Being able to remember and retain the most current information in a dynamically changing world requires the capacity to update one’s memory in a goal-directed manner. Updating occurs when some information is downgraded as outdated or irrelevant, and newer information is promoted as its replacement. Examples range from something as simple as trying to remember one’s new home address and phone number, to replacing maladaptive memories with adaptive ones in a therapeutic setting. Successful updating of memory involves strengthening the more current memory trace, weakening the older information, and/or differentiating old and new memories, in order to ensure that the old memory is less interfering.

Ample behavioral evidence suggests that although memory updating can be fostered by repeatedly studying the new replacement information, memory is more successfully updated by the act of retrieving the new knowledge via self-tests, a process called retrieval practice (Roediger and Butler, 2011). Compared to simple restudy of the same material again, retrieval of learnt information leads not only to better retention of relevant memory when no obvious interference is involved (Karpicke and Roediger, 2008; Pyc and Rawson, 2010), but also to better inhibition of competing, outdated memories (Anderson et al., 1994), reduced proactive interference (Szpunar et al., 2008), enhanced memory integration (Hupbach et al., 2007), and greater susceptibility to subsequent modification (Chan et al., 2009). These findings indicate that retrieval practice modifies the state of memory according to mnemonic goals. Despite this ubiquitous behavioral effect, the neural basis for memory updating and the mechanism that underlies the retrieval-practice benefit remain unclear.

A large body of work on memory reconsolidation (Dudai, 2006) suggests that consolidated memories, when retrieved and reactivated, enter into a transient, labile state, rendering them vulnerable to modification (Lee et al., 2017). Electrical shock (Kroes et al., 2014) and pharmacological treatments (Kindt et al., 2009; Nader et al., 2000) that block protein synthesis can cause long-lasting impairment of existing memories when they are reactivated. Reactivated memories can also be disrupted by behavioral methods, including retrieval-extinction (Schiller et al., 2010; Xue et al., 2012), counterconditioning (Goltseker et al., 2017), and interference approaches (James et al., 2015), which have been found to reduce the expression of fear or drug memories and amygdala responses associated with fear (Agren et al., 2012). Using representational similarity analysis, recent studies further show that the reactivated memories could be selectively strengthened (Jonker et al., 2018; Lu et al., 2015; Xue et al., 2010), weakened (Wimber et al., 2015), integrated (Schlichting et al., 2015), and/or differentiated (Hulbert and Norman, 2015), depending on the mnemonic goals and the characteristics of the reactivated memory (Tambini and Davachi, 2019).

Emerging studies suggest that the medial prefrontal cortex (MPFC) probably plays an important role in the rapid formation of cortical memories, especially during retrieval practice. Specifically, it has been hypothesized that retrieval practice will reactivate related memory traces, and that the MPFC is able to develop integrated neocortical representations of these memory traces rapidly, in a way that resembles the characteristics of rapid system consolidation (Antony et al., 2017). Consistently, the MPFC has been found to be involved in the integration and updating of reactivated memory traces (Gilboa and Marlatte, 2017; Preston and Eichenbaum, 2013). For example, the MPFC is critically involved in encoding novel but related information into existing knowledge (Sommer, 2016), in representing overlapping memories (Tompary and Davachi, 2017), and in inferring relationships between distinct events that share common features (Zeithamova et al., 2012). In these studies, the reactivation of related memories has been found to be important for the MPFC-mediated processes. Nevertheless, it is unclear how the MPFC would be involved when competing memories were reactivated.

Several studies suggest that the lateral prefrontal cortex (LPFC) may play a role in regulating reactivated memories to support memory changes. First, the LPFC could bias the competition and reduce the intrusion of unwanted memory in later memory retrieval (Kuhl et al., 2012). Intentional suppression of memory retrieval reduces hippocampal activity via control mechanisms mediated by the LPFC (Hulbert et al., 2016). These studies suggest an important role of LPFC in control of memory. Second, studies examining retrieval-induced forgetting (Anderson et al., 1994; Norman et al., 2007) have found that retrieval practice could suppress (Wimber et al., 2015) or differentiate (Hulbert and Norman, 2015) the neural representation of competitive memories in the sensory cortices or hippocampus, and that these processes are mediated by the LPFC. Third, reactivation during wakefulness has been found to destabilize memories and has been associated with LPFC activation (Diekelmann et al., 2011), supporting its role in modifying reactivated memory representations.

The present study compared the neural reactivation during retrieval practice with that during restudy, and examined how the reactivated memory representations interact with the lateral PFC to achieve subsequent memory updating. In the Restudy condition, we simply asked subjects to study new replacement memories repeatedly. In the retrieval practice (RetPrac) condition, we asked subjects to retrieve the new replacement memory repeatedly. Feedback was provided as previous studies have shown that it could boost the behavioral performance (Butler and Roediger, 2008; Pashler et al., 2005). We hypothesized that during updating, retrieval practice would elicit greater reactivation of the outdated competitor (i.e., B) than would restudy exposures, reflecting the greater tendency of retrieval to trigger competition that requires correction. Despite these added difficulties, we predicted that during the final memory test, retrieval practice would lead to a better long-term effect: stronger reactivation of the replacement memory (i.e., the target, C) as well as weaker reactivation of the outdated competitor.

We used a multi-day design to examine both the short-term and long-term effects of memory updating under these two conditions. On Day 1, we extensively trained 19 subjects on associations between words (A) and pictures (B). On Day 2, we introduced new replacement A-C associations (B and C were of different visual categories) and asked subjects to replace the old associations with the new ones before entering the scanner. In the scanner, subjects encountered the new A-C associations three times, either via retrieval practice (RetPrac) or extra study exposures (Restudy). For RetPrac, the cue word was paired with a black rectangle, and subjects were asked to recall the picture associated with the word. When the black rectangle turned red after 2 s, they were asked to judge the category of picture C by pressing one of the four buttons corresponding to Face, Object, Scene, and Don’t know. Then the correct picture C was shown on the screen for 1 s as a feedback (Figure 1A, upper panel). For Restudy, the procedure was identical except the cue word was paired with the correct picture C at the beginning of the trial and subjects were asked to memorize the association and then make the category judgment. After these updating trials on Day 2, we waited for 24 hr to probe how successful retrieval practice and restudy were at accomplishing long-lasting memory updating. On Day 3, we scanned subjects again using a cued recall test. Subjects were asked to recall the visual details of the picture associated with the cue word presented on the screen, and then to respond by pressing one of the four buttons corresponding to Face, Object, Scene, and Don’t know. They were then asked to perform a perceptual orientation judgment task for 8 s (Figure 1A, lower panel) before the next trial started.

Figure 1

Download asset Open asset

Figure 1—source data 1

Memory performance during Updating (Exp1).

https://cdn.elifesciences.org/articles/57023/elife-57023-fig1-data1-v2.csv

Download elife-57023-fig1-data1-v2.csv

Figure 1—source data 2

Memory performance during final test (Exp1).

https://cdn.elifesciences.org/articles/57023/elife-57023-fig1-data2-v2.csv

Download elife-57023-fig1-data2-v2.csv

Retrieval practice facilitated memory updating via differentiation

The above analysis suggests that retrieval practice could enhance the accessibility of A-C memories and reduce that of A-B memories in the A-C memory test. Several mechanisms could account for this result: (1) strengthening of A-C memory (Karpicke and Roediger, 2008; Roediger and Butler, 2011); (2) weakening of A-B memory (Anderson et al., 1994; Levy and Anderson, 2002); (3) differentiation of A-C and A-B memories (Hulbert and Norman, 2015; Storm et al., 2008). Previous studies suggest that reactivating the old memory and detecting and remembering the difference would help to resolve the proactive interference (van den Honert et al., 2016; Wahlheim and Jacoby, 2013). Thus, our results could be contributed by (4) integration and differentiation, which is a more specific version of the differentiation mechanism. To differentiate these hypotheses, we performed two additional behavioral experiments (Exp. 2 and 3) to examine how retrieval-practice affected A-B memory performance.

The inhibition hypothesis would predict greater C memory and weaker B memory intrusion in the A-C memory test, but worse B memory and greater C intrusion in the A-B memory test under the RetPrac condition than under the Restudy condition. By contrast, both the differentiation hypothesis and the integration and differentiation hypothesis would predict weaker B memory intrusion in the A-C memory test, and weaker or comparable C memory intrusion (given C memory was strengthen) in the A-B memory test.

In the new experiment (n = 46, Exp. 2), the procedure was nearly identical to the main experiment except that during the Day 3 test, we asked subjects to do both A-B and A-C memory tests, without the perceptual orientation judgment task between trials. To examine the effect of test order, half of the subjects did the A-B memory test first and the other half did the A-C memory test first. For both A-C and A-B memory tests and each response type (target, competitor, 'other'), test order (A-C first, A-B first) by update method (RetPrac, Restudy) two-way ANOVA revealed no significant main effect of test order (p-values>0.43, except for a trend towards a significant effect in 'other' responses during the A-B test, p=0.07, FDR corrected), nor of the test order by update method interaction (p-values >0.54, FDR corrected) (Supplementary file 1). Given our focus on the effect of updating method and the lack of updating method by test order interaction, we thus combined the data from both test order groups in the following analyses to increase the statistical power. Replicating the main effect, we found greater A-C memory under the RetPrac condition than under the Restudy condition during the A-C memory test, as indicated by the significantly higher correct recall of targets (t₍₄₅₎ = 3.93, p<0.001, Cohen’s d = 0.57) and fewer competitor intrusions (t₍₄₅₎ = −2.75, p=0.008, Cohen’s d = 0.36) under the RetPrac than under the Restudy condition (Figure 2A). However, we found that A-B memory was very strong during A-B recall (55.9% correct) and showed no effect of updating method on either B memory (t₍₄₅₎ = 1.43, p=0.19) or C intrusion (t₍₄₅₎ = 0.80, p=0.43).

Figure 2 with 1 supplement see all

Download asset Open asset

Figure 2—source data 1

Memory performance during final test (Exp2).

https://cdn.elifesciences.org/articles/57023/elife-57023-fig2-data1-v2.csv

Download elife-57023-fig2-data1-v2.csv

Figure 2—source data 2

Memory performance during final test (Exp3).

https://cdn.elifesciences.org/articles/57023/elife-57023-fig2-data2-v2.csv

Download elife-57023-fig2-data2-v2.csv

This pattern was consistent with the differentiation hypothesis. That is, although C memory was strengthened by retrieval practice, no greater C intrusion was found in the A-B memory test. On the other hand, although B memory was comparable between the two conditions, there were fewer B intrusions in the A-C memory test. We noticed that the number of C memory intrusions was overall low, perhaps because of the weak A-C memory. We therefore did a further analysis to focus on strong A-C memory trials (correct trials in the A-C memory test), which would produce more intrusions. Consistently, we found that the correct trials in the A-C memory test (reflecting strong A-C memory) showed more C memory intrusions overall during the A-B test than did the incorrect trials (t₍₄₅₎ = 3.38, p=0.001, Cohen’s d = 0.71). Interestingly, the intrusion rate was numerically smaller in the RetPrac condition than in the Restudy condition (t₍₄₅₎ = −1.59, p=0.12), which is consistent with the differentiation hypothesis (Figure 2B).

To further test the differentiation hypothesis, we did a joint analysis to examine the subjects’ answers in both A-B and A-C memory tests given the same word cues. The differentiation hypothesis would predict more correct trials in both tests (i.e., C response in the A-C memory test and B response in the A-B memory test) under the RetPrac condition, i.e., that subjects maintained stronger and nonoverlapping representations of both A-B and A-C memories. Meanwhile, we would predict fewer trials in which subjects responded with old B memory in both tests (due to differentiation), and also fewer (due to differentiation) or comparable (due to strengthening of C and differentiation) trials in which subjects responded with new C memory in both memory tests. Our data supported all three predictions. We found that subjects made more correct responses in both tests under the RetPrac condition than under the Restudy condition (24.0% vs 19.7%, t₍₄₅₎ = 3.81, p<0.001, Cohen’s d = 0.48), but showed less B memory (21.0% vs 23.2%, t₍₄₅₎ = −2.61, p=0.012, Cohen’s d = 0.25), and comparable C memory (11.7% vs 11.5%, t₍₄₅₎ = 0.18, p=0.86) in both memory tests (Figure 2C).

To further test whether RetPrac was able to modify the details of memory representation, in a third experiment (n = 28, Exp.3), we asked the subjects to write down the name of the associated picture (or any associated details if they could not recall the exact name) for each cue word, instead of choosing one of the three categories by pressing a button. Only answers with correct picture name or specific details were considered as a correct item recall. Our findings again replicated the retrieval-practice effect on A-C memory, as indicated by the significantly higher correct item recall (t₍₂₇₎ = 8.06, p<0.001, Cohen’s d = 0.50) under the RetPrac condition than under the Restudy condition, but comparable performance on the A-B memory test (t₍₂₇₎ = 0.81, p=0.43) (Figure 2D). These results are consistent with the hypothesis that RetPrac helped to achieve better memory updating by differentiation and do not favor the idea that updating is accomplished by inhibition.

The effect of retrieval practice on target and competitor representations

To understand the neural basis of the retrieval-practice advantage, we used fMRI and MVPA to track the reactivation of the outdated and replacement memories during the two types of updating, and to link those patterns to memory performance on the final test. First, we trained neural classifiers to differentiate three categories of materials, i.e., faces, scenes, and objects, on the basis of independent functional localizer data. We then used them to examine the degree of memory reactivation during Day 2 updating and Day 3 final test (Kuhl et al., 2012; Figure 3—figure supplement 1A). We focused our analysis on the medial prefrontal cortex (MPFC), ventral temporal cortex (VTC), angular gyrus (AG), and hippocampus (HPC), which overlap with the core recollection network (Rugg and Vilberg, 2013) and have consistently shown a neural reinstatement effect during memory retrieval (Kuhl and Chun, 2014; Wimber et al., 2015; Xiao et al., 2017). Confirming the relevance of these regions to our task, we found that the classifier output could predict subjects’ categorical judgments during the final test with significantly above-chance accuracy in the MPFC, VTC, and AG (ranging from 39.6% to 42.6%, all p-values <0.001, all survived FDR correction), but not in the HPC (35.5%, p=0.35) (Figure 3—figure supplement 1B). The following analysis thus focused on the first three regions (Figure 3A).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Classifier evidence during the final test.

https://cdn.elifesciences.org/articles/57023/elife-57023-fig3-data1-v2.csv

Download elife-57023-fig3-data1-v2.csv

Next, we used these classifiers to examine how retrieval practice could shape the target and competitor representations in the brain during the final memory test. To further examine how the representations in these regions were differentially modulated by retrieval practice and behavioral performance, we separately examined the correct trials (i.e., when targets were chosen) and the incorrect trials (i.e., when competitors were chosen). Three-way repeated-measures ANOVA was conducted, with evidence (Target vs Competitor), outcome (Correct vs Incorrect) and updating method (RetPrac vs Restudy) as within-subject factors. Both the inhibition and differentiation hypothesis would predict stronger target representation and weaker competitor representation for RetPrac than for Restudy, whereas the integration and differentiation hypothesis would predict strong reactivation of both target and competitor representations for RetPrac, despite the superior behavioral performance.

In the MPFC, three-way ANOVA revealed a significant evidence-by-outcome interaction (F_(1,18) = 23.25, p=0.0001, survived FDR correction, Figure 3B), suggesting that the activation in the MPFC tracked behavioral performance. We also found a significant method-by-evidence interaction (F_(1,18) = 6.50, p=0.02, survived FDR correction). No three-way interaction or method-by-outcome interaction was found (p-values >0.08, Supplementary file 2a). We then did two separate two-way ANOVAs for target and competitor evidence reactivation. For target reactivation, there were significant main effects of updating method (RetPrac vs Restudy) (F_(1,18) = 7.01, p=0.02, Supplementary file 2b) and outcome (chosen targets vs chosen competitors) (F_(1,18) = 15.10, p=0.001, survived FDR correction), suggesting that correct responses were associated with stronger target evidence reactivation, and that retrieval practice was able to boost the target reactivation for both correct and incorrect trials. For competitor evidence, however, we found that there was stronger competitor evidence reactivation for incorrect trials than for correct trials under the Restudy condition (t₍₁₈₎ = 3.46, p=0.003, Cohen’s d = 0.86, survived FDR correction), whereas no such difference was found for the RetPrac condition (t₍₁₈₎ = 0.35, p=0.73), although the outcome-by-condition interaction did not reach significance (F_(1,18) = 2.53, p=0.13, Supplementary file 2b). The latter result indicated that even when correct responses were made, there was still strong and comparable competitor reactivation under the RetPrac condition, suggesting that RetPrac integrated and differentiated competitor and target evidence in the MPFC.

In the AG, three-way ANOVA also revealed significant a evidence-by-outcome interaction (F_(1,18) = 18.46, p=0.0004, survived FDR correction, Figure 3B), suggesting that the representation in the AG tracked behavioral performance. We also found a significant method-by-evidence interaction (F_(1,18) = 7.45, p=0.01, survived FDR correction). No other main effect or interaction was found (p-values >0.36, Supplementary file 2a). Once again, we performed two separate two-way ANOVAs for target and competitor reactivation. For target reactivation, there was a significant main effect of response type (F_(1,18) = 7.23, p=0.02, survived FDR correction, Supplementary file 2b), with stronger target evidence for correct responses than for incorrect responses. For competitor reactivation, there was a significant main effect of response type (F_(1,18) = 6.04, p=0.02, Supplementary file 2b), with stronger competitor evidence for incorrect responses than for correct responses. Together, these results suggested that the AG representation mainly tracked the behavioral performance.

In the VTC, we also found a significant evidence-by-outcome interaction (F_(1,18) = 26.51, p<0.0001, survived FDR correction, Figure 3B), again suggesting that the representation in the VTC tracked behavioral performance. Interestingly, we found a significant main effect of method (F_(1,18) = 5.58, p=0.03), which did not interact with other factors (p-values >0.20, Supplementary file 2a), suggesting that retrieval practice significantly suppressed competitor evidence (F_(1,18) = 4.68, p=0.04) and marginally reduced the target evidence (F_(1,18) = 3.26, p=0.09) in the VTC.

Retrieval practice and restudy were associated with distinct subsequent memory effects

The improved memory that arises from retrieval practice may be supported by neural mechanisms that are distinct from those involved in restudy, and the former mechanisms could produce more resilient traces than the latter. To test this possibility, we performed an analysis to determine whether the two updating methods were associated with different subsequent memory effects. In particular, we examined the pattern of target reactivation during updating and subsequent memory.

This analysis revealed distinct patterns for the Restudy and RetPrac conditions: correctly recalled items (i.e., target) showed stronger target activation than did incorrectly recalled ones (i.e., competitor) in the VTC (t₍₁₈₎ = 4.29, p<0.001, Cohen’s d = 0.79, survived FDR correction) during restudy (all other ROIs, p-values >0.13), whereas retrieval practiced items that were later correctly recalled were associated with stronger target activation than incorrectly recalled ones in the MPFC (t₍₁₈₎ = 2.66, p=0.016, Cohen’s d = 0.60, survived FDR correction, Figure 4A) (all other ROIs, p-values >0.12). This finding suggests that successful memory updating may involve different representations under the RetPrac and Restudy conditions. The greater association of MPFC representation with enduring retention is consistent with its putative involvement in the consolidation process (Antony et al., 2017; Tompary and Davachi, 2017), and suggests that retrieval practice improves updating by driving consolidation more successfully than does restudy.

Figure 4 with 3 supplements see all

Download asset Open asset

Figure 4—source data 1

Classifier evidence during memory updating by subsequent memory performance.

https://cdn.elifesciences.org/articles/57023/elife-57023-fig4-data1-v2.csv

Download elife-57023-fig4-data1-v2.csv

Figure 4—source data 2

Classifier evidence during memory updating.

https://cdn.elifesciences.org/articles/57023/elife-57023-fig4-data2-v2.csv

Download elife-57023-fig4-data2-v2.csv

The LPFC contributed to MPFC memory updating under the RetPrac condition

To identify processes that contributed to goal-directed modulation of reactivated memories, we compared brain activity during updating between the RetPrac and Restudy conditions. We found that updating by retrieval practice engaged the left lateral prefrontal cortex (LPFC), dorsal anterior cingulate gyrus (dACC), bilateral anterior insular cortex (AI), and caudate nucleus (Figure 5A, Supplementary file 3a) more than did updating by restudy, whereas the hippocampus and other regions showed significantly weaker activation (Figure 5—figure supplement 1, Supplementary file 3b). These findings are consistent with prior work showing the engagement of cognitive control during retrieval (Wimber et al., 2015). To further probe the function of these regions in overcoming intrusions from outdated competitors, we examined how these prefrontal and striatal activations varied with updating performance during retrieval practice. We distinguished between those retrieval-practiced trials that subjects recalled incorrectly (incorrect trials, IC), those they got correct the first time a given pair was shown (First Correct, FC), and those that they got correct the second or third time for a given pair was shown (Later Correct, LC). The rationale is that compared to LC trials. IC and FC trials should have a greater need for competition resolution. In addition, the FC trials may engage stronger reward-based learning than IC and LC trials because of the former’s greater positive prediction error. Both mechanisms could contribute to representational change and memory differentiation.

Figure 5 with 2 supplements see all

Download asset Open asset

Figure 5—source data 1

Percent signal change by conditions during memory updating.

https://cdn.elifesciences.org/articles/57023/elife-57023-fig5-data1-v2.csv

Download elife-57023-fig5-data1-v2.csv

Consistently, we found that the left LPFC activation for LC trials was significantly lower than that for FC trials (t₍₁₈₎ = −7.17, p<0.001) and IC trials (t₍₁₈₎ = −4.76, p<0.001) (Figure 5B), suggesting that the left LPFC activation was mainly driven by the extent of memory competition. A similar pattern was also found in the dACC and AI (Figure 5—figure supplement 2, Supplementary file 3c). By contrast, the caudate activation for the FC trials was significantly greater than that for the other two types of trials (all p-values <0.004, Supplementary file 3c), consistent with its role in prediction-error-based processing.

To link the LPFC and caudate activation to representational change during updating, we examined whether the caudate and LPFC activation in the current repetition was related to competitor suppression in the subsequent repetition. Owing to the caudate’s role in prediction error, we focused on the first correct trial (FC). This revealed that strong caudate activation was associated with greater competitor evidence reduction in the next repetition in the VTC (χ²₍₁₎=5.86, p=0.015), but not in MPFC or AG (all p-values >0.15), suggesting that the VTC evidence could be temporary weakened by reinforcement learning. No such effect was found for LPFC when focusing on the incorrect and first correct trials (p-values >0.41), suggesting that LPFC did not temporally suppress the competitor evidence.

We further examined whether the LPFC and caudate activity during retrieval practice was associated with long-term memory updating on Day 3. This analysis revealed that trials with greater LPFC activity during retrieval practice ultimately showed superior memory updating (i.e., target – competitor evidence) during the final test on Day 3 in the MPFC (χ²₍₁₎=4.62, p=0.032), but no effect was found in the caudate. Together, these results suggest that the LPFC is involved in resolving retrieval competition between targets and competitors, and ultimately contributes to successful long-term memory updating in the MPFC. By contrast, the caudate may suppress short-term representation through reward-based supervised learning.

Memory updating serves an adaptive role in ensuring that the most relevant information is accessible in memory. Behavioral studies have long emphasized the role of retrieval in memory updating, yet the neural mechanisms behind this process are barely understood. We found that, compared to simple restudy, retrieval practice was associated with better memory updating without suppressing the old memories. Furthermore, by tracking the neural evidence of old and new memories during both final memory and updating, we demonstrated that superior memory updating under retrieval practice could be achieved by multiple mechanisms. These results provide important insights into the neural mechanisms of memory updating.

When updating memory with replacement information, one needs to enhance the new target memory, inhibit the outdated competing traces, and/or differentiate the old and new memory traces. These different mechanisms could be examined by testing A-B memory. We found that retrieval practice had no effect on A-B memory, albeit it significantly enhanced new memory and reduced old memory intrusions in the A-C memory test. Further supporting the differentiation hypothesis, retrieval practice increased the number of trials in which subjects appropriately chose the targets in different test conditions (an indication of differentiation), and reduced the number of the trials in which the same responses were made in both test conditions (an indication of indifferentiation). At least two factors might contribute to the lack of suppression of old memory trace. First, some studies have shown that retrieval-induced forgetting was more pronounced after a short delay (minutes to hours) than after a long delay (days) (Abel and Bäuml, 2014; Liu and Ranganath, 2019; Murayama et al., 2014). Second, the old memory was extensively trained and consolidated, which made it harder to inhibit. In any case, our results suggest that reduced intrusions could be achieved without significantly suppressing the old memories, but by strengthening the new memory traces and differentiating the old and new memory traces.

The current study revealed several neural mechanisms that could account for the advantages of retrieval practice in memory updating. First, we found that retrieval practice could shift the neural substrates from VTC to MPFC, which is involved in fast system consolidation (Antony et al., 2017). Existing studies have shown that during retrieval, item-specific reactivation is generally not found in the VTC (Favila et al., 2018; Xiao et al., 2017). As a result, with repeated retrieval practice, the brain may rely less on the sensory information for mnemonic decisions. These features are consistent with the behavioral findings that retrieval practice does not improve the quality of sensory memory (Sutterer and Awh, 2016), and may promote gist-based false memory (McDermott, 2006). Our study suggests that the MPFC may be responsible for this gist-based memory given its role in schema-based learning (Gilboa and Marlatte, 2017; Preston and Eichenbaum, 2013).

Second, feedback was provided during retrieval practice in the current study. Although existing studies have found a significant effect of retrieval practice when no feedback was provided (Karpicke and Roediger, 2008), feedback has been consistently shown to improve memory performance (Butler and Roediger, 2008; Pashler et al., 2005). The current study also found greater caudate activation under the RetPrac condition, in particular for the first correct trial, which is consistent with its role in processing positive prediction error (O'Doherty et al., 2004). Recent studies suggest that prediction error plays an important role in memory updating (Kim et al., 2014) and reconsolidation (Lee et al., 2017), and that the caudate is involved in modifying and re-encoding the retrieved memory representation (Scimeca and Badre, 2012). Extending these observations, we found that the caudate’s activation during first correct response was associated with reduced competitor evidence in the visual cortex, lending support to the idea that the supervised-learning mechanism could lead to representational changes (Ritvo et al., 2019).

Third, retrieval practice is an effortful process as compared to simple restudy. Consistently, neuroimaging studies have found greater neural activity during retrieval than during restudy (Wing et al., 2013). In addition, retrieval practice could also potentiate subsequent learning, which is associated with greater frontoparietal activity (Nelson et al., 2013). Our results are highly consistent with these observations, revealing stronger activation in the LPFC, dACC and insula. We further found that LPFC activation was greater when there was a greater competitor intrusion, which is consistent with its role in controlled memory retrieval among competitors (Badre et al., 2005). Previous studies have further implicated the LPFC in reducing the intrusion of competitors in memory retrieval (Kuhl et al., 2012), and in reducing competition memories through cortical pattern suppression (Wimber et al., 2015). The current study did not find a strong association between LPFC activation and competitor suppression during updating, possibly because the to-be-suppressed competing memories in the current study were well trained and consolidated by overnight sleep. We however found that the LPFC activation was associated with long-term memory updating in the MPFC, suggesting that the LPFC might help to resolve the interference between old and new memories.

Critically, the current study identified a region-specific relationship between competitor reactivation during updating and later memory changes. Consistent with previous studies (Kuhl et al., 2012; Wimber et al., 2015), we found significant VTC competitor reactivation during updating and competitor suppression during the final test. One difference is that the current study also found a trend of target suppression in this region, whereas previous studies found target enhancement during retrieval practice. Furthermore, we found a U-shaped relationship between competitor reactivation in the VTC during updating and the final test, which is consistent with the nonmonotonic plasticity mechanism (Ritvo et al., 2019).

A different pattern was found in the MPFC, where the target evidence was strengthened. More importantly, it was integrated but differentiated from the competitor evidence, as indicated by the comparable competitor reactivation for correct and incorrect responses. Furthermore, we found that the MPFC’s memory updating was not predicted by the nonmonotonic plasticity principle, but was rather associated with LPFC activity driven by the competitor reactivation. The MPFC has been implicated in memory integration and updating (Preston and Eichenbaum, 2013; Zeithamova et al., 2012). Our results replicated and extended these observations by showing that memory integration could occur even when competing memories were simultaneously reactivated. This fits very well with the hypothesis that MPFC is able to develop rapidly integrated neocortical representations of reactivated memory traces during retrieval practice (Antony et al., 2017).

These results suggest that during retrieval practice, the co-activation of old and new memories might provide a unique opportunity to modify these representations and to facilitate memory updating. The MPFC could form integrated representations of co-activated and competing memories, while the LPFC control mechanism might contribute to memory updating by selectively strengthening the reactivated target memory and differentiating the old and new memory representations in the MPFC. The differentiation could be achieved by adding contextual representations into the memory trace, thus forming more unique representations of old and new memories, and/or by linking old and new memory representations to different aspects of cue representations. These processes could be further enhanced by feedback-driven supervised learning during retrieval practice, as well as by non-supervised Hebbian learning that involves nonmonotonic plasticity (Ritvo et al., 2019). Future studies should further examine how the LPFC and MPFC could contribute to the representational change of the reactivated old and new memories.

In the current study, we also found significant memory reactivation in the angular gyrus during both updating and the final test. Unlike memory reactivation in the VTC or MPFC, we found that memory reactivation in the angular gyrus tracked closely the behavioral performance. Consistently, other studies have shown that the angular gyrus exhibits abstract yet item-specific mnemonic representations (Kuhl and Chun, 2014; Xiao et al., 2017), which is modulated by mnemonic goals (Favila et al., 2018). Through its connection with the more anterior region, i.e., the lateral intraparietal sulcus (latIPS), the representations in the angular gyrus can serve as a mnemonic buffer that helps to make mnemonic decisions (Sestieri et al., 2017; Wagner et al., 2005). Our results support the idea that AG functions as a multimodal convergent zone to combine memory signals from multiple brain regions and to form a memory representation that is closely related to subjective experience and memory decisions.

Successful memory retrieval is often associated with greater hippocampal activity. Interestingly, we observed weaker hippocampal activation under the RetPrac condition than under the Restudy condition. Previous studies suggested that the hippocampus might be inhibited when subjects were required to suppress thoughts and memories (Benoit and Anderson, 2012; Hulbert et al., 2016). It is thus tempting to speculate that the hippocampus might be inhibited as a result of strong competition from the old memory. The deactivation itself, however, might not be sufficient to support the inhibition hypothesis. For example, although the MPFC also showed weaker activation under the RetPrac condition, MVPA analysis suggested that the MPFC represented task-related information that was related to subsequent memory performance (Figure 4A), suggesting that it played an important role in memory updating. Several major factors might account for the chance-level decoding of memory information in the hippocampus during retrieval. On the one hand, the classifier accuracy during training was lower in the hippocampus than in other regions. This might be due to the sparse nature of hippocampal representation (Quiroga et al., 2008), the low signal-to-noise ratio in this region, and/or the weak categorical representation. Consistently, previous studies also found hippocampal representation when using representational similarity analysis to probe item-level representations (e.g., Jonker et al., 2018; Tompary and Davachi, 2017; Wimber et al., 2015; Xiao et al., 2017). On the other hand, we trained the classifiers during perception and applied them to memory retrieval. Previous studies have shown that memory representation could be transformed from encoding to retrieval (Chen et al., 2017; Xiao et al., 2017; Xue, 2018), and that this transformation could have further reduced the classifier performance (Albers et al., 2013). Future studies should use an optimized design and item-level analysis to further elucidate the role of the hippocampus in memory updating.

Memory is a dynamic process that depends on memory reactivation. On the one hand, reactivation of a target memory will strengthen the activated memory (Tambini and Davachi, 2019; Xue, 2018). On the other hand, reactivation of unwanted memories (e.g., competitors) can facilitate the suppression of those unwanted memories (Lee et al., 2017). Beyond the mechanisms of strengthening and weakening memory representations to achieve memory updating, the current study adds to the growing literature reporting that reactivation (via retrieval practice) can also integrate and differentiate co-activated memories (Chan et al., 2009; Hulbert and Norman, 2015; Schlichting et al., 2015; Zeithamova et al., 2012) and hence can increase the flexibility of memory updating in different contexts. A better understanding of these diverse mechanisms can be leveraged to develop more effective behavioral interventions to modify maladaptive memories in some psychiatric conditions.

All fMRI data collected in this study is available on OpenNeuro under the accession number 002773 (https://openneuro.org/datasets/ds002773).

1. 1. Ye Z
   2. Xue G

   (2020) OpenNeuro

   Retrieval practice facilitates memory updating by enhancing and differentiating medial prefrontal cortex representations.

   https://doi.org/10.18112/openneuro.ds002773.v1.0.0

1. 1. Abel M
   2. Bäuml KH

   (2014) The roles of delay and retroactive interference in retrieval-induced forgetting

   Memory & Cognition 42:141–150.

   https://doi.org/10.3758/s13421-013-0347-0

   • PubMed
   • Google Scholar
2. 1. Agren T
   2. Engman J
   3. Frick A
   4. Björkstrand J
   5. Larsson EM
   6. Furmark T
   7. Fredrikson M

   (2012) Disruption of reconsolidation erases a fear memory trace in the human amygdala

   Science 337:1550–1552.

   https://doi.org/10.1126/science.1223006

   • PubMed
   • Google Scholar
3. 1. Albers AM
   2. Kok P
   3. Toni I
   4. Dijkerman HC
   5. de Lange FP

   (2013) Shared representations for working memory and mental imagery in early visual cortex

   Current Biology 23:1427–1431.

   https://doi.org/10.1016/j.cub.2013.05.065

   • PubMed
   • Google Scholar
4. 1. Anderson MC
   2. Bjork RA
   3. Bjork EL

   (1994) Remembering can cause forgetting: retrieval dynamics in long-term memory

   Journal of Experimental Psychology: Learning, Memory, and Cognition 20:1063–1087.

   https://doi.org/10.1037/0278-7393.20.5.1063

   • Google Scholar
5. 1. Antony JW
   2. Ferreira CS
   3. Norman KA
   4. Wimber M

   (2017) Retrieval as a fast route to memory consolidation

   Trends in Cognitive Sciences 21:573–576.

   https://doi.org/10.1016/j.tics.2017.05.001

   • PubMed
   • Google Scholar
6. 1. Avants BB
   2. Tustison NJ
   3. Song G
   4. Cook PA
   5. Klein A
   6. Gee JC

   (2011) A reproducible evaluation of ANTs similarity metric performance in brain image registration

   NeuroImage 54:2033–2044.

   https://doi.org/10.1016/j.neuroimage.2010.09.025

   • PubMed
   • Google Scholar
7. 1. Badre D
   2. Poldrack RA
   3. Paré-Blagoev EJ
   4. Insler RZ
   5. Wagner AD

   (2005) Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex

   Neuron 47:907–918.

   https://doi.org/10.1016/j.neuron.2005.07.023

   • PubMed
   • Google Scholar
8. 1. Benoit RG
   2. Anderson MC

   (2012) Opposing mechanisms support the voluntary forgetting of unwanted memories

   Neuron 76:450–460.

   https://doi.org/10.1016/j.neuron.2012.07.025

   • PubMed
   • Google Scholar
9. 1. Butler AC
   2. Roediger HL

   (2008) Feedback enhances the positive effects and reduces the negative effects of multiple-choice testing

   Memory & Cognition 36:604–616.

   https://doi.org/10.3758/MC.36.3.604

   • PubMed
   • Google Scholar
10. 1. Chan JC
    2. Thomas AK
    3. Bulevich JB

    (2009) Recalling a witnessed event increases eyewitness suggestibility: the reversed testing effect

    Psychological Science 20:66–73.

    https://doi.org/10.1111/j.1467-9280.2008.02245.x

    • PubMed
    • Google Scholar
11. 1. Chen J
    2. Leong YC
    3. Honey CJ
    4. Yong CH
    5. Norman KA
    6. Hasson U

    (2017) Shared memories reveal shared structure in neural activity across individuals

    Nature Neuroscience 20:115–125.

    https://doi.org/10.1038/nn.4450

    • PubMed
    • Google Scholar
12. 1. Detre GJ
    2. Natarajan A
    3. Gershman SJ
    4. Norman KA

    (2013) Moderate levels of activation lead to forgetting in the think/no-think paradigm

    Neuropsychologia 51:2371–2388.

    https://doi.org/10.1016/j.neuropsychologia.2013.02.017

    • PubMed
    • Google Scholar
13. 1. Diekelmann S
    2. Büchel C
    3. Born J
    4. Rasch B

    (2011) Labile or stable: opposing consequences for memory when reactivated during waking and sleep

    Nature Neuroscience 14:381–386.

    https://doi.org/10.1038/nn.2744

    • PubMed
    • Google Scholar
14. 1. Dudai Y

    (2006) Reconsolidation: the advantage of being refocused

    Current Opinion in Neurobiology 16:174–178.

    https://doi.org/10.1016/j.conb.2006.03.010

    • PubMed
    • Google Scholar
15. 1. Fan R-E
    2. Chang K-W
    3. Hsieh C-J
    4. Wang X-R
    5. Lin C-J

    (2008) LIBLINEAR: a library for large linear classification

    Journal of Machine Learning Research : JMLR 9:1871–1874.

    https://doi.org/10.1145/1390681.1442794

    • Google Scholar
16. 1. Favila SE
    2. Samide R
    3. Sweigart SC
    4. Kuhl BA

    (2018) Parietal representations of stimulus features are amplified during memory retrieval and flexibly aligned with Top-Down goals

    The Journal of Neuroscience 38:7809–7821.

    https://doi.org/10.1523/JNEUROSCI.0564-18.2018

    • PubMed
    • Google Scholar
17. 1. Gilboa A
    2. Marlatte H

    (2017) Neurobiology of schemas and Schema-Mediated memory

    Trends in Cognitive Sciences 21:618–631.

    https://doi.org/10.1016/j.tics.2017.04.013

    • PubMed
    • Google Scholar
18. 1. Goltseker K
    2. Bolotin L
    3. Barak S

    (2017) Counterconditioning during reconsolidation prevents relapse of cocaine memories

    Neuropsychopharmacology 42:716–726.

    https://doi.org/10.1038/npp.2016.140

    • PubMed
    • Google Scholar
19. 1. Hulbert JC
    2. Henson RN
    3. Anderson MC

    (2016) Inducing amnesia through systemic suppression

    Nature Communications 7:11003.

    https://doi.org/10.1038/ncomms11003

    • PubMed
    • Google Scholar
20. 1. Hulbert JC
    2. Norman KA

    (2015) Neural differentiation tracks improved recall of competing memories following interleaved study and retrieval practice

    Cerebral Cortex 25:3994–4008.

    https://doi.org/10.1093/cercor/bhu284

    • PubMed
    • Google Scholar
21. 1. Hupbach A
    2. Gomez R
    3. Hardt O
    4. Nadel L

    (2007) Reconsolidation of episodic memories: a subtle reminder triggers integration of new information

    Learning & Memory 14:47–53.

    https://doi.org/10.1101/lm.365707

    • PubMed
    • Google Scholar
22. 1. James EL
    2. Bonsall MB
    3. Hoppitt L
    4. Tunbridge EM
    5. Geddes JR
    6. Milton AL
    7. Holmes EA

    (2015) Computer game play reduces intrusive memories of experimental trauma via Reconsolidation-Update mechanisms

    Psychological Science 26:1201–1215.

    https://doi.org/10.1177/0956797615583071

    • PubMed
    • Google Scholar
23. 1. Jonker TR
    2. Dimsdale-Zucker H
    3. Ritchey M
    4. Clarke A
    5. Ranganath C

    (2018) Neural reactivation in parietal cortex enhances memory for episodically linked information

    PNAS 115:11084–11089.

    https://doi.org/10.1073/pnas.1800006115

    • PubMed
    • Google Scholar
24. 1. Karpicke JD
    2. Roediger HL

    (2008) The critical importance of retrieval for learning

    Science 319:966–968.

    https://doi.org/10.1126/science.1152408

    • PubMed
    • Google Scholar
25. 1. Kim G
    2. Lewis-Peacock JA
    3. Norman KA
    4. Turk-Browne NB

    (2014) Pruning of memories by context-based prediction error

    PNAS 111:8997–9002.

    https://doi.org/10.1073/pnas.1319438111

    • PubMed
    • Google Scholar
26. 1. Kindt M
    2. Soeter M
    3. Vervliet B

    (2009) Beyond extinction: erasing human fear responses and preventing the return of fear

    Nature Neuroscience 12:256–258.

    https://doi.org/10.1038/nn.2271

    • PubMed
    • Google Scholar
27. 1. Kroes MC
    2. Tendolkar I
    3. van Wingen GA
    4. van Waarde JA
    5. Strange BA
    6. Fernández G

    (2014) An electroconvulsive therapy procedure impairs reconsolidation of episodic memories in humans

    Nature Neuroscience 17:204–206.

    https://doi.org/10.1038/nn.3609

    • PubMed
    • Google Scholar
28. 1. Kuhl BA
    2. Bainbridge WA
    3. Chun MM

    (2012) Neural reactivation reveals mechanisms for updating memory

    Journal of Neuroscience 32:3453–3461.

    https://doi.org/10.1523/JNEUROSCI.5846-11.2012

    • PubMed
    • Google Scholar
29. 1. Kuhl BA
    2. Chun MM

    (2014) Successful remembering elicits event-specific activity patterns in lateral parietal cortex

    Journal of Neuroscience 34:8051–8060.

    https://doi.org/10.1523/JNEUROSCI.4328-13.2014

    • PubMed
    • Google Scholar
30. 1. Lee JLC
    2. Nader K
    3. Schiller D

    (2017) An update on memory reconsolidation updating

    Trends in Cognitive Sciences 21:531–545.

    https://doi.org/10.1016/j.tics.2017.04.006

    • PubMed
    • Google Scholar
31. 1. Levy BJ
    2. Anderson MC

    (2002) Inhibitory processes and the control of memory retrieval

    Trends in Cognitive Sciences 6:299–305.

    https://doi.org/10.1016/S1364-6613(02)01923-X

    • PubMed
    • Google Scholar
32. Preprint

    1. Liu X
    2. Ranganath C

    (2019) Temporal context and sleep-dependent memory consolidation mediate the boundary between retrieval induced forgetting and facilitation

    PsyArXiv.

    https://doi.org/10.31234/osf.io/4nfue

    • Google Scholar
33. 1. Lu Y
    2. Wang C
    3. Chen C
    4. Xue G

    (2015) Spatiotemporal neural pattern similarity supports episodic memory

    Current Biology 25:780–785.

    https://doi.org/10.1016/j.cub.2015.01.055

    • PubMed
    • Google Scholar
34. 1. McDermott KB

    (2006) Paradoxical effects of testing: repeated retrieval attempts enhance the likelihood of later accurate and false recall

    Memory & Cognition 34:261–267.

    https://doi.org/10.3758/BF03193404

    • PubMed
    • Google Scholar
35. 1. Murayama K
    2. Miyatsu T
    3. Buchli D
    4. Storm BC

    (2014) Forgetting as a consequence of retrieval: a meta-analytic review of retrieval-induced forgetting

    Psychological Bulletin 140:1383–1409.

    https://doi.org/10.1037/a0037505

    • PubMed
    • Google Scholar
36. 1. Nader K
    2. Schafe GE
    3. Le Doux JE

    (2000) Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval

    Nature 406:722–726.

    https://doi.org/10.1038/35021052

    • PubMed
    • Google Scholar
37. 1. Nelson SM
    2. Arnold KM
    3. Gilmore AW
    4. McDermott KB

    (2013) Neural signatures of test-potentiated learning in parietal cortex

    Journal of Neuroscience 33:11754–11762.

    https://doi.org/10.1523/JNEUROSCI.0960-13.2013

    • PubMed
    • Google Scholar
38. 1. Newman EL
    2. Norman KA

    (2010) Moderate excitation leads to weakening of perceptual representations

    Cerebral Cortex 20:2760–2770.

    https://doi.org/10.1093/cercor/bhq021

    • PubMed
    • Google Scholar
39. 1. Norman KA
    2. Newman EL
    3. Detre G

    (2007) A neural network model of retrieval-induced forgetting

    Psychological Review 114:887–953.

    https://doi.org/10.1037/0033-295X.114.4.887

    • PubMed
    • Google Scholar
40. 1. O'Doherty J
    2. Dayan P
    3. Schultz J
    4. Deichmann R
    5. Friston K
    6. Dolan RJ

    (2004) Dissociable roles of ventral and dorsal striatum in instrumental conditioning

    Science 304:452–454.

    https://doi.org/10.1126/science.1094285

    • PubMed
    • Google Scholar
41. 1. Pashler H
    2. Cepeda NJ
    3. Wixted JT
    4. Rohrer D

    (2005) When does feedback facilitate learning of words?

    Journal of Experimental Psychology: Learning, Memory, and Cognition 31:3–8.

    https://doi.org/10.1037/0278-7393.31.1.3

    • Google Scholar
42. 1. Preston AR
    2. Eichenbaum H

    (2013) Interplay of Hippocampus and prefrontal cortex in memory

    Current Biology 23:R764–R773.

    https://doi.org/10.1016/j.cub.2013.05.041

    • PubMed
    • Google Scholar
43. 1. Pyc MA
    2. Rawson KA

    (2010) Why testing improves memory: mediator effectiveness hypothesis

    Science 330:335.

    https://doi.org/10.1126/science.1191465

    • PubMed
    • Google Scholar
44. 1. Quiroga RQ
    2. Kreiman G
    3. Koch C
    4. Fried I

    (2008) Sparse but not 'grandmother-cell' coding in the medial temporal lobe

    Trends in Cognitive Sciences 12:87–91.

    https://doi.org/10.1016/j.tics.2007.12.003

    • PubMed
    • Google Scholar
45. 1. Ritvo VJH
    2. Turk-Browne NB
    3. Norman KA

    (2019) Nonmonotonic plasticity: how memory retrieval drives learning

    Trends in Cognitive Sciences 23:726–742.

    https://doi.org/10.1016/j.tics.2019.06.007

    • PubMed
    • Google Scholar
46. 1. Roediger HL
    2. Butler AC

    (2011) The critical role of retrieval practice in long-term retention

    Trends in Cognitive Sciences 15:20–27.

    https://doi.org/10.1016/j.tics.2010.09.003

    • PubMed
    • Google Scholar
47. 1. Rolls ET
    2. Joliot M
    3. Tzourio-Mazoyer N

    (2015) Implementation of a new parcellation of the orbitofrontal cortex in the automated anatomical labeling atlas

    NeuroImage 122:1–5.

    https://doi.org/10.1016/j.neuroimage.2015.07.075

    • PubMed
    • Google Scholar
48. 1. Rugg MD
    2. Vilberg KL

    (2013) Brain networks underlying episodic memory retrieval

    Current Opinion in Neurobiology 23:255–260.

    https://doi.org/10.1016/j.conb.2012.11.005

    • PubMed
    • Google Scholar
49. 1. Schaefer A
    2. Kong R
    3. Gordon EM
    4. Laumann TO
    5. Zuo XN
    6. Holmes AJ
    7. Eickhoff SB
    8. Yeo BTT

    (2018) Local-Global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI

    Cerebral Cortex 28:3095–3114.

    https://doi.org/10.1093/cercor/bhx179

    • PubMed
    • Google Scholar
50. 1. Schiller D
    2. Monfils MH
    3. Raio CM
    4. Johnson DC
    5. Ledoux JE
    6. Phelps EA

    (2010) Preventing the return of fear in humans using reconsolidation update mechanisms

    Nature 463:49–53.

    https://doi.org/10.1038/nature08637

    • PubMed
    • Google Scholar
51. 1. Schlichting ML
    2. Mumford JA
    3. Preston AR

    (2015) Learning-related representational changes reveal dissociable integration and separation signatures in the Hippocampus and prefrontal cortex

    Nature Communications 6:8151.

    https://doi.org/10.1038/ncomms9151

    • PubMed
    • Google Scholar
52. 1. Scimeca JM
    2. Badre D

    (2012) Striatal contributions to declarative memory retrieval

    Neuron 75:380–392.

    https://doi.org/10.1016/j.neuron.2012.07.014

    • PubMed
    • Google Scholar
53. 1. Sestieri C
    2. Shulman GL
    3. Corbetta M

    (2017) The contribution of the human posterior parietal cortex to episodic memory

    Nature Reviews Neuroscience 18:183–192.

    https://doi.org/10.1038/nrn.2017.6

    • PubMed
    • Google Scholar
54. 1. Smith SM
    2. Jenkinson M
    3. Woolrich MW
    4. Beckmann CF
    5. Behrens TE
    6. Johansen-Berg H
    7. Bannister PR
    8. De Luca M
    9. Drobnjak I
    10. Flitney DE
    11. Niazy RK
    12. Saunders J
    13. Vickers J
    14. Zhang Y
    15. De Stefano N
    16. Brady JM
    17. Matthews PM

    (2004) Advances in functional and structural MR image analysis and implementation as FSL

    NeuroImage 23 Suppl 1:S208–S219.

    https://doi.org/10.1016/j.neuroimage.2004.07.051

    • PubMed
    • Google Scholar
55. 1. Sommer T

    (2016) The emergence of knowledge and how it supports the memory for novel related information

    Cerebral Cortex 107:bhw031.

    https://doi.org/10.1093/cercor/bhw031

    • Google Scholar
56. 1. Storm BC
    2. Bjork EL
    3. Bjork RA

    (2008) Accelerated relearning after retrieval-induced forgetting: the benefit of being forgotten

    Journal of Experimental Psychology: Learning, Memory, and Cognition 34:230–236.

    https://doi.org/10.1037/0278-7393.34.1.230

    • Google Scholar
57. 1. Sutterer DW
    2. Awh E

    (2016) Retrieval practice enhances the accessibility but not the quality of memory

    Psychonomic Bulletin & Review 23:831–841.

    https://doi.org/10.3758/s13423-015-0937-x

    • PubMed
    • Google Scholar
58. 1. Szpunar KK
    2. McDermott KB
    3. Roediger HL

    (2008) Testing during study insulates against the buildup of proactive interference

    Journal of Experimental Psychology: Learning, Memory, and Cognition 34:1392–1399.

    https://doi.org/10.1037/a0013082

    • Google Scholar
59. 1. Tambini A
    2. Davachi L

    (2019) Awake reactivation of prior experiences consolidates memories and biases cognition

    Trends in Cognitive Sciences 23:876–890.

    https://doi.org/10.1016/j.tics.2019.07.008

    • PubMed
    • Google Scholar
60. 1. Tompary A
    2. Davachi L

    (2017) Consolidation promotes the emergence of representational overlap in the Hippocampus and medial prefrontal cortex

    Neuron 96:228–241.

    https://doi.org/10.1016/j.neuron.2017.09.005

    • PubMed
    • Google Scholar
61. 1. van den Honert RN
    2. McCarthy G
    3. Johnson MK

    (2016) Reactivation during encoding supports the later discrimination of similar episodic memories

    Hippocampus 26:1168–1178.

    https://doi.org/10.1002/hipo.22598

    • PubMed
    • Google Scholar
62. 1. Wagner AD
    2. Shannon BJ
    3. Kahn I
    4. Buckner RL

    (2005) Parietal lobe contributions to episodic memory retrieval

    Trends in Cognitive Sciences 9:445–453.

    https://doi.org/10.1016/j.tics.2005.07.001

    • PubMed
    • Google Scholar
63. 1. Wahlheim CN
    2. Jacoby LL

    (2013) Remembering change: the critical role of recursive remindings in proactive effects of memory

    Memory & Cognition 41:1–15.

    https://doi.org/10.3758/s13421-012-0246-9

    • PubMed
    • Google Scholar
64. 1. Wimber M
    2. Alink A
    3. Charest I
    4. Kriegeskorte N
    5. Anderson MC

    (2015) Retrieval induces adaptive forgetting of competing memories via cortical pattern suppression

    Nature Neuroscience 18:582–589.

    https://doi.org/10.1038/nn.3973

    • PubMed
    • Google Scholar
65. 1. Wing EA
    2. Marsh EJ
    3. Cabeza R

    (2013) Neural correlates of retrieval-based memory enhancement: an fMRI study of the testing effect

    Neuropsychologia 51:2360–2370.

    https://doi.org/10.1016/j.neuropsychologia.2013.04.004

    • PubMed
    • Google Scholar
66. 1. Xiao X
    2. Dong Q
    3. Gao J
    4. Men W
    5. Poldrack RA
    6. Xue G

    (2017) Transformed neural pattern reinstatement during episodic memory retrieval

    The Journal of Neuroscience 37:2986–2998.

    https://doi.org/10.1523/JNEUROSCI.2324-16.2017

    • PubMed
    • Google Scholar
67. 1. Xue G
    2. Dong Q
    3. Chen C
    4. Lu Z
    5. Mumford JA
    6. Poldrack RA

    (2010) Greater neural pattern similarity across repetitions is associated with better memory

    Science 330:97–101.

    https://doi.org/10.1126/science.1193125

    • PubMed
    • Google Scholar
68. 1. Xue YX
    2. Luo YX
    3. Wu P
    4. Shi HS
    5. Xue LF
    6. Chen C
    7. Zhu WL
    8. Ding ZB
    9. Bao YP
    10. Shi J
    11. Epstein DH
    12. Shaham Y
    13. Lu L

    (2012) A memory retrieval-extinction procedure to prevent drug craving and relapse

    Science 336:241–245.

    https://doi.org/10.1126/science.1215070

    • PubMed
    • Google Scholar
69. 1. Xue G

    (2018) The neural representations underlying human episodic memory

    Trends in Cognitive Sciences 22:544–561.

    https://doi.org/10.1016/j.tics.2018.03.004

    • PubMed
    • Google Scholar
70. 1. Zeithamova D
    2. Dominick AL
    3. Preston AR

    (2012) Hippocampal and ventral medial prefrontal activation during retrieval-mediated learning supports novel inference

    Neuron 75:168–179.

    https://doi.org/10.1016/j.neuron.2012.05.010

    • PubMed
    • Google Scholar

Author details

1. Zhifang Ye

   State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute of Brain Research, Beijing Normal University, Beijing, China

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0489-2619

2. Liang Shi

   State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute of Brain Research, Beijing Normal University, Beijing, China

   Contribution

   Investigation

   Competing interests

   No competing interests declared

3. Anqi Li

   State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute of Brain Research, Beijing Normal University, Beijing, China

   Contribution

   Investigation

   Competing interests

   No competing interests declared

4. Chuansheng Chen

   Department of Psychological Science, University of California, Irvine, Irvine, United States

   Contribution

   Writing - review and editing

   Competing interests

   No competing interests declared

5. Gui Xue

   State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute of Brain Research, Beijing Normal University, Beijing, China

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   gxue@bnu.edu.cn

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7891-8151

• 3,151

  Page views

• 456

  Downloads

• 16

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.