Skip to main content
NCBI home page
Human Brain Mapping logoLink to Human Brain Mapping
. 2010 Apr 16;32(2):171–181. doi: 10.1002/hbm.21014

Early parietal response in episodic retrieval revealed with MEG

Tyler M Seibert 1,2,✉, Donald J Hagler Jr 2, James B Brewer 2,3
PMCID: PMC3107600  NIHMSID: NIHMS295067  PMID: 20623759

Abstract

Recent neuroimaging and lesion studies have led to competing hypotheses for potential roles of the left lateral parietal lobe in episodic memory retrieval. These hypotheses may be dissociated by whether they imply a role in preretrieval or postretrieval processes. For example, one hypothesis is the left parietal cortex (particularly in more ventral subregions) forms part of an “episodic buffer” that supports the online representation of the retrieved target, a role that is, by definition, postretrieval. An alternate view maintains parietal activity (particularly in more dorsal subregions) contributes to top‐down orientation of attention to retrieval search, a preretrieval role. The present investigation seeks to reveal the earliest onset of lateral parietal activity in three anatomically‐defined subregions of the left lateral parietal cortex to identify any preretrieval activation. Subjects performed a pair‐cued recall task while neural activity was recorded with magnetoencephalography (MEG) at millisecond temporal resolution. MEG data were then mapped to each subject's cortical surface using dynamic statistical parametric mapping (dSPM). Both dorsal and ventral regions showed retrieval‐related activations beginning within ∼100 ms of the cue to retrieve and lasting up to 400 ms. We conclude that this early and transient pattern of activity in lateral parietal cortex is most consistent with a preretrieval role, possibly in directing attention to episodic memory retrieval. Hum Brain Mapp, 2011. © 2010 Wiley‐Liss, Inc.

Keywords: long‐term memory, cued recall, attention, magnetoencephalography, dSPM, human

INTRODUCTION

Retrieval of episodic memory is consistently associated with increases in activity of the lateral parietal cortex. Converging evidence of this association is found in studies using positron emission tomography (PET) [Owen et al., 1996; Rugg et al., 1998; Tulving et al., 1994; for review, see Cabeza and Nyberg, 2000], event‐related functional MRI (fMRI) [for review, see Cabeza, 2008; Ciaramelli et al., 2008; Vilberg and Rugg, 2008a], and event‐related potentials (ERPs) measured by electroencephalography (EEG) [for review, see Friedman and Johnson, 2000; Rugg and Curran, 2007]. While parietal damage is not as commonly associated with memory deficits as damage to the medial temporal or prefrontal cortices, it has long been known that parietal lesions can result in deficient attention to spatial aspects of memory [Bisiach and Luzzatti, 1978], and recent studies suggest some patients with parietal lesions may have poor free recall of detailed autobiographical memories [Berryhill et al., 2007; for review, see Olson and Berryhill, 2009].

For decades electrophysiologists have described an increase in activity, most prominent in electrodes over the left parietal cortex, for items correctly labeled as previously studied (“old”) relative to items correctly labeled as novel (“new”) [Curran, 2004; Duarte et al., 2004; Neville et al., 1986; Wilding and Rugg, 1996; for review, see Rugg and Curran, 2007; Rugg et al., 2002]. The left parietal old/new effect onsets ∼400–450 ms after presentation of the test item, and typical average response times in recognition tests are ∼800 ms or greater [Vilberg and Rugg, 2008a]. Even allowing a few hundred milliseconds for execution of the response, the parietal effect begins prior to response selection and therefore sufficiently early to play a role in episodic retrieval [Vilberg and Rugg, 2008a].

The magnitude of this “left parietal old/new effect” is modulated by the type of memory retrieval. For example, it is greater for recollection (memory of a studied item as well as contextual details from the study event) than for familiarity (memory of a studied item alone, without recalling the precise study event) [Vilberg and Rugg, 2008a; for review of the “dual‐process” description of recognition memory see Rugg and Yonelinas, 2003; Yonelinas, 2001]. When the quality of the study experience is varied, the parietal old/new effect measured during retrieval is greater for items believed to have been deeply encoded than for those shallowly encoded [Vilberg and Rugg, 2008a]. Additionally, retrieval events believed to reflect a larger retrieved load have been shown to elicit a larger left parietal old/new effect [Vilberg and Rugg, 2008a, 2008b].

Event‐related fMRI studies confirmed the previous ERP findings of left parietal activation in retrieval tasks [Buckner and Wheeler, 2001; Henson et al., 1999; Kahn et al., 2004; Rugg et al., 2002; Wagner et al., 2005; Wheeler and Buckner, 2003], providing converging evidence and leading to several competing hypotheses for the still unclear role this region plays in retrieving episodic memories. One hypothesis supported by evidence from a large number of studies is that the lateral parietal cortex forms part of an “episodic buffer” acting to hold retrieved information [Baddeley, 2000; Vilberg and Rugg, 2008a; Wagner et al., 2005]. Another hypothesis, also consistent with much of the current literature, is that lateral parietal cortex is involved in directing attention internally to memory [Cabeza, 2008; Ciaramelli et al., 2008; Wagner et al., 2005]. At least two other ideas have also been described: the first maintains that parietal activity does not reflect retrieval of episodic information, but rather is associated with the subjective experience of recollection [Ally et al., 2008; Olson and Berryhill, 2009]; the second suggests the parietal region acts as a “mnemonic accumulator,” integrating the accumulated evidence for a memory until a threshold is reached that signals recognition [Wagner et al., 2005].

With its much higher spatial resolution, studies using fMRI have produced considerable evidence suggesting that the left lateral parietal cortex may be functionally dissociated with respect to episodic retrieval into dorsal and ventral subregions [Cabeza, 2008; Ciaramelli et al., 2008; Vilberg and Rugg, 2008a]. As two of the leading theories for left parietal involvement in recognition memory make distinct hypotheses for the dorsal and ventral subregions, the present article will treat these subregions as having potentially distinct roles. The dorsal parietal cortex (DPC) lies in and superior to the intraparietal sulcus—corresponding most consistently to the lateral portion of Brodmann Area (BA) 7 [Cabeza, 2008; Ciaramelli et al., 2008; Vilberg and Rugg, 2008b, 2008a], but also reported as including adjacent portions of BA 19 [Ciaramelli et al., 2008] or BA 40 [Vilberg and Rugg, 2008a]. The ventral parietal cortex (VPC) most consistently includes the angular gyrus (BA 39) [Cabeza, 2008; Ciaramelli et al., 2008; Vilberg and Rugg, 2008b], but may also include adjacent posterior and inferior areas, BA 40 [Cabeza, 2008; Ciaramelli et al., 2008] and BA 19 [Vilberg and Rugg, 2008b].

Left Dorsal Parietal Cortex (DPC)

One view maintains that DPC activity in memory tasks may reflect goal‐driven, top‐down direction of attention to retrieval [Cabeza, 2008; Ciaramelli et al., 2008]. Familiarity judgments may require greater effort to direct attention to retrieval than recollection judgments [Ciaramelli et al., 2008], and at least two meta‐analyses of left parietal fMRI activations have found the DPC to be more closely linked to familiarity than to recollection [Ciaramelli et al., 2008; Vilberg and Rugg, 2008a]. Similarly, items for which subjects have lower confidence might also require greater top‐down attention than items assigned high confidence judgments and therefore result in greater activation of the DPC [Cabeza, 2008; Ciaramelli et al., 2008]. This assertion is supported by several studies of recognition memory where subject confidence ratings were obtained [Cabeza, 2008; Ciaramelli et al., 2008; Daselaar et al., 2006; Fleck et al., 2006; Kim and Cabeza, 2007, 2009]. Finally, two meta‐analyses found greater activation increases in the DPC for specific source recall relative to subjective recollection [Ciaramelli et al., 2008; Spaniol et al., 2009]. As the authors point out, source recall requires retrieving specific contextual details, while subjective recollection in the “remember/know” paradigm requires only some contextual details to come to mind; thus, source recall is expected to result in greater top‐down retrieval effort and greater activity in the DPC.

Another view asserts that activity in the DPC during episodic memory tasks does not contribute directly to retrieval at all, but rather reflects “processes downstream of retrieval” and depends on the salience or behavioral relevance of the presented item [Vilberg and Rugg, 2008a]. The authors note DPC activity in recognition tasks may not be specific for familiarity, as DPC activity has also been observed to accompany recollection in some studies [Henson et al., 1999; Vilberg and Rugg, 2008a]. Additionally, one experiment showed the DPC recognition effect to be sensitive to the relative frequency of old and new test items, which the authors suggest would not be expected if the region contributed directly to retrieval [Herron et al., 2004].

Left Ventral Parietal Cortex (VPC)

There is greater agreement that the VPC plays a direct role in retrieval. One proposal for the function of the VPC is that of an episodic buffer (or part of the network supporting a buffer), whose activity serves to hold retrieved information [Baddeley, 2000; Vilberg and Rugg, 2008a; Wagner et al., 2005]. Greater activations found in the VPC for recollection relative to familiarity support a VPC role in recollection [Vilberg and Rugg, 2008a]. In contrast to DPC effects, VPC retrieval effects were not shown to be sensitive to the relative frequencies of old and new test items [Herron et al., 2004]. Perhaps most interesting, two studies have shown VPC activity to be greater when subjects are believed to have recollected more information [Vilberg and Rugg, 2008b, 2007]. Vilberg and Rugg [2008a] maintain that this dependence of VPC activity on the size of the retrieved “load” is most consistent with an episodic buffer.

The attention hypothesis assigns the VPC a role reflecting automatic, bottom‐up attention that is “captured” by the retrieved output [Cabeza, 2008; Ciaramelli et al., 2008]. As with the episodic buffer hypothesis, VPC activation is expected to have increased activity for recollection relative to familiarity, and that activity is expected to vary with the size of the retrieved load. However, in the attention model, VPC activity reflects the increased attentional demands of retrieved information rather than buffering of that information [Cabeza, 2008; Ciaramelli et al., 2008]. High confidence responses in recognition studies lead to greater VPC activation [Cabeza, 2008; Ciaramelli et al., 2008; Daselaar et al., 2006; Fleck et al., 2006; Kim and Cabeza, 2007, 2009], a finding that may be explained by stronger bottom‐up attention paid to stronger memories [Cabeza, 2008]. As recollection judgments tend to be associated with higher confidence than familiarity judgments, it follows that VPC activity is greater for recollection than familiarity [Cabeza, 2008]. Additional evidence supporting the bottom‐up attention to memory hypothesis comes from a meta‐analysis showing greater VPC activity for deeply encoded items [Ciaramelli et al., 2008]. Finally, the bottom‐up attention explanation finds support in a study of patients with bilateral parietal infarcts that included VPC [Berryhill et al., 2007]. The spontaneous autobiographical memories produced by these patients were poor in detail, but they were able to produce a normal amount of detail of the same memories when probed with specific questions [Berryhill et al., 2007]. Interpreted in terms of the attention hypotheses, answering specific questions may rely more on top‐down attention, whereas these patients' deficient spontaneous retrieval may reflect impaired bottom‐up attention to the memories [Cabeza, 2008].

Temporal dynamics

The respective timing of DPC and VPC activity could inform the evaluation of their proposed mechanistic roles. Some proposed roles (e.g., episodic buffer, mnemonic accumulator, and attentional capture by retrieved information) would best explain activity that begins after retrieval has occurred, while a very early activation would be more consistent with a role in directing attention to memory. However, the temporal dynamics of left parietal subregions remain unknown. Methodological constraints have limited direct investigation of the timing of activity in the DPC and VPC: fMRI lacks sufficient temporal resolution to dissociate pre‐ and postretrieval processes, and the ERP parietal old/new effect has not been localized to specific cortical subregions. However, a relatively recent method, dynamic statistical parametric mapping (dSPM), allows combination of magnetoencephalography (MEG) and structural MRI to calculate maps of estimated subregional cortical activity with millisecond temporal resolution [Dale et al., 2000].

In the present study we used MEG and dSPM to investigate the early response of left parietal cortex in episodic retrieval. Prior ERP and fMRI studies still leave at least two key points unknown. The first is whether parietal activity is present in the first 400 ms of episodic retrieval. The second is whether early parietal activity in retrieval localizes to DPC, VPC, or both. Answers to each of these questions have consequences for the hypotheses described above. For example, if retrieval‐related activity in either region begins after sufficient time has passed for retrieval to occur; this finding would support roles downstream of retrieval. Alternatively, a very early pattern of activity (especially if transient) may be most consistent with orientation of attention to memory search.

To isolate whether parietal lobe subregions participate in the earliest retrieval processes, such as orientation to memory search, we examined the first 500 ms after cue to retrieve. Healthy subjects studied pairs of drawings of common objects prior to scanning and were tested during MEG recording. A single item from one of the pairs was presented in each trial of the test phase. In the control (“classify”) condition, subjects made a living/nonliving judgment on the presented item; in the retrieval (“recall‐classify”) condition, subjects made a living/nonliving judgment on the absent pair of the presented item, which they had to retrieve from memory. The dSPM method [Dale et al., 2000] was used to estimate relative increases in neural activity associated with retrieval in specific, anatomically‐defined subregions of the left lateral parietal cortex.

MATERIALS AND METHODS

Participants

Eleven healthy, right‐handed adults (mean age: 23.7 years; six male) participated in this study, which was approved by the institutional review board of the University of California, San Diego. Informed consent was obtained from each subject prior to participation. Subjects received $40 for their participation.

Task

Prior to the MEG session, subjects studied 128 pairs of drawings of common objects and animals on a computer screen. Each pair was displayed for 3 s on three separate occasions, and subjects were instructed to memorize the pairs for subsequent testing.

Approximately 45 min after completing the study phase, subjects began the test phase while MEG signals were recorded. In all test phase trials, a single drawing from one of the studied pairs was presented for 500 ms in one of the two boxes (see Fig. 1), followed by an additional 2,750‐ms response period. During “classify” trials, subjects simply indicated by a finger response whether the presented stimulus was a living object. During “recall‐classify” trials, subjects indicated whether the absent associate of the presented stimulus was a living object, requiring recall of the paired associate. A colored box, present from 1,000 ms prior to stimulus onset, designated the trial type—green for classify and red for recall‐classify. A fixation cross, flanked by two black boxes, was shown for the first 250 ms of each trial. Subjects were instructed to respond as quickly and accurately as possible.

Figure 1.

Figure 1

Open in a new tab

Pair‐cued recall task. Subjects viewed each pair for 3 s during the study phase (repeated in random order three times). MEG recordings were acquired during the test phase. In classify trials subjects made a simple living/nonliving judgment on the presented item. In recall‐classify trials subjects retrieved the absent associate and then made a living/nonliving judgment on the item in memory. In both conditions the test item was equally likely to appear on the left and right sides. The timeline at the bottom of the figure represents trial timing for the test phase. During the period of the timeline represented in black, the boxes and fixation cross were presented with both boxes in black. The cue period is enlarged only for display in the figure.

The test phase comprised 256 trials, presented in eight runs of 32 trials each. Five subjects were given “trial list A,” and six subjects were given “trial list B.” Each trial list was constructed by pseudorandomly choosing the order of the stimulus pairs, then, for each trial, pseudorandomly choosing which stimulus of the pair to present, which side to present it on, and which condition (classify or recall‐classify) the trial would correspond to. The lists were then manually adjusted to ensure a balance between the two conditions in each run, as well as to remove any long streaks of a single condition, stimulus presentation side, or correct response (living or nonliving).

Both phases of the task were created and displayed using Presentation software (Neurobehavioral Systems, Albany, CA). During MEG acquisition, stimuli were presented using a three‐mirror DLP projector. Subjects indicated their responses by lifting a finger that was otherwise blocking a laser. For two subjects, a single run had to be excluded from analysis where subjects' finger movements were inadequate for response recording.

MEG Acquisition

MEG data were recorded via a 306 channel whole‐head MEG system (Elekta Neuromag, Elekta, Helsinki, Finland), consisting of 204 planar gradiometers and 102 magnetometers. Two electro‐oculograph (EOG) electrodes were placed, one each above and below the left eye, to monitor eye movements and blinks. To facilitate registration of the MEG sensor locations to each subject's structural MRI, three head position coils were placed, and the locations of ∼150 fiducial landmarks were recorded with a Polhemus Fastrak digitizer (Polhemus, Colchester, VT). MEG signals were sampled at 1,000 Hz, with an antialiasing filter of 333 Hz.

A band‐pass filter was applied offline between 0.2 and 33 Hz, and the data were then downsampled to 100 Hz. Individual trials containing eye blinks or other artifacts during the baseline period or time period of interest (or immediately before or after either of these periods) were excluded from analysis. No other trials were excluded. An average MEG recording (event‐related field, or ERF) was calculated for each subject from all remaining trials of the same condition.

MRI Acquisition

Two high‐resolution, three‐dimensional, T1‐weighted volumes (TE: 4.9 ms, TR: 10.7 ms, TI: 1,000 ms, flip angle: 8°, matrix: 256 × 256, voxel size: 1 mm × 1 mm × 1 mm) were acquired for each subject on a General Electric 1.5 T Signa Excite HDx using an eight‐channel phased‐array head coil (General Electric Healthcare, Waukesha, WI). Image intensities were corrected for spatial sensitivity inhomogeneities in the eight‐channel head coil by normalizing with the ratio of a body coil scan to a head coil scan.

Cortical Surface Reconstruction

A model of each subject's cortical surface was generated [Dale et al., 1999; Fischl et al., 1999a] to serve as the source space for the locations of dipoles in the MEG analysis. The two T1‐weighted MRI volumes were first corrected for spatial distortion due to gradient nonlinearity [Jovicich et al., 2006], registered to each other, and then averaged to improve the signal‐to‐noise ratio. The FreeSurfer software package (version 3.0.5, http://surfer.nmr.mgh.harvard.edu) was used to create a high‐resolution surface mesh for each hemisphere, representing the gray–white matter boundary. This folded surface was subsampled to define the assumed cortical dipole locations—∼2,500 dipoles per hemisphere, about 7‐mm apart. For group analysis, individual surfaces were aligned to a spherical representation of the FreeSurfer average subject [Fischl et al., 1999b].

MEG Activity Estimates

For each condition, a time series of estimated activity at each dipole was calculated from the gradiometer data of each subject's average MEG using the dynamic statistical parametric mapping (dSPM) method described by Dale et al. [2000]. The dSPM method involves computing a noise‐normalized, L 2 minimum‐norm, linear inverse to estimate the strength of each dipole's contribution to the average MEG recording at every time point in the series. The dSPM estimate is a measure of the MEG signal to noise ratio (SNR) at each spatial location; MEG SNR is related to neural activity, and so, in reference to MEG data, the terms activity and dSPM amplitude are used interchangeably in this article. Forward solutions were calculated using the boundary element method [de Munck, 1992; Mosher et al., 1999; Oostendorp and van Oosterom, 1989]. A surface tesselation was created for the inner skull from the same high‐resolution T1‐weighted MRI volume used for cortical surface reconstruction (SegLab, from Elekta). Baseline correction was performed on individual trial sensor waveforms and on the average waveforms using the period from −1,090 ms to −950 ms (see Fig. 1).

A noise covariance matrix was calculated from the same baseline period of individual trials. Three orthogonal vector components of each dipole were estimated simultaneously at every time point, thus allowing the dipole orientation to freely vary, and the corresponding vector magnitude, after normalizing by noise sensitivity, was taken as the estimated activity for that dipole. The result of the dSPM analysis was two time series (one for recall‐classify, and one for classify) for each subject, representing the estimated activity at each dipole (i.e., cortical location). Because the MEG recordings were downsampled to 100 Hz, the interval between time points was 10 ms. Time series from the recall‐classify and classify conditions were compared to identify activity differences attributable to episodic retrieval.

ROI Time Course Analysis

MEG activity was estimated at dipoles aligned to the FreeSurfer average subject, allowing anatomical ROIs to be defined using the cortical parcellation available in FreeSurfer. Individual subject data from dipoles within the left superior parietal ROI (ROIs shown in Fig. 2A) were combined to create an average ROI time series for each condition. This procedure was repeated for the other two ROIs: left inferior parietal and left supramarginal. Each ROI comprised ∼160 dipoles.

Figure 2.

Figure 2

Open in a new tab

Three anatomical ROIs (A) on the left cortical surface and their estimated activity time courses (note baseline dSPM amplitude is ∼1.0) (B–D): Superior Parietal (green, B), Supramarginal (orange, C), Inferior Parietal (blue, D). Recall‐classify activity was significantly greater than classify activity for 100‐ms time period indicated in yellow; *: P < 0.01, **: P < 0.001, ***: P < 10−4, ****: P < 10−7 (P‐values from paired t‐tests).

A three‐way repeated measures ANOVA with repeated factors of location (three ROIs), condition, and time was performed to examine the data for significant (P < 0.05) main effects and interactions (PASW 18, SPSS, Chicago, IL). ROI time series data were separated into subperiods of 100 ms (i.e., 100–190 ms, 200–290 ms, etc.), and, for the ANOVA, activity estimates from each subject were averaged across the time points in each subperiod. Five subperiods were initially included in the analysis, corresponding to the first 500 ms after cue onset. Subject blink frequency, possibly increased at stimulus offset, and disparities in response time made exploratory analysis of the entire length of the −1,000‐ms to 3,250‐ms task impractical due to reduced power, and thus our analysis remained focused on testing for an early response in the parietal lobe.

To further explore the temporal dynamics of the activity difference between the recall‐classify and classify conditions, paired t‐tests were performed on data from individual subperiods. Subperiods extending from 300 ms prior to cue onset to 500 ms after cue onset were examined, and activity differences were considered significant at the P < 0.01 level, after applying a Bonferroni correction for multiple comparisons. For the t‐tests, all time points from each subject's 100 Hz data were included.

A supplementary analysis was performed to assess whether a potential activity difference between the two conditions immediately prior to stimulus onset could partially account for activity differences observed after stimulus onset. Activity levels in the immediate prestimulus period were subtracted from the time series, and statistical comparisons were repeated. A detailed description of these methods is provided in the Supporting Information.

While the left lateral parietal cortex was the a priori area of interest, a post‐hoc analysis of FreeSurfer ROIs from lateral and medial cortex of both hemispheres was also performed and is described further in the Supporting Information available online.

Individual Dipole Analysis

Left hemisphere MEG data were analyzed at each dipole on the lateral surface to visualize individual parietal dipole activity across the entirety of the lateral hemisphere. For each dipole, the average classify time series was subtracted from the recall‐classify time series, giving an estimated activity difference for every time point. The activity difference at each dipole was displayed on the folded cortical surface to create an image at each time point. Together these images form a movie that shows the evolution of estimated activity on the surface of the left hemisphere in 10‐ms intervals over the entire period of interest.

RESULTS

Behavioral Results

On average, subjects took about half a second longer to respond in recall‐classify trials than in classify trials. The mean (±standard deviation) response time was 1701 ± 228 ms for recall‐classify trials and 1244 ± 181 ms for classify trials, representing a significant difference (P < 10−5, two‐tailed t‐test).

A subject response was recorded within the specified response period in 89% of trials. Of these trials, subjects responded correctly in 96% ± 6% of classify trials and 85% ± 7% of recall‐classify trials. A technical malfunction in linking one subject's recorded responses to each presented stimulus prevented exact measurement of his accuracy. Despite this, the recorded responses indicate that his performance was similar to that of other subjects.

ROI Time Course Analysis Results

A three‐way repeated measures ANOVA established differences in activity between ROIs, time subperiods, and conditions. There was a significant main effect of location, F(2,20) = 5.0, P < 0.05; of time, F(4,40) = 4.4, P < 0.01; and of condition, F(1,10) = 4.6, P < 0.05. There was also a significant interaction of time with condition, F(4,40) = 3.1, P < 0.05. Interactions of location with time and location with condition were not significant.

Time series plots show that in all three left parietal ROIs, average MEG activity in the recall‐classify condition was greater than in the classify condition during the first 100 ms after stimulus onset (Fig. 2B–D). In the left superior parietal ROI, recall‐classify activity remained greater than classify activity for ∼400 ms (Fig. 2B), with a peak difference at ∼150 ms. Recall‐classify activity was also greater than classify activity in the left supramarginal ROI (Fig. 2C), though the effect started earlier (100 ms prior to stimulus onset) and was of shorter duration, resolving by 250 ms after the stimulus. Left inferior parietal recall‐classify activity (Fig. 2D) differed from classify activity during the first 100 ms, then temporarily approached that of the classify condition, but soon diverged again and remains elevated above the classify time series until ∼300 ms.

Testing 100 ms subperiods by paired t‐tests revealed left superior parietal recall‐classify activity was significantly greater than classify activity over 0–100 ms (t = 4.4, P < 0.001), 100–200 ms (t = 6.5, P < 10−7), and 200–300 ms (t = 6.6, P < 10−7), with a trend toward significance over 300–400 ms (t = 3.6, P < 0.05). In the left supramarginal ROI, recall‐classify activity was significantly greater over −100–0 ms (t = 4.2, P < 0.01), 0–100 ms (t = 4.8, P < 0.001), and 100–200 ms (t = 5.0, P < 10−4). In the left inferior parietal ROI, recall‐classify activity was significantly greater over 0–100 ms (t = 4.4, P < 0.001) and 200–300 ms (t = 4.9, P < 10−4), with a trend over 100–200 ms (t = 3.2, P < 0.05). There was also a trend toward significantly greater classify activity in the left superior parietal ROI over −200 to −100 ms (t = −3.6, P < 0.05). None of the ROIs show a significant activity difference over 400–500 ms. All P‐values reported for these t‐tests were corrected for multiple comparisons using the Bonferroni method for 24 comparisons (eight subperiods in three ROIs).

In a supplementary analysis, all significant between‐condition activity differences in the left superior parietal and left inferior ROIs remained significant after subtraction of immediate prestimulus activity. In the left supramarginal ROI, however, no significant activity difference remained in any of the 100 ms subperiods after subtraction of the prestimulus activity. Detailed results are included in the Supporting Information.

Individual Dipole Analysis Results

From visual inspection of the left hemisphere dSPMs, the parietal effect appears to be the most prominent retrieval‐related activity difference revealed by the MEG recordings in this early time period (see Fig. 3). Other, smaller differences were observed in caudal temporal and caudal frontal lobes.

Figure 3.

Figure 3

Open in a new tab

Early retrieval‐related activity for left hemisphere. Baseline image is at −1,000 ms and is representative of the other images from the baseline period. Overlay shows the activity difference between recall‐classify (RC) and classify (C) conditions for baseline and from 150 to 230 ms following cue. Threshold was chosen for display purposes only; differences shown are >1.0 in dSPM amplitude.

DISCUSSION

Our results demonstrate an early and transient activation in both dorsal and ventral regions of the left lateral parietal cortex in response to a cue to retrieve episodic memory. With an onset within 100 ms of the retrieval cue, this increase in parietal activity precedes recall of the target and therefore would not be consistent with manipulation or representation of retrieved information. Though the parietal lobe may be involved in such later processes as well, these data suggest the parietal lobe has at least an early involvement that is consistent with a role in orienting attention to retrieval.

One lateral parietal ROI (left supramarginal) had significantly greater activity in the recall‐classify condition just prior to stimulus onset, and all three ROIs had recall‐associated differences immediately following stimulus onset. Additionally, results from a supplementary analysis suggest poststimulus activity differences in the left supramarginal ROI may be partially accounted for by the prestimulus activity difference in this region. The differences preceding stimulus onset must be dependent on the condition type generally, rather than an effort to retrieve a particular item. Recall‐associated differences occurring after stimulus onset, though, may additionally depend on specific retrieval efforts. In any case, both anticipatory and reactionary parietal responses in these MEG recordings occur prior to retrieval of the target item.

These findings support a broader role for parietal lobe involvement in directing attention not only to external stimuli, but also to internal processes [e.g., Desmurget and Sirigu, 2009]. MEG findings of anticipatory and reactionary responses in parietal cortex also converge with a recent fMRI study that attributes activity in left lateral parietal cortex to processes both preceding and during an old/new recognition task [Phillips et al., 2009]. The present study attempts to isolate episodic retrieval, rather than familiarity or general recognition, by using stimuli in the retrieval and control (classify) conditions that are equally familiar. This design allows for the possibility of incidental retrieval in control trials, which might have reduced the measured retrieval effect. Despite this potential decrease in power, significantly greater activity was observed for trials that required episodic retrieval.

It is possible that both anticipatory and reactionary effects simply reflect dissociable levels of arousal or effort in the two conditions. The recall‐classify task is, overall, more difficult than the classify task because it requires retrieval of the absent associate, and subjects may therefore exhibit increased arousal or effort in the retrieval task. However, while increased general arousal might be expected to result in increased activity in the recall‐classify condition throughout the duration of the trial, it is harder to explain why increased general arousal would result in the specific, transitory activity difference measured with MEG. As the effects presented here begin just before stimulus onset and last only a few hundred milliseconds, the relevant questions are whether a difficulty difference exists in the peristimulus onset period and, if so, whether such a difference is specifically related to episodic retrieval demands. Immediately following stimulus onset, subjects must either direct attention toward searching for episodic memory or toward searching for semantic memory. It is unknown whether subjects had more difficulty directing attention to semantic retrieval or to episodic retrieval, but the present results demonstrate that the left parietal lobe is more active when attention is directed toward search for episodic memory.

Another possible interpretation of these parietal responses is that they may represent part of a “retrieval mode” or “retrieval orientation,” tonic states that begin after instructions to retrieve and are maintained throughout retrieval effort [Rugg and Wilding, 2000; Tulving, 1983]. Subjects were instructed to retrieve 1,000 ms prior to stimulus onset, allowing time to adopt a retrieval state in anticipation of the retrieval task. There are several reasons to doubt this explanation, though. Most importantly, the retrieval mode, as defined by Tulving et al., is a tonic state maintained while episodic retrieval is required [Düzel et al., 1999; see also Rugg and Wilding, 2000]. The marked transience of the present parietal lobe findings argues against effects due to a sustained retrieval mode in the recall‐classify condition. Further, the design of the trial would be expected to minimize such effects. There is support in the EEG literature to suggest that retrieval mode activity is poorly detected when alternating task designs are employed rather than blocks of retrieval [Herron and Wilding, 2006; Morcom and Rugg, 2002]. Given that our design was a pseudorandom mixed event‐related design, subjects are less likely to enter such a tonic retrieval state. Finally, prior EEG studies have not identified prominent retrieval mode effects in the left parietal lobe. Rather, these effects appear to be greatest in the right hemisphere [Düzel et al., 1999; Herron and Wilding, 2004, 2006; Morcom and Rugg, 2002; Nyberg et al., 1995; Rugg and Wilding, 2000], with some evidence for left frontal activity in retrieval orientations associated with certain types of target information [Morcom and Rugg, 2002].

Cabeza et al. [2008] and Ciaramelli et al. [2008] suggest, based on fMRI experiments, that the parietal lobe is involved in top‐down attention to memory retrieval [Cabeza, 2008; Ciaramelli et al., 2008]. This hypothesis pertains specifically to the DPC. The present data lend support to this hypothesis, as the time course for our left superior parietal ROI does indeed show activation shortly after stimulus onset that could represent orientation to memory search. Regional anticipatory activity is also consistent with attention to retrieval in our task, as the subjects were instructed whether to retrieve well before stimulus onset. The transience of the activation (<400‐ms duration) is not consistent with top‐down attention to postretrieval processes—for instance, attention to the retrieved target during classification in the recall‐classify task—but the DPC activity may be involved in initiating search and retrieval.

Another hypothesis suggests that DPC activity may depend on the salience or task relevance of the presented test item and that the DPC role is “downstream” of retrieval [Vilberg and Rugg, 2008a]. The present data do not preclude the possibility of a later parietal activation that represents such information, but the very early DPC activation identified here, which begins over 1,500 ms prior to the average behavioral response, cannot represent processes occurring only after successful completion of retrieval. Furthermore, the presented item in each condition is equally relevant to the behavioral task at hand, so the activity difference cannot easily be explained by behavioral relevance.

The early DPC activation, though likely upstream of retrieval, might nonetheless be partially driven by increased salience of the item presented under the recall‐classify condition. In both conditions, the presented item had been previously studied, so they were equally “old,” but it is difficult to determine whether salience of the item was greater in the classify condition or in the recall‐classify condition. On one hand, items presented in the classify condition were, themselves, to be classified, which might increase their salience. On the other hand, items presented in the recall‐classify condition were cues that triggered a process of recall, requiring more effort and more attention. Under recall‐classify conditions, this additional effort and attention is quickly directed away from the item and toward retrieval. A rapid and transient activation following the cue, such as that seen here, might therefore be consistent with such an interpretation. Activity prior to the cue cannot be explained by cue salience, but might possibly be attributed to anticipation of a more salient item.

The early VPC activity seen in MEG also suggests a preretrieval role for this region, though hypotheses based on fMRI focus on a post‐retrieval role, including that VPC serves as a buffer for retrieved episodic memories. An episodic buffer [Baddeley, 2000; Vilberg and Rugg, 2008a; Wagner et al., 2005] whose function is to represent retrieved information would be expected to perform this function beginning once some information is retrieved, and would presumably have sustained activity as long as that retrieved information is held in mind. It is possible that a region contributing to an episodic buffer network might activate prior to actual retrieval in preparation for the coming retrieved load, but the transience of the MEG VPC effect is, again, difficult to reconcile with a role of holding retrieved information for manipulation during the task.

Current attentional hypotheses for VPC activity do not offer satisfactory interpretations of the present VPC findings. Existing attentional interpretations of VPC activity posit that this region reflects the capture of bottom‐up attention by the targets of retrieval [Cabeza, 2008; Ciaramelli et al., 2008]. Given that cues presented during each task were equally “old” and that the VPC response observed using MEG preceded retrieval of the target, this hypothesis is not consistent with the current findings, unless one posits an additional, later VPC response. Another possibility, however, is that subjects, having been cued that the trial would involve retrieval, showed a differential early response in VPC elicited by the studied cue in recall‐classify trials relative to those in classify trials. Nevertheless, given the timing of the response and the clearest difference between the conditions (requiring recall or not), the results of this study suggest that early VPC activity, like early DPC activity, might represent direction of attention toward memory retrieval.

CONCLUSIONS

We conclude that the left lateral parietal cortex is activated within 100 ms of a signal to retrieve episodic memories. In our cued‐recall task, this earliest activation is also transient, persisting for only a few hundred milliseconds. While there are some differences in the timing of activity in different subregions of the lateral parietal lobe, each of our three ROIs (left superior parietal, left inferior parietal, and left supramarginal) displayed an early and transient increase in estimated MEG activity when subjects were prompted to initiate retrieval. This very early response thus reflects processes prior to retrieval of the target and is most consistent with an attentional role in episodic retrieval. The finding of an early, robust, and transient activation neither precludes a later role, nor a later dissociation between subregions. Nevertheless, a comprehensive account of left lateral parietal function in episodic retrieval should include the early, transient activation revealed by MEG.

Supporting information

Additional Supporting Information may be found in the online version of this article.

Supporting Information Table 1. While the left lateral parietal cortex was the a priori area of interest, a post‐hoc analysis of the remaining FreeSurfer ROIs of the lateral cortex was also performed using the same methods as described for the ROI analysis in the manuscript. Several medial ROIs were also included in the analysis to rule out the possibility that the lateral parietal effects reflect crosstalk from sources in the medial temporal lobe or in primary visual cortex. Regions showing a sustained ‘cluster’ of significant activation are parietal and superior frontal in the left hemisphere, and dorsolateral prefrontal, parietal, and pars opercularis in the right hemisphere. P‐values were corrected for multiple (294) comparisons using the Bonferroni method.

Click here for additional data file. (116KB, doc)

Supporting Information Materials.

Click here for additional data file. (22KB, doc)

Acknowledgements

The authors thank the following individuals: Mingxiong Huang and Tao Song for assistance with MEG acquisition; Sanja Kovacevic, Ksenija Marinkovic, Eric Halgren, and Anders Dale for helpful suggestions for data analysis; and Linda McEvoy for very useful comments on early drafts of the manuscript.

REFERENCES

  1. Ally BA, Simons JS, McKeever JD, Peers PV, Budson AE ( 2008): Parietal contributions to recollection: Electrophysiological evidence from aging and patients with parietal lesions. Neuropsychologia 46: 1800–1812. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Baddeley A ( 2000): The episodic buffer: A new component of working memory? Trends Cogn Sci (Regul Ed) 4: 417–423. [DOI] [PubMed] [Google Scholar]
  3. Berryhill ME, Phuong L, Picasso L, Cabeza R, Olson IR ( 2007): Parietal lobe and episodic memory: Bilateral damage causes impaired free recall of autobiographical memory. J Neurosci 27: 14415–14423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bisiach E, Luzzatti C ( 1978): Unilateral neglect of representational space. Cortex 14: 129–133. [DOI] [PubMed] [Google Scholar]
  5. Buckner RL, Wheeler ME ( 2001): The cognitive neuroscience of remembering. Nat Rev Neurosci 2: 624–634. [DOI] [PubMed] [Google Scholar]
  6. Cabeza R ( 2008): Role of parietal regions in episodic memory retrieval: The dual attentional processes hypothesis. Neuropsychologia 46: 1813–1827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cabeza R, Nyberg L ( 2000): Imaging cognition II: An empirical review of 275 PET and fMRI studies. J Cogn Neurosci 12: 1–47. [DOI] [PubMed] [Google Scholar]
  8. Ciaramelli E, Grady CL, Moscovitch M ( 2008): Top‐down and bottom‐up attention to memory: A hypothesis (AtoM) on the role of the posterior parietal cortex in memory retrieval. Neuropsychologia 46: 1828–1851. [DOI] [PubMed] [Google Scholar]
  9. Curran T ( 2004): Effects of attention and confidence on the hypothesized ERP correlates of recollection and familiarity. Neuropsychologia 42: 1088–1106. [DOI] [PubMed] [Google Scholar]
  10. Dale AM, Fischl B, Sereno MI ( 1999): Cortical surface‐based analysis. I. Segmentation and surface reconstruction. Neuroimage 9: 179–194. [DOI] [PubMed] [Google Scholar]
  11. Dale AM, Liu AK, Fischl BR, Buckner RL, Belliveau JW, Lewine JD, Halgren E ( 2000): Dynamic statistical parametric mapping: Combining fMRI and MEG for high‐resolution imaging of cortical activity. Neuron 26: 55–67. [DOI] [PubMed] [Google Scholar]
  12. Daselaar SM, Fleck MS, Cabeza R ( 2006): Triple dissociation in the medial temporal lobes: Recollection, familiarity, and novelty. J Neurophysiol 96: 1902–1911. [DOI] [PubMed] [Google Scholar]
  13. de Munck JC ( 1992): A linear discretization of the volume conductor boundary integral equation using analytically integrated elements. IEEE Trans Biomed Eng 39: 986–990. [DOI] [PubMed] [Google Scholar]
  14. Desmurget M, Sirigu A ( 2009): A parietal‐premotor network for movement intention and motor awareness. Trends Cogn Sci (Regul Ed) 13: 411–419. [DOI] [PubMed] [Google Scholar]
  15. Duarte A, Ranganath C, Winward L, Hayward D, Knight RT ( 2004): Dissociable neural correlates for familiarity and recollection during the encoding and retrieval of pictures. Brain Res Cogn Brain Res 18: 255–272. [DOI] [PubMed] [Google Scholar]
  16. Düzel E, Cabeza R, Picton TW, Yonelinas AP, Scheich H, Heinze HJ, Tulving E ( 1999): Task‐related and item‐related brain processes of memory retrieval. Proc Natl Acad Sci USA 96: 1794–1799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fischl B, Sereno MI, Dale AM ( 1999a): Cortical surface‐based analysis. II: Inflation, flattening, and a surface‐based coordinate system. Neuroimage 9: 195–207. [DOI] [PubMed] [Google Scholar]
  18. Fischl B, Sereno MI, Tootell RB, Dale AM ( 1999b): High‐resolution intersubject averaging and a coordinate system for the cortical surface. Hum Brain Mapp 8: 272–284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Fleck MS, Daselaar SM, Dobbins IG, Cabeza R ( 2006): Role of prefrontal and anterior cingulate regions in decision‐making processes shared by memory and nonmemory tasks. Cereb Cortex 16: 1623–1630. [DOI] [PubMed] [Google Scholar]
  20. Friedman D, Johnson R ( 2000): Event‐related potential (ERP) studies of memory encoding and retrieval: A selective review. Microsc Res Tech 51: 6–28. [DOI] [PubMed] [Google Scholar]
  21. Henson RNA, Rugg MD, Shallice T, Josephs O, Dolan RJ ( 1999): Recollection and familiarity in recognition memory: An event‐related functional magnetic resonance imaging study. J Neurosci 19: 3962–3972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Herron JE, Wilding EL ( 2004): An electrophysiological dissociation of retrieval mode and retrieval orientation. Neuroimage 22: 1554–1562. [DOI] [PubMed] [Google Scholar]
  23. Herron JE, Wilding EL ( 2006): Brain and behavioral indices of retrieval mode. Neuroimage 32: 863–870. [DOI] [PubMed] [Google Scholar]
  24. Herron JE, Henson RNA, Rugg MD ( 2004): Probability effects on the neural correlates of retrieval success: An fMRI study. Neuroimage 21: 302–310. [DOI] [PubMed] [Google Scholar]
  25. Jovicich J, Czanner S, Greve D, Haley E, van der Kouwe A, Gollub R, Kennedy D, Schmitt F, Brown G, Macfall J, Fischl B, Dale A ( 2006): Reliability in multi‐site structural MRI studies: effects of gradient non‐linearity correction on phantom and human data. Neuroimage 30: 436–443. [DOI] [PubMed] [Google Scholar]
  26. Kahn I, Davachi L, Wagner AD ( 2004): Functional‐neuroanatomic correlates of recollection: Implications for models of recognition memory. J Neurosci 24: 4172–4180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kim H, Cabeza R ( 2007): Trusting our memories: Dissociating the neural correlates of confidence in veridical versus illusory memories. J Neurosci 27: 12190–12197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kim H, Cabeza R ( 2009): Common and specific brain regions in high‐ versus low‐confidence recognition memory. Brain Res 1282: 103–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Morcom AM, Rugg MD ( 2002): Getting ready to remember: The neural correlates of task set during recognition memory. Neuroreport 13: 149–152. [DOI] [PubMed] [Google Scholar]
  30. Mosher JC, Leahy RM, Lewis PS ( 1999): EEG and MEG: Forward solutions for inverse methods. IEEE Trans Biomed Eng 46: 245–259. [DOI] [PubMed] [Google Scholar]
  31. Neville HJ, Kutas M, Chesney G, Schmidt AL ( 1986): Event‐related brain potentials during initial encoding and recognition memory of congruous and incongruous words. J Mem Lang 25: 75–92. [Google Scholar]
  32. Nyberg L, Tulving E, Habib R, Nilsson LG, Kapur S, Houle S, Cabeza R, McIntosh AR ( 1995): Functional brain maps of retrieval mode and recovery of episodic information. Neuroreport 7: 249–252. [PubMed] [Google Scholar]
  33. Olson IR, Berryhill M ( 2009): Some surprising findings on the involvement of the parietal lobe in human memory. Neurobiol Learn Mem 91: 155–165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Oostendorp TF, van Oosterom A ( 1989): Source parameter estimation in inhomogeneous volume conductors of arbitrary shape. IEEE Trans Biomed Eng 36: 382–391. [DOI] [PubMed] [Google Scholar]
  35. Owen AM, Milner B, Petrides M, Evans AC ( 1996): Memory for object features versus memory for object location: A positron‐emission tomography study of encoding and retrieval processes. Proc Natl Acad Sci USA 93: 9212–9217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Phillips JS, Velanova K, Wolk DA, Wheeler ME ( 2009): Left posterior parietal cortex participates in both task preparation and episodic retrieval. Neuroimage 46: 1209–1221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rugg MD, Wilding EL ( 2000): Retrieval processing and episodic memory. Trends Cogn Sci (Regul Ed) 4: 108–115. [DOI] [PubMed] [Google Scholar]
  38. Rugg M, Yonelinas A ( 2003): Human recognition memory: A cognitive neuroscience perspective. Trends Cogn Sci (Regul Ed) 7: 313–319. [DOI] [PubMed] [Google Scholar]
  39. Rugg MD, Curran T ( 2007): Event‐related potentials and recognition memory. Trends Cogn Sci (Regul Ed) 11: 251–257. [DOI] [PubMed] [Google Scholar]
  40. Rugg MD, Fletcher PC, Allan K, Frith CD, Frackowiak RS, Dolan RJ ( 1998): Neural correlates of memory retrieval during recognition memory and cued recall. Neuroimage 8: 262–273. [DOI] [PubMed] [Google Scholar]
  41. Rugg MD, Otten LJ, Henson RNA ( 2002): The neural basis of episodic memory: Evidence from functional neuroimaging. Philos Trans R Soc Lond B Biol Sci 357: 1097–1110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Spaniol J, Davidson PSR, Kim ASN, Han H, Moscovitch M, Grady CL ( 2009): Event‐related fMRI studies of episodic encoding and retrieval: Meta‐analyses using activation likelihood estimation. Neuropsychologia 47: 1765–1779. [DOI] [PubMed] [Google Scholar]
  43. Tulving E ( 1983): Elements of Episodic Memory. Oxford University Press: New York. [Google Scholar]
  44. Tulving E, Kapur S, Markowitsch HJ, Craik FI, Habib R, Houle S ( 1994): Neuroanatomical correlates of retrieval in episodic memory: Auditory sentence recognition. Proc Natl Acad Sci USA 91: 2012–2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Vilberg KL, Rugg MD ( 2007): Dissociation of the neural correlates of recognition memory according to familiarity, recollection, and amount of recollected information. Neuropsychologia 45: 2216–2225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Vilberg KL, Rugg MD ( 2008a): Memory retrieval and the parietal cortex: A review of evidence from a dual‐process perspective. Neuropsychologia 46: 1787–1799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Vilberg KL, Rugg MD ( 2008b): Functional significance of retrieval‐related activity in lateral parietal cortex: Evidence from fMRI and ERPs. Hum Brain Mapp 30: 1490–1501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wagner AD, Shannon BJ, Kahn I, Buckner RL ( 2005): Parietal lobe contributions to episodic memory retrieval. Trends Cogn Sci 9: 445–453. [DOI] [PubMed] [Google Scholar]
  49. Wheeler ME, Buckner RL ( 2003): Functional dissociation among components of remembering: Control, perceived oldness, and content. J Neurosci 23: 3869–3880. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wilding EL, Rugg MD ( 1996): An event‐related potential study of recognition memory with and without retrieval of source. Brain 119 ( Part 3): 889–905. [DOI] [PubMed] [Google Scholar]
  51. Yonelinas AP ( 2001): Components of episodic memory: The contribution of recollection and familiarity. Philos Trans R Soc Lond B Biol Sci 356: 1363–1374. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Additional Supporting Information may be found in the online version of this article.

Supporting Information Table 1. While the left lateral parietal cortex was the a priori area of interest, a post‐hoc analysis of the remaining FreeSurfer ROIs of the lateral cortex was also performed using the same methods as described for the ROI analysis in the manuscript. Several medial ROIs were also included in the analysis to rule out the possibility that the lateral parietal effects reflect crosstalk from sources in the medial temporal lobe or in primary visual cortex. Regions showing a sustained ‘cluster’ of significant activation are parietal and superior frontal in the left hemisphere, and dorsolateral prefrontal, parietal, and pars opercularis in the right hemisphere. P‐values were corrected for multiple (294) comparisons using the Bonferroni method.

Click here for additional data file. (116KB, doc)

Supporting Information Materials.

Click here for additional data file. (22KB, doc)

Articles from Human Brain Mapping are provided here courtesy of Wiley

RESOURCES