Lingual gyrus 19 −26 −62 4.14 16
Inferior frontal gyrus 47 −24 22 −20 4.06 35
47 −22 17 −13 3.15
Middle frontal gyrus 38 37 35 3.99 12
Postcentral gyrus 51 −24 55 3.94 19
Medial frontal gyrus 12 47 3.91 65
Superior frontal gyrus 14 53 3.4
Medial frontal gyrus 50 3.29
Superior frontal gyrus −20 68 3.89 20
Precuneus −2 −79 45 3.89 13
Middle temporal gyrus 19 50 −60 14 3.88 24
Cingulate gyrus 32 −12 21 34 3.85 18
Middle temporal gyrus 37 −51 −52 3.83 17
Inferior frontal gyrus 13 −36 22 3.82 15
Inferior frontal gyrus 47 50 16 −1 3.79 24
Superior frontal gyrus 24 66 3.76 24
Fusiform gyrus 37 40 −44 −16 3.7 20
Cuneus 17 14 −77 13 3.65 22
Posterior cingulate 29 14 −42 13 3.65 40
Inferior frontal gyrus 47 34 25 −15 3.64 10
Inferior frontal gyrus 13 32 −12 3.63 10
Posterior cingulate 23 −2 −32 22 3.57 10
Superior temporal gyrus 38 −55 −9 3.56 14
Cingulate gyrus 32 14 21 28 3.53 13
Parahippocampal gyrus 28 18 −1 −12 3.5 27
IPL 40 51 −41 41 3.45 24
Hippocampus* n/a −30 −7 −16 3.26
DA > FA in YA: DA-I > FA for Hit > Miss Precentral gyrus 44 −46 18 5.86 27
Middle temporal gyrus 37 −44 −54 5.35 60
Subgyral 37 −46 −49 −4 3.28
Medial frontal gyrus 18 −13 50 5.35 93
Cingulate gyrus 24 20 −12 39 5.18
Middle temporal gyrus 39 36 −53 21 5.04 28
Insula 13 40 −39 26 4.63 24
Middle occipital gyrus 19 −34 −74 3.85 20
Insula 13 −42 −15 3.58 11
−34 −40 22 3.55 19
DA-R > FA for Hit > Miss Middle occipital gyrus 19 −36 −75 4.49 11
Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
FA > DA in YA: FA > DA-I for Hit > Miss Posterior cingulate 29 14 −42 18 6.17 38
Medial frontal gyrus −20 29 25 4.16 20
FA > DA-R for Hit > Miss Cuneus 18 −14 −75 17 6.62 216
Superior temporal gyrus 22 −61 −50 17 5.52 106
Middle temporal gyrus 39 −57 −64 11 3.74
Superior frontal gyrus −4 36 50 4.97 34
Middle frontal gyrus 10 −30 38 17 4.89 33
Middle frontal gyrus −42 33 32 4.81 149
IPL 40 −61 −41 39 4.8 82
−57 −33 48 3.5
Inferior temporal gyrus 20 −51 −55 −12 4.57 23
Inferior temporal gyrus 20 51 −11 −33 4.47 38
Middle frontal gyrus 11 −28 46 −14 4.43 77
Middle frontal gyrus 10 −24 54 −9 3.55
Precentral gyrus −38 −11 52 4.43 40
Superior temporal gyrus 22 −38 −52 17 4.38 16
Superior frontal gyrus −12 49 45 4.32 38
Precentral gyrus 38 37 4.28 29
Middle frontal gyrus −24 20 56 4.21 25
Superior frontal gyrus −20 16 51 3.87
Lingual gyrus 19 −26 −62 4.14 16
Inferior frontal gyrus 47 −24 22 −20 4.06 35
47 −22 17 −13 3.15
Middle frontal gyrus 38 37 35 3.99 12
Postcentral gyrus 51 −24 55 3.94 19
Medial frontal gyrus 12 47 3.91 65
Superior frontal gyrus 14 53 3.4
Medial frontal gyrus 50 3.29
Superior frontal gyrus −20 68 3.89 20
Precuneus −2 −79 45 3.89 13
Middle temporal gyrus 19 50 −60 14 3.88 24
Cingulate gyrus 32 −12 21 34 3.85 18
Middle temporal gyrus 37 −51 −52 3.83 17
Inferior frontal gyrus 13 −36 22 3.82 15
Inferior frontal gyrus 47 50 16 −1 3.79 24
Superior frontal gyrus 24 66 3.76 24
Fusiform gyrus 37 40 −44 −16 3.7 20
Cuneus 17 14 −77 13 3.65 22
Posterior cingulate 29 14 −42 13 3.65 40
Inferior frontal gyrus 47 34 25 −15 3.64 10
Inferior frontal gyrus 13 32 −12 3.63 10
Posterior cingulate 23 −2 −32 22 3.57 10
Superior temporal gyrus 38 −55 −9 3.56 14
Cingulate gyrus 32 14 21 28 3.53 13
Parahippocampal gyrus 28 18 −1 −12 3.5 27
IPL 40 51 −41 41 3.45 24
Hippocampus* n/a −30 −7 −16 3.26
DA > FA in YA: DA-I > FA for Hit > Miss Precentral gyrus 44 −46 18 5.86 27
Middle temporal gyrus 37 −44 −54 5.35 60
Subgyral 37 −46 −49 −4 3.28
Medial frontal gyrus 18 −13 50 5.35 93
Cingulate gyrus 24 20 −12 39 5.18
Middle temporal gyrus 39 36 −53 21 5.04 28
Insula 13 40 −39 26 4.63 24
Middle occipital gyrus 19 −34 −74 3.85 20
Insula 13 −42 −15 3.58 11
−34 −40 22 3.55 19
DA-R > FA for Hit > Miss Middle occipital gyrus 19 −36 −75 4.49 11

Regions significant at uncorrected p < .005 with a cluster extent threshold of k ≥ 10 (for *, k ≥ 5). R = right; L = left; BA = approximate Brodmann's area based on coordinates.

One of our hypotheses was that the attention task requiring relational processing would selectively affect the neural correlates of successful RM encoding in YAs, whereas the attention task involving minimal relational processing would not. To test this hypothesis, we used a sequential masking procedure to examine whether the subsequent memory effects found in the FA condition were selectively attenuated in either the DA-I or the DA-R conditions in YAs (see Uncapher & Rugg, 2008, for the same masking procedure). First, to identify effects that were selectively attenuated by the relational attention task (i.e., DA-R), we used the interaction contrast for the task of interest (i.e., FA(Hit > Miss) × DA-R(Hit > Miss)) as an inclusive mask (p < .05) for the regions demonstrating subsequent memory effects in the FA condition (i.e., FA(Hit > Miss), thresholded at p < .005, k ≥ 10). The outcome of this procedure was then exclusively masked by the alternate interaction (i.e., FA(Hit > Miss) × DA-I(Hit > Miss); p < .05). A parallel analysis was conducted to identify regions where the subsequent memory effects in the FA condition were selectively attenuated by the item attention task (i.e., DA-I).

As shown in Table 6, a number of regions demonstrating subsequent memory effects in the FA condition were selectively attenuated by the relational attention task compared with the item attention task. These regions included bilateral VLPFC (BA 11/45/47), bilateral DLPFC (BA 6/9/10/46), and left IPL (BA 40). Interestingly, no brain regions showed selectively reduced activity by the item attention task compared with the relational attention task. Of note, dual-task performance on the two attention tasks was equivalent under the same encoding load as revealed in the behavioral data from the two tasks. Together, these results support our hypothesis that the concurrent relational attention task attenuates neural activity observed during RM encoding in YAs, whereas an attention task involving minimal relational processing does not.

Table 6.

Subsequent Memory Effects Selectively Attenuated by Relational Attention Task in YAs

Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
Attenuated by DA-R condition only Middle frontal gyrus 32 60 5.18 91
38 55 5.15
46 53 3.98
Superior frontal gyrus 10 13 56 4.84 23
Middle temporal gyrus 37 48 −60 4.82 32
Inferior frontal gyrus 11 −24 26 −20 4.71 31
Cuneus 19 −16 −92 27 4.63 18
Cuneus 30 −6 −70 4.56 95
17 −10 −83 13
Middle frontal gyrus 42 10 38 4.5 49
34 37 3.55
Precentral gyrus −38 −11 48 4.4 35
Inferior frontal gyrus 45 44 22 15 4.25 36
Middle frontal gyrus 10 −30 38 15 4.24 20
Superior frontal gyrus −18 −12 61 4.13 22
Superior frontal gyrus 68 4.1 18
Inferior frontal gyrus 47 50 16 −1 4.09 35
Superior frontal gyrus −40 35 30 4.05 26
Supramarginal gyrus 40 −59 −48 21 3.7 13
Middle frontal gyrus 46 50 45 3.33 10
Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
Attenuated by DA-R condition only Middle frontal gyrus 32 60 5.18 91
38 55 5.15
46 53 3.98
Superior frontal gyrus 10 13 56 4.84 23
Middle temporal gyrus 37 48 −60 4.82 32
Inferior frontal gyrus 11 −24 26 −20 4.71 31
Cuneus 19 −16 −92 27 4.63 18
Cuneus 30 −6 −70 4.56 95
17 −10 −83 13
Middle frontal gyrus 42 10 38 4.5 49
34 37 3.55
Precentral gyrus −38 −11 48 4.4 35
Inferior frontal gyrus 45 44 22 15 4.25 36
Middle frontal gyrus 10 −30 38 15 4.24 20
Superior frontal gyrus −18 −12 61 4.13 22
Superior frontal gyrus 68 4.1 18
Inferior frontal gyrus 47 50 16 −1 4.09 35
Superior frontal gyrus −40 35 30 4.05 26
Supramarginal gyrus 40 −59 −48 21 3.7 13
Middle frontal gyrus 46 50 45 3.33 10

Regions significant at uncorrected p < .005 with a cluster extent threshold of k ≥ 10. R = right; L = left; BA = approximate Brodmann's area based on coordinates.

#### Neural Effect of Different Attention Tasks on Successful RM Encoding in OAs

As noted in Table 7, OAs showed greater activity in the FA condition compared with the DA-I condition in left inferior temporal gyrus (BA 20), right middle temporal gyrus (BA 39), right VLPFC (BA 47), right middle frontal gyrus (BA 6), right cingulated gyrus (BA 31), and left medial frontal gyrus (BA 10). Interestingly, the reverse contrast (i.e., FA(Hit > Miss) < DA-I(Hit − Miss)) revealed significant activity in additional regions observed during successful RM formation, including bilateral DLPFC (BA 6/9), right inferior frontal gyrus (BA 44), and left SPL. This neural pattern fits with our behavioral results found in OAs. Unlike YAs, OAs did not show decreased RM performance under the DA-I than under the FA condition. The additional recruitment of frontal and parietal regions in OAs may have mediated the behavioral performance observed in the FA and the DA-I conditions.

Table 7.

Neural Effects of Attention on Successful RM Encoding in OAs

Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
FA > DA in OA: FA > DA-I for Hit > Miss Inferior temporal gyrus 20 −44 −11 −33 5.47 69
Middle temporal gyrus 39 50 −75 13 4.75 45
Inferior frontal gyrus 47 26 21 −14 4.36 10
Cingulate gyrus 31 −27 35 4.28 52
Middle frontal gyrus 26 −10 41 4.1 12
Medial frontal gyrus 10 −16 49 18 3.94 10
FA > DA-R for Hit > Miss Amygdala/hippocampus n/a −16 −8 −10 5.82 30
Superior temporal gyrus 38 42 18 −33 5.57 57
Insula 13 42 −8 5.25 42
Superior temporal gyrus 38 30 −42 5.23 17
Inferior frontal gyrus 13 −32 −12 5.15 92
Medial frontal gyrus 49 42 4.53 10
Anterior cingulate 25 17 −1 4.5 29
Superior frontal gyrus −10 47 46 4.05 17
Inferior frontal gyrus 47 36 30 −17 3.67 12
Lingual gyrus 28 26 −30 3.64 20
DA > FA in OA: DA-I > FA for Hit > Miss Fusiform gyrus 37 38 −45 −13 5.64 25
SPL −30 −59 56 5.49 48
Middle temporal gyrus 39 −42 −52 14 4.89 38
Precuneus 31 −28 −71 22 4.89 13
Inferior frontal gyrus 44 55 16 10 4.45 20
Superior temporal gyrus 21 −51 −26 −5 4.44 16
Posterior cingulate 23 −30 20 4.21 26
Supramarginal gyrus 40 −40 −45 32 4.16 30
Middle frontal gyrus 38 11 57 4.12 11
Supramarginal gyrus 40 −61 −47 23 4.08 11
Cingulate gyrus 32 −14 21 39 4.01 11
Anterior cingulate 32 18 34 24 3.93 11
Superior frontal gyrus 20 40 29 3.34
Medial frontal gyrus 57 3.66 13
Middle frontal gyrus −30 −11 59 3.43 11
DA-R > FA for Hit > Miss Cingulate gyrus 32 24 11 34 5.31 87
Precuneus −20 −52 43 5.16 130
Cingulate gyrus 31 −20 −23 40 5.15 44
Middle frontal gyrus 11 −30 46 −9 4.33 33
Contrast
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
FA > DA in OA: FA > DA-I for Hit > Miss Inferior temporal gyrus 20 −44 −11 −33 5.47 69
Middle temporal gyrus 39 50 −75 13 4.75 45
Inferior frontal gyrus 47 26 21 −14 4.36 10
Cingulate gyrus 31 −27 35 4.28 52
Middle frontal gyrus 26 −10 41 4.1 12
Medial frontal gyrus 10 −16 49 18 3.94 10
FA > DA-R for Hit > Miss Amygdala/hippocampus n/a −16 −8 −10 5.82 30
Superior temporal gyrus 38 42 18 −33 5.57 57
Insula 13 42 −8 5.25 42
Superior temporal gyrus 38 30 −42 5.23 17
Inferior frontal gyrus 13 −32 −12 5.15 92
Medial frontal gyrus 49 42 4.53 10
Anterior cingulate 25 17 −1 4.5 29
Superior frontal gyrus −10 47 46 4.05 17
Inferior frontal gyrus 47 36 30 −17 3.67 12
Lingual gyrus 28 26 −30 3.64 20
DA > FA in OA: DA-I > FA for Hit > Miss Fusiform gyrus 37 38 −45 −13 5.64 25
SPL −30 −59 56 5.49 48
Middle temporal gyrus 39 −42 −52 14 4.89 38
Precuneus 31 −28 −71 22 4.89 13
Inferior frontal gyrus 44 55 16 10 4.45 20
Superior temporal gyrus 21 −51 −26 −5 4.44 16
Posterior cingulate 23 −30 20 4.21 26
Supramarginal gyrus 40 −40 −45 32 4.16 30
Middle frontal gyrus 38 11 57 4.12 11
Supramarginal gyrus 40 −61 −47 23 4.08 11
Cingulate gyrus 32 −14 21 39 4.01 11
Anterior cingulate 32 18 34 24 3.93 11
Superior frontal gyrus 20 40 29 3.34
Medial frontal gyrus 57 3.66 13
Middle frontal gyrus −30 −11 59 3.43 11
DA-R > FA for Hit > Miss Cingulate gyrus 32 24 11 34 5.31 87
Precuneus −20 −52 43 5.16 130
Cingulate gyrus 31 −20 −23 40 5.15 44
Middle frontal gyrus 11 −30 46 −9 4.33 33

Regions significant at uncorrected p < .005 with a cluster extent threshold of k ≥ 10. R = right; L = left; BA = approximate Brodmann's area based on coordinates.

In contrast, compared with the FA condition, OAs showed greater behavioral impairment on the RM task when a relational attention task was imposed during encoding. Consistent with this behavioral pattern, OAs exhibited attenuated activation in bilateral inferior prefrontal gyrus (BA 13/47), left superior frontal gyrus (BA 8), right superior temporal gyrus (BA 38), and the left anterior hippocampus in the DA-R than in the FA condition. Additionally, OAs showed greater activity in the FA than in the DA-R condition in right medial frontal gyrus (BA 8), right insula (BA 13), and right anterior cingulate (BA 25). Unlike in the DA-I condition, no regions related to successful RM formation showed greater activity in the DA-R than the FA condition in OAs.

#### The Common Effect of Aging and Relational Attention on Successful RM Encoding

Finally, to examine whether the effect of reduced relational attention and the effect of aging on the RM encoding are similar at the neural level, conjunction analyses were conducted with the data from the two sample t tests between YAs and OAs under their FA and the data from the paired sample t tests between YAs' FA and each DA condition. The conjunction analysis between YAs under the DA-I condition and OAs under the FA condition revealed no regions showing significant reduction by both age and attention factor compared with YAs under FA. In contrast, a subset of the neural regions associated with YAs' successful RM encoding, including bilateral DLPFC (BA 6/9/10), right VLPFC (BA 11/45/47), and left IPL (BA 40), were significantly attenuated both by relational attention load and by aging (Table 8; Figure 4).

Table 8.

Common Regions Where YAs under DA with Relation Task and OAs under FA Failed to Show Activation Compared with YAs under FA

Conjunction
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
YA, FA > DA-R (Hit > Miss) inclusively masked with YA > OA, FA (Hit > Miss) Middle frontal gyrus 10 −30 38 17 4.89 16
Middle frontal gyrus −42 33 32 4.81 15
Precentral gyrus 38 37 4.28 105
Superior frontal gyrus −20 68 3.89 20
Cuneus 17 −12 −81 13 3.83 14
Inferior frontal gyrus 47 50 16 −1 3.79 44
Precentral gyrus 44 53 18 2.47
Inferior frontal gyrus 45 59 22 14 2.23
IPL 40 −65 −28 31 3.63 11
IPL 40 −63 −33 37 3.19
Middle frontal gyrus 11 −30 46 −14 3.62 23
Superior frontal gyrus 24 64 3.54 60
Middle frontal gyrus 30 11 58 3.2
Insula 13 40 26 19 3.36 39
Inferior frontal gyrus 45 46 22 15 2.55
Subgyral 18 57 3.05 15
Superior frontal gyrus 68 2.97 35
Middle frontal gyrus 51 14 44 2.64 38
Middle frontal gyrus 46 50 2.57
Conjunction
Region of Activation
Hemisphere
BA
Talairach Coordinates
t
k
x
y
z
YA, FA > DA-R (Hit > Miss) inclusively masked with YA > OA, FA (Hit > Miss) Middle frontal gyrus 10 −30 38 17 4.89 16
Middle frontal gyrus −42 33 32 4.81 15
Precentral gyrus 38 37 4.28 105
Superior frontal gyrus −20 68 3.89 20
Cuneus 17 −12 −81 13 3.83 14
Inferior frontal gyrus 47 50 16 −1 3.79 44
Precentral gyrus 44 53 18 2.47
Inferior frontal gyrus 45 59 22 14 2.23
IPL 40 −65 −28 31 3.63 11
IPL 40 −63 −33 37 3.19
Middle frontal gyrus 11 −30 46 −14 3.62 23
Superior frontal gyrus 24 64 3.54 60
Middle frontal gyrus 30 11 58 3.2
Insula 13 40 26 19 3.36 39
Inferior frontal gyrus 45 46 22 15 2.55
Subgyral 18 57 3.05 15
Superior frontal gyrus 68 2.97 35
Middle frontal gyrus 51 14 44 2.64 38
Middle frontal gyrus 46 50 2.57

Regions significant at the conjoint probability of p < .00025 (k ≥ 10) following inclusive masking. An explicit, inclusive t mask was imposed, and the mask was created with regions of significant in “YA(Hit > Miss) > OA(Hit > Miss)” at uncorrected p < .005, with a cluster extent threshold of k ≥ 10. R = right; L = left; BA = approximate Brodmann's area based on coordinates.

Figure 4.

Common neural effects of aging and reduced relational attention on successful RM encoding. The results from the two sample t test, “YA(Hit > Miss) > OA(Hit > Miss)” under the FA condition (p < .005 uncorrected, k ≥ 10) were subsequently used as an inclusive mask for identifying common effects of aging and reduced relational attention on successful RM encoding at p < .05 with 10 contiguous voxels, resulting in the conjoint probability approached p < .00025, k ≥ 10. L = left; R = right.

Figure 4.

Common neural effects of aging and reduced relational attention on successful RM encoding. The results from the two sample t test, “YA(Hit > Miss) > OA(Hit > Miss)” under the FA condition (p < .005 uncorrected, k ≥ 10) were subsequently used as an inclusive mask for identifying common effects of aging and reduced relational attention on successful RM encoding at p < .05 with 10 contiguous voxels, resulting in the conjoint probability approached p < .00025, k ≥ 10. L = left; R = right.

## DISCUSSION

### Behavioral Performance

The behavioral results replicated our previous finding that reduced resources for relational attention disproportionately impaired RM performance in YAs. Specifically, imposing attention tasks during RM encoding impaired YAs' RM performance, but the degree of impairment was different depending on the type of attention task: YAs showed significantly lower accuracy on the RM task when the attention task required relational attention than when it required item attention. Furthermore, the reduction in relational attention during encoding equated YAs' RM performance to that of OAs under the FA condition. The current study also extended our previous findings to OAs by examining the effects of DA on RM performance in OAs. In line with previous findings, the age-related RM deficits were evident in our study. Moreover, we found that OAs showed significantly greater RM impairment when the relational attention task was imposed during RM encoding compared with the FA condition.

The OA results in the current study dovetail with findings from a prior behavioral study. Kilb and Naveh-Benjamin (2007) tested both IM and RM performance in YAs and OAs and imposed an attention task, which did not require relational processing, during encoding in both age groups. The authors examined whether RM deficits observed in OAs under the FA condition would become exacerbated under the DA condition. However, the authors found that OAs did not show a larger RM deficit under the DA condition, contrary to the predictions of the reduced attentional resources hypothesis. In their study, OAs showed smaller RM deficits under the DA than the FA condition.

Consistent with Kilb and Naveh-Benjamin (2007), the current study showed no differential impairment in OA's RM under the DA with item attention condition than under the FA condition. Unlike YAs who exhibited reduced RM performance under the DA-I condition, OAs, in fact, overcame the obstacle (i.e., a reduction in attentional resources in item attention) and showed equivalent RM performance between the FA and the DA-I conditions. This behavioral pattern also accords well with the current fMRI results, showing that OAs recruited more neural regions for successful RM encoding under the DA-I than the FA condition. However, in contrast to the DA-I condition, a reduction in relational attention during encoding significantly impaired RM performance in OAs to a greater extent than did the FA condition. This result is different from those of Kilb and Naveh-Benjamin (2007), who used a similar attention task as our item attention task. Rather, our finding with the relational attention task is in line with predictions of the reduced attentional resources hypothesis (Craik & Byrd, 1982), because the lack of resources for relational attention during RM encoding significantly worsened OA's RM performance. Together with findings in YAs, which demonstrate larger RM impairments under the DA-R condition than under other attention conditions, the current behavioral results suggest that a reduction in relational attention plays a critical role in the RM deficit observed in OAs.

### fMRI Findings

The present study extended our behavioral findings by examining the neural correlates of the effects of DA with relational attention at encoding on later RM performance. Furthermore, our study also investigated whether reduced resources for relational attention plays a key role in age-related RM deficits at the neural level. The fMRI results are discussed in terms of each hypothesis tested in the current study.

#### The Neural Correlates of Successful RM Formation

Using the subsequent memory procedure (Wagner et al., 1998), we identified brain regions mediating successful RM encoding in YAs. Previous research on the neural correlates of RM encoding has demonstrated that successful RM encoding is often accompanied with activation in the VLPFC, DLPFC, and left hippocampus and parahippocampal gyrus. The current neuroimaging results replicated these previous findings: bilateral VLPFC (BA 11/45/47), bilateral DLPFC (BA 6/9/46), left parahippocampal gyrus (BA 36), and the left anterior hippocampus all showed subsequent memory effects for RM encoding in YAs under the FA condition.

Previous researchers have suggested that PFC plays a strategic role in episodic memory formation by generating associations among different information (Addis & McAndrews, 2006). Furthermore, numerous neuroimaging studies have reported significant activity in VLPFC during successful memory encoding (Park & Rugg, 2008; Prince, Daselaar, & Cabeza, 2005; Cabeza et al., 1997). Specifically, Prince and colleagues (2005) compared encoding and retrieval related activity during RM tasks and demonstrated that left VLPFC is strongly associated with successful encoding of RM for semantic information. Moreover, Park and Rugg (2008) also demonstrated robust involvement of left VLPFC in successful memory encoding for interitem associations for both semantically and phonologically related information.

In addition to VLPFC, recent neuroimaging studies have reported significant activation in DLPFC during RM encoding. Specifically, Murray and Ranganath (2007) directly compared neural correlates of IM versus RM encoding and demonstrated that DLPFC activity was greater during RM compared with IM encoding. The authors also found that DLPFC activity predicted successful memory for associations, whereas VLPFC activation predicted successful memory for both RM and IM. Moreover, a recent review on the role of PFC on episodic memory suggests distinct roles of DLPFC and VLPFC (Blumenfeld & Ranganath, 2007). Specifically, the researchers reviewed lesion and neuroimaging studies and concluded that VLPFC is crucial for directing attention toward goal-relevant items, which enhances representation for both IM and RM formation, whereas DLPFC mediates the ability to organize and chunk multiple pieces of information, which enhances memory for associations. In relation to this account, we also found significant involvement of both DLPFC and VLPFC for successful RM encoding.

Along with PFC activation, we also found significant activation in the left anterior hippocampus in YAs under the FA condition. Such engagement of anterior hippocampus is in line with recent findings in RM (Giovanello, Schnyer, & Verfaellie, 2009; Sperling et al., 2003). For example, using fMRI during the encoding of novel face–name associations, Sperling and colleagues (2003) found significantly greater activation in the anterior hippocampus for successfully remembered pairs compared with forgotten pairs. Addis and McAndrews (2006) also reported significant activation in the left anterior hippocampus when a task required generation of relational information during encoding. Finally, Giovanello and colleagues (2009) found that different regions along the long axis of the hippocampus make distinct contributions to RM processing. Specifically, Giovanello et al. (2009) found that the posterior hippocampus was involved in the reinstatement of perceptual information, whereas the anterior hippocampus was involved in flexible relational operations at retrieval. The finding that the anterior hippocampus is important for flexible retrieval of RM fits well with the current findings. In the current paradigm, participants encoded word pairs, which switched their location after initial presentation. This manipulation was included to prevent the potential attentional capture by new number stimuli presented with the word pairs. However, this manipulation may have also forced participants to use a flexible encoding strategy to memorize each word pair, which changed their location after initial presentation (i.e., A–B, then B–A). Overall, the current fMRI results extended previous evidence that the anterior hippocampus is associated with successful RM formation, especially when the task requires the flexible encoding of relational information.

In the present study, we also found significant activation in the left SPL (BA 7) and IPL (BA 40) associated with successful RM encoding in YAs. Although the role of PFC and hippocampus in episodic memory encoding has been demonstrated in numerous imaging and lesion studies, recent neuroimaging studies suggest that a complete story of the functional neurobiology of episodic memory may require appreciation of possible contributions from the parietal cortex (e.g., Cabeza, Ciaramelli, Olson, & Moscovitch, 2008). For example, recent studies have shown that the engagement of posterior parietal cortex (PPC) is consistently active during episodic retrieval (for a review, see Cabeza et al., 2008). Furthermore, in a recent meta-analysis study, Uncapher and Wagner (2009) explored the role of PPC on episodic memory encoding. The authors concluded that dorsal PPC, including SPL, supports the allocation of goal-directed attention, which may, in turn, increase the probability that the attended information is encoded into episodic memory. The researchers also suggested that ventral PPC, including IPL, may contribute to successful episodic memory by fostering the encoding of surface event details that impact retrieval over short retention intervals. Given that the current study did not include deep elaborative encoding instructions and that the retention interval was relatively short, the recruitment of both SPL and IPL during successful memory encoding is consistent with the conclusion by Uncapher and Wagner (2009). Although it is not clear whether the recruitment of parietal regions is for episodic memory encoding in general or for RM encoding in particular, it is worth noting that the involvement of these regions likely reflects the attentional processing required for RM encoding.

In summary, findings from YAs under the FA condition indicated the neural correlates of successful RM formation: Successful RM encoding was associated with significant activation in the VLPFC, DLPFC, left anterior hippocampus, and left SPL and IPL.

#### The Effect of Aging on Successful RM Encoding at the Neural Level

In accordance with our behavioral findings documenting age-related RM deficits, we found reduced activity in OAs compared with YAs in the neural regions for successful RM encoding. Specifically, using the same threshold, OAs did not show significant activation in the DLPFC, VLPFC, hippocampus, or PPC. Instead, OAs showed significant activation for RM encoding in right perirhinal cortex, right insula, and left inferior and superior temporal gyrus. The activity in perirhinal cortex is worth noting. The activation in rhinal cortex may indicate OAs' tendency to rely more on familiarity, as opposed to recollection-based processing when performing RM tasks, as previous studies have demonstrated that perirhinal activation during encoding correlates with later item recognition, without specific memory of its source (Davachi, Mitchell, & Wagner, 2003). Alternatively, the engagement of perirhinal cortex and other MTL areas may reflect compensation for an age-related reduction in the ability to engage the hippocampus during successful RM. Although prior research suggests that the hippocampus mediates RM encoding, more so than does the neighboring MTL cortex (Davachi, 2006), a recent study demonstrated significant activation in rhinal cortex and medial frontal cortex in cognitively intact healthy OAs during RM retrieval (Braskie, Small, & Bookheimer, 2009), suggesting a possible compensatory mechanism for those regions in OAs. A future study with confidence ratings during recognition would further elucidate whether the activation in the perirhinal cortex reflect familiarity or compensatory processing by separating the familiarity versus recollection-based processing.

A direct comparison of neural activity between YA and OAs under the FA condition revealed significant age-related attenuation in several of the core regions observed during RM encoding. Specifically, OAs showed significantly reduced activity in bilateral DLPFC, right VLPFC, left SPL, and IPL, relative to YAs. The reduction in activity in lateral PFC in OAs compared with that of YAs during RM encoding has been documented previously in the literature. For example, Logan, Sanders, Snyder, Morris, and Buckner (2002) found underrecruitment of left PFC in OAs compared with YAs during intentional memory encoding. Other researchers have also demonstrated age-related decrease in PFC activity during RM encoding for semantic information (Anderson et al., 2000; Cabeza et al., 1997), spatial association (Mitchell et al., 2000), pictorial stimuli (Iidaka et al., 2001), and face–scene association (Dennis et al., 2008). As discussed above, VLPFC and DLPFC appear to play distinct roles during successful RM encoding. Although the VLPFC is involved in the selection of goal-relevant features, which in turn may enhance representations of goal-relevant items, the DLPFC controls organization of different information, thereby enhancing association among different items. Thus, the age-related reductions in VLPFC and DLPFC observed in the current study may reflect reductions in selection and control processes, which then contribute to age-related RM impairments.

Although OAs failed to show significant activation in the hippocampus during RM encoding, a direct comparison between YAs and OAs did not reveal significant differences in the magnitude of hippocampal activity. This result is counter to some prior reports indicating significantly reduced hippocampal activation in OAs relative to YAs during RM encoding (Mitchell et al., 2000; Anderson et al., 2000). However, other studies have found that OAs activate the hippocampus to a similar degree as YAs during RM encoding. For example, Miller and colleagues (2008) demonstrated a similar magnitude and extent of hippocampal activation in both YAs and OAs during successful RM encoding for face–name pairs. Using different materials, other researchers have also reported similar hippocampal activity in YAs and OAs during RM encoding (Leshikar, Gutchess, Hebrank, Sutton, & Park, 2010; Rand-Giovannetti et al., 2006). These conflicting findings about hippocampal activity and age-related RM deficits may stem from different task paradigms and comparisons among different studies. For example, studies that have found greater hippocampal activity in YAs than OAs used both IM and RM tasks and compared the hippocampal activation between the IM and RM process in both age groups. Specifically, Mitchell and colleagues (2000) found that YAs showed greater left hippocampal activity for combination of object and location pairs than for item trials, whereas OAs did not show the same pattern of activity in the hippocampus. Thus, it is possible that YAs differentially activated the hippocampus for RM compared with IM formation, whereas OAs activated the hippocampus for both IM and RM encoding. Because the current paradigm did not include the IM task, it is not possible to examine whether OAs proportionally activated the hippocampus during both RM and IM encoding. Further studies that include both IM and RM tests will help to clarify whether hippocampal dysfunction can account for age-related RM encoding deficits.

#### The Neural Effects of Different Attention Tasks on Successful RM Encoding

In the current study, we manipulated the type of attentional load imposed during RM encoding and demonstrated that a reduction in a specific type of attentional resource (i.e., relational attention) significantly impaired RM performance both in YAs and OAs. The effects of reduced relational attention on RM were also observed at the neural level: The neural regions mediating successful RM formation in YAs, including DLPFC, VLPFC, IPL, SPL, and the anterior hippocampus showed significantly attenuated activation when YAs encoded relational information under the DA with relational attention condition compared with their activation under the FA condition. In contrast, the attention task requiring little or no relational processing (i.e., DA-I condition) did not significantly attenuate activity in any RM encoding regions.

Using PET, Anderson and colleagues (2000) also reported a significant reduction in PFC activity during RM encoding under a DA condition. However, the researchers did not find significant reductions in the hippocampus when the attentional load was imposed during RM encoding. Although the previous study provides important insight into the neural effects of attention on RM, it is noteworthy that the previous study was not able to test the effect of attention on successful RM encoding because the study used a blocked design with PET imaging. To our knowledge, there has been no event-related fMRI study using a subsequent memory procedure to investigate the effect of attention on successful interitem RM encoding, although recent studies have reported the effect of attention on successful IM encoding or episodic memory recollection, which is relevant to item–context associations (Uncapher & Rugg, 2005, 2008; Kensinger et al., 2003). Thus, the current study provides the first report that reduced attentional resources during encoding significantly attenuates activity in the neural correlates of successful interitem RM in the prefrontal, parietal, and hippocampal regions.

In addition, the current study provides another novel finding regarding the effects of qualitatively different attentional loads on the neural correlates of interitem RM encoding. In the current experiment, the key manipulation of attention tasks was not based on task difficulty but rather on the type of resources that each attention task required. With an item attention task and a relational attention task during RM encoding, we demonstrated that only the attentional load involving relational attention significantly attenuated activity in the neural regions of successful RM encoding (i.e., VLPFC, DLPFC, SPL, IPL, and the anterior hippocampus), whereas a qualitatively, not quantitatively, different item attention task did not.

Finally, results from the conjunction analyses confirmed the similar effect of relational attention and aging on RM: The neural regions mediating successful RM that showed attenuated activity by aging were strikingly similar to the regions showing attenuation by reduced relational attention. In contrast, there was no overlap between the effect of aging and the effect of reduced item attention on RM. Together with our behavioral data, this result indicates that the reduction in relational attention contributes to age-related RM deficits both at behavioral and neural levels.

### Conclusion

The current study contributes several novel findings to the memory, aging, and cognitive neuroscience literatures. Collectively, the findings provide evidence that a reduction in relational attention in OAs is a critical factor for age-related RM deficits. Behaviorally, reduced relational attention in YAs equated their performance during an RM task to that of OAs under FA. A reduction in relational attention during RM encoding in YAs also attenuated brain activity in the neural regions mediating RM formation, as was the case for OAs under the FA condition. Within the broader domain of episodic memory, these results suggest that reductions in a specific type of attentional resource can explain at least some aspect of age-related episodic memory decline in binding separate information together. Understanding the nature and source of age-related impairments in episodic memory is important for both theoretical and ecological reasons, because the ability to connect and associate separate information into contextual representation is essential for coherent memories of everyday events. By examining both behavioral and neural data regarding the source of RM deficits in OAs, the current study contributes to a more comprehensive picture of cognitive aging, especially in the domain of episodic memory decline.

## Acknowledgments

We thank Amber Abernethy, Emilie Kearns, Hannah McNamara, Kathy Wilber, Kristin Dellinger, and Maggie Fitch for help with data collection and acknowledge NIA AG028774.

Reprint requests should be sent to So-Yeon Kim, Center for Mind and Brain, University of California at Davis, 267 Cousteau Place, Davis, CA 95618, or via e-mail: soykim@ucdavis.edu.

## Notes

1.

Although we did not assess performance on the two attention tasks under the FA condition necessary to measure task difficulty, participants in this study, as well as in a prior study (Kim & Giovanello, 2011), showed equivalent dual-task performance for the two attention tasks under the same memory encoding load, which should affect the two attention tasks in the same manner.

2.

As described in the Methods section, we matched performance between YAs under the DA-R condition and OAs under the FA condition. We did not intend to match memory performance between YAs and OAs under FA conditions because one of our hypotheses was that OAs have a fundamental RM deficit compared with YAs. Rugg and Morcom (2005) have indicated that a performance difference between two age groups is not as problematic for an encoding study as for a retrieval study, if the subsequent memory procedure is employed. We agree with this assertion because age-related differences would not be confounded with the neural correlates of effective versus ineffective encoding.

## REFERENCES

REFERENCES
,
D. R.
, &
McAndrews
,
M. P.
(
2006
).
Prefrontal and hippocampal contributions to the generation and binding of semantic associations during successful encoding.
Neuroimage
,
33
,
1194
1206
.
Allman
,
J. M.
,
Hakeem
,
A.
,
Erwin
,
J. M.
,
Nimchinsky
,
E.
, &
Hof
,
P.
(
2001
).
The anterior cingulate cortex. The evolution of an interface between emotion and cognition.
Annals of the New York Academy of Sciences
,
935
,
107
117
.
Anderson
,
N. D.
,
Craik
,
F. I.
, &
Naveh-Benjamin
,
M.
(
1998
).
The attentional demands of encoding and retrieval in younger and older adults: 1. Evidence from divided attention costs.
Psychology and Aging
,
13
,
405
423
.
Anderson
,
N. D.
,
Iidaka
,
T.
,
Cabeza
,
R.
,
Kapur
,
S.
,
McIntosh
,
A. R.
, &
Craik
,
F. I. M.
(
2000
).
The effects of divided attention on encoding- and retrieval-related brain activity: A PET study of younger and older adults.
Journal of Cognitive Neuroscience
,
12
,
775
792
.
Blumenfeld
,
R. S.
, &
Ranganath
,
C.
(
2007
).
Prefrontal cortex and long-term memory encoding: An integrative review of findings from neuropsychology and neuroimaging.
The Neuroscientist
,
13
,
280
291
.
,
M. N.
,
Small
,
G. W.
, &
Bookheimer
,
S. Y.
(
2009
).
Entorhinal cortex structure and functional MRI response during an associative verbal memory task.
Human Brain Mapping
,
30
,
3981
3992
.
Cabeza
,
R.
,
Ciaramelli
,
E.
,
Olson
,
I. R.
, &
Moscovitch
,
M.
(
2008
).
The parietal cortex and episodic memory: An attentional account.
Nature Reviews Neuroscience
,
9
,
613
625
.
Cabeza
,
R.
,
,
C. L.
,
Nyberg
,
L.
,
McIntosh
,
A. R.
,
Tulving
,
E.
,
Kapur
,
S.
,
et al
(
1997
).
Age-related differences in neural activity during memory encoding and retrieval: A positron emission tomography study.
Journal of Neuroscience
,
17
,
391
400
.
Chua
,
E. F.
,
Schacter
,
D. L.
,
Rand-Giovannetti
,
E.
, &
Sperling
,
R. A.
(
2006
).
Understanding metamemory: Neural correlates of the cognitive process and subjective level of confidence in recognition memory.
Neuroimage
,
29
,
1150
1160
.
Craik
,
F. I.
,
Govoni
,
R.
,
Naveh-Benjamin
,
M.
, &
Anderson
,
N. D.
(
1996
).
The effects of divided attention on encoding and retrieval processes in human memory.
Journal of Experimental Psychology: General
,
125
,
159
180
.
Craik
,
F. I. M.
, &
Byrd
,
M.
(
1982
).
Aging and cognitive deficits: The role of attentional resources.
In F. I. M. Craik & S. Trehub (Eds.),
Aging and cognitive processes
(pp.
199
211
).
New York
:
Plenum Press
.
Darby
,
D.
(
2006
).
MacStim.
Melbourne
:
WhiteAnt Occasional Publishing
.
Davachi
,
L.
(
2006
).
Item, context and relational episodic encoding in humans.
Current Opinion in Neurobiology
,
16
,
693
700
.
Davachi
,
L.
,
Mitchell
,
J. P.
, &
Wagner
,
A. D.
(
2003
).
Multiple routes to memory: Distinct medial temporal lobe processes build item and source memories.
Proceedings of the National Academy of Sciences, U.S.A.
,
100
,
2157
2162
.
Davachi
,
L.
, &
Wagner
,
A. D.
(
2002
).
Hippocampal contributions to episodic encoding: Insights from relational and item-based learning.
Journal of Neurophysiology
,
88
,
982
990
.
Dennis
,
N. A.
,
Hayes
,
S. M.
,
Prince
,
S. E.
,
,
D. J.
,
Huettel
,
S. A.
, &
Cabeza
,
R.
(
2008
).
Effects of aging on the neural correlates of successful item and source memory encoding.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
34
,
791
808
.
Diana
,
R. A.
,
Yonelinas
,
A. P.
, &
Ranganath
,
D.
(
2007
).
Imaging recollection and familiarity in the medial temporal lobe: A three-component model.
Trends in Cognitive Sciences
,
11
,
379
386
.
Eichenbaum
,
H.
,
Yonelinas
,
A. R.
, &
Ranganath
,
C.
(
2007
).
The medial temporal lobe and recognition memory.
Annual Review of Neuroscience
,
30
,
123
152
.
Fisher
,
R. A.
(
1950
).
Statistical methods for research workers.
London
:
Oliver & Boyd
.
Francis
,
W. N.
, &
Kucera
,
H.
(
1982
).
Frequency analysis of English usage: Lexicon and grammar.
Boston
:
Houghton Mifflin Co
.
Giovanello
,
K. S.
,
Kensinger
,
E. A.
,
Wong
,
A. T.
, &
Schacter
,
D. L.
(
2010
).
Age-related neural changes during memory conjunction errors.
Journal of Cognitive Neuroscience
,
22
,
1348
1361
.
Giovanello
,
K. S.
,
Schnyer
,
D.
, &
Verfaellie
,
M.
(
2009
).
Distinct hippocampal regions make unique contributions to relational memory.
Hippocampus
,
19
,
111
117
.
Hoyer
,
W. J.
, &
Verhaeghen
,
P.
(
2006
).
Memory aging.
In J. E. Birren & K. W. Schaie (Eds.),
Handbook of the psychology of aging
(6th ed., pp.
209
232
).
San Diego
:
Elsevier
.
Iidaka
,
T.
,
,
N.
,
,
H.
,
Murata
,
T.
,
Omori
,
M.
, &
Yonekura
,
Y.
(
2001
).
An fMRI study of the functional neuroanatomy of picture encoding in younger and older adults.
Cognitive Brain Research
,
11
,
1
11
.
Kausler
,
D. H.
,
Salthouse
,
T. A.
, &
Saults
,
J. S.
(
1988
).
Temporal memory over the adult lifespan.
The American Journal of Psychology
,
101
,
207
215
.
Kensinger
,
E. A.
,
Clarke
,
R. J.
, &
Corkin
,
S.
(
2003
).
What neural correlates underlie successful encoding and retrieval? A functional magnetic resonance imaging study using a divided attention paradigm.
Journal of Neuroscience
,
23
,
2407
2415
.
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.
Memory & Cognition
,
35
,
1162
1174
.
Kim
,
S.-Y.
, &
Giovanello
,
K. S.
(
2011
).
The effects of attention on age-related relational memory deficits: Evidence from a novel attentional manipulation.
Psychology and Aging
,
26
,
678
688
.
Leshikar
,
E. D.
,
Gutchess
,
A. H.
,
Hebrank
,
A. C.
,
Sutton
,
B. P.
, &
Park
,
D. C.
(
2010
).
The impact of increased relational encoding demands on frontal and hippocampal function in older adults.
Cortex
,
46
,
507
521
.
Lieberman
,
M. D.
, &
Cunningham
,
W. A.
(
2009
).
Type I and Type II error concerns in fMRI research: Re-balancing the scale.
Social Cognitive and Affective Neuroscience
,
4
,
423
.
Logan
,
J. M.
,
Sanders
,
A. L.
,
Snyder
,
A. Z.
,
Morris
,
J. C.
, &
Buckner
,
R. L.
(
2002
).
Under-recruitment and nonselective recruitment: Dissociable neural mechanisms associated with aging.
Neuron
,
33
,
827
840
.
,
D. J.
,
Spaniol
,
J.
,
Bucur
,
B.
, &
Whiting
,
W. L.
(
2005
).
Adult age differences in the implicit and explicit components of top–down attentional guidance during visual search.
Psychology and Aging
,
20
,
317
329
.
Miller
,
S. L.
,
Celone
,
K.
,
DePeau
,
K.
,
Diamond
,
E.
,
Dickerson
,
B. C.
,
Rentz
,
D.
,
et al
(
2008
).
Age-related memory impairment associated with loss of parietal deactivation but preserved hippocampal activation.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
2181
2186
.
Mitchell
,
K. J.
,
Johnson
,
M. K.
,
Raye
,
C. L.
, &
D'Esposito
,
M.
(
2000
).
fMRI evidence of age-related hippocampal dysfunction in feature binding in working memory.
Cognitive Brain Research
,
10
,
197
206
.
Murray
,
L. J.
, &
Ranganath
,
C.
(
2007
).
The dorsolateral prefrontal cortex contributes to successful relational memory encoding.
Journal of Neuroscience
,
27
,
5515
5522
.
Naveh-Benjamin
,
M.
(
2000
).
Adult age differences in memory performance: Tests of an associative deficit hypothesis.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
26
,
1170
1187
.
Naveh-Benjamin
,
M.
,
Hussain
,
Z.
,
Guez
,
J.
, &
Bar-On
,
M.
(
2003
).
Adult age differences in episodic memory: Further support for an associative-deficit hypothesis.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
29
,
826
837
.
Old
,
S. R.
, &
Naveh-Benjamin
,
M.
(
2008
).
Differential effects of age on item and associative measures of memory: A meta-analysis.
Psychology and Aging
,
23
,
104
118
.
Park
,
D. C.
,
Puglisi
,
J. T.
, &
Lutz
,
R.
(
1982
).
Spatial memory in older adults: Effects of intentionality.
Journal of Gerontology
,
37
,
330
335
.
Park
,
H.
, &
Rugg
,
M. D.
(
2008
).
Neural correlates of successful encoding of semantically and phonologically mediated inter-item associations.
Neuroimage
,
43
,
165
172
.
Prince
,
S. E.
,
Daselaar
,
S. M.
, &
Cabeza
,
R.
(
2005
).
Neural correlates of relational memory: Successful encoding and retrieval of semantic and perceptual associations.
Journal of Neuroscience
,
25
,
1203
1210
.
Rand-Giovannetti
,
E.
,
Chua
,
E. F.
,
Driscoll
,
A. E.
,
Schacter
,
D. L.
,
Albert
,
M. S.
, &
Sperling
,
R. A.
(
2006
).
Hippocampal and neocortical activation during repetitive encoding in older persons.
Neurobiology of Aging
,
27
,
173
182
.
Rugg
,
M. D.
, &
Morcom
,
A. M.
(
2005
).
The relationship between brain activity, cognitive performance and aging: The case of memory.
In R. Cabeza, L. Nyberg, & D. C. Park (Eds.),
The cognitive neuroscience of aging.
New York
:
Oxford University Press
.
Sperling
,
R.
,
Chua
,
E.
,
Cocchiarella
,
A.
,
Rand-Giovannetti
,
E.
,
Poldrack
,
R.
,
Schacter
,
D. L.
,
et al
(
2003
).
Putting names to faces: Successful encoding of associative memories activates the anterior hippocampal formation.
Neuroimage
,
20
,
1400
1410
.
Tulving
,
E.
(
1983
).
Elements of episodic memory.
Oxford, UK
:
Clarendon Press
.
Uncapher
,
M. R.
, &
Rugg
,
M. D.
(
2005
).
Effects of divided attention on fMRI correlates of memory encoding.
Journal of Cognitive Neuroscience
,
17
,
1923
1935
.
Uncapher
,
M. R.
, &
Rugg
,
M. D.
(
2008
).
Fractionation of the component processes underlying successful episodic encoding: A combined fMRI and divided-attention study.
Journal of Cognitive Neuroscience
,
20
,
240
254
.
Uncapher
,
M. R.
, &
Wagner
,
A. D.
(
2009
).
Posterior parietal cortex and episodic encoding: Insights from fMRI subsequent memory effects and dual-attention theory.
Neurobiology of Learning and Memory
,
91
,
139
154
.
Wagner
,
A. D.
,
Schacter
,
D. L.
,
Rotte
,
M.
,
Koutstaal
,
W.
,
Maril
,
A.
,
Dale
,
A. M.
,
et al
(
1998
).
Building memories: Remembering and forgetting of verbal experiences as predicted by brain activity.
Science
,
281
,
1188
1191
.