Retrieval practice prevents stress-induced inference impairment and preserves rapid bridge-related memory reactivation

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife Assessment

    This study presents valuable findings on how retrieval practice protects memory inferences from stress via covert memory reactivation. Via two EEG experiments manipulating stress and retrieval practice, the authors provide solid evidence showing that retrieval practice buffered negative impact of stress on behaviour and neural activities. This work will be of interest to cognitive and affective neuroscientists working on the intersection between memory and stress.

This article has been Reviewed by the following groups

Read the full article

Abstract

A hallmark of human memory is its ability to form novel inferences by linking discrete but related events. We examined whether acute stress impairs memory inference and if retrieval practice can buffer this effect. Participants were trained on image pairs, AB and BC, to establish interconnected triads (ABC) with a shared bridge element B. Twenty-four hours later, we induced acute stress in half of the participants and then tested their capacity to infer the indirect AC associations. Behavioral results indicated that acute stress-induced memory inference accuracy and speed, and retrieval practice was associated with attenuation of stress-induced impairment in A-C inference accuracy and speed. Using multivariate decoding analysis of human electroencephalogram (EEG) recordings, we found neural evidence that bridge element B is rapidly reactivated during the inferential process, a neural signature predictive of subsequent successful inference. Importantly, stress disrupts this rapid neural reactivation of the bridge element, but retrieval practice buffers the stress effect and enhances the strength of reactivation signals beyond the non-stress condition. Time-frequency analyses of theta oscillations mirrored these effects of stress and retrieval practice on inference performance and neural reactivation. Collectively, our findings pinpoint rapid reactivation of the bridge element as an essential neural mechanism for memory-based inference. Although susceptible to stress, this mechanism can be enhanced through retrieval practice, suggesting that building robust memory traces renders subsequent inferences resilient to stress.

Article activity feed

  1. eLife Assessment

    This study presents valuable findings on how retrieval practice protects memory inferences from stress via covert memory reactivation. Via two EEG experiments manipulating stress and retrieval practice, the authors provide solid evidence showing that retrieval practice buffered negative impact of stress on behaviour and neural activities. This work will be of interest to cognitive and affective neuroscientists working on the intersection between memory and stress.

  2. Reviewer #1 (Public review):

    Summary:

    The authors combine a two-day associative-inference task with EEG in two experiments involving 136 participants. Participants learn overlapping AB and BC pairs and subsequently infer AC relationships after a stress or control procedure. The second experiment adds retrieval practice after learning. Stress-related reductions in inference performance and bridge-category decoding in Experiment 1 are attenuated or reversed in Experiment 2.

    The revision provides useful additional analyses and more explicit acknowledgment of the design's limitations. I no longer interpret Figure 4A as evidence of below-chance inference. However, several conclusions in the title, abstract, and Discussion remain stronger than the revised evidence supports.

    Strengths:

    The study addresses a worthwhile question by combining behavioral measures, an experimentally induced stress manipulation, and temporally resolved EEG. The independent visual localizer and decoding during the pre-choice period provide useful evidence that information related to the absent bridge category is available during inference.

    Importantly, the evidence is not limited to a significant stress effect in one experiment and a nonsignificant effect in the other. The manuscript reports direct stress-by-experiment interactions for inference performance, response time, and decoding. The behavioral interaction also remains significant after adjustment for both-correct premise-triad counts. The exploratory mediation analysis, Bayesian comparison, and trial-level model improve the transparency of the evidence and deserve explicit credit.

    Weaknesses:

    (1) Premise memory strength and inference specific protection remain difficult to separate

    The authors now acknowledge that strengthening AB/BC memories may contribute to better AC performance, and the exploratory mediation analysis supports this possibility. This is a useful clarification. However, a residual direct association after adjustment for premise memory does not establish an additional inference-specific mechanism. Premise memory was measured after inference using a coarse 2AFC test, and the mediation concerns the overall association with retrieval practice rather than specifically explaining the stress-by-practice interaction. Similarly, restricting analyses to both-correct trials or adjusting their number cannot equate underlying memory strength. The Discussion should therefore avoid treating the nonsignificant difference between the two control groups as evidence that the benefit transcends strengthened associations or specifically preserves online integration.

    (2) The cross-experiment comparison supports an association with attenuation

    The new interaction analyses strengthen the evidence that the stress effect differs between experiments. Nevertheless, retrieval practice was introduced in a separate follow-up cohort, so its effect remains confounded with experiment. The authors appropriately acknowledge this limitation; it should also guide the title and broader conclusions. "Associated with attenuation" is better supported than "prevents" or "restores." Likewise, the reported BF01 of 2.77 provides only weak support for no stress effect in Experiment 2 and does not establish equivalent performance. A randomized factorial study would provide a stronger causal test, but the present findings remain informative with appropriately limited claims.

    (3) The neural findings support category level information with limited mechanistic specificity

    The shift toward "bridge-related category information" is appropriate because the decoder distinguishes faces from buildings. It cannot identify reinstatement of the particular B item linking a given AB-BC pair. The association with inference success is meaningful, but category imagery, retrieval strategy, and item-specific recollection are not dissociated. Nor does this association establish that the decoded activity is necessary for inference or mediates stress-related impairment. Despite the response letter's clarification, the abstract and conclusion still describe rapid bridge-element reactivation as an essential or causal mechanism. Consistent category-level, correlational language would align these passages with the analyses. The claim that retrieval practice prevents the theta reduction is also tentative: the reported within-experiment tests and overall, between-experiment comparison do not directly test a stress-by-practice interaction for theta (Section 3.6).

    (4) Retrieval specific and item specific benefits remain unresolved

    The lack of a matched restudy or additional-exposure condition limits conclusions about the distinctive contribution of retrieval practice. The authors now acknowledge this appropriately. They also explain that practice responses cannot be reliably scored, so the proposed successful-versus-unsuccessful practice analysis is unavailable. The new participant-and-triad mixed-effects model is a useful robustness check, but its interaction is uncertain (p = .058), and it cannot establish whether protection follows successful retrieval of particular items or a broader change in strategy. These limitations remain unresolved by the new model.

  3. Reviewer #2 (Public review):

    Summary:

    Guo et al. investigate how acute stress affects memory-based inference and whether retrieval practice protects against these effects. Across two experiments involving 136 participants, the authors report that stress impairs inference performance and alters rapid neural reactivation of bridge-related memories. Retrieval practice appears to preserve both inference performance and the associated neural reactivation.

    Strengths:

    (1) Two independent experiments strengthen confidence in the behavioral findings.

    (2) Time-resolved EEG decoding provides valuable temporal information about bridge-related memory reactivation during inference.

    (3) The findings connect a practical learning strategy with behavioral and neural measures of resilience to acute stress.

    Weaknesses:

    (1) The physiological interpretation requires some clarification. Although the manuscript acknowledges that cortisol was not measured, wording referring to a confirmed cortisol peak remains. The timing reflects an expected physiological response rather than a measured peak, and the available measurements cannot distinguish the respective contributions of glucocorticoid and adrenergic activity.

    (2) Significant comparisons in Figures 4 and 6 remain difficult to identify in the supplied revised manuscript. Clear annotations and corresponding legend definitions would improve interpretation.

    The authors have acknowledged these concerns in their response, but the corresponding revisions are not fully reflected in the manuscript provided. These remaining issues can be addressed through minor revisions.

    Appraisal and Impact

    This study provides evidence linking stress-related impairment in memory-based inference to altered rapid reactivation of bridge-related memories. The combination of behavioral experiments and time-resolved EEG offers insight into the processes supporting inference under stress. The finding that retrieval practice can preserve performance and associated neural reactivation is relevant to cognitive neuroscience and applied learning research.

  4. Reviewer #3 (Public review):

    Summary:

    In this study, Guo and colleagues investigated the effects of stress and retrieval practice on memory inference. In the first experiment, they found that memory inference was significantly worse after induced stress. Conversely, when participants received retrieval practice in the second experiment, they found no significant differences between these conditions. They monitored EEG during the inference phase and applied multivariate decoding analysis to examine evidence of neural reactivation. Complementing the behavioural findings of the first experiment, they found that they were able to decode the stimulus category of the inference item with more fidelity in the no stress condition. Surprisingly, they found the opposite direction when participants had retrieval practice, with stronger evidence of reactivation in the stress condition than in the control condition.

    Strengths:

    (1) With a basis in strong theoretical frameworks, the authors have carefully designed two studies investigating the effects of stress and memory retrieval on memory inference.

    (2) The use multivariate decoding on the inference phase data combined with behavioural data sheds new light on how stress and retrieval may impact the neural signatures of inference processing.

    Weaknesses:

    (1) In places, some clarifications are needed about their statistical approach. For example, no multiple comparisons corrections have been applied to the follow-up tests even though they should be considered within the same 'family'. Therefore, the results from the uncorrected analyses should be interpreted with caution, but this should not affect the main claims of the paper.

  5. Author response:

    The following is the authors’ response to the original reviews.

    Public Reviews:

    Reviewer #1 (Public review):

    (1) Insufficient evidence that retrieval practice specifically protects inference rather than strengthening associative memories

    Sorry for the writing led to misunderstanding,we did not aim to argue that retrieval practice protects inference without affecting associative memories. Instead, we found associative memories mediated the effect of retrieval practice on A-C inference. Specifically, we added an exploratory mediation analysis to quantify the extent to which strengthened premise memory accounted for the association between retrieval practice and later A-C inference. Retrieval practice was associated with better AB/BC premise-memory performance (a = 2.13), and premise-memory performance predicted A-C inference after retrieval practice was controlled (b = 0.012). The bootstrapped indirect effect was significant (ab = 0.026, 95% CI [0.005, 0.049]), while the direct effect of retrieval practice remained significant (c' = 0.028, 95% CI [0.002, 0.055]). We have added this analysis to Section 3.2.3, “Retrieval practice enhanced direct memory retrieval but not baseline inference,” in the revised manuscript. Thus, premise memory (i.e., AB/BC memory) partially, but not completely, mediated the association. Because premise memory was assessed after inference with a 2AFC test, we explicitly describe this analysis as exploratory and interpret it cautiously.

    (2) Importantly, the direct memory retrieval test used a two-alternative forced-choice (2AFC) format, which inherently allows a substantial proportion of correct responses to arise from guessing. Consequently, trials classified as 'successfully retrieved' may still include weak associative memory traces, making it difficult to conclude that failures in inference reflect deficits in integration rather than incomplete associative learning.

    We agree with the reviewer1 that A correct response in a 2AFC test is not equivalent to unequivocal retrieval of a strong associative trace. We therefore replaced wording such as 'recalled' or 'successfully retrieved' with the more precise phrase 'answered correctly' when referring to the AB/BC direct-memory test. We also added this limitation to the Methods(2.2.6 Memory inference phase)and clarified that restricting the analysis to both-correct AB/BC trials reduces, but does not eliminate, uncertainty due to guessing. The revised Discussion notes that future studies should use cued recall, confidence ratings, or continuous measures of premise-memory strength to distinguish incomplete associative learning from a deficit in integration more sensitively.

    (3) Apparent below-chance inference performance raises interpretational concerns

    Thank you for identifying this ambiguity. We did not present the raw ACC in the figure,therefore, the presented value cannot be compare to 0.33 to get above/below chance conclusion. Figure 4A displayed the primary normalized/conditional inference measure, which relates successful A-C performance to the availability of the directly learned AB/BC premise associations; it is therefore not a raw 3AFC proportion and should not be compared directly with 0.33. We revised the Figure 4 legend to define the plotted measure explicitly and to state that it does not represent raw 3AFC accuracy.

    (4) Between-experiment implementation of retrieval practice weakens causal inference

    We admit that experiment 2 was conducted as a follow-up experiment after Experiment 1, and the retrieval-practice factor was therefore not randomized within a single fully crossed experiment. Although the two experiments were closely matched in recruitment, participant population, laboratory setting, EEG system, task structure, and experimenter team, cohort-related confounds cannot be excluded. We have accordingly revised causal wording throughout the manuscript, described the cross-experiment interaction as an association, and added this limitation to the Discussion. We also state that a within-experiment 2 x 2 design (Stress x Retrieval Practice) is needed to provide stronger causal evidence.

    (5) Lack of an appropriate comparison condition limits the interpretation of retrieval-practice mechanisms

    The present study did not include a re-exposure or restudy control condition. Therefore, the current design cannot determine whether the observed effect was specifically attributable to retrieval practice or could also be produced by re-exposure or restudy. We have added this limitation to the Discussion and noted that future studies should include a re-exposure or restudy control condition to clarify the specific contribution of retrieval practice.

    (6) Furthermore, it is unclear whether retrieval practice operates at the trial level or the participant level. Retrieval practice could enhance memory representations for specific practiced items, making those trials more resistant to stress, or it could induce a more global change in cognitive strategy or stress resilience across participants. One way to address this issue would be to analyze inference performance separately for trials that were successfully retrieved during the retrieval practice phase versus those that were not

    We agree with reviewer1 that additional analyses of retrieval performance during retrieval could help improve. However, in Experiment 2, retrieval practice was implemented using a free-recall procedure. Because participants' responses during this phase were not recorded, retrieval-practice performance could not be reliably coded for individual AB/BC associations. Consequently, the present data do not allow us to compare later A-C inference performance between associations that were successfully versus unsuccessfully retrieved during retrieval practice.

    As an indirect robustness check, we fitted a mixed-effects model to trial-level A-C inference accuracy, including fixed effects of stress, retrieval practice, and their interaction, as well as random intercepts for participants and ABC triads. The stress-by-retrieval-practice interaction showed a positive trend consistent with the participant-level analysis, although it did not reach the conventional significance threshold (b = 0.116, SE = 0.061, z = 1.90, p = .058). We have added this analysis to Section 3.2.3, “Retrieval practice enhanced direct memory retrieval but not baseline inference,” in the revised manuscript. Future studies should adopt a standardized retrieval-practice procedure and systematically record retrieval-practice performance for each association. This would allow researchers to determine whether the protective effect of retrieval practice is specific to successfully retrieved trials or reflects a broader participant-level effect.

    (7) Interpretation of EEG decoding as bridge-element reactivation may be overstated

    Yes, our decoding targeted at category classification, therefore, we cannot sure specific picture (i.e., the bridge-element was reactivated). We revised the terminology throughout the manuscript to use “bridge-related category information” (or equivalent category-level wording) and removed claims that the analysis decoded a specific bridging memory.

    Reviewer #2 (Public review):

    (1) The authors correctly timed the inference task to begin approximately 20 minutes after the onset of the stressor. While this window aligns with the expected peak of the glucocorticoid (HPA) response, it also represents a period where the rapid adrenergic (SAM) response, confirmed by heart rate elevation, is still highly influential. As the authors acknowledge, because they did not collect saliva samples due to safety protocols, they cannot definitively separate the influence of peak cortisol from the tail-end of the adrenergic surge on the observed memory impairments.

    The approximately 20-minute interval was selected on the basis of prior stress-and-retrieval studies and corresponds to the expected window of the delayed HPA response, while the heart-rate data also indicate that autonomic arousal remained relevant. Saliva collection was omitted because the study was conducted under COVID-19 safety protocols, and the present data therefore cannot dissociate glucocorticoid from adrenergic contributions. We revised the manuscript to avoid implying that cortisol was measured or that a cortisol peak was confirmed, and we now state explicitly that future studies should collect hormonal measures, when safe, to separate these mechanisms.

    (2) Figures 4 and 6: Without asterisks is really difficult to compare the significant group differences.

    Thank you for this suggestion, we added asterisks to Figures 4 and 6. At the same time, statistical significance was already conveyed in the original figures by whether the confidence intervals included zero.

    Reviewer #3 (Public review):

    (1) There are some key gaps in the reporting of the data. In particular, data is missing on how many trials were included in the inference phase and how many were retrieved in the direct memory task. This is important to know as the main conclusions are based on inference trials proportional to the direct retrieval trials. Considering that the direct retrieval performance differs significantly between the experiments, there could be issues with floor/ceiling effects (in the behaviour) and statistical power (in the EEG results) that confound the comparisons between experiments. Without the data, it is difficult to draw conclusions.

    We agree with Reviewer 3 and have added Table S1 to the Supplementary Material. The table reports group-level direct-memory correct counts, the number of triads for which both AB and BC were answered correctly, and conditional A-C accuracy. Importantly, the number of both-correct AB/BC triads did not differ significantly between stress and control groups in either Experiment 1 (t (66) = -1.65, p = .10) or Experiment 2 (t(66) = -0.28, p = .78). Moreover, the cross-experiment stress-by-retrieval-practice interaction remained significant after controlling for both-correct triad count (b = 0.051, SE = 0.024, t = 2.17, p = .031, 95% CI [0.005, 0.098]). We have added this analysis to Section 3.2.2, “Retrieval practice protects direct memory retrieval from stress without affecting premise pairs counts. These additions make the available trial numbers transparent and indicate that the primary behavioral pattern was not explained by unequal numbers of premise-supported trials.

    (2) Additionally, one-tailed t-tests have been used in follow-up tests, and, as I understand it, no multiple comparisons corrections have been applied to the post-hoc tests, suggesting that these findings should be interpreted with caution.

    The one-tailed follow-up tests were used because the relevant hypotheses were directional a priori. Based on previous findings and our theoretical framework, we specifically predicted that acute stress would impair inference-related performance and neural reactivation, whereas retrieval practice would mitigate these effects. These follow-up tests were therefore planned comparisons used to decompose the significant omnibus effects in the predicted directions, rather than exploratory post-hoc comparisons. For this reason, we considered one-tailed testing appropriate. Similarly, we did not apply an additional multiple-comparison correction to these planned follow-up tests because they were limited in number, theory-driven, and designed to evaluate specific directional predictions rather than to search broadly across multiple possible contrasts. Importantly, our conclusions do not depend on any single follow-up comparison, but on the consistent pattern observed across behavioral inference performance, neural decoding of bridge-related category information, and theta-band activity. We have revised the manuscript to clarify the rationale for these analyses.

    (3) In places, the structure is unclear, making the narrative difficult to follow, often making it necessary for the reader to go back and forth between the sections to understand the study and analyses.

    We thank Reviewer 3 for this helpful suggestion. We have thoroughly reorganized the manuscript based on the detailed recommendations to improve the overall narrative flow and readability, ensuring that the study and its analyses are easy to follow.

    Reviewer #3 (Recommendations for the authors):

    (1) There is no mention of why the delay between learning and test is important. The delay may relate to consolidation, but the rationale is unclear.

    We added the rationale for the using the 24-hour delay between learning and test to the Introduction. First, separating Day 1 learning/retrieval practice from Day 2 stress induction and testing minimizes direct effects of stress on encoding and aligns the design with prior work on stress-related impairment of delayed retrieval (Kuhlmann et al., 2005; Schönfeld et al., 2014; Smith et al., 2016). Second, the delay allows the longer-term advantage of retrieval practice to emerge; prior work showed that repeated study can be advantageous on an immediate test, whereas retrieval practice produces better retention after longer delays (Roediger and Karpicke, 2006). The interval therefore permits assessment of memory after offline processing, sleep, and natural forgetting. We also state explicitly that consolidation was neither manipulated nor measured here, so the design cannot establish that consolidation caused the stress-buffering effect.

    (2) Additionally, it is unclear what happened between Day 1 and Day 2 (did you ask the participants to do anything between the sessions, e.g., refrain from alcohol, sleep at a normal time, if so, was this recorded, etc?).

    Participants received no additional task, intervention, or behavioral restrictions between sessions, and alcohol use and sleep schedules were not recorded. They returned to the laboratory approximately 24 hours after Day 1 under otherwise naturalistic conditions. We added this information explicitly to the Experimental Procedure section. We agree that monitoring sleep and substance use would improve control over potential intersession variability in future studies.

    (3) Section 2.5 Time-frequency analyses: The rationale for investigating theta power is unclear. Did the authors only investigate theta, or were other frequency bands often associated with memory retrieval also explored (e.g., alpha or beta)? If so, these should be reported. If not, the rationale should be clearer

    We clarified that theta was selected a priori because theta oscillations have been linked to associative memory, retrieval, hippocampal-prefrontal communication, and memory integration in associative-inference paradigms (including Backus et al., 2016). The time-frequency analysis therefore focused on theta power (2-8 Hz) rather than conducting an exploratory search across frequency bands. Alpha and beta bands were not analyzed, and this is now stated explicitly in the Methods. In the Discussion, we now relate the theta-band findings to the behavioral and decoding results, noting that acute stress reduced theta power during the later stage of inference, whereas retrieval practice prevented this reduction and enhanced theta activity during the earlier stage of inference.

    (4) 2.6.2 Analysis of Memory Inference Performance: There seems to be a disconnect between the hypothesis, which predicts that Experiment 2 conditions will be the same (i.e., confirming the null) and the independent t-test, which would tell you if the alternative can be accepted (rather than if it's a true null). Bayes or Equivalence tests would instead link to the hypothesis

    Because a non-significant frequentist t-test cannot directly provide evidence for the null hypothesis, we additionally conducted a Bayesian independent-samples t-test. The analysis yielded BF10 = 0.361 (BF01 = 2.77), providing evidence in support of the null hypothesis that memory inference performance did not differ between the stress and control groups in Experiment 2. We have added this analysis to the Methods and Results.

    (5) Line 508: It is not clear where the 207 +/- 50 ms time window came from.

    Thank you for identifying this error. The value 207 ms was a typographical mistake and the correct value is 198 ms. We rewrote the procedure in detail 2.6.4 Optimal time window identification and validation). In short, participant-level peak latencies were first identified in the localizer training data; their group mean was 197.46 ms (SD = 245.22 ms), which was rounded to define the primary 198 +/- 50 ms training window. An independent cross-task time-generalization analysis yielded a peak latency of 246.01 ms (SD = 220.08 ms). This did not differ significantly from the localizer-derived latency, t(133) = 1.66, p = .10, supporting use of the fixed 198 +/- 50 ms window. The Methods now describe all three steps and the corrected value consistently.

    (6) Section 3.5: It is not quite clear to me why the choice phase is of interest for the decoder. As I understand it, the participants are asked to select the image during this 3-second time window, meaning that they will be moving (artifacts). Therefore, I am unsure what question this analysis is addressing and would recommend clarification

    We clarified that our primary theoretical interest was in the pre-choice inference period, during which participants considered and integrated the relevant information before making a response. We were also interested, secondarily, in the choice period to examine whether bridge-related category reactivation persisted into response selection. Thus, the inference period served as our primary neural window of interest, with the choice period providing a complementary analysis.

    We agree that button presses create a risk of movement-related contamination. EEG trials were visually inspected, and ICA was used to remove ocular and movement-related components; nevertheless, we now interpret the choice-period findings cautiously.

    (7) The title boldly suggests that the restoration of the impairment is due to memory reactivation. Although the evidence hints that this may be the case, there could be many other factors contributing to the results, and the title can be misleading. I would therefore recommend changing the title to suggest a link with memory reactivation.

    Thank you for this suggestion. We have revised the title to “Retrieval practice prevents stress-induced inference impairment and preserves rapid bridge-related memory reactivation” to avoid implying a causal role of memory reactivation.

    Figure 1A: illustrates the stimuli as words, whereas the stimuli were images. To avoid confusion, I would recommend replacing the words with images (or adding images alongside).

    We have replaced the words with images in Figure 1A to accurately reflect the stimuli used in the experiment.

    Line 654: I presume the authors meant to say "memory inference phase" instead of "memory reference phase".

    Sorry for that. We apologize for this typographical error. We have corrected “memory reference phase” to “memory inference phase.”

    Specific comments to the authors are listed below in the order of the manuscript:

    For the remaining minor comments, we have made the corresponding revisions throughout the manuscript. To keep the response letter focused on the major scientific issues and avoid an overly lengthy point-by-point description of minor editorial changes, we do not elaborate on each of these changes here. All corresponding revisions can be found in the revised manuscript.

  6. eLife Assessment

    This study presents valuable findings on how retrieval practice protects memory inferences from stress via covert memory reactivation. Via two EEG experiments manipulating stress and retrieval practice, the authors provide solid evidence supporting the conclusion. This work will be of interest to cognitive and affective neuroscientists working on the intersection between memory and stress.

  7. Reviewer #1 (Public review):

    Summary:

    This manuscript examines whether retrieval practice protects memory-based inference from acute stress and proposes rapid neural reactivation of a bridging memory element as the underlying mechanism. Using a two-day associative inference paradigm combined with EEG decoding, the authors report that stress impairs inference accuracy and speed, while retrieval practice eliminates these deficits and restores neural signatures associated with bridge-element reactivation. The study addresses an important and timely question by integrating research on retrieval-based learning, stress effects on memory, and neural dynamics of inference. While the work provides promising multi-level evidence linking behavioral and neural findings, limitations in experimental design, causal interpretation, and decoding specificity weaken the strength of the mechanistic claims and suggest that further work is needed to disentangle strengthened associative memory from inference-specific protection effects

    Strengths:

    (1) Strong theoretical integration
    The study integrates three influential frameworks: memory integration through associative inference, stress-induced retrieval impairment, and the testing effect. The authors present a clear theoretical narrative linking these domains and derive testable hypotheses that retrieval practice protects inference by strengthening neural reactivation of a bridge element. The conceptual framing is well-grounded in prior literature and addresses an important gap regarding neural dynamics during inference.

    (2) Multi-level evidence
    The study provides converging behavioral and neural evidence. The authors demonstrate that stress reduces inference accuracy and speed, while retrieval practice eliminates these deficits. EEG decoding further suggests that bridge element reactivation predicts successful inference. The combination of behavioral performance and neural decoding strengthens the overall argument.

    (3) Transparent experimental implementation
    The procedures are described in substantial detail, including stimulus construction, stress manipulation, and decoding pipelines. Data and code availability are also strengths, facilitating reproducibility.

    Weaknesses:

    (1) Insufficient evidence that retrieval practice specifically protects inference rather than strengthening associative memories

    A central claim of the manuscript is that retrieval practice specifically protects inference ability rather than simply strengthening underlying associative memories. However, the current data do not convincingly distinguish between these possibilities. Although the authors limited analyses to trials in which AB and BC pairs were correctly retrieved in the subsequent memory test, this procedure does not fully rule out the possibility that improved inference performance reflects stronger base associative memories rather than enhanced integrative processes.

    Importantly, the direct memory retrieval test used a two-alternative forced-choice (2AFC) format, which inherently allows a substantial proportion of correct responses to arise from guessing. Consequently, trials classified as "successfully retrieved" may still include weak associative memory traces, making it difficult to conclude that failures in inference reflect deficits in integration rather than incomplete associative learning.

    The authors further argue that retrieval practice does not improve inference in the absence of stress, suggesting independence between inference and associative memory strength. However, this null effect does not sufficiently rule out mediation through strengthened premise memory. A factorial design and/or mediation analysis would be necessary to determine whether inference resilience emerges independently of premise memory strength.

    (2) Apparent below-chance inference performance raises interpretational concerns

    One surprising aspect of the results is that inference performance across experiments and groups appears to fall below the theoretical chance level (0.33) in Figure 4A. This is particularly unexpected because analyses were restricted to trials in which participants correctly retrieved both AB and BC associations.

    If performance is indeed below chance, this raises concerns regarding whether participants fully understood the task instructions or whether other methodological factors influenced performance. Additionally, below-chance performance complicates the interpretation of subsequent behavioral and neural analyses. It is possible that this reflects my misunderstanding of the figure; therefore, clarification from the authors regarding how inference accuracy is calculated and presented would be helpful.

    (3) Between-experiment implementation of retrieval practice weakens causal inference

    The retrieval practice manipulation was implemented as a separate experiment rather than as part of a factorial design. Experiment 2 was conducted after results from Experiment 1 were known, and the authors acknowledge this post hoc decision. This design introduces several potential confounds, including cohort differences between experiments, possible differences in participant motivation or task familiarity, and reduced ability to rigorously test interaction effects.

    Although the authors combined data across experiments to test interactions between stress and retrieval practice, such post hoc aggregation cannot fully substitute for a factorial design. A within-experiment 2 × 2 design (Stress × Retrieval Practice) would provide substantially stronger causal evidence and reduce confounding influences.

    (4) Lack of an appropriate comparison condition for retrieval practice limits the interpretation of the mechanism

    Although acknowledged briefly in the discussion, the absence of an appropriate comparison condition for retrieval practice represents a critical limitation. Without a matched re-exposure or restudy control condition, it remains unclear whether observed benefits are attributable specifically to retrieval practice or to additional exposure to AB and BC associations.

    Furthermore, it is unclear whether retrieval practice operates at the trial level or the participant level. Retrieval practice could enhance memory representations for specific practiced items, making those trials more resistant to stress, or it could induce a more global change in cognitive strategy or stress resilience across participants. One way to address this issue would be to analyze inference performance separately for trials that were successfully retrieved during the retrieval practice phase versus those that were not.

    (5) Interpretation of EEG decoding as bridge-element reactivation may be overstated

    The neural decoding results form the mechanistic foundation of the manuscript; however, the interpretation that decoding reflects reactivation of specific bridging memories may be overstated. The classifier distinguishes between face and building categories, and because the bridging element belongs to one of these categories, successful decoding may reflect category-level semantic activation rather than reinstatement of item-specific episodic representations.

    Alternative explanations include category-level retrieval, strategic task differences, or even attentional biases. Because only two categories were used, the decoding analysis lacks the specificity necessary to distinguish between category-level and item-level reactivation. As such, conclusions regarding the reinstatement of specific bridging memories should be tempered or supported with additional analyses.

  8. Reviewer #2 (Public review):

    Summary:

    Guo et al. investigate the neural and behavioral mechanisms of stress-induced impairments in memory-based inference. Across two well-powered experiments (N=136), the authors demonstrate that acute stress disrupts the rapid neural reactivation of "bridge" elements necessary for novel inferences. Crucially, they identify retrieval practice as a robust behavioral buffer that restores both inferential performance and the underlying neural signatures of memory reactivation.

    Strengths:

    (1) The use of two independent experiments provides high confidence in the behavioral findings.

    (2) Utilizing time-resolved EEG decoding allows the authors to pinpoint the "online" moment of inferential failure, a significant advancement over the lower temporal resolution of fMRI.

    Weaknesses:

    (1) The authors correctly timed the inference task to begin approximately 20 minutes after the onset of the stressor. While this window aligns with the expected peak of the glucocorticoid (HPA) response, it also represents a period where the rapid adrenergic (SAM) response, confirmed by heart rate elevation, is still highly influential. As the authors acknowledge, because they did not collect saliva samples due to safety protocols, they cannot definitively separate the influence of peak cortisol from the tail-end of the adrenergic surge on the observed memory impairments.

    (2) Figures 4 and 6: Without asterisks is really difficult to compare the significant group differences.

    Appraisal and Impact:

    This study provides high-quality evidence that acute stress impairs the rapid neural reactivation of "bridge" elements necessary for novel memory-based inferences. By leveraging the high temporal resolution of EEG decoding, the authors identify the specific neural "chokepoint" where inferential failure occurs. The research is strengthened by two independent experiments and the identification of retrieval practice as a powerful buffer that not only preserves but also enhances neural reactivation under pressure. The findings have significant implications for both cognitive neuroscience and applied learning science.

  9. Reviewer #3 (Public review):

    Summary:

    In this study, Guo and colleagues investigated the effects of stress and retrieval practice on memory inference. In the first experiment, they found that memory inference was significantly worse after induced stress. Conversely, when participants received retrieval practice in the second experiment, they found no significant differences between these conditions. They monitored EEG during the inference phase and applied multivariate decoding analysis to examine evidence of neural reactivation. Complementing the behavioural findings of the first experiment, they found that they were able to decode the stimulus category of the inference item with more fidelity in the no stress condition. Surprisingly, they found the opposite direction when participants had retrieval practice, with stronger evidence of reactivation in the stress condition than in the control condition.

    Strengths:

    (1) The authors have carefully designed two studies investigating the effects of stress and memory retrieval on memory inference.

    (2) The use of multivariate decoding on the inference phase data sheds new light on how stress and retrieval may impact the neural signatures of inference processing.

    Weaknesses:

    (1) There are some key gaps in the reporting of the data. In particular, data is missing on how many trials were included in the inference phase and how many were retrieved in the direct memory task. This is important to know as the main conclusions are based on inference trials proportional to the direct retrieval trials. Considering that the direct retrieval performance differs significantly between the experiments, there could be issues with floor/ceiling effects (in the behaviour) and statistical power (in the EEG results) that confound the comparisons between experiments. Without the data, it is difficult to draw conclusions.

    (2) There are some relatively strong conclusions drawn without the data to support them. An important example is the title suggesting a mechanistic role of memory reactivation for these effects; however, the data instead suggest a relationship between successful inference and evidence of reactivation. Additionally, one-tailed t-tests have been used in follow-up tests, and, as I understand it, no multiple comparisons corrections have been applied to the post-hoc tests, suggesting that these findings should be interpreted with caution.

    (3) In places, the structure is unclear, making the narrative difficult to follow, often making it necessary for the reader to go back and forth between the sections to understand the study and analyses. I have made some recommendations for how to improve this.

  10. Author response:

    Public reviews:

    Reviewer #1 (Public review):

    (1) We agree that the current design does not allow us to cleanly dissociate whether the beneficial effect of retrieval practice on AC inference under stress reflects a selective enhancement of inferential processing or, instead, stronger memory for the underlying AB and BC premise pairs that supports later inference. We plan to revise the manuscript to remove wording that could be read as claiming that retrieval practice specifically protects inference independently of associative-memory strengthening.

    Our intended interpretation is more modest. As shown in Section 3.2.3, retrieval practice improved direct premise-memory performance, consistent with the well-established testing effect. In the present paradigm, successful AC inference necessarily depends on access to the AB and BC premise associations. Accordingly, strengthened premise memory is not an alternative explanation that can be excluded by our data, but rather a plausible mechanism through which retrieval practice may promote more resilient inference performance under stress.

    Because AC inference in our paradigm necessarily depends on retrieving and linking the AB and BC premise pairs, strengthened premise memory is not merely a competing explanation that can be separated from inference performance in the current dataset. Rather, it is a plausible mechanism through which retrieval practice may support inference, especially under stress. We therefore will revise the manuscript to avoid implying that retrieval practice protects inferential processing independently of associative-memory strengthening, and instead interpret the effect more conservatively as reflecting enhanced premise representations and/or more effective reactivation of bridge information during inference.

    We also agree that the post-inference direct memory test, which used a 2AFC format, provides only a coarse measure of premise-memory strength and allows some proportion of correct responses to arise from guessing. Therefore, restricting analyses to trials in which AB and BC were later answered correctly does not fully guarantee that those trials were supported by strong associative memories. We will acknowledge this limitation explicitly in the manuscript and have tempered our interpretation of these “successfully retrieved” premise trials accordingly. More stringent measures, such as cued recall, confidence-based memory judgments, or other continuous indices of premise-memory strength, would be better suited to this question in future work.

    Finally, we agree that the absence of a retrieval-practice benefit in the non-stress condition does not by itself rule out mediation through strengthened premise memory. Because the retrieval-practice manipulation was introduced in a follow-up study after completion of Study 1, the present dataset was not designed as a single fully crossed factorial experiment. In response to the reviewer’s suggestion, we will add an exploratory mediation analysis testing whether premise-memory performance statistically accounts for the relationship between retrieval practice and inference performance. We will report this analysis cautiously, given that premise memory was assessed using a post-inference 2AFC measure, and we note in the manuscript that a future fully crossed design with more sensitive premise-memory measures will be needed for a stronger test.

    (2) We apologize that the presentation of Figure 4A was not sufficiently clear and may have created the impression of below-chance inference performance. The values shown in Figure 4A do not represent raw 3-alternative forced-choice (3AFC) A-C inference accuracy, for which the theoretical chance level would be 0.33. Instead, Figure 4A plots a normalized inference index, calculated as inference performance relative to direct retrieval performance, to account for individual differences in the availability of the directly learned premise pairs. Therefore, the raw 3AFC chance level is not the appropriate reference for interpreting this measure. To avoid this confusion, we will clarify in the revised manuscript and figure legend that Figure 4A shows a normalized inference index rather than raw inference accuracy.

    (3) We agree that implementing retrieval practice in a separate experiment, rather than within a single 2 × 2 factorial design, limits the strength of the causal inference regarding retrieval practice and reduces our ability to formally test the retrieval practice × stress interaction within one unified design.

    In response, we will revise the manuscript to more explicitly acknowledge this limitation and to temper our interpretation throughout. Specifically, we now avoid overstating retrieval practice as definitively preventing the effects of stress, and instead describe the findings more cautiously as evidence that retrieval practice was associated with attenuation of stress-related inference impairments across experiments. We also will add a limitation statement in the Discussion noting that the current design cannot fully rule out cohort-related confounds and that a fully crossed factorial design will be necessary in future work to provide a more rigorous test of the interaction between retrieval practice and stress.

    At the same time, we have clarified that the two experiments were conducted under closely matched conditions: participants were recruited using the same protocol from the same campus population, demographic characteristics were matched, and both experiments were run in the same laboratory using the same EEG system, task procedures, and experimenter team. We agree, however, that these procedural consistencies reduce but do not eliminate the concern about between-experiment confounds.

    (4) We agree that the absence of a matched re-exposure/restudy control condition limits the mechanistic interpretation of the retrieval-practice effect. In the revised manuscript, we will make this limitation more explicit in the Discussion and temper our conclusions accordingly. Specifically, we clarify that the present design shows that a post-encoding retrieval-practice intervention buffered the impact of acute stress on later inference, but it does not allow us to determine whether this benefit is specific to retrieval practice per se, rather than to additional exposure to the AB and BC associations.

    We also agree that it is important to distinguish whether the effect operates at the level of specific practiced items or reflects a more global participant-level effect. In the current study, however, the retrieval-practice phase in Experiment 2 was implemented as a brief timed free-recall procedure rather than a trial-by-trial cued retrieval task, and the available records do not allow us to reliably link retrieval-practice success for individual associations to specific later AC inference trials. Therefore, we cannot directly compare later inference performance for successfully versus unsuccessfully retrieved items on a trial-by-trial basis.

    To address this issue as far as possible with the current dataset, we instead plan to conduct an additional item-level robustness analysis using mixed-effects models that accounted for variability across ABC associations. Specifically, we tested whether the critical stress-by-retrieval-practice effect remained after modeling triad-level variability, and whether there was evidence that this effect differed substantially across triads. This analysis does not provide a direct test of whether successfully retrieved items benefit more than unsuccessfully retrieved items, but it does help assess whether the observed effect is broadly distributed across associations or driven by only a small subset of items.

    (5) We agree that our current decoding approach does not justify a strong claim of item-specific reinstatement of a unique bridge memory. The classifier was trained to discriminate stimulus categories (faces vs. buildings) in the independent localizer and then applied during the inference phase. Therefore, the present analysis is better interpreted as indexing reactivation of bridge-related category information, rather than reinstatement of an item-specific episodic representation.

    Importantly, however, we believe this signal remains theoretically informative for the inferential process examined here. In our design, the bridge element B belonged to one of the trained categories, and the classifier was applied during the cue period when no face or building stimulus was physically present. Thus, successful decoding in this time window suggests that task-relevant bridge-related information was re-expressed online during inference, rather than reflecting concurrent perceptual processing. At the same time, we agree that, because only two categories were used, the decoding analysis cannot fully dissociate bridge-related category reactivation from broader category-level retrieval, strategic task differences, or attentional contributions.

    To address this concern, we plan to revise the manuscript in three ways. First, we will soften the interpretation throughout the Results and Discussion to avoid claims of item-specific bridge-memory reinstatement. Second, we now refer to the decoding effect more conservatively as bridge-related or category-level mnemonic reactivation during inference. Third, we have added an explicit limitation stating that the current design does not allow us to distinguish item-specific episodic reinstatement from category-level reactivation, and that future work using more fine-grained representational analyses and/or a larger stimulus set will be needed to resolve this issue more directly.

    Reviewer #2 (Public review):

    (1) We agree with this important point. The inference task was scheduled to begin approximately 20 minutes after stress onset based on prior human stress literature, with the intention of probing a time window commonly associated with glucocorticoid effects. However, as the reviewer notes, this period may also still reflect residual adrenergic/SAM influences. Because salivary cortisol was not collected due to the COVID-19-related safety protocol, we cannot disentangle the relative contributions of glucocorticoid and adrenergic responses to the observed stress-related effects on inference and neural reactivation. We will revise the manuscript to make this limitation more explicit in the Discussion and to avoid attributing the effects to a specific physiological component of the stress response.

    (2) In the revised manuscript, we will add asterisks (or equivalent significance annotations) to Figures 4 and 6 to improve clarity and readability.

    Reviewer #3 (Public review):

    (1) We thank the reviewer for highlighting this important reporting issue. We agree that the number of trials contributing to the behavioral and EEG analyses should be reported more explicitly, particularly because inference performance was analyzed in relation to direct retrieval performance and because direct retrieval differed across experiments.

    In the revised manuscript, we will report, for each group and experiment, the number of trials presented in the AC inference phase, the number of trials retained for the behavioral analyses, and the number of successfully retrieved direct-memory trials in the AB and BC tasks. These values will be summarized in the revised Results section and in Supplementary Tables.

    To directly address the reviewer’s concern, we will also compared trial counts across groups/experiments and evaluated whether differences in direct retrieval performance could account for the inference and EEG effects. To further address the concern about potential unequal trial numbers, we plan to repeat the analyses such as trial-count-matched subsets analyses to see whether results remained qualitatively unchanged.

    (2) We thank the reviewer for this important comment. We agree that our original title and some parts of the manuscript used language that was stronger than warranted by the data. Our results show that rapid reactivation of the bridge element is associated with successful inference and is modulated by stress and retrieval practice, but they do not by themselves establish a causal mechanistic role for reactivation. We therefore plan to revise the title and softened the relevant wording throughout the manuscript to better reflect the correlational nature of this evidence.

    Specifically, we plan to change the title from “Retrieval practice prevents stress-induced inference impairment by restoring rapid memory reactivation” to “for example, Retrieval practice prevents stress-induced inference impairment and preserves rapid bridge-item memory reactivation” We also revised the Abstract, Results, and Discussion to replace stronger mechanistic wording such as “prevents,” “restoring,” and “essential neural mechanism” with more cautious phrasing such as “buffers” or “attenuates,” “preserves” or “is associated with,” and “neural correlate” or “candidate process,” as appropriate. This revision will led us to temper the overall interpretation of the EEG findings: rather than claiming that reactivation is the mechanism by which retrieval practice prevents stress-related inference deficits, we now conclude that rapid bridge-item reactivation is a neural correlate of successful inference that is sensitive to stress and enhanced by retrieval practice.

    We also appreciate the reviewer’s concern regarding the use of one-tailed follow-up tests and the absence of multiple-comparison correction. With respect to the one-tailed t-tests, these follow-up comparisons were conducted because the relevant hypotheses were directional a priori. Based on prior work and our theoretical framework, we specifically predicted that acute stress would impair inference-related performance and neural reactivation, and that retrieval practice would mitigate these effects. The follow-up tests were therefore not exploratory post-hoc comparisons, but planned tests used to decompose the significant omnibus effects in the predicted direction. For this reason, we considered one-tailed testing appropriate for these comparisons.

    Similarly, we did not apply an additional multiple-comparison correction to these planned follow-up tests because they were limited in number, theory-driven, and conducted to evaluate specific directional predictions rather than to search broadly across many possible contrasts. Importantly, our interpretation does not depend on any isolated post-hoc comparison, but on the consistency of the results across behavioral inference measures, neural decoding of bridge-item reactivation, and theta-band analyses. We have revised the manuscript to make this rationale clearer and to ensure that the follow-up results are interpreted in the context of the full pattern of evidence.

    (3) We agree that, in the previous version, parts of the manuscript were not structured clearly enough, which may have made it difficult for readers to follow the logic of the study and the sequence of analyses without moving back and forth across sections. In the revised manuscript, we will reorganize the presentation to improve the overall narrative flow and readability. Specifically, we plan to clarify the study logic and analysis sequence, strengthened transitions between sections, and revised the relevant text in line with the #reviewer3’s detailed suggestions.

  11. Authors used AB in their study. Including #RRIDs will make this less ambiguous. SciScore made a table with this resource, see “Automated Services” module (download as csv, xml or #jats) #RRID #methodsmatter

  12. Authors published a paper in Biorxiv, they used MAGE in the study. Including #RRIDs will make this less ambiguous. SciScore made a table with this resource, see “Automated Services” module (download as csv, xml or #jats) #methodsmatter #STMpublishing

  13. Retrieval practice prevents stress-induced inference impairment by restoring rapid memory reactivation https://www.biorxiv.org/content/10.1101/2025.03.14.643187v1

  14. Retrieval practice prevents stress-induced inference impairment by restoring rapid memory reactivation https://www.biorxiv.org/content/10.1101/2025.03.14.643187v1