Abstract

The hippocampus and the striatum are thought to play distinct roles in learning and memory, each supporting an independent memory system. A fundamental question is whether, and how, these systems interact to jointly contribute to learning and memory. In particular, it remains unknown whether the striatum contributes selectively to implicit, habitual learning, or whether the striatum may also contribute to long-term episodic memory. Here, we show with functional magnetic resonance imaging (fMRI) that the hippocampus and the striatum interact cooperatively to support episodic memory formation. Participants were scanned during a memory encoding paradigm and, subsequently, were tested for memory of encoded items. fMRI data revealed that successful memory was associated with greater activity in both the hippocampus and the striatum (putamen) during encoding. Furthermore, activity in the hippocampus and the striatum was correlated within subjects for items that were later remembered, but not for items that were forgotten. Finally, across subjects, the strength of the correlation between the hippocampus and the striatum predicted memory success. These findings provide novel evidence for contributions of both the striatum and the hippocampus to successful episodic encoding and for a cooperative interaction between them.

INTRODUCTION

Decades of research indicate that there are multiple forms of learning and memory that depend on discrete memory systems (Squire, 2004; Gabrieli, 1998; Squire & Zola, 1996; Cohen & Squire, 1980). A declarative memory system for events or episodes, referred to as episodic memory, is known to depend on the medial temporal lobe (MTL; including the hippocampus and surrounding MTL cortices; Eichenbaum, 2004; Squire, 2004; Paller & Wagner, 2002). A distinct and independent habit learning system is thought to support incremental learning of stimulus–response associations and to depend on the striatum (Shohamy, Myers, Grossman, et al., 2004; Shohamy, Myers, Onlaor, & Gluck, 2004; Poldrack et al., 2001; Packard & Teather, 1998; Knowlton, Mangels, & Squire, 1996; for reviews, see Shohamy, Myers, Kalanithi, & Gluck, 2008; Yin & Knowlton, 2006).

Recent data, however, suggest that the hippocampus and the striatum may interact in some cases, with reports of both competitive and cooperative interactions between them (Doeller, King, & Burgess, 2008; Lee, Duman, & Pittenger, 2008; Foerde, Knowlton, & Poldrack, 2006; Seger & Cincotta, 2006; Voermans et al., 2004; Poldrack et al., 2001; see also Hartley & Burgess, 2005). When and how these two memory systems interact is a topic of active investigation, and many open questions remain. As detailed below, two factors which may be important to consider jointly are (a) the anatomical subregions of the striatum and (b) the nature of the task being learned.

The striatum is a heterogeneous structure which is often divided into distinct anatomical and functional subregions. Studies in animals highlight a role for the ventral striatum (nucleus accumbens) in goal-directed learning of stimulus–reward associations (Haber, Kim, Mailly, & Calzavara, 2006; Alexander, DeLong, & Strick, 2003). By contrast, the dorsolateral striatum in rodents (thought to be the analog of the dorsal putamen in humans) is thought to support gradual acquisition of habitual behaviors (Yin & Knowlton, 2006; Faure, Haberland, Conde, & Massioui, 2005; Yin, Knowlton, & Balleine, 2004). Yet a third functional network involving the dorsomedial striatum (considered an analog to the caudate in humans) is thought to support goal-directed learning of action–outcome contingencies (Williams & Eskandar, 2006; Yin & Knowlton, 2006; Faure et al., 2005; Yin et al., 2004; Alexander et al., 2003). Anatomical and behavioral studies suggest that the hippocampus has a direct cooperative interaction with the ventral striatum (Goto & Grace, 2008; Goto & O'Donnell, 2001), whereas the relation between the hippocampus and the dorsomedial and dorsolateral striatum is less well characterized.

Studies in humans are somewhat consistent with these findings. One particular point of consistency is the demonstration of a cooperation between the hippocampus and the ventral striatum during learning (Adcock, Thangavel, Whitfield-Gabrieli, Knutson, & Gabrieli, 2006; Wittmann et al., 2005). However, although there is some evidence for parallel organization of subregions of the striatum in different forms of learning in humans (Cincotta & Seger, 2007; Seger & Cincotta, 2005, 2006; Tricomi, Delgado, & Fiez, 2004), the data are less consistent across studies and suggest that there may be more functional overlap between medial and lateral portions of the striatum in humans. Indeed, despite compelling evidence for a dorsal versus ventral divide, it remains uncertain precisely how striatal subregions align across species. Further, it is important to consider that there is much overlap and interaction between multiple parallel loops in the striatum (Haber et al., 2006; Haber, Fudge, & McFarland, 2000), suggesting that these distinctions might be viewed more as gradations rather than subregions per se.

Studies in humans suggest that the nature of the interaction between the hippocampus and the striatum may also differ depending on the kind of learning processes that are engaged. To date, most studies of the role of the striatum in learning have focused on multitrial feedback-driven learning, such as probabilistic category learning (e.g., Foerde et al., 2006; Shohamy, Myers, Grossman, et al., 2004; Poldrack et al., 2001; Knowlton et al., 1996; for a review, see Shohamy et al., 2008). In such incremental learning paradigms, participants are asked to predict outcomes for a set of stimuli using trial-by-error feedback, and each stimulus is associated with each outcome with a given probability. Learning is driven by feedback, and occurs gradually, over many trials. Neuropsychological and imaging studies indicate that this kind of incremental learning depends on the striatum (e.g., Foerde et al., 2006; Shohamy, Myers, Grossman, et al., 2004; Poldrack et al., 2001; Knowlton et al., 1996; for a review, see Shohamy et al., 2008), and elicits a competitive interaction between the striatum (dorsal caudate) and the hippocampus during learning (Poldrack et al., 2001; Poldrack, Prabhakaran, Seger, & Gabrieli, 1999). A competitive interaction between the hippocampus and the striatum has also been demonstrated in animals with navigation studies that involve repeated incremental learning of associations between actions and rewards. Importantly, lesion studies of rodents and imaging studies of humans with striatal-degenerative disease indicate that when one of the two structures is impaired, the function of the intact structure is enhanced (Lee et al., 2008; Hartley & Burgess, 2005; Moody, Bookheimer, Vanek, & Knowlton, 2004; Voermans et al., 2004; Dagher, Owen, Boecker, & Brooks, 2001; McDonald & White, 1993; Packard, Hirsh, & White, 1989).

Beyond its documented role in multitrial learning, recent data suggest that the striatum may also contribute to single-trial episodic memory formation, and may do so jointly with the hippocampus (Albouy et al., 2008; Shohamy & Wagner, 2008; Tricomi & Fiez, 2008; DeCoteau et al., 2007; Adcock et al., 2006; Schott et al., 2006; Staresina & Davachi, 2006; Degonda et al., 2005; Wittmann et al., 2005; Schendan, Searl, Melrose, & Stern, 2003; Rose, Haider, Weiller, & Büchel, 2002). Much recent progress has been made into understanding the neural bases of episodic memory in humans using the subsequent memory paradigm. This paradigm enables the examination of the neural correlates of successful encoding in episodic memory by contrasting brain activity during encoding of items subsequently remembered with encoding activity of items subsequently forgotten. This approach has shown that successful encoding of episodic memories depends primarily on the hippocampus and prefrontal cortex (PFC; for review see Paller & Wagner, 2002). Interestingly, a review of a large number of fMRI subsequent memory studies reveals numerous cases where activation in multiple subregions of the striatum is found to show a parallel pattern, being more highly activated during encoding of subsequently remembered items compared to subsequently forgotten items. Notably, such effects have been reported both in the putamen (Adcock et al., 2006; Prince, Daselaar, & Cabeza, 2005; Sperling et al., 2003) and in the caudate (Axmacher, Schmitz, Weinreich, Elger, & Fell, 2008; Ritchey, Dolcos, & Cabeza, 2008; Dennis, Daselaar, & Cabeza, 2007; Nichols, Kao, Verfaellie, & Gabrieli, 2006; Staresina & Davachi, 2006; Prince et al., 2005; Chua, Rand-Giovannetti, Schacter, Albert, & Sperling, 2004; Schon, Hasselmo, LoPresti, Tricarico, & Stern, 2004). These effects thus appear to go beyond the known cooperative interactions between the ventral striatum and the hippocampus: They are found in other regions of the striatum, and are found in tasks that do not involve reward. Thus, these findings suggest that at least in some cases the striatum and the hippocampus may jointly, and perhaps cooperatively, contribute to the formation of long-term episodic memories for unique single-trial events.

Like subsequent memory studies, data from neurologically impaired individuals also suggest involvement of the striatum in episodic encoding. Patients with striatal degeneration following Parkinson or Huntington disease, well known to have impairments in gradual trial-by-error learning (Shohamy, Myers, Grossman, et al., 2004; Knowlton et al., 1996), also have subtle but consistent impairments in episodic memory (Beste, Saft, Gunturkun, & Falkenstein, 2008; Altgassen, Phillips, Kopp, & Kliegel, 2007). Episodic memory impairments are also found in patients with focal lesions to the striatum (mainly caudate; Vakil, Blachstein, & Soroker, 2004).

What cognitive circumstances might trigger the involvement of the dorsal striatum in episodic encoding—and thereby its cooperation with the hippocampus? Both theoretical and empirical considerations led us to believe that the dorsal striatum may be involved in encoding in tasks which heavily tax complex working memory operations. Specifically, it has been suggested recently that the dorsal striatum supports operations in which certain stimuli need to be filtered out, whereas others need to be actively maintained in working memory (McNab & Klingberg, 2008). Theoretical models further suggest that, in such contexts, the striatum may interact with the hippocampus (as well as other cortical regions such as PFC; Hazy, Frank, & O'Reilly, 2007).

We therefore sought an encoding task which would involve both maintenance of relevant stimuli and filtering out of irrelevant stimuli. To this end, we used a phoneme substitution task in which participants were presented with two words and had to create a third word by combining the first syllable of the first word with the second syllable of the second word, while disregarding the remaining syllables. Thus, this task involved both maintenance of relevant stimuli (the two syllables combining the third word) and filtering out of irrelevant stimuli (the remaining syllables). Importantly, we wished to explicitly test whether the involvement of the striatum in working memory functions underlies its suggested role in explicit memory. Therefore, our task was designed such that the working memory operations of maintenance and filtering were critical for episodic encoding, as the to-be-remembered stimuli were only generated when the working memory task was performed successfully.

Using this encoding task, we found that activity in both the hippocampus and the striatum predicted successful subsequent memory. We also found a positive correlation between the hippocampus and the striatum during encoding that was correlated with memory success and was selective to items that were later remembered. These findings imply that memory systems in the hippocampus and the striatum may cooperatively interact to support successful formation of episodic memories.

METHODS

Participants

Fifteen subjects (8 women; ages 22–33 years, mean = 26.3 years) participated in this study. Data from five additional subjects were excluded due to poor task compliance or excessive motion. All subjects reported themselves as native Hebrew speakers, were neurologically intact, and had normal or corrected-to-normal vision. Subjects were paid for their participation in the experiment. Informed consent was obtained in a manner approved by the Tel-Aviv Medical Center on Clinical Investigation.

Behavioral Procedure

Both study and test phases of the experiment were conducted in the scanner. In each trial of the encoding task, participants were presented with two words (via headphones) and simultaneously shown visual images corresponding to each of the words (via a mirror which reflected an image of the projection screen). Because the input words were presented aurally in the noisy scanner environment, the purpose of the corresponding images was to prevent possible ambiguity regarding the identity of the input words (e.g., to help participants distinguish between “butter” and “cutter”). Participants were assigned the task of combining the first syllable of the first word with the second syllable of the second word. In order to track the incidental encoding task performance, subjects indicated at each trial whether or not the combination of the two relevant syllables created a semantically meaningful word (“word” trials, 50% of the encoding trials, or “nonwords” trials, 50% of the encoding trials). There were 180 encoding trials (90 “words” and 90 “nonwords”). Each study trial lasted 4 sec, where presentation of the first word and image coincided with the beginning of the trial for 1200 msec, the second word and visual image then presented for 1200 msec (the image corresponding to the first word remained on the screen for this period), and a fixation cross then appeared for an additional 1600 msec when participants made their word/nonword judgment. Their responses were made with the index and ring fingers of the left hand (the mapping of response to finger was counterbalanced across participants).

Baseline trials, in which a fixation cross was presented, were interleaved among the trials using a rapid event-related design (Dale & Buckner, 1997). The duration of the baseline trials randomly varied between 2 and 8 sec (mean ITI = 3.66 sec), totaling one third of the study duration.

Participants were carefully instructed and trained on the study task prior to entering the scanner. This training, conducted outside the scanner, was then repeated verbatim once each participant was inside the scanner. The items used for the training session were taken from a different pool of words than those used for the encoding task itself. The instructions encouraged participants to combine the syllables by “keeping in mind” the first syllable until the appearance of the second word.

When the study phase was finished there was an approximately 20-min break during which participants underwent anatomical scans while listening to classical music. After the break, instructions for the surprise recognition test were projected on the screen. Participants were then given a short training of the test phase, before proceeding to the test itself.

The test included 180 words: 90 old internally generated words and 90 new words. The order of presentation was pseudorandom, with the proviso that each quarter of the encoding list was represented equally in each quarter of the recognition test. For each word, participants were required to determine whether it was a word that they had generated during the study phase, choosing one of three fixed options: (i) “definitely yes,” for words remembered with high confidence; (ii) “yes,” for words remembered with low confidence; or (iii) “no,” for words that they did not recognize as having been generated earlier. Participants were required to indicate their choice by pressing on one of three buttons of the response box. These responses were used to categorize each of the participants' encoding trials as one of three trial types: high-confidence hits (items subsequently remembered with high confidence; H-hit), low-confidence hits (items subsequently remembered with low confidence), or misses (items subsequently forgotten; Miss).

Each 3-sec trial of the test phase was presented both visually and aurally—the visual presentation lasting 2600 msec—followed by a 400-msec fixation cross. The duration of baseline trials for the test phase randomly varied between 3 and 9 sec.

Materials

All 540 word stimuli were two-syllable Hebrew words, three to seven letters long (mean = 3.9; SD = 0.7). For the study phase, two sets of stimuli were compiled, each presented to half of the subjects. Each set contained 90 pairs of concrete nouns which together formed a semantically meaningful word, and 90 pairs of concrete nouns which together formed a nonword (see Behavioral Procedure). The semantically meaningful words were of various parts of speech (nouns, adjectives, and verbs; an item analysis of the semantically meaningful words according to subsequent memory condition revealed no item effect). Nonword stimuli were 180 three- to five-letter strings (mean = 3.8; SD = 0.68). The nonwords were created by combining the first and second syllables of the 180 noun-pairs from the two noun-pair sets. For each of the two noun-pair sets, a set of 90 nonwords was matched such that the nonwords were syllable combinations of the complementary set of noun-pairs. For aural presentation of the noun-pairs, we obtained male voice recording of each of the nouns. These recordings were thoroughly tested in several pilots. Images corresponding to the nouns in each of the pairs were 3 × 3 inch colored drawings adapted from www.clipart.com.

Two sets of word stimuli were created, framing each word as a target in the following recognition test in one set and as a foil in the other. These two sets were counterbalanced across participants.

Imaging Procedure

Whole-brain T2*-weighted EPI functional images were acquired with a GE 3-T Signa Horizon LX 9.1 echo speed scanner (Milwaukee, WI). The study phase consisted of three sequential scanning sessions in each of which 165 volumes were acquired (TR = 2000 msec, 200 mm FOV, 64 × 64 matrix, TE = 35, 30 pure axial slices, 3.125 × 3.125 × 4 mm voxel size, no gap). For the test phase, there were two sequential scanning sessions in each of which 124 volumes were acquired (TR = 3000 msec, 200 mm FOV, 64 × 64 matrix, TE = 35, 30 pure axial slices, 3.125 × 3.125 × 4 mm voxel size, no gap). In both phases, slices were collected in an interleaved order. At the beginning of each scanning session, four additional volumes were acquired to allow for T1 equilibration (they were not included in the analysis). Before the study phase, high-resolution anatomical images (SPGR; 1 mm sagittal slices) were obtained. Head motion was minimized by using cushions arranged around each participant's head, and by explicitly guiding the participants prior to entering the scanner.

Imaging data were preprocessed and analyzed using SPM2 (Wellcome Department of Cognitive Neurology, London). A slice-timing correction to the first slice was performed followed by realignment of the images to the first image. Next, data were spatially normalized to an EPI template based upon the MNI305 stereotactic space (Cocosco, Kollokian, Kwan, & Evans, 1997). The images were then resampled into 2-mm3 cubic voxels, and finally smoothed with an 8-mm FWHM isotropic Gaussian kernel.

In order to model task-related activity in each of the relevant conditions, the canonical hemodynamic response was convolved with the onset of each trial. The general linear model was used for statistical analyses. For each subject, a fixed-effect model was implemented to linearly contrast brain activity for the effects of interest. We then computed the second-level analyses (subjects treated as random effects) with one-sample t tests (threshold p = .001; extent threshold of 5 contiguous voxels).

Because our a priori hypothesis concerned the interaction between the hippocampus and the striatum, effects within these a priori regions were small-volume corrected using an anatomical mask. The mask was created using the Anatomical Automatic Labeling brain atlas (Tzourio-Mazoyer et al., 2002) and the Wake Forest University Pick Atlas Tool. This mask included the bilateral hippocampus, amygdala, parahippocampal gyri, and fusiform gyri (to cover the MTL regions), and the bilateral putamen, caudate, and globus pallidus (to cover the striatal regions). The resulting ROIs were summed and used as a single mask during analyses. These ROIs allowed for relatively conservative small-volume correction. To explore the differences between conditions, we used these ROIs as masks for identifying clusters that survived the thresholding criteria in the contrast of H-hit > Miss.

Beta-series Node-based Connectivity Analysis

A beta-series node-based connectivity analysis (Rissman, Gazzaley, & D'Esposito, 2004) was applied in order to calculate the within-subject correlations between memory condition-related activation in clusters of activity in the hippocampus and the striatum. This method allowed us to extract an individual beta value for each trial by modeling each as an individual condition (thus obtaining 180 beta values for each subject). To avoid spurious correlations driven by extreme values, the most extreme-lowest and highest-beta values were removed as outliers, leading to an exclusion of 1% of the data (10 data points; removal of outliers revealed the same pattern of results as the data including them). The beta values were then sorted according to the four relevant task conditions (“high-confidence hits,” “low-confidence hits,” “misses,” and “nonword”). The correlation coefficients between activation corresponding to each of the memory conditions in each region could now be calculated for each subject. We then applied a Fisher transformation on the correlation coefficients, aiming to make their sampling distribution approach that of the normal distribution, and divided the results by the coefficients' standard deviations (1/√n − 3)—creating transformed correlation coefficients. These steps were taken to allow for the use of t tests on the transformed coefficients. Finally, t tests were applied to check whether the mean correlation (across subjects) was significantly greater than zero.

RESULTS

Behavioral Data

Results of the recognition test are presented in Table 1. An average of 41 ± 10% of the words were remembered with high confidence, whereas an average of 30 ± 9% of words were forgotten (n = 15, mean ± SD).

Table 1. 

Mean (SD) Proportions of Responses in the Recognition Test

Response
Item Type
Old
New
Old–high confidence 0.41 (0.10) 0.05 (0.04) 
Old–low confidence 0.29 (0.08) 0.16 (0.09) 
New 0.30 (0.09) 0.79 (0.11) 
Response
Item Type
Old
New
Old–high confidence 0.41 (0.10) 0.05 (0.04) 
Old–low confidence 0.29 (0.08) 0.16 (0.09) 
New 0.30 (0.09) 0.79 (0.11) 

Table 2 presents accuracy levels and response latencies (in the word/nonword decision) under each of the subsequent memory conditions (high-confidence hits, low-confidence hits, and misses). A within-subject analysis of variance revealed no significant differences between the subsequent memory conditions for accuracies or response latencies [F(2, 28) = 1.47, p = .25 for accuracies; F < 1 for response latencies].

Table 2. 

Accuracy and Response Latencies in the Word/Nonword Decision Task


High-confidence Hits
Low-confidence Hits
Misses
Accuracy 0.90 (0.07) 0.87 (0.10) 0.86 (0.11) 
Response latencies 2673 (200) 2673 (216) 2684 (214) 

High-confidence Hits
Low-confidence Hits
Misses
Accuracy 0.90 (0.07) 0.87 (0.10) 0.86 (0.11) 
Response latencies 2673 (200) 2673 (216) 2684 (214) 

Mean (SD) for each of the subsequent memory conditions.

Finally, an item analysis of the internally generated words revealed no item effect which could account for memory performance.

Imaging Data

Subsequent Memory Analysis

A subsequent memory analysis was conducted in which trials corresponding to high-confidence hits (H-hit) were directly compared to trials corresponding to forgotten target words (Miss). In line with our a priori hypothesis, this contrast revealed both striatal and hippocampal activations. Specifically, regions in the bilateral putamen [right: (A) x = 28, y = 0, z = 10; (B) x = 24, y = −6, z = 16; left: x = −22, y = −2, z = 14; Figure 1] and in the left hippocampus (x = −24, y = −26, z = −6; Figure 2) were more activated during encoding of subsequently remembered items, in comparison to subsequently forgotten ones (p < .001, extent threshold 5 voxels; p < .05, small-volume corrected for the MTL and striatal regions). To further explore brain activity during encoding, we used a more liberal statistical threshold (p < .001, extent threshold 5 voxels; uncorrected) to search for additional regions which were activated by successful encoding (for the full set of regions see Table 3 and Figure 3). This analysis revealed activation in a region in left inferior prefrontal cortex (iPFC; x = −56, y = 30, z = 2)—a region generally activated during successful encoding (Paller & Wagner, 2002).

Figure 1. 

Bilateral putamen [right: (A) x = 28, y = 0, z = 10; (B) x = 24, y = −6, z = 16; left: x = −22, y = −2, z = 14] observed in the H-hit > Miss contrast. Statistical parametric maps rendered on an average MNI template brain (p < .001, extent threshold 5 voxels; p < .05, small-volume corrected for the MTL and striatal regions).

Figure 1. 

Bilateral putamen [right: (A) x = 28, y = 0, z = 10; (B) x = 24, y = −6, z = 16; left: x = −22, y = −2, z = 14] observed in the H-hit > Miss contrast. Statistical parametric maps rendered on an average MNI template brain (p < .001, extent threshold 5 voxels; p < .05, small-volume corrected for the MTL and striatal regions).

Figure 2. 

Left hippocampus (x = −24, y = −26, z = −6) observed in the H-hit > Miss contrast. Statistical parametric maps rendered on an average MNI template brain (p < .001, extent threshold 5 voxels; p < .05, small-volume corrected for the MTL and striatal regions).

Figure 2. 

Left hippocampus (x = −24, y = −26, z = −6) observed in the H-hit > Miss contrast. Statistical parametric maps rendered on an average MNI template brain (p < .001, extent threshold 5 voxels; p < .05, small-volume corrected for the MTL and striatal regions).

Table 3. 

Regions Observed in the Encoding H-hit > Miss Contrast (p < .001, uncorrected, Cluster > 5 Voxels)

Region
Number of Voxels
t
MNI Coordinates
x
y
z
Frontal 
L Inferior frontal gyrus 72 7.86 −62 −12 12 
16 5.88 −62 20 14 
4.19 −56 30 
L Precentral gyrus 79 4.53 −4 −20 68 
52 4.63 −14 −26 72 
38 4.44 −40 −18 46 
18 4.09 −48 −18 52 
R Precentral gyrus 180 8.5 24 −28 62 
R Subcentral gyrus 55 4.96 46 −14 26 
L Subcentral gyrus 16 4.44 −44 −14 28 
 
Temporal 
L Hippocampus 50 5.44 −24 −26 −6 
L Superior temporal gyrus 29 4.49 −66 −30 
39 4.42 −66 −18 
 
Parietal 
L Postcentral gyrus 73 5.34 −20 −34 70 
56 4.67 −10 −36 70 
R Postcentral gyrus 119 6.26 16 −36 62 
R Supramarginal gyrus 48 5.02 50 −26 22 
R Insula 168 5.57 32 −8 
38 4.68 48 −14 14 
L Cingulate gyrus 27 4.34 −4 −34 54 
 
Subcortical 
L Thalamus 25 4.28 −28 −10 
R Thalamus 120 6.27 28 −10 −6 
L Putamen 84 5.81 −22 −2 14 
R Putamen 97 5.47 28 10 
64 4.9 24 −6 16 
R Isthmus 46 4.72 14 −28 −10 
L Cerebellum 44 4.8 −24 −44 −34 
27 4.32 −36 −48 −38 
R Cerebellum 18 4.84 −40 
11 4.11 30 −56 −26 
Region
Number of Voxels
t
MNI Coordinates
x
y
z
Frontal 
L Inferior frontal gyrus 72 7.86 −62 −12 12 
16 5.88 −62 20 14 
4.19 −56 30 
L Precentral gyrus 79 4.53 −4 −20 68 
52 4.63 −14 −26 72 
38 4.44 −40 −18 46 
18 4.09 −48 −18 52 
R Precentral gyrus 180 8.5 24 −28 62 
R Subcentral gyrus 55 4.96 46 −14 26 
L Subcentral gyrus 16 4.44 −44 −14 28 
 
Temporal 
L Hippocampus 50 5.44 −24 −26 −6 
L Superior temporal gyrus 29 4.49 −66 −30 
39 4.42 −66 −18 
 
Parietal 
L Postcentral gyrus 73 5.34 −20 −34 70 
56 4.67 −10 −36 70 
R Postcentral gyrus 119 6.26 16 −36 62 
R Supramarginal gyrus 48 5.02 50 −26 22 
R Insula 168 5.57 32 −8 
38 4.68 48 −14 14 
L Cingulate gyrus 27 4.34 −4 −34 54 
 
Subcortical 
L Thalamus 25 4.28 −28 −10 
R Thalamus 120 6.27 28 −10 −6 
L Putamen 84 5.81 −22 −2 14 
R Putamen 97 5.47 28 10 
64 4.9 24 −6 16 
R Isthmus 46 4.72 14 −28 −10 
L Cerebellum 44 4.8 −24 −44 −34 
27 4.32 −36 −48 −38 
R Cerebellum 18 4.84 −40 
11 4.11 30 −56 −26 
Figure 3. 

Statistical parametric maps of activation from the H-hit > Miss contrast (unmasked; p < .001, uncorrected) are rendered on a canonical brain template. For a list of all regions showing a significant subsequent memory effect, see Table 3.

Figure 3. 

Statistical parametric maps of activation from the H-hit > Miss contrast (unmasked; p < .001, uncorrected) are rendered on a canonical brain template. For a list of all regions showing a significant subsequent memory effect, see Table 3.

Correlation Analyses

Our a priori interest with regard to putative interactions during encoding concerned the hippocampus and the striatum. Therefore, our analysis focused on the correlations between the hippocampus and the three striatal regions detected in the subsequent memory analysis. To explore the pattern of correlations between the hippocampus and the striatum, we conducted the following steps of analysis:

  • Step 1: A beta-series node-based connectivity analysis (described in the Methods section) was conducted to investigate the within-subject interaction between the ROIs (see also Axmacher et al., 2008; Ritchey et al., 2008; Rissman et al., 2004). For each of the three striatal regions, we required that (a) a correlation significantly larger than zero be found between the hippocampus and the striatal region for the H-hits trials and (b) this correlation be significantly different than the one for the Miss trials. Thus, our requirement was for the conjunction of effects (a) and (b), in the sense of the logical AND operator. We therefore report—in addition to the p values of each effect individually—the joint p value of both effects. Because this step of analysis regarded the correlation between the hippocampus and three different striatal regions, we adjusted for multiple comparisons by correcting the joint p values for the three analyses (one for each striatal region) using the Bonferroni method (Fleiss, 1986).

  • Step 2: We next examined whether, across subjects, the correlation between the striatum and the hippocampus during successful encoding was related to memory strength. To this end, we checked the across-subject correlation between the d′ scores of the participants and their transformed correlation coefficients. This step of analysis was only conducted for the striatal region(s) that survived the first step of analysis.

Step 1

For the H-hit condition, the correlations between both regions in the right putamen (see Figure 1) and the left hippocampus were significantly larger than zero [A: mean r = .23, t(14) = 3.99, p < .001; B: mean r = .29, t(14) = 4.31, p < .001]. In contrast, the same correlations were not significantly greater than zero for the Miss condition [A: mean r = .0134; t(14) < 1; B: mean r = .16, t(14) = 1.47, p = .08]. Thus, a distinct pattern of interaction between the putamen and the hippocampus for each of the two memory conditions was observed. A direct comparison of the mean correlation of the H-hit to the Miss conditions revealed a statistically significant difference between the conditions for Region A [t(14) = 1.8, p = .04; joint p value = 0.001; corrected]. To illustrate the interaction between the two structures, we present data for a typical subject showing the pattern of correlation between this right putamen region and the hippocampus for the H-hit condition (Figure 4).

Figure 4. 

Single-subject correlation between beta series in the right putamen (A) and in the left hippocampus for the H-hit condition (r = .5).

Figure 4. 

Single-subject correlation between beta series in the right putamen (A) and in the left hippocampus for the H-hit condition (r = .5).

For the left putamen (which was also observed in the H-hit vs. Miss contrast), the correlation with the hippocampus was significantly greater than zero for the H-hit condition [mean r = .15, t(14) = 1.95, p = .04], but not for the Miss condition [mean r = .07, t(14) = 0.69, p = .25]. However, the difference between the two conditions did not reach statistical significance [t(14) = 0.99, p = .17].

Step 2

This step of the analysis was only conducted for Region A of the right putamen, as only this region survived the conjunction of both effects (a: significant correlation for H-hit condition; b: different pattern of correlations between H-hit and Miss conditions). This analysis found a positive across-subject correlation between the striatum–hippocampus transformed correlation coefficients and the d′ scores of the participants (r = .52, p = .046; Figure 5).

Figure 5. 

Across-subject correlation between the right putamen (A)–left hippocampus transformed correlation coefficients and the d′ scores.

Figure 5. 

Across-subject correlation between the right putamen (A)–left hippocampus transformed correlation coefficients and the d′ scores.

In light of the rich literature connecting iPFC to episodic encoding (Paller & Wagner, 2002), we also examined whether activity in both the hippocampus and the three striatal regions correlated with activity in a prefrontal region that showed the subsequent memory effect. This analysis followed the same steps as the analysis of our a priori hypothesis (namely, the above Steps 1 and 2). Note that because this analysis regarded the correlation between iPFC and four regions (the hippocampus and three different striatal regions), we adjusted for multiple comparisons by correcting the probabilities for four analyses.

Step 1: PFC

The correlation between the left hippocampus and left iPFC did not reach significance for the H-hit condition [mean r = .10, t(14) = 1.36, p = .20]. The same correlation was found significant for the Miss condition [mean r = .27, t(14) = 2.59, p = .02]. The correlation between both right putamen regions and left iPFC were significant for the H-hit condition but not for the Miss condition [Region (A): mean r = .29, t(14) = 3.78, p = .002 for the H-hit condition; mean r = −.06, t(14) = −0.55, p = .58 for the miss condition; Region (B): mean r = .23, t(14) = 3.12, p = .008 for the H-hit condition; mean r = −.09, t(14) = −0.89, p = .38 for the Miss condition]. Further, the difference between the two conditions reached significance for both Region (A) and Region (B) [t(14) = 3.46, p = .004, joint p value < .001, corrected; t(14) = 2.96, p = .01, joint p value < .001, corrected].

With regard to the correlation between the left putamen and iPFC, the pattern of results was similar to that found for the right putamen [mean r = .23, t(14) = 2.38, p = .04 for the H-hit condition; mean r = .05, t(14) = 0.29, p = .78 for the Miss condition; t(14) = 1.97, p = .06 for the difference between conditions, joint probability = .0096, corrected].

Step 2: PFC

We next investigated the across-subject correlations between both the right– and the left putamen–iPFC transformed correlation coefficients and the d′ score of each participant, none of which reached statistical significance.

Whole-brain Connectivity Analysis

Finally, in order to examine whether the hippocampus–putamen and iPFC–putamen correlations were selective to these regions, we conducted a whole-brain voxel-level beta-series connectivity analysis (Rissman et al., 2004) once using the hippocampus as a seed and once using the right putamen (A) as a seed. Each of these analyses was conducted separately for the H-hit condition and for the Miss condition. Because of the exploratory nature of these analyses, we used a conservative threshold (correcting for the family-wise error rate). Using the hippocampus as a seed, this analysis found functional correlations only with clusters within the hippocampus itself both for the H-hit condition and for the Miss conditions. The same pattern was found when using the right putamen as a seed (namely, only correlations with the right putamen itself were found).

Data from the retrieval phase of the experiment did not reveal any finding of importance with regard to the encoding findings reported here, and thus, will not be discussed further.

DISCUSSION

Our findings provide novel evidence for cooperative contributions of the striatum and the hippocampus to episodic encoding. First, using an episodic encoding task, we found that activation in the striatum (bilateral putamen) predicted subsequent memory. Moreover, we found a positive intrasubject correlation between encoding activation in the putamen and in the hippocampus, specifically for subsequently remembered words, with no such correlation for subsequently forgotten items. Thus, when supporting operations beneficial for successful episodic encoding (i.e., during encoding of subsequently remembered items), the striatum operated in concert with the hippocampus.

The cooperative interaction between the striatum and the hippocampus was further supported by the positive intersubject correlation between memory performance (indicated by d′) and the strength of the correlation between a right striatal region and the left hippocampus (indicated by the transformed correlation coefficients). That is, the stronger the interaction between the left hippocampus and this striatal region, the more items were subsequently remembered.

Independent vs. Interactive Memory Systems

To the best of our knowledge, the current findings are the first report of a positive interaction between the striatum and the hippocampus during episodic memory formation. These findings extend the prevailing view according to which each of these regions contributes separately to distinct aspects of learning and memory. Indeed, the traditional emphasis on the mnemonic functional division between the hippocampus and the striatum may not fully account for the complexity of human memory. Instead, our data suggest that the hippocampus and the striatum are each involved in additional mnemonic functions, and interact in multiple ways with each other, depending on the cognitive circumstances. In particular, although previous studies have emphasized the role of the striatum in incremental stimulus–response learning, but not in episodic memory, our data indicate that the striatum may contribute importantly to episodic memory. In doing so, the striatum interacts with the hippocampus, suggesting that at least in some circumstances both structures may be involved in mutual aspects of learning, and may operate in concert to support such learning.

Our finding of a cooperative interaction between the dorsal striatum and the hippocampus should be considered vis-à-vis the widely reported competitive interaction between these structures. Importantly, the studies reporting a competitive interaction between the two structures used multitrial incremental learning tasks (Seger & Cincotta, 2006; Moody et al., 2004; Poldrack & Gabrieli, 2001; Poldrack et al., 1999, 2001). In contrast, the current study involved a nonincremental, episodic learning task in which each trial was a unique learning event, independent of the previous trials. In this episodic task, the two structures interacted cooperatively to support learning. The present findings thus suggest that the nature of the interaction between the hippocampus and the striatum may differ depending on the task demands and the kind of learning taking place. Future studies are necessary to gain further leverage on understanding how different learning circumstances modulate the contribution of the striatum and the hippocampus and the interaction between them.

Episodic Memory, Working Memory, and The Striatum

The present findings are consistent with numerous previous reports of subsequent memory effects in the striatum (Axmacher et al., 2008; Ritchey et al., 2008; Dennis, Kim, & Cabeza, 2007; Qin et al., 2007; Kensinger & Schacter, 2006; Nichols et al., 2006; Staresina & Davachi, 2006; Prince et al., 2005; Chua et al., 2004; Schon et al., 2004; Sperling et al., 2003). These studies have all reported higher activity in the striatum during episodic encoding of items subsequently remembered compared to items subsequently forgotten. Although these studies also report a subsequent memory effect in the hippocampus, the interaction between the hippocampus and the striatum was not reported in any of these studies. Further analyses of data from these studies could potentially provide valuable insights regarding the interaction between the striatum and the hippocampus at encoding across multiple paradigms.

Indeed, an important question for future research is under what circumstances the striatum contributes to episodic memory. The above subsequent memory studies have found striatal contributions to episodic memory using a variety of different paradigms, which do not appear to share a clear commonality as far as the cognitive task involved. However, a different line of research concerning the cognitive role of the striatum highlights the importance of this region in complex working memory processes (Frank, Cohen, & Sanfey, 2009; Grahn, Parkinson, & Owen, 2008; McNab & Klingberg, 2008; Hazy et al., 2007; O'Reilly & Frank, 2006), with the implication that it may contribute to episodic memory when such processes are engaged.

For example, McNab and Klingberg (2008) found that the striatum is engaged in preparing to ignore or filter-out irrelevant information (what is called filtering set) in a task whose structure is similar to the encoding task used here. In their paradigm, subjects viewed visuospatial arrays containing red and yellow squares, and had to perform a task that on some trials required ignoring the yellow squares, and on the other trials these yellow squares were part of the target stimulus. The authors reported higher activity in the striatum during the distractor condition (“ignore yellow squares”), compared to the nondistractor condition (“attend yellow squares”). Based on these findings, McNab and Klingberg argue that the striatum plays a critical role in working memory by operating to bias representations held in working memory, a notion that is consistent with work in computational modeling (Frank et al., 2009; Hazy et al., 2007; O'Reilly & Frank, 2006). Together, these models and data suggest that the striatum may play an important role in maintaining relevant information in mind, while filtering out or updating irrelevant information.

These same operations may play a key function in episodic memory success by determining which information enters mnemonic processing and storage. Indeed, these operations are essential components of the task used in the current study: Subjects received one input word which they had to break down, discarding the irrelevant part and, at the same time, keeping in mind the part to be used in the next step. This operation was repeated when the second word was presented, and finally, the two nondiscarded elements were integrated to form a word (or a nonword), memory for which would later be tested. Thus, the findings reported here are consistent with considerable empirical and theoretical data indicating a role for the striatum in maintenance of relevant stimuli and filtering of irrelevant stimuli in working memory. The present findings extend these results and suggest that these working memory operations may also be critical in determining the long-term memory for items that were held in mind.

Importantly, the abovementioned models regarding the involvement of the striatum in complex working memory operations highlight the interaction between PFC and the striatum during such operations. Indeed, in the current study too, we found a positive correlation between the striatum and iPFC, which was selective for successful encoding. In contrast, the correlation between the hippocampus and iPFC was not selective for successful encoding. Taken together, these data raise the intriguing possibility that, at least in the current task, the contribution of iPFC to episodic encoding is mediated by the striatum (and is not via a direct communication with the hippocampus). However, a better understanding of the dynamics between iPFC and other brain regions at encoding is a subject for future research.

Characterization of the Interplay between the Hippocampus and the Striatum

We now turn to discuss the nature of the cooperation between the striatum and the hippocampus in the current study. We first note that because fMRI can provide correlative information about activated brain structures, a strong definition of cooperation could only be demonstrated using other invasive methods which allow disrupting the circuit through which these brain structures interact. Still, the idea that, in our study, the two structures cooperate is supported by the fact that the positive correlation between the two structures appears to bear functional significance: Not only do these two structures interact only and more strongly during “Hit” trials, but the coherence between them predicts behavioral memory performance (the d′ scores of participants).

The current experimental design and data do not permit characterization of a specific interplay between the two structures in the encoding process. In other words, our findings do not necessarily imply a direct interaction between the striatum and the hippocampus. Still, anatomical studies indicate that a direct interaction between the striatum and the hippocampus is possible. For example, tract-tracing methods have provided evidence for a direct connection from entorhinal cortex—the main input structure of the hippocampus (Finch, 1996)—to the dorsal striatum (Sørensen & Witter, 1983). Electrophysiological studies further reveal that stimulation of both entorhinal cortex and the hippocampus evokes responses in the dorsal striatum (Finch, 1996; Finch, Gigg, Tan, & Kosoyan, 1995). Data supporting projections in the opposite direction—namely, from the dorsal striatum to the hippocampus, are available primarily in electrophysiological studies showing that striatal stimulation reduces hippocampal spike frequency and/or amplitude by inducing theta-field activity in the hippocampus (Hallworth & Bland, 2004; Sabatino et al., 1989; La Grutta & Sabatino, 1988; Sabatino, Ferraro, Liberti, Vella, & La Grutta, 1985).

Striatum–Hippocampus–Behavior Interactive Pattern across Striatal Regions

Finally, although our primary prediction concerning the interaction between the hippocampus and the striatum was confirmed overall, the three striatal regions did not show an identical pattern of correlations with the hippocampus in the additional analyses we conducted—specifically, the comparison of correlation coefficients of remembered and forgotten words. That is, although all three regions showed effects in the same direction, some differences did not reach statistical significance in all regions. Whether this may reflect incidental signal strength differences or a somewhat differential contribution of each striatal region to the episodic encoding process remains a question for future research. Nevertheless, the fact that all three striatal regions showed a positive correlation with the hippocampus specifically during successful encoding raises our confidence in the robustness of this finding. We thus provide novel evidence for a cooperative interaction between the striatum and the hippocampus during episodic encoding, thereby expanding the array of possible interactions between these two core memory structures.

Acknowledgments

We thank O. Hegedish for helpful comments, as well as M. Zuckerman and R. Mandel for assisting in data collection and analysis. This work was supported by The Israel Science Foundation (in part; Grant no. 1418/06 to A. M.); the European Community under the Marie Curie International Reintegration Grant (in part; MIRG-CT-2007-046457 to A. M.); The National Institute for Psychobiology in Israel—Founded by The Charles E. Smith Family (in part, to A. M.); T. S. was supported in part by The Levy Edersheim Gitter Institute for Neuroimaging.

Reprint requests should be sent to Anat Maril, Department of Psychology, The Hebrew University in Jerusalem, Mt. Scopus, Jerusalem, 91905, Israel, or via e-mail: marila@pluto.huji.ac.il.

REFERENCES

REFERENCES
Adcock
,
R. A.
,
Thangavel
,
A.
,
Whitfield-Gabrieli
,
S.
,
Knutson
,
B.
, &
Gabrieli
,
J. D. E.
(
2006
).
Reward-motivated learning: Mesolimbic activation precedes memory formation.
Neuron
,
50
,
507
517
.
Albouy
,
G.
,
Sterpenich
,
V.
,
Balteau
,
E.
,
Vandewalle
,
G.
,
Desseilles
,
M.
,
Dang-Vu
,
T.
,
et al
(
2008
).
Both the hippocampus and striatum are involved in consolidation of motor sequence memory.
Neuron
,
58
,
261
272
.
Alexander
,
G. E.
,
DeLong
,
M. R.
, &
Strick
,
P. L.
(
2003
).
Parallel organization of functionally segregated circuits linking basal ganglia and cortex.
Annual Review of Neuroscience
,
9
,
357
.
Altgassen
,
M.
,
Phillips
,
L.
,
Kopp
,
U.
, &
Kliegel
,
M.
(
2007
).
Role of working memory components in planning performance of individuals with Parkinson's disease.
Neuropsychologia
,
45
,
2393
2397
.
Axmacher
,
N.
,
Schmitz
,
D. P.
,
Weinreich
,
I.
,
Elger
,
C. E.
, &
Fell
,
J.
(
2008
).
Interaction of working memory and long-term memory in the medial temporal lobe.
Cerebral Cortex
,
18
,
2868
2878
.
Beste
,
C.
,
Saft
,
C.
,
Gunturkun
,
O.
, &
Falkenstein
,
M.
(
2008
).
Increased cognitive functioning in symptomatic Huntington's disease as revealed by behavioral and event-related potential indices of auditory sensory memory and attention.
Journal of Cognitive Neuroscience
,
28
,
11695
11702
.
Chua
,
E. F.
,
Rand-Giovannetti
,
E.
,
Schacter
,
D. L.
,
Albert
,
M. S.
, &
Sperling
,
R. A.
(
2004
).
Dissociating confidence and accuracy: Functional magnetic resonance imaging shows origins of the subjective memory experience.
Journal of Cognitive Neuroscience
,
16
,
1131
1142
.
Cincotta
,
C. M.
, &
Seger
,
C. A.
(
2007
).
Dissociation between striatal regions while learning to categorize via feedback and via observation.
Journal of Cognitive Neuroscience
,
19
,
249
265
.
Cocosco
,
C. A.
,
Kollokian
,
V.
,
Kwan
,
R. K.-S.
, &
Evans
,
A. C.
(
1997
).
BrainWeb: Online interface to a 3DMRI simulated.
Neuroimage
,
5
,
S425
.
Cohen
,
N. J.
, &
Squire
,
L. R.
(
1980
).
Preserved learning and retention of pattern-analyzing skill in amnesia: Dissociation of knowing how and knowing that.
Science
,
210
,
207
210
.
Dagher
,
A.
,
Owen
,
A. M.
,
Boecker
,
H.
, &
Brooks
,
D. J.
(
2001
).
The role of the striatum and hippocampus in planning: A PET activation study in Parkinson's disease.
Brain
,
124
,
1020
1032
.
Dale
,
A. M.
, &
Buckner
,
R. L.
(
1997
).
Selective averaging of rapidly presented individual trials using fMRI.
Human Brain Mapping
,
5
,
329
340
.
DeCoteau
,
W. E.
,
Thorn
,
C.
,
Gibson
,
D. J.
,
Courtemanche
,
R.
,
Mitra
,
P.
,
Kubota
,
Y.
,
et al
(
2007
).
Learning-related coordination of striatal and hippocampal theta rhythms during acquisition of a procedural maze task.
Proceedings of the National Academy of Sciences, U.S.A.
,
104
,
5644
.
Degonda
,
N.
,
Mondadori
,
C. R. A.
,
Bosshardt
,
S.
,
Schmidt
,
C. F.
,
Boesiger
,
P.
,
Nitsch
,
R. M.
,
et al
(
2005
).
Implicit associative learning engages the hippocampus and interacts with explicit associative learning.
Neuron
,
46
,
505
520
.
Dennis
,
N. A.
,
Daselaar
,
S.
, &
Cabeza
,
R.
(
2007
).
Effects of aging on transient and sustained successful memory encoding activity.
Neurobiology of Aging
,
28
,
1749
1758
.
Dennis
,
N. A.
,
Kim
,
H.
, &
Cabeza
,
R.
(
2007
).
Effects of aging on true and false memory formation: An fMRI study.
Neuropsychologia
,
45
,
3157
3166
.
Doeller
,
C. F.
,
King
,
J. A.
, &
Burgess
,
N.
(
2008
).
Parallel striatal and hippocampal systems for landmarks and boundaries in spatial memory.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
5915
5920
.
Eichenbaum
,
H.
(
2004
).
Hippocampus: Cognitive processes and neural representations that underlie declarative memory.
Neuron
,
44
,
109
120
.
Faure
,
A.
,
Haberland
,
U.
,
Conde
,
F.
, &
Massioui
,
N. E.
(
2005
).
Lesion to the nigrostriatal dopamine system disrupts stimulus–response habit formation.
Journal of Neuroscience
,
25
,
2771
2780
.
Finch
,
D. M.
(
1996
).
Neurophysiology of converging synaptic inputs from the rat prefrontal cortex, amygdala, midline thalamus, and hippocampal formation onto single neurons of the caudate/putamen and nucleus accumbens.
Hippocampus
,
6
,
495
512
.
Finch
,
D. M.
,
Gigg
,
J.
,
Tan
,
A. M.
, &
Kosoyan
,
O. P.
(
1995
).
Neurophysiology and neuropharmacology of projections from entorhinal cortex to striatum in the rat.
Brain Research
,
670
,
233
247
.
Fleiss
,
J. L.
(
1986
).
The design and analysis of clinical experiments.
New York
:
Wiley
.
Foerde
,
K.
,
Knowlton
,
B. J.
, &
Poldrack
,
R. A.
(
2006
).
Modulation of competing memory systems by distraction.
Proceedings of the National Academy of Sciences, U.S.A.
,
103
,
11778
11783
.
Frank
,
M. J.
,
Cohen
,
M. X.
, &
Sanfey
,
A. G.
(
2009
).
Multiple systems in decision making: A neurocomputational perspective.
Current Directions in Psychological Science
,
18
,
73
77
.
Gabrieli
,
J. D. E.
(
1998
).
Cognitive neuroscience of human memory.
Annual Review of Psychology
,
49
,
87
115
.
Goto
,
Y.
, &
Grace
,
A. A.
(
2008
).
Limbic and cortical information processing in the nucleus accumbens.
Trends in Neurosciences
,
31
,
552
.
Goto
,
Y.
, &
O'Donnell
,
P.
(
2001
).
Synchronous activity in the hippocampus and nucleus accumbens in vivo.
Journal of Neuroscience
,
21
,
RC131
.
Grahn
,
J. A.
,
Parkinson
,
J. A.
, &
Owen
,
A. M.
(
2008
).
The cognitive functions of the caudate nucleus.
Progress in Neurobiology
,
86
,
141
.
Haber
,
S. N.
,
Fudge
,
J. L.
, &
McFarland
,
N. R.
(
2000
).
Striatonigrostriatal pathways in primates form an ascending spiral from the shell to the dorsolateral striatum.
Journal of Neuroscience
,
20
,
2369
2382
.
Haber
,
S. N.
,
Kim
,
K.-S.
,
Mailly
,
P.
, &
Calzavara
,
R.
(
2006
).
Reward-related cortical inputs define a large striatal region in primates that interface with associative cortical connections, providing a substrate for incentive-based learning.
Journal of Neuroscience
,
26
,
8368
8376
.
Hallworth
,
N. E.
, &
Bland
,
B. H.
(
2004
).
Basal ganglia–hippocampal interactions support the role of the hippocampal formation in sensorimotor integration.
Experimental Neurology
,
188
,
430
443
.
Hartley
,
T.
, &
Burgess
,
N.
(
2005
).
Complementary memory systems: Competition, cooperation and compensation.
Trends in Neurosciences
,
28
,
169
.
Hazy
,
T. E.
,
Frank
,
M. J.
, &
O'Reilly
,
R. C.
(
2007
).
Towards an executive without a homunculus: Computational models of the prefrontal cortex/basal ganglia system.
Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences
,
362
,
1601
1613
.
Kensinger
,
E. A.
, &
Schacter
,
D. L.
(
2006
).
Amygdala activity is associated with the successful encoding of item, but not source, information for positive and negative stimuli.
Journal of Neuroscience
,
26
,
2564
2570
.
Knowlton
,
B. J.
,
Mangels
,
J. A.
, &
Squire
,
L. R.
(
1996
).
A neostriatal habit learning system in humans.
Science
,
273
,
1399
1402
.
La Grutta
,
V.
, &
Sabatino
,
M.
(
1988
).
Focal hippocampal epilepsy: Effect of caudate stimulation.
Experimental Neurology
,
99
,
38
49
.
Lee
,
A. S.
,
Duman
,
R. S.
, &
Pittenger
,
C.
(
2008
).
A double dissociation revealing bidirectional competition between striatum and hippocampus during learning.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
17163
17168
.
McDonald
,
R. J.
, &
White
,
N. M.
(
1993
).
A triple dissociation of memory systems: Hippocampus, amygdala and dorsal striatum.
Behavioural Neuroscience
,
107
,
3
22
.
McNab
,
F.
, &
Klingberg
,
T.
(
2008
).
Prefrontal cortex and basal ganglia control access to working memory.
Nature Neuroscience
,
11
,
103
107
.
Moody
,
T. D.
,
Bookheimer
,
S. Y.
,
Vanek
,
Z.
, &
Knowlton
,
B. J.
(
2004
).
An implicit learning task activates medial temporal lobe in patients with Parkinson's disease.
Behavioural Neuroscience
,
118
,
438
442
.
Nichols
,
E. A.
,
Kao
,
Y.-C.
,
Verfaellie
,
M.
, &
Gabrieli
,
J. D. E.
(
2006
).
Working memory and long-term memory for faces: Evidence from fMRI and global amnesia for involvement of the medial temporal lobes.
Hippocampus
,
16
,
604
616
.
O'Reilly
,
R. C.
, &
Frank
,
M. J.
(
2006
).
Making working memory work: A computational model of learning in the prefrontal cortex and basal ganglia.
Neural Computation
,
18
,
283
328
.
Packard
,
M. G.
,
Hirsh
,
R.
, &
White
,
N. M.
(
1989
).
Differential effects of fornix and caudate nucleus lesions on two radial maze tasks: Evidence for multiple memory systems.
Journal of Neuroscience
,
9
,
1465
1472
.
Packard
,
M. G.
, &
Teather
,
L. A.
(
1998
).
Amygdala modulation of multiple memory systems: Hippocampus and caudate–putamen.
Neurobiology of Learning and Memory
,
69
,
163
.
Paller
,
K. A.
, &
Wagner
,
A. D.
(
2002
).
Observing the transformation of experience into memory.
Trends in Cognitive Sciences
,
6
,
93
102
.
Poldrack
,
R. A.
,
Clark
,
J.
,
Pare-Blagoev
,
E. J.
,
Shohamy
,
D.
,
Creso Moyano
,
J.
,
Myers
,
C.
,
et al
(
2001
).
Interactive memory systems in the human brain.
Nature
,
414
,
546
550
.
Poldrack
,
R. A.
, &
Gabrieli
,
J. D. E.
(
2001
).
Characterizing the neural mechanisms of skill learning and repetition priming: Evidence from mirror reading.
Brain
,
124
,
67
82
.
Poldrack
,
R. A.
,
Prabhakaran
,
V.
,
Seger
,
C. A.
, &
Gabrieli
,
J. D. E.
(
1999
).
Striatal activation during acquisition of a cognitive skill.
Neuropsychology
,
13
,
564
574
.
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
.
Qin
,
S.
,
Piekema
,
C.
,
Petersson
,
K. M.
,
Han
,
B.
,
Luo
,
J.
, &
Fernández
,
G.
(
2007
).
Probing the transformation of discontinuous associations into episodic memory: An event-related fMRI study.
Neuroimage
,
38
,
212
222
.
Rissman
,
J.
,
Gazzaley
,
A.
, &
D'Esposito
,
M.
(
2004
).
Measuring functional connectivity during distinct stages of a cognitive task.
Neuroimage
,
23
,
752
763
.
Ritchey
,
M.
,
Dolcos
,
F.
, &
Cabeza
,
R.
(
2008
).
Role of amygdala connectivity in the persistence of emotional memories over time: An event-related fMRI investigation.
Cerebral Cortex
,
18
,
2494
2504
.
Rose
,
M.
,
Haider
,
H.
,
Weiller
,
C.
, &
Büchel
,
C.
(
2002
).
The role of medial temporal lobe structures in implicit learning: An event-related fMRI study.
Neuron
,
36
,
1221
1231
.
Sabatino
,
M.
,
Ferraro
,
G.
,
Liberti
,
G.
,
Vella
,
N.
, &
La Grutta
,
V.
(
1985
).
Striatal and septal influence on hippocampal theta and spikes in the cat.
Neuroscience Letters
,
61
,
55
59
.
Sabatino
,
M.
,
Gravante
,
G.
,
Ferraro
,
G.
,
Vella
,
N.
,
Grutta
,
G. L.
, &
Grutta
,
V. L.
(
1989
).
Striatonigral suppression of focal hippocampal epilepsy.
Neuroscience Letters
,
98
,
285
290
.
Schendan
,
H. E.
,
Searl
,
M. M.
,
Melrose
,
R. J.
, &
Stern
,
C. E.
(
2003
).
An fMRI study of the role of the medial temporal lobe in implicit and explicit sequence learning.
Neuron
,
37
,
1013
1025
.
Schon
,
K.
,
Hasselmo
,
M. E.
,
LoPresti
,
M. L.
,
Tricarico
,
M. D.
, &
Stern
,
C. E.
(
2004
).
Persistence of parahippocampal representation in the absence of stimulus input enhances long-term encoding: A functional magnetic resonance imaging study of subsequent memory after a delayed match-to-sample task.
Journal of Neuroscience
,
24
,
11088
11097
.
Schott
,
B. H.
,
Seidenbecher
,
C. I.
,
Fenker
,
D. B.
,
Lauer
,
C. J.
,
Bunzeck
,
N.
,
Bernstein
,
H.-G.
,
et al
(
2006
).
The dopaminergic midbrain participates in human episodic memory formation: Evidence from genetic imaging.
Journal of Neuroscience
,
26
,
1407
1417
.
Seger
,
C. A.
, &
Cincotta
,
C. M.
(
2005
).
The roles of the caudate nucleus in human classification learning.
Journal of Neuroscience
,
25
,
2941
2951
.
Seger
,
C. A.
, &
Cincotta
,
C. M.
(
2006
).
Dynamics of frontal, striatal, and hippocampal systems during rule learning.
Cerebral Cortex
,
16
,
1546
1555
.
Shohamy
,
D.
,
Myers
,
C. E.
,
Grossman
,
S.
,
Sage
,
J.
,
Gluck
,
M. A.
, &
Poldrack
,
R. A.
(
2004
).
Cortico-striatal contributions to feedback-based learning: Converging data from neuroimaging and neuropsychology.
Brain
,
127
,
851
859
.
Shohamy
,
D.
,
Myers
,
C. E.
,
Kalanithi
,
J.
, &
Gluck
,
M. A.
(
2008
).
Basal ganglia and dopamine contributions to probabilistic category learning.
Neuroscience & Biobehavioral Reviews
,
32
,
219
.
Shohamy
,
D.
,
Myers
,
C. E.
,
Onlaor
,
S.
, &
Gluck
,
M. A.
(
2004
).
Role of the basal ganglia in category learning: How do patients with Parkinson's disease learn?
Behavioral Neuroscience
,
118
,
676
.
Shohamy
,
D.
, &
Wagner
,
A. D.
(
2008
).
Integrating memories in the human brain: Hippocampal–midbrain encoding of overlapping events.
Neuron
,
60
,
378
389
.
Sørensen
,
K.
, &
Witter
,
M.
(
1983
).
Entorhinal efferents reach the caudato-putamen.
Neuroscience Letters
,
35
,
259
264
.
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
.
Squire
,
L. R.
(
2004
).
Memory systems of the brain: A brief history and current perspective.
Neurobiology of Learning and Memory
,
82
,
171
177
.
Squire
,
L. R.
, &
Zola
,
S.
(
1996
).
Structure and function of declarative and nondeclarative memory.
Proceedings of the National Academy of Sciences, U.S.A.
,
93
,
13515
13522
.
Staresina
,
B. P.
, &
Davachi
,
L.
(
2006
).
Differential encoding mechanisms for subsequent associative recognition and free recall.
Journal of Neuroscience
,
26
,
9162
9172
.
Tricomi
,
E.
, &
Fiez
,
J. A.
(
2008
).
Feedback signals in the caudate reflect goal achievement on a declarative memory task.
Neuroimage
,
41
,
1154
1167
.
Tricomi
,
E. M.
,
Delgado
,
M. R.
, &
Fiez
,
J. A.
(
2004
).
Modulation of caudate activity by action contingency.
Neuron
,
41
,
281
292
.
Tzourio-Mazoyer
,
N.
,
Landeau
,
B.
,
Papathanassiou
,
D.
,
Crivello
,
F.
,
Etard
,
O.
,
Delcroix
,
N.
,
et al
(
2002
).
Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain.
Neuroimage
,
15
,
273
289
.
Vakil
,
E.
,
Blachstein
,
H.
, &
Soroker
,
N.
(
2004
).
Differential effect of right and left basal ganglionic infarctions on procedural learning.
Cognitive and Behavioral Neurology
,
17
,
62
73
.
Voermans
,
N. C.
,
Petersson
,
K. M.
,
Daudey
,
L.
,
Weber
,
B.
,
van Spaendonck
,
K. P.
,
Kremer
,
H. P. H.
,
et al
(
2004
).
Interaction between the human hippocampus and the caudate nucleus during route recognition.
Neuron
,
43
,
427
.
Williams
,
Z. M.
, &
Eskandar
,
E. N.
(
2006
).
Selective enhancement of associative learning by microstimulation of the anterior caudate.
Nature Neuroscience
,
9
,
562
568
.
Wittmann
,
B. C.
,
Schott
,
B. H.
,
Guderian
,
S.
,
Frey
,
J. U.
,
Heinze
,
H.-J.
, &
Düzel
,
E.
(
2005
).
Reward-related fMRI activation of dopaminergic midbrain is associated with enhanced hippocampus-dependent long-term memory formation.
Neuron
,
45
,
459
467
.
Yin
,
H. H.
, &
Knowlton
,
B. J.
(
2006
).
The role of the basal ganglia in habit formation.
Nature Reviews Neuroscience
,
7
,
464
.
Yin
,
H. H.
,
Knowlton
,
B. J.
, &
Balleine
,
B. W.
(
2004
).
Lesions of dorsolateral striatum preserve outcome expectancy but disrupt habit formation in instrumental learning.
European Journal of Neuroscience
,
19
,
181
189
.