Complementary stable and dynamic prelimbic ensembles encode learned threat value underlying generalization and discrimination

  1. Marc E Normandin
  2. Pedro M Ogallar
  3. Matthew R Lopez
  4. Manuel M Ramos-Alvarez
  5. Isabel A Muzzio

  • Curated by eLife

    eLife logo

    eLife Assessment

    This valuable study examines how the rodent prelimbic cortex represents learned and generalized threat over time and identifies distinct stable and dynamic neuronal populations that contribute to these representations. The evidence is convincing, supported by longitudinal calcium imaging, appropriate control groups, and sophisticated analyses showing that the relevant neural signals cannot be explained simply by freezing behavior. The work provides a conceptual framework for understanding how stable threat-related representations supporting memory generalization and discrimination can be maintained despite ongoing changes in neuronal ensemble composition.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Adaptive behavior requires assigning emotional value to sensory cues and generalizing to similar stimuli. The prelimbic cortex (PL) is critical for threat expression and discrimination, yet its neuronal ensembles undergo substantial turnover over time. How stable population codes emerge despite this reorganization remains unknown. Using longitudinal calcium imaging in freely moving mice, we tracked PL activity for 30 days during auditory discriminative fear learning and retrieval. Despite substantial ensemble turnover, stable graded representations persisted in PL populations. A generalized linear model showed that the tone threat value explained graded activity better than freezing, indicating that PL primarily encodes stimulus significance. Population similarity across stimulus presentations was highest for shock-associated and highly generalized tones, suggesting a stable population code underlying threat generalization. Stable neurons preserved graded gradients across sessions, whereas learning modulated the rate of dynamic frequency-selective neurons. Together, these populations enabled stable threat-value coding despite ongoing ensemble reorganization.

Article activity feed

  1. Mentioned by @bertogliolab.bsky.social on Bluesky

    Complementary stable and dynamic prelimbic ensembles encode learned threat value underlying generalization and discrimination elifesciences.org/articles/111...

  2. Version published to 10.7554/elife.111177.4 on eLife
  3. Version published to 10.7554/elife.111177 on eLife
  4. Version published to 10.7554/elife.111177.3 on eLife
  5. eLife

    eLife Assessment

    This valuable study examines how the rodent prelimbic cortex represents learned and generalized threat over time and identifies distinct stable and dynamic neuronal populations that contribute to these representations. The evidence is convincing, supported by longitudinal calcium imaging, appropriate control groups, and sophisticated analyses showing that the relevant neural signals cannot be explained simply by freezing behavior. The work provides a conceptual framework for understanding how stable threat-related representations supporting memory generalization and discrimination can be maintained despite ongoing changes in neuronal ensemble composition.

  6. eLife

    Reviewer #1 (Public review):

    [Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. We appreciate the care with which the authors have addressed the remaining concerns. The additional analyses and the revisions to the framing and interpretation have strengthened the manuscript and clarified several of the issues raised during the previous round of review. In particular, the added analyses in Figure 4 and the more precise treatment of repeated retrieval, learned threat value, and the relationship between neural activity and freezing have improved the paper substantially.]

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a no-shock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure.

    To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences.

  7. eLife

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under examined in the field.

  8. eLife

    Author response:

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

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a noshock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure.

    To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences. However, in its current form, the manuscript does not yet fully support the strength of the conceptual claims. Several issues limit confidence in the interpretation, including the possibility that repeated testing itself contributes to changes across days, uncertainty about the relationship between neural activity and freezing behavior, limited quantitative documentation of longitudinal cell registration, and a number of problems in figure clarity and statistical framing. Overall, the study contains promising observations, but the claims should be narrowed, and several analyses or controls would be needed to fully support the proposed framework.

    Comments on revised version.

    The authors have addressed my previous concerns well, and the revised manuscript is substantially improved. In particular, the additional analyses strengthen the conclusion that prelimbic cortical activity reflects learned threat value rather than simply freezing behavior, while the revised framing and additional controls clarify the interpretation of the longitudinal neural dynamics. This paper represents an important contribution to our understanding of the neural mechanisms supporting aversive learning, memory, and generalization.

    We thank Reviewer 1 for his/her constructive comments, which significantly contributed to strengthen our conclusions.

    In response to the reviewer’s recommendations, we added two new analyses to Fig. 4, presented in Fig. 4c–g.

    Reviewer #2 (Public review):

    The authors have substantially revised the paper in response to the original review, which is greatly appreciated. It is clear that it will eventually make a nice contribution to the literature. This being said, the following points are somewhere between major and minor in term of their implications for interpretation of the study results. If they were to be addressed, the paper would be again improved.

    We thank Reviewer 2 for the careful second review, which helped us sharpen our arguments and improve the manuscript. The reviewer’s additional comments prompted us to refine the scope and precision of several interpretations. Accordingly, we have removed or clarified potentially confusing language, reframed the longitudinal findings in terms of repeated retrieval rather than the passage of time, and explicitly acknowledged the limitations of relying exclusively on freezing as a behavioral measure. We now also recognize repeated nonreinforced testing as the most direct explanation for the sharpening of the generalization gradients and the responses of newly recruited neurons, while clarifying below how we view this additional learning as contributing to memory updating. We believe these revisions have substantially improved the manuscript, and we are grateful for the reviewer’s thoughtful and detailed feedback.

    There are a few remnants of the past language that are not helpful for the interpretation of the study results: 1) "Specifically, the observed population gradients could emerge either from the pooled activity of frequency-selective neurons that respond to individual tones or from neuronal subpopulations that integrate information across tones to encode their learned threat-value."; and 2) "Together, these findings suggest that the PL integrates sensory similarity with learned threat value to generate stable representations that support adaptive generalization and discrimination." Neither of these statement follows what has been shown in the study, even with inclusion of the results from the GLM analysis (see point 4 below).

    We directly address this point in response to the reviewer’s point 4

    (1) This paragraph in the Discussion is difficult to follow: "Generalization has traditionally been explained by perceptual similarity (Shepard, 1987), whereby stimuli resembling a conditioned cue recruit overlapping sensory representations and evoke similar behavioral responses (Corches et al., 2019; Grosso et al., 2018). Although perceptual similarity clearly influences the extent of generalization, accumulating evidence indicates that it cannot fully account for generalized responding (Verra et al., 2026). More recent frameworks propose that associative learning assigns learned value to novel stimuli by integrating their sensory similarity with previous experience, allowing behavior to scale according to predicted biological significance (Verra et al., 2026; Zaman et al., 2023). Our findings provide a neural framework consistent with these ideas. Sensory similarity promoted consistent neuronal population responses across tones, whereas associative learning organized these responses into graded representations that tracked learned threat value across the stimulus continuum. Thus, sensory similarity appears to define the neuronal substrate upon which associative learning constructs value-based representations that support graded behavioral generalization."

    While the revisions have removed the many unnecessary references to inference and integration, this paragraph seems like it is adhering to the original idea of how the authors wished to present their work. If the authors wished to talk about something more than perceptual similarity in the context of generalization, they should have used a task that lends itself to a more-than-perceptual-similarity explanation. Again, the inclusion of the GLM analysis is suggestive for some of what the authors wish to say, but doesn't justify the statements that: "Sensory similarity promoted consistent neuronal population responses across tones, whereas associative learning organized these responses into graded representations that tracked learned threat value across the stimulus continuum." In short, the analysis does not substitute for the design that could have and should have been used to assess learned threat value independently of sensory similarity.

    We thank the reviewer for this comment, which helped us identify ambiguity in our terminology. By “learned threat value,” we did not intend to imply that the PL independently infers or computes the value of each novel tone, separate from perceptual similarity. Rather, we used this term to denote the associative significance acquired by the CS+ and CS− during conditioning. Novel cues may then be evaluated according to their perceptual similarity to these learned associations. We have now defined “learned threat value” when it first appears in the manuscript.

    Furthermore, we agree that perceptual similarity can account for the graded generalization of responses from the CS+ to neighboring frequencies. We therefore do not consider learned associations and perceptual similarity to be competing explanations. Instead, conditioning determines which stimulus anchors the gradient and its direction according to their acquired “threat value”, whereas perceptual similarity governs how the response extends across intermediate frequencies. This interpretation is supported by our reversed conditioning design: the same frequency continuum produced opposite neural gradients depending on whether 3 or 15 kHz served as the CS+, and no gradients emerged in control animals exposed to the same tones without conditioning. Thus, the physical relationships among the tones shape the graded response, but mnemonic experience determines its organization along the frequency continuum.

    The GLM analysis was not intended to demonstrate threat value independently of sensory similarity. Rather, it showed that the learning-dependent gradient remained after accounting for freezing and therefore could not be explained solely by this behavioral response (we address the point of other potential fear responses below).

    “Generalization has traditionally been explained in terms of perceptual similarity (Shepard, 1987): stimuli resembling a conditioned cue recruit overlapping sensory representations and therefore evoke similar behavioral responses (Corches et al., 2019; Grosso et al., 2018). More recent frameworks propose that responses to novel stimuli depend not only on sensory similarity but also on mnemonic representations of learned threat value, allowing behavior to scale with the predicted significance of each stimulus (Verra et al., 2026; Zaman et al., 2023). This mnemonic contribution should be especially important when generalization is assessed long after conditioning because responding to a novel cue requires retrieval of the associations formed during learning. Our design did not manipulate sensory similarity and learned threat value independently. However, the reversed conditioning procedure dissociated them in one respect: depending on whether 3 or 15 kHz served as the CS+, the neural gradients ran in opposite directions along the same frequency continuum, with each anchored to the threat-associated tone. A gradient determined by spectral features alone would have run in the same direction in both groups. Thus, the learned associations and their mnemonic representations must have determined the anchor and direction of the neural gradient. By contrast, the contribution of sensory similarity is inferred rather than directly demonstrated: intermediate tones were represented in an order that followed their spectral distance from the CS+. Because discriminative conditioning tends to narrow generalization gradients (Dunsmoor & LaBar, 2013; Herzog et al., 2021; Jenkins & Harrison, 1960; Lommen et al., 2017), tracking neural representations over 30 days allowed us to examine how PL populations evolved as behavioral discrimination became progressively more precise during repeated retrieval. Whereas stable neuronal populations retained consistent response profiles across retrieval sessions, neurons recruited after conditioning exhibited changes that paralleled the behavioral sharpening. These results suggest that memory, new learning, and sensory similarity interact to shape generalization gradients.”

    (2) The next paragraph in the Discussion is also confusing. "Such reorganization has been proposed to provide flexibility by allowing new information to be incorporated into existing cortical representations while preserving stable behavioral performance (Mau et al., 2020; Zaki & Cai, 2024). Several mechanisms could contribute to this turnover, including systems consolidation, retrieval-induced reconsolidation or memory updating, and repeated nonreinforced stimulus exposure (Lacagnina et al., 2019; Mau et al., 2020; Sangha, 2015; Zaki & Cai, 2024). Although our experiments cannot distinguish between the first two possibilities, the behavioral data argue against extinction as the primary explanation. Extinction is generally associated with the formation of new CS+-safety associations (Bouton et al., 2021), whereas discrimination ratios increased across retrieval sessions, indicating that animals progressively improved their discrimination between threat-associated and safe stimuli rather than acquiring generalized safety responses. This pattern is consistent with previous work showing that discrimination learning sharpens stimulus representations and narrows behavioral generalization gradients (Dunsmoor & LaBar, 2013; Herzog et al., 2021; Jenkins & Harrison, 1960; Lommen et al., 2017). Importantly, turnover was not uniform across the population. Graded neurons retained remarkably consistent response profiles across retrieval sessions, and their activity remained more strongly associated with learned threat value than with freezing behavior. These observations indicate that stable components of the population code can coexist with extensive reorganization of surrounding neuronal ensembles."

    The issue with repeated testing is not caused by extinction per se. The issue is that nonreinforcement across the repeated testing should differentially affect the CS+ and CS-. Specifically, it should extinguish responding to the CS- stimulus at a rate that matches its distance from the CS+, thereby sharpening the CS+ versus CS- discrimination in precisely the ways that have been observed. Ergo, the repeated testing is a problem for inferences that might be drawn about the way that generalization gradients change with; and is a problem for statements regarding "dynamic reorganization of cortical activity patterns over time." There is nothing in the study that allows one to comment on the reorganization of cortical activity patterns over time. The reorganization can and should be attributed to the repeated testing, which is confounded with time. Nonetheless, the reorganization must be due to the repeated testing and NOT time as the present findings are inconsistent with the well-documented broadening of generalization gradients with time.

    We agree with the reviewer that discrimination between the CS− and CS+ sharpened across retrieval sessions and that repeated testing may have contributed to this effect. Our discussion of extinction was included specifically to address a concern previously raised by Reviewer 1 and to clarify that the observed sharpening did not reflect a uniform reduction in conditioned responding. We now only left this explanation in the results section.

    We also agree that later retrieval sessions are not passive readouts of the original memory. Repeated presentation can provide new opportunities for learning, update reactivated memory traces, and initiate reconsolidation. The intermediate tones illustrate this point particularly well: they are novel during the first retrieval session but become familiar through subsequent exposure, and their representations may therefore change across sessions. Thus, new learning, memory updating, and reconsolidation are not mutually exclusive explanations but interacting processes engaged by repeated retrieval. We have revised the Discussion to acknowledge these contributions more explicitly.

    We also agree that our use of phrases such as “over time” was imprecise. We have replaced this language with “across retrieval sessions” to avoid implying that the passage of time alone produced ensemble turnover.

    However, we respectfully disagree that cortical reorganization cannot be evaluated across repeated retrieval sessions or that repeated testing alone necessarily accounts for all the observed changes. Multiple studies have documented turnover within cortical ensembles following learning (Lacagnina et al., 2019; Mau et al., 2020; Sangha, 2015; Zaki & Cai, 2024). Importantly, we observed ensemble turnover by test day 1, before repeated testing could have exerted a cumulative effect, and similar early turnover has been reported by other laboratories (Kitamura et al., 2017; DeNardo et al., 2019). These previous studies have also interpreted changes in ensemble composition between conditioning and subsequent retrieval sessions, including remote retrieval, as evidence of cortical reorganization. Our paradigm differs from these studies because we introduced novel tones to examine how representations of the original CS+/CS− associations interact with representations of perceptually similar cues.

    In summary, we agree with the reviewer that repeated testing may generate additional learning that contributes to the sharpening of behavioral generalization gradients and the responses of newly recruited neurons. However, we view this learning as inseparable from memory updating and reconsolidation, because information acquired during retrieval must be incorporated into existing memory representations to persist across sessions.

    “The ensemble turnover observed here is consistent with previous studies demonstrating dynamic reorganization of cortical activity patterns across temporally separated retrieval sessions (DeNardo et al., 2019; Gallego et al., 2020; Kitamura et al., 2017; Tome et al., 2024). Such reorganization has been proposed to provide flexibility by allowing new information to be incorporated into existing cortical representations while preserving stable behavioral performance (Mau et al., 2020; Zaki & Cai, 2024). Several processes could contribute to this turnover, including additional learning during repeated nonreinforced testing, memory updating, and reconsolidation (Lacagnina et al., 2019; Mau et al., 2020; Sangha, 2015; Zaki & Cai, 2024). In our experiments, turnover was already evident at the first retrieval session, before repeated testing could have exerted a cumulative effect, indicating that this reorganization emerges early after learning. Nevertheless, repeated testing provides a plausible explanation for changes during subsequent sessions. Each retrieval session reactivated the original CS+/CS− associations while presenting the conditioned and intermediate tones without reinforcement, thereby creating opportunities for additional discrimination learning. These experiences may have progressively sharpened differentiation between the CS+ and CS-, modifying the retrieved memory representations and changing which neurons participated in representing the refined associations. Although our data do not establish a mechanistic link between neuronal turnover and additional learning, they raise the possibility that dynamic ensemble membership provides the flexibility needed to incorporate new information.”

    (3) In the next paragraph, the authors state: "At the same time, narrower generalization gradients and improved discrimination across retrieval sessions suggests ongoing memory updating. These observations are consistent with contemporary theories proposing that systems consolidation and retrieval-dependent updating are complementary processes through which memories continue to evolve after learning (Mau et al., 2020; Tome et al., 2024; Zaki & Cai, 2024)."

    In general, I'm not sure why one would invoke systems consolidation or retrieval-induced reconsolidation as an explanation for any of the present findings: they are not explanations of much at all. In this specific text, the authors seem to be implying an updating process that occurs independently of what is learned across the repeated sessions of testing. Why? The changes that occur in the behaviour and neuronal representations are perfectly explicable in terms of additional learning that occurs - of the sort that I hope to have made clear in my previous comment. Why invoke more than what is needed to explain the observed pattern of results?

    We appreciate the reviewer’s clarification. As discussed in our response to the preceding comment, we agree that repeated nonreinforced testing provided additional learning opportunities and offers the most direct explanation for the progressive sharpening of behavioral discrimination and neuronal responses. We recognize that our original wording could have been interpreted as proposing memory updating as a process occurring independently of learning during the test sessions. This was not our intention. By “memory updating,” we meant that new information acquired during repeated retrieval modified the existing representations of the CS+ and CS−, allowing these stimuli to become more clearly differentiated. Thus, the additional learning that occurs through repeated testing provides the experience through which the original associations are refined.

    We agree that systems consolidation and retrieval-induced reconsolidation are not required as direct explanations for the observed sharpening and that our design cannot isolate their contributions. However, ensemble turnover was evident by day 1, before repeated testing could have exerted cumulative effects, indicating that repeated testing cannot fully explain the turnover. We view the additional learning acquired during subsequent retrieval sessions as refining the represented values of the CS+ and CS− and, if these changes persist, as requiring incorporation into existing memory representations through updating and reconsolidation. These processes therefore complement, rather than compete with, additional learning. The revised Discussion distinguishes repeated testing as the most direct explanation for the sharpening of discrimination from the broader memory processes that may accompany this learning and contribute to cortical ensemble reorganization.

    “In summary, PL population responses formed generalization gradients that paralleled behavioral generalization and were primarily organized according to tone-learned threat value. The overlap among responses to conditioned and novel tones was consistent with their perceptual similarity, whereas repeated nonreinforced testing was accompanied by progressive sharpening of both behavioral discrimination and the response gradients expressed by neurons recruited after conditioning. Together, these findings provide a potential neural framework through which stored threat associations may guide the evaluation of perceptually similar cues.”

    “The coexistence of these stable neurons with dynamic ensembles may allow repeated retrieval to refine a memory while preserving core features of its original representation (Mau et al., 2020; Tome et al., 2024; Zaki & Cai, 2024). Determining whether these stable graded neurons are causally required for memory storage or retrieval—and whether they constitute a persistent cortical memory trace—will require longitudinal imaging combined with selective manipulation of this population.”

    (4) Re the GLM analysis - The authors write that: "the fact that the GLM analysis indicates that these neurons reflect learned threat value more than freezing behavior, suggests that they encode an abstract property of the learned stimulus rather than simply mirroring behavioral output."

    This is fine if freezing fully indexes the state of conditioned fear and there are no other behaviours in which animals express their fear. If, however, fear is expressed in a range of other behaviours that are likely coordinated by the PL (e.g., startle, vigilance, scanning, orienting to source of danger), this interpretation of the GLM analysis is unwarranted. This is an important point and would be worth noting somewhere in the paragraph where the statement appears.

    We agree with the reviewer. Our GLM demonstrates that the graded neuronal responses cannot be explained by freezing behavior alone, but it does not exclude contributions from other unmeasured manifestations of conditioned fear, including vigilance, scanning, orienting, startle, or physiological responses. We have therefore removed the claim that these neurons primarily encode threat value and revised the Discussion to acknowledge this limitation explicitly.

    “Freezing, however, does not exhaust the conditioned defensive state, and we cannot rule out contributions from unmeasured behavioral or physiological correlates of fear, including vigilance, scanning, orienting, startle, and changes in autonomic state. Our findings are therefore consistent with PL neurons representing the learned significance of the stimuli, but they do not establish that this representation is independent of all fear-related behavioral and physiological states.”

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under examined in the field.

    Comments on revised version.

    The authors have convincingly and thoroughly addressed my concerns. I have no further issues regarding this study.

    We thank reviewer 3 for his thoughtful and constructive comments and the suggestion to use the GLM which greatly improved the interpretation of our data.

    Recommendations for the authors:

    Reviewer #1 (Recommendations for the authors):

    One small point is that the authors could report the beta coefficients for the tone and freezing predictors from the GLM analysis shown in Figure 4. Providing these values would make it easier to directly compare the relative contributions of tone-related and freezing-related activity to the model.

    We now include this analysis in Figure 4c and 4g

  9. eLife

    eLife Assessment

    This valuable study examines how the rodent prelimbic cortex represents learned and generalized threat over time and identifies distinct stable and dynamic neuronal populations that contribute to these representations. The evidence is convincing, supported by longitudinal calcium imaging, appropriate control groups, and sophisticated analyses showing that the relevant neural signals cannot be explained simply by freezing behavior. The work provides a conceptual framework for understanding how stable threat-related representations supporting memory generalization and discrimination can be maintained despite ongoing changes in neuronal ensemble composition.

  10. eLife

    Reviewer #1 (Public review):

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a no-shock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure.

    To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences. However, in its current form, the manuscript does not yet fully support the strength of the conceptual claims. Several issues limit confidence in the interpretation, including the possibility that repeated testing itself contributes to changes across days, uncertainty about the relationship between neural activity and freezing behavior, limited quantitative documentation of longitudinal cell registration, and a number of problems in figure clarity and statistical framing. Overall, the study contains promising observations, but the claims should be narrowed, and several analyses or controls would be needed to fully support the proposed framework.

    Comments on revised version.

    The authors have addressed my previous concerns well, and the revised manuscript is substantially improved. In particular, the additional analyses strengthen the conclusion that prelimbic cortical activity reflects learned threat value rather than simply freezing behavior, while the revised framing and additional controls clarify the interpretation of the longitudinal neural dynamics. This paper represents an important contribution to our understanding of the neural mechanisms supporting aversive learning, memory, and generalization.

  11. eLife

    Reviewer #2 (Public review):

    The authors have substantially revised the paper in response to the original review, which is greatly appreciated. It is clear that it will eventually make a nice contribution to the literature. This being said, the following points are somewhere between major and minor in term of their implications for interpretation of the study results. If they were to be addressed, the paper would be again improved.

    There are a few remnants of the past language that are not helpful re interpretation of the study results: 1) "Specifically, the observed population gradients could emerge either from the pooled activity of frequency-selective neurons that respond to individual tones or from neuronal subpopulations that integrate information across tones to encode their learned threat-value."; and 2) "Together, these findings suggest that the PL integrates sensory similarity with learned threat value to generate stable representations that support adaptive generalization and discrimination." Neither of these statement follows what has been shown in the study, even with inclusion of the results from the GLM analysis (see point 4 below).

    (1) This paragraph in the Discussion is difficult to follow: "Generalization has traditionally been explained by perceptual similarity (Shepard, 1987), whereby stimuli resembling a conditioned cue recruit overlapping sensory representations and evoke similar behavioral responses (Corches et al., 2019; Grosso et al., 2018). Although perceptual similarity clearly influences the extent of generalization, accumulating evidence indicates that it cannot fully account for generalized responding (Verra et al., 2026). More recent frameworks propose that associative learning assigns learned value to novel stimuli by integrating their sensory similarity with previous experience, allowing behavior to scale according to predicted biological significance (Verra et al., 2026; Zaman et al., 2023). Our findings provide a neural framework consistent with these ideas. Sensory similarity promoted consistent neuronal population responses across tones, whereas associative learning organized these responses into graded representations that tracked learned threat value across the stimulus continuum. Thus, sensory similarity appears to define the neuronal substrate upon which associative learning constructs value-based representations that support graded behavioral generalization."

    While the revisions have removed the many unnecessary references to inference and integration, this paragraph seems like it is adhering to the original idea of how the authors wished to present their work. If the authors wished to talk about something more than perceptual similarity in the context of generalization, they should have used a task that lends itself to a more-than-perceptual-similarity explanation. Again, the inclusion of the GLM analysis is suggestive for some of what the authors wish to say, but doesn't justify the statements that: "Sensory similarity promoted consistent neuronal population responses across tones, whereas associative learning organized these responses into graded representations that tracked learned threat value across the stimulus continuum." In short, the analysis does not substitute for the design that could have and should have been used to assess learned threat value independently of sensory similarity.

    (2) The next paragraph in the Discussion is also confusing. "Such reorganization has been proposed to provide flexibility by allowing new information to be incorporated into existing cortical representations while preserving stable behavioral performance (Mau et al., 2020; Zaki & Cai, 2024). Several mechanisms could contribute to this turnover, including systems consolidation, retrieval-induced reconsolidation or memory updating, and repeated nonreinforced stimulus exposure (Lacagnina et al., 2019; Mau et al., 2020; Sangha, 2015; Zaki & Cai, 2024). Although our experiments cannot distinguish between the first two possibilities, the behavioral data argue against extinction as the primary explanation. Extinction is generally associated with the formation of new CS+-safety associations (Bouton et al., 2021), whereas discrimination ratios increased across retrieval sessions, indicating that animals progressively improved their discrimination between threat-associated and safe stimuli rather than acquiring generalized safety responses. This pattern is consistent with previous work showing that discrimination learning sharpens stimulus representations and narrows behavioral generalization gradients (Dunsmoor & LaBar, 2013; Herzog et al., 2021; Jenkins & Harrison, 1960; Lommen et al., 2017). Importantly, turnover was not uniform across the population. Graded neurons retained remarkably consistent response profiles across retrieval sessions, and their activity remained more strongly associated with learned threat value than with freezing behavior. These observations indicate that stable components of the population code can coexist with extensive reorganization of surrounding neuronal ensembles."

    The issue with repeated testing is *not* caused by extinction per se. The issue is that non-reinforcement across the repeated testing should differentially affect the CS+ and CS-. Specifically, it should extinguish responding to the CS- stimulus at a rate that matches its distance from the CS+, thereby sharpening the CS+ versus CS- discrimination in precisely the ways that have been observed. Ergo, the repeated testing *is* a problem for inferences that might be drawn about the way that generalization gradients change with time; and *is* a problem for statements regarding "dynamic reorganization of cortical activity patterns over time." There is nothing in the study that allows one to comment on the reorganization of cortical activity patterns over time. The reorganization can and should be attributed to the repeated testing, which is confounded with time. Nonetheless, the reorganization must be due to the repeated testing and NOT time as the present findings are inconsistent with the well-documented broadening of generalization gradients with time.

    (3) In the next paragraph, the authors state: "At the same time, narrower generalization gradients and improved discrimination across retrieval sessions suggests ongoing memory updating. These observations are consistent with contemporary theories proposing that systems consolidation and retrieval-dependent updating are complementary processes through which memories continue to evolve after learning (Mau et al., 2020; Tome et al., 2024; Zaki & Cai, 2024)."

    In general, I'm not sure why one would invoke systems consolidation or retrieval-induced reconsolidation as an explanation for any of the present findings: they are not explanations of much at all. In this specific text, the authors seem to be implying an updating process that occurs independently of what is learned across the repeated sessions of testing. Why? The changes that occur in the behaviour and neuronal representations are perfectly explicable in terms of additional learning that occurs - of the sort that I hope to have made clear in my previous comment. Why invoke more than what is needed to explain the observed pattern of results?

    (4) Re the GLM analysis - The authors write that: "the fact that the GLM analysis indicates that these neurons reflect learned threat value more than freezing behavior, suggests that they encode an abstract property of the learned stimulus rather than simply mirroring behavioral output."

    This is fine if freezing fully indexes the state of conditioned fear and there are no other behaviours in which animals express their fear. If, however, fear is expressed in a range of other behaviours that are likely coordinated by the PL (e.g., startle, vigilance, scanning, orienting to source of danger), this interpretation of the GLM analysis is unwarranted. This is an important point and would be worth noting somewhere in the paragraph where the statement appears.

  12. eLife

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under examined in the field.

    Comments on revised version.

    The authors have convincingly and thoroughly addressed my concerns. I have no further issues regarding this study.

  13. eLife

    Author response:

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

    Public review:

    Reviewer #1 (Public review):

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a no-shock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure. To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences. However, in its current form, the manuscript does not yet fully support the strength of the conceptual claims. Several issues limit confidence in the interpretation, including the possibility that repeated testing itself contributes to changes across days, uncertainty about the relationship between neural activity and freezing behavior, limited quantitative documentation of longitudinal cell registration, and a number of problems in figure clarity and statistical framing. Overall, the study contains promising observations, but the claims should be narrowed, and several analyses or controls would be needed to fully support the proposed framework.

    Detailed Comments

    (1) A general concern is that the repeated test procedure itself may contribute to extinction. Because the animals are exposed to multiple CS frequencies across multiple test days, and each tone is presented three times per session, some of the reported changes in behavior and neural activity across days could reflect extinction or repeated nonreinforced retrieval rather than the passage of time per se. This is especially relevant given that the manuscript makes claims about recent versus remote representations and representational drift over 30 days. At a minimum, the authors should discuss this limitation explicitly and temper claims about time-dependent changes. Ideally, they would include a control group in which animals are tested only once or twice (e.g., at an early and later time point with fewer CS frequencies), or a reduced-frequency testing design that minimizes extinction while still allowing evaluation of recent versus remote memory.

    We agree with the reviewer that repeated testing is an inherent limitation of longitudinal memory studies and may itself contribute to neural changes across sessions. Repeated retrieval can induce memory updating (reconsolidation) or extinction, the latter involving the formation of a new association between the CS+ and safety. Although memory updating may have contributed to the ensemble reorganization observed here, several aspects of our findings argue against extinction as the primary explanation for the observed neural changes.

    First, we observed substantial neuronal ensemble turnover beginning with the first retrieval session. This early turnover is consistent with previous observations in the prefrontal cortex [1, 2] and with growing evidence that cortical memory representations remain dynamic throughout systems consolidation [3, 4]. Longitudinal studies have shown that neurons are continuously recruited into and removed from cortical memory ensembles while memory expression remains stable [1-4].

    Second, we calculated discrimination ratios to quantify discrimination of each tone relative to the CS+ across retrieval sessions (Figure S1). These analyses showed that discrimination increased, rather than decreased, over successive retrieval sessions, a pattern inconsistent with the behavioral profile expected if repeated testing had induced extinction.

    Finally, one of the most novel findings of our study is that ensemble turnover does not affect all neuronal populations equally. The graded neurons identified by our clustering analysis maintained their identity and functional organization across retrieval sessions, and their activity was better explained by tone threat value than by freezing behavior (Figure 8). This selective stability indicates that ensemble reorganization is not a uniform process but instead preferentially affects specific neuronal subpopulations while preserving a stable threat-value generalization gradient. Thus, although repeated retrieval may contribute to ongoing ensemble reorganization, our results demonstrate that this process is selective and largely spares the neuronal subpopulations that encode graded threat-value representations.

    Accordingly, we have revised the Discussion to explicitly acknowledge these points as follows:

    “The ensemble turnover observed here is consistent with previous studies demonstrating dynamic reorganization of cortical activity patterns over time [1-3, 5]. Such reorganization has been proposed to provide flexibility by allowing new information to be incorporated into existing cortical representations while preserving stable behavioral performance [4, 6]. Several mechanisms could contribute to this turnover, including systems consolidation, retrieval-induced reconsolidation or memory updating, and repeated nonreinforced stimulus exposure [4, 6-8]. Although our experiments cannot distinguish between the first two possibilities, the behavioral data argue against extinction as the primary explanation. Extinction is generally associated with the formation of new CS+-safety associations [9], whereas discrimination ratios increased across retrieval sessions, indicating that animals progressively improved their discrimination between threat-associated and safe stimuli rather than acquiring generalized safety responses. This pattern is consistent with previous work showing that discrimination learning sharpens stimulus representations and narrows behavioral generalization gradients [10-13]. Importantly, turnover was not uniform across the population. Graded neurons retained remarkably consistent response profiles across retrieval sessions, and their activity remained more strongly associated with learned threat value than with freezing behavior. These observations indicate that stable components of the population code can coexist with extensive reorganization of surrounding neuronal ensembles.” Pg. 19

    (2) More generally, some of the reported learning-related neural differences may be driven by behavioral differences, particularly freezing, rather than by learning or generalization per se. For example, animals that freeze more to certain frequencies may show corresponding neural response differences simply because freezing alters PL activity. The authors should examine this possibility more directly. Analyses testing whether recorded cells encode freezing behavior, or whether tone frequency-related neural differences remain robust when comparing high- and low-freezing epochs, would help determine whether the reported effects reflect learned stimulus value rather than behavioral state differences.

    This is an important point, which was also highlighted by the other reviewers. To directly address this concern, we implemented the generalized linear model (GLM) analysis suggested by Reviewer 3. We modeled the neuronal activity time series using both tone identity and freezing behavior as simultaneous predictors. Because tone identity was fixed across trials whereas freezing varied from trial to trial, the GLM allowed us to dissociate their independent contributions to neuronal activity.

    As described in the original submission, freezing was estimated from the miniscope's onboard inertial measurement unit (IMU), which measures body acceleration along three axes. Rather than classifying freezing using a fixed threshold, we estimated the continuous probability of freezing from the accelerometer signal using a Gaussian mixture model. This probabilistic estimate was incorporated directly into the GLM together with tone identity, providing a conservative test of whether neuronal activity was better explained by freezing behavior or by the auditory stimulus.

    We applied the GLM both to all sound-responsive neurons contributing to the population response curves (Figure 4) and to the graded and frequency-selective neuronal subpopulations identified by our clustering analysis (Figure 8). Across both experimental groups and all analyses, the median regression coefficients (β) associated with tone identity were consistently larger than those associated with freezing, indicating that tone identity contributed more strongly to neuronal activity. Moreover, tone coefficients exhibited graded monotonic profiles that closely tracked the learned threat value of each tone, with graded neurons showing the strongest gradients (Figures 4a, 8a, and 8e). Consistent with previous reports [14, 15] freezing accounted for a modest but significant component of PL activity. However, only 6–8% of graded neurons were classified as freezing-dominant, indicating that for the vast majority of these neurons, tone identity was the stronger predictor. Together, these findings demonstrate that the graded representation of learned threat value persists after accounting for freezing behavior, supporting our conclusion that PL activity reflects learned threat value rather than merely the behavioral expression of fear.

    (3) A central feature of the manuscript is the analysis of neural response properties over an extended period of time, up to 30 days after learning. However, aside from a brief mention in the Methods that spatial registration was used, the manuscript provides very little quantitative information about this critical aspect of the study. The paper would be strengthened by including explicit metrics describing longitudinal cell tracking, such as the number and proportion of ROIs retained across all sessions, distributions of spatial-footprint correlations or centroid distances across days, and representative examples of matched imaging fields over time. Without this information, it is difficult to assess how strongly the longitudinal claims are supported.

    We thank the reviewer for this suggestion. We now include measures of registration quality in the resubmission. Specifically, we calculated shifts in centroid distances, proportion of ROIs retained across all sessions, and representative examples of matched imaging fields over time (Fig, S3).

    (4) The text states that "Figs. 1c and 1d show GCaMP6f expression in PL, representative calcium footprints, and activity traces". However, the figure as presented does not clearly show all of these elements, at least not in a way that matches the description in the Results. The correspondence between text and figure should be corrected.

    We corrected correspondence between text and Figure.

    (5) The labeling of Figure 2a is insufficient for interpretation. The legend states that the panel shows raster plots of sound responsiveness, but the axes and scaling are not clearly defined. It is not clear from the figure what the x-axis represents, whether the y-axis corresponds to individual neurons, where the CS period occurs, or what the activity scale at the right denotes. Also, the term 'rasters' implies that spikes were analyzed. It seems that the spike inference approach (CASCADE) was only used for later analyses. Perhaps 'heat-plot' would be more accurate here? Generally, this figure should be annotated more clearly so that the reader can understand it without referring back to the Methods.

    We clarified the labelling of the Figure 2a and call the graphs “activity-plots”.

    (6) In relation to Figure 3, the analysis of population-averaged responses across tone frequencies is useful, but the manuscript would be stronger with additional statistical analyses across time and across groups. For example, if the authors want to argue that learning induces graded changes in neural responses and that these evolve across time, they should directly compare within-group responses across days and also compare matched frequencies between the conditioned groups and the no-shock controls. These analyses would help establish whether the observed differences are genuinely learning dependent and whether they change significantly over time.

    For Figure 3, we maintained the previous one-way ANOVAs assessing changes in AUC per day to be able to note significance on the Figure panels. However, we added a three-way mixed-effects analysis, using group (CS15, CS3, no shocks), frequency (3, 7, 11, 15), and day of testing (2, 15, 30) as variables, with frequency and day of testing as repeated measures. The results were described as follows (statistical details Table S1):

    “To determine how AUC varied across groups over time, we performed a three-way mixed-effects ANOVA with group (CS+15, CS+3, and no shock), frequency (3, 7, 11, and 15 kHz), and time (test days 1, 15, and 30) as factors, with repeated measures on frequency and time. For positive responder neurons, the analysis revealed significant main effects of group (p < 0.001) and time (p < 0.05), as well as a significant group × frequency interaction (p < 0.001), whereas the time × frequency and group × time × frequency interactions were not significant (p > 0.05; Table S2a). Tukey-corrected post hoc comparisons showed that, in the CS+15 group, AUC differed between all frequency pairs except 11 and 15 kHz (p < 0.05). In the CS+3 group, the AUC at 3 kHz differed from those at 7, 11, and 15 kHz (p < 0.05), whereas no significant frequency differences were observed in the no-shock controls (p > 0.05). For negative responder neurons, the only significant effect was a time × frequency interaction (p < 0.01). Tukey-corrected simple-effects analyses revealed that, on day 30, the AUC at 15 kHz differed from those at 3, 7, and 11 kHz (p < 0.05; Table S2b). Because this pattern was observed across all experimental groups, including the no-shock controls, it is unlikely to reflect associative learning. These results indicate that although the AUC exhibited modest changes over time, these changes were not group-specific and therefore do not support learning-dependent alterations in neuronal responses. Together, these results show that despite substantial neuronal turnover, PL population responses encode generalization gradients, closely matching behavioral expression.” Pg. 9

    (7) The inclusion of two different CS+ frequencies and a no-shock control is a strength of the study and substantially improves the interpretation that graded neural responses are related to learning and generalization rather than to simple sensory processing or passage of time. That said, I am not entirely comfortable with the use of the term "inference" throughout the manuscript. What is being measured here appears closer to sensory generalization than inference in a stronger cognitive sense. The current task does not clearly require that animals infer hidden structure or stimulus value through abstract reasoning; rather, the generalized stimulus may simply be treated as similar to the conditioned cue. The terminology should therefore be reconsidered or softened.

    We thank the reviewer for appreciating the strengths of the experimental design and for this thoughtful suggestion regarding terminology. We agree that the term inference may overstate the cognitive processes engaged by the current task. Accordingly, we revised the terminology throughout the manuscript to describe these effects as graded generalization of threat value across stimuli. The new GLM analyses further support this interpretation by demonstrating that, in the conditioned groups, neuronal activity at both the population and single-neuron levels is explained substantially better by tone identity than by freezing behavior (Figures 4 and 8). We therefore retained the term threat value, as our results indicate that PL activity primarily reflects learned threat value rather than simply the expression of freezing behavior, but removed inference.

    (8) I also found the use of the term "valence" somewhat problematic. The manuscript appears to use valence to refer to graded responding across tones with different aversive significance, but valence typically refers more broadly to distinctions between appetitive and aversive value. Here, terms such as "threat value," "aversive value," may be more precise. The authors should consider revising this language throughout.

    We corrected the language and replaced valence for “threat value”

    Reviewer #2 (Public review):

    Summary:

    The following points are those that occurred to me across readings of the paper. They are listed in what I take to be the order of their significance. Many of the points relate to the loose use of language and invocation of concepts that are not warranted, given the study design and results obtained.

    Major Comments:

    (1) The concept of ensemble turnover is interesting - the way it is introduced and discussed implies some type of spontaneous change in the neural underpinnings of fear discrimination and generalization in the PL. But, of course, every trial involves an opportunity to learn about the threat CS or the generalization test stimuli, and I am troubled by the thought that stability in the neural underpinnings of fear discrimination and generalization will actually reflect the level of defensive behaviours evoked on different trial types and/or the discrepancy between those behaviours and the outcome of a given trial in the generalization test. That is, stability in the neural underpinnings may be related to an animal's certainty or uncertainty in the contingency between a stimulus and danger; or, put another way, an animal's confidence that danger will or won't occur given the presence of some stimulus. This is not uninteresting. It is, however, not considered anywhere in the paper, which is overloaded with references to inferred threat values and integration of information across different types of stimuli. The protocol is not one that requires inference about anything or integration across anything.

    We thank the reviewer for this thoughtful comment. We agree that our original wording may have implied that turnover was a spontaneous process. Repeated retrieval provides opportunities for updating the learned contingencies associated with both the conditioned and generalization stimuli, and therefore changes in ensemble composition across sessions need not arise independently of experience. We also agree that the stability of graded neuronal representations may be related to the animal's certainty about the learned contingencies. However, in our data the graded neuronal population remained remarkably stable across retrieval sessions, whereas changes occurred primarily within the dynamic, frequency-selective neuronal populations. This suggests that stable ensembles preserve representations of learned threat value while updating is concentrated in a distinct neuronal subpopulation. We have now incorporated these ideas into the Discussion.

    (2) I appreciate the link to Gu and Johansen in paragraph 3 of the Introduction, but the type of generalization under investigation here is not the same as the type of 'generalization' studied by Gu and Johansen [who used a sensory preconditioning protocol]. Nonetheless, the authors have forced the language used by Gu and Johansen into their paper, and this has created tension [at least for this reader] as the concepts introduced by Gu and Johansen [inference, integration] are simply not relevant given the generalization protocol used here. Here are a few examples of points where the tension might interfere with a reader's understanding:

    We thank the reviewer for these specific criticisms. We revised the manuscript throughout to remove or redefine terms like "inferred valence" and "integration," replacing them with clearer, more accurate descriptions of gradient generalization of threat value. Below we address each point raised by the reviewer regarding terminology clarifications.

    (a) 'We hypothesized that generalization to novel stimuli depends on stable subnetwork organization that enables comparisons between learned and inferred valence, as well as population-level features that reduce variability across related representations.'

    I understand the words in the hypothesis, but can't form a representation of what is being said because of the reference to terms that stand in need of clarification [inferred valence, variability across related representations], but, ultimately, won't be clarified. This needs to be re-expressed so that the reader can appreciate what is being said.

    (a) We hypothesized that the PL generates representations of learned threat value that support threat generalization and discrimination, and that these representations emerge from the coordinated activity of stable and dynamic neuronal subnetworks, preserving consistent relationships among stimuli despite ongoing cellular turnover.

    (b) 'Our results show that stable cortical subnetworks integrate the emotional "gist" of memory and inferred valence for novel cues over time, despite ongoing ensemble reorganization, and that population-level firing rate similarity across stimulus presentations determines threat generalization.'

    Again, what does this mean? How is the gist of a memory integrated with inferred valence for novel cues over time? The statement simply doesn't make sense. This needs to be rewritten for clarity.

    (b) The summary statement was rewritten: " Together, these findings provide a neural framework for understanding how the PL supports adaptive threat generalization and discrimination.” pg. 4

    (c) 'In CS+ 15 mice, positively modulated sound-responsive neurons exhibited graded tone activity reflecting the contingency learned valence as well as the inferred valence of novel tones across testing days...'.

    Can this be rewritten as 'In CS+15 mice, positively modulated sound-responsive neurons exhibited graded activity to the tone CS and its variants that were used to assess generalization.'? The overloading of the text with references to 'contingency learned valence' and 'inferred valence' is unnecessary and makes it much harder to understand what has been shown in the results.

    We adopted the reviewer's suggested rewording: " In CS+ 15 mice, positively modulated sound-responsive neurons exhibited graded tone activity reflecting learned contingency value across testing days" pg. 9

    We will systematically review the entire manuscript to ensure consistency with this revised framing.

    (3) Re the same passage of text as in 2c:

    Is it the case that these neurons are simply tracking the expression of freezing to the various tones? The same question applies to the results obtained for the CS+3 mice. If this is the case, then why should the results be taken to support the banner statement that 'Sound-modulated PL population responses encode learned and inferred valence' - these analyses do not support that statement. And, as indicated, I don't believe that the language of learned and inferred valence is appropriate to such statements, given the nature of the protocol used and results obtained. It is a study looking at how populations of neurons in the PL respond during presentations of auditory stimuli that were subject to discriminative conditioning, and during tests of generalized freezing to other [intermediate] auditory stimuli.

    The reviewer is correct that the graded population responses observed in PL could reflect freezing behavior across tone frequencies rather than encoding an abstract threat-value representation. This important concern was also raised by other reviewers. To address it directly, we followed Reviewer 3’s suggestion and implement a Generalized Linear Model (GLM) using the time series activity derived from the Ca2+ signals, with both tone identity and freezing behavior included as predictors. This analysis allowed us to dissociate the respective contributions of tone frequency and freezing to the graded neural responses. Based on the outcome of this analysis, we concluded that tone identity was a stronger predictor of neuronal activity than freezing. These results are summarized in Figures 4 for all cells contributing to population responses and Figure 8 for the main neuron types identified in the clustering analysis (frequency-selective and graded neurons). All details of this extensive new analysis are shown in red in the revised resubmission.

    In addition, we revised the text to remove the terminology of “learned and inferred valence” throughout the manuscript.

    (4) It is stated that:

    'In no-shock controls, although both positive and negative responses were present, population activity was not modulated by tone frequency or valence'.

    What does this mean? I can understand that population activity was not modulated by tone frequency. But what does it mean to say that it was not modulated by valence? Why should it have been when none of the tones were conditioned in this group and, hence, mice were responding to all the tones equally? And given that this is true, I don't understand the use of 'valence' here, or the subsequent statements in this paragraph that 'graded responses require associative learning' and that 'PL population responses encode graded sound-valence associations that reflect both learning and inference, closely matching behavioral generalization.' The latter statement is particularly unwarranted and, again, highlights a major issue with the paper. It could and should be rewritten as 'PL population responses reflect behavioral generalization.' There is nothing in the additional language that adds to the reader's understanding of what has been shown. The reference to 'graded sound-valence associations that reflect both learning and inference' is completely unwarranted, given the nature of this study. It is anathema to the vast literature on stimulus generalization. If the authors wished to make statements of this sort, they should have taken a different approach, perhaps using protocols like those featured in Gu and Johansen.

    We thank the reviewer for this helpful comment. We agree that our use of the term valence in describing the no-shock controls was imprecise. Because none of the tones was associated with reinforcement in this group, there was no learned valence that could modulate neuronal activity. Our intention was simply to convey that, although both positive and negative sound-responsive neurons were present, the population responses did not vary systematically across tone frequencies. We have revised this section accordingly.

    We also agree that our original wording overstated the interpretation of the graded population responses. Our data do not demonstrate that associative learning is required for sound responsiveness itself; rather, they show that associative learning is required for the emergence of graded population responses that distinguish tones according to their learned threat value. We have revised the text to make this distinction explicit.

    Finally, we agree that our previous references to "learning and inference" were not justified by the behavioral paradigm. We have removed this language throughout the manuscript and now describe the findings more directly as graded representations of learned threat value that closely parallel the observed behavioral generalization gradients.

    (5) The section titled, 'Consistently active neurons preserve valence representations as newly recruited neurons sharpen remote memory traces' ends with the following summary:

    'Together, these results indicate that consistently active neurons maintain stable representations of learned and inferred sound associations across time, whereas neurons recruited after conditioning progressively acquire graded tuning at later retrieval stages. This dynamic refinement suggests that cortical memory representations become increasingly selective during systems consolidation, while a stable neuronal subpopulation preserves the core emotional content of the memory.'

    Once again, the summary is not in keeping with the results obtained. The 'dynamic refinement' of representations is far more likely to reflect the repeated testing across days 1, 15, and 30 rather than anything to do with systems consolidation - at the very least, it is the simplest interpretation of the results. The impact of repeated testing is evident in the sharpening of generalization gradients over time, which is contrary to what is otherwise observed in the literature - the incredibly well -documented broadening of generalization gradients with time. Given this impact of repeated testing, surely the changes in the neuronal population that underlie performance are more likely to reflect the learning that occurs on days 1, 15, and 30, which is reflected in reduced freezing to the non-conditioned tones. If this is a reasonable take on the results, then I don't see the basis for invoking systems consolidation at all, and I don't see the basis for inferring a stable neuronal subpopulation that preserves the emotional content of the memory. Rather, non-reinforced presentations of 'never-reinforced' tones result in recruitment of additional neurons that result in suppression of freezing responses to those stimuli.

    We thank the reviewer for this thoughtful comment. We agree that repeated retrieval is an inherent limitation of longitudinal memory studies and that repeated non-reinforced presentations of the tones provide opportunities for memory updating. Accordingly, we have revised the Discussion to explicitly acknowledge that repeated retrieval may contribute to the ensemble reorganization observed across sessions through memory updating or reconsolidation processes (Discussion, pg. 19).

    We also agree that the progressive sharpening of the behavioral generalization gradients across retrieval sessions is consistent with memory updating. Both the behavioral data (increased discrimination ratios) and the neuronal data (progressively sharper population generalization gradients among neurons active after conditioning) indicate that the memory representation became more precise over time. We now discuss this possibility explicitly in the revised Discussion. We also agree that fear generalization often broadens with time; however, this is not universal. Under discriminative conditioning paradigms, repeated retrieval can instead produce progressively narrower generalization gradients [11]. We have revised the Discussion to clarify this distinction and added the appropriate references (pg. 19).

    While the reviewer's interpretation is therefore plausible, we do not believe it fully accounts for our observations. If repeated non-reinforced presentations were the sole driver of the observed neuronal changes, one might expect a more uniform reorganization across the neuronal populations engaged by the task. Instead, the reorganization was highly selective. Neurons encoding graded threat value remained remarkably stable across retrieval sessions, whereas neuronal turnover occurred primarily within the frequency-selective subpopulations. Thus, although repeated retrieval may update the memory representation, the neuronal substrate supporting graded threat-value coding is largely preserved while refinement occurs within a distinct neuronal subpopulation.

    Moreover, we observed substantial neuronal turnover beginning with the first retrieval session, consistent with previous longitudinal studies showing that cortical memory ensembles remain dynamic despite stable memory [1-4]. This early emergence of turnover suggests that repeated testing alone is unlikely to account for the continuous population dynamics observed throughout the experiment.

    Rather than viewing these findings as evidence exclusively for either memory updating or systems consolidation, we believe they are more consistent with current models proposing that these processes occur in parallel. Several influential frameworks argue that memories are continuously modified through retrieval while simultaneously undergoing systems-level reorganization [4, 6, 16, 17].We have therefore revised the Discussion to interpret the longitudinal changes more conservatively as reflecting the combined influence of retrieval-dependent memory updating and systems-level reorganization.

    In summary, we have revised the manuscript to better acknowledge the contribution of repeated retrieval while emphasizing what we believe is the principal finding of our study: despite substantial turnover within the overall ensemble, the neuronal population encoding graded threat value remained remarkably stable, whereas refinement occurred primarily within dynamic frequency-selective neuronal populations.

    (6) In the section titled, 'Population vector similarity at stimulus onset determines degree of generalization', it is stated that:

    'Because population similarity peaked shortly after stimulus onset, we quantified similarity during the first 5 s after tone onset relative to the CS+. In CS+15 mice, population similarity was highest for 15/15 and 15/11 tone pairs with no differences between them.'

    Isn't this consistent with the view that the population response in the PL simply reflects the level of freezing? Freezing to the 15-15 and 15-11 tones is most likely to be similar on their first presentation prior to the effects of extinction on the 11 Hz tone; hence the results obtained. That is, these results appear to clearly indicate that neuronal responses in the PL reflect the degree of stimulus generalization, as evidenced in freezing behavior. Given all that we know about the involvement of the PL in expressing fear responses, it is not appropriate to claim that 'population vector similarity at stimulus onset *determines* the degree of generalization. The PL responses simply reflect the varying levels of performance displayed to the different types of tones. What have I missed that could be taken to support additional statements?

    We agree that, because population similarity is highest for the 3/3, 15/15, and 15/11 tone pairs and freezing is also greatest for these same stimuli, the neural data could, in principle, reflect a correlate of behavioral expression rather than an independent representation of learned threat value.

    To directly address this possibility, we implemented a generalized linear model (GLM) to dissociate the contributions of tone identity and freezing behavior to neuronal activity. Across all analyses, tone identity consistently explained substantially more variance in neuronal activity than freezing behavior. Importantly, this finding held not only for the full population of sound-responsive neurons used to generate the population similarity analyses (Figure 4), but also for both the stable graded neurons and the dynamic tone-selective neuronal populations identified by our clustering analysis (Figure 8). Thus, although freezing behavior contributes modestly to PL activity, it cannot account for the enhanced similarity of population vectors across stimulus presentations or the graded population responses that form the basis of our conclusions.

    In addition, the temporal dynamics of the population vector similarity analysis are not entirely consistent with the interpretation that PL activity simply reflects the expression of freezing behavior. Population vector similarity peaked during the first 5 seconds following tone onset, whereas freezing occurred intermittently throughout the tone presentations. Although this temporal relationship does not establish causality, it is consistent with the interpretation that PL activity reflects the learned threat value associated with each tone rather than merely tracking the magnitude of freezing.

    Finally, we have revised the manuscript to more clearly acknowledge the correlational nature of these analyses. Specifically, we now state that population vector similarity is associated with, rather than determines, the degree of threat generalization.

    Later in the same section, it is stated that 'population-level similarity at stimulus onset scales with behavioral threat generalization and is maximal for tones associated with robust threat responses.' For simplicity and, therefore, clarity, this should be rewritten as 'population-level similarity at stimulus onset reflects behavioral threat generalization.'

    We made this correction. (“These findings indicate that population-level similarity at stimulus onset scales with behavioral threat generalization”. pg. 13)

    (7) In the section titled, 'Different subnetworks encode acoustic versus learned properties of sound association', it is stated that:

    'Our previous analyses show that learned and inferred associations are represented at the population level. However, these results do not resolve whether graded responses arise from pooled activity of frequency-selective neurons or from subnetworks encoding integrated learned valence across tones.'

    What does it mean to say 'integrated learned valence across tones'? As it presently stands, the meaning of the phrase is unclear. It only makes sense if one supposes that generalized freezing responses to the 11 and 7 kHZ tones reflect separate associations between those tones and the aversive foot shock US. This supposition is inconsistent with the rich literature on generalization of Pavlovian conditioned fear responses. Specifically, it is inconsistent with the many theories of fear generalization, which attribute the reduction in fear as one moves away from the specific conditioned stimulus to a decrement in the ability of the test stimulus to activate the trained CS-US association. My strong impression is that the authors would do well to ground their findings in theories of stimulus/fear generalization, of which there are many. This would better serve the results obtained [and the reader's appreciation of them] - at present, the unnecessary invocation of concepts does very little to enhance the reader's appreciation or understanding of what has been found in the study.

    We agree that the phrase "integrated learned valence" is unnecessarily opaque and we replaced it with more precise language “Our previous analyses demonstrated that threat-value generalization gradients are represented at the population level. However, these findings do not reveal how these representations arise. Specifically, the observed population gradients could emerge either from the pooled activity of frequency-selective neurons that respond to individual tones or from neuronal subnetworks that integrate information across tones to encode their learned threat-value.” (Pg. 13)

    (8) Another example of what has been a common theme in this review:

    '...we hypothesized that the PL active ensemble segregates into functionally distinct subnetworks: one encoding tone-specific sensory features with dynamic characteristics, and another responding to all frequencies encoding stable core memory content and inferred emotional valence.'

    What does it mean to say 'all frequencies encoding stable core memory content and inferred emotional valence'? Do the authors mean to say '...and another that tracks freezing/defensive responses regardless of whether they were elicited by the trained CS or one of the generalization test stimuli'?

    We thank the reviewer for pointing out that this section was unclear. We agree that our original wording was imprecise and could be interpreted as implying cognitive processes that were not directly tested in the present study. Accordingly, we have revised the terminology throughout the manuscript. We no longer refer to "inferred emotional valence" or "core memory content" and instead describe these neurons more specifically as exhibiting graded representations of learned threat value.

    This is not the interpretation we intended. To determine whether these neurons primarily reflected defensive behavior rather than learned stimulus value, we implemented a Generalized Linear Model (GLM) that dissociates the contributions of tone identity and freezing behavior to neuronal activity. Across the entire neuronal population, as well as within the stable graded and dynamic tone-selective neuronal subpopulations, tone identity consistently explained substantially more variance than freezing behavior (Figures 4 and 8). Furthermore, after accounting for freezing, the regression coefficients of the graded neurons continued to follow the learned threat value of the tones, exhibiting opposite monotonic gradients in the CS+15 and CS+3 groups. If these neurons simply tracked defensive behavior irrespective of the stimulus presented, this relationship would not be expected to persist after accounting for freezing. We therefore conclude that the activity of this stable neuronal subpopulation is better explained by graded representations of learned threat value than by defensive behavior alone, and we have revised the manuscript accordingly.

    (9) It is stated that - 'Graded clusters encode emotional valence but constitute only a fraction of the active population; yet valence coding at the population level remains accurate and precise. This indicates that neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.'

    What does this mean? Are the authors trying to say that - 'Some clusters of PL neurons track freezing responses. In spite of the fact that these are only a fraction of the total active neuronal population, the population-level response of PL neurons also tracks the levels of fear to the trained tone and its variants used in the test for generalization.' If this is what one wants to say, then the final statement in the reproduced section does not follow. That is, there is no indication that 'neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.' As noted, the characteristics of other ensembles that become active across the repeated tests on days 1, 15, and 30 are more likely to reflect learning from non-reinforcement that occurs within and across those sessions. Perhaps this is what is meant by the phrase, 'shaped by associative processes'? If so, it should be stated explicitly instead of left to the reader to work out.

    We thank the reviewer for highlighting that this section was unclear. We agree that the original phrasing was insufficiently precise. Our intention was to convey that only a subset of PL neurons displays graded tuning that tracks behavioral generalization across tones. Nevertheless, despite constituting only a fraction of the total active population, this graded coding is also reflected at the population level. This observation led us to hypothesize that neurons recruited into the active population after conditioning— likely dynamic, frequency-selective neurons—also contribute to these graded population responses through modulation of their firing rates.

    The reviewer correctly notes that the phrase "shaped by associative processes" was too vague. By this we meant that the firing properties of these neurons are modified by the animal's associative history, including both the original conditioning experience and any retrieval-dependent updating that may occur during subsequent test sessions. We have revised the manuscript to make this interpretation explicit rather than leaving it to the reader to infer.

    To test this hypothesis, the GLM analysis we implemented dissociated the contributions of tone identity and freezing behavior to neuronal activity. After accounting for freezing, tone identity (i.e., learned threat value) remained a significant predictor of neuronal responses. Importantly, this was also true for the dynamic, frequency-selective neurons (Fig. 8e–f), indicating that these neurons contribute to population-level representations of learned threat value through firing-rate modulation rather than simply reflecting defensive behavior.

    To clarify our interpretation, we have rewritten the relevant section as follows:

    "Graded clusters encode generalization gradients but constitute only a subset of the active neuronal population. Nevertheless, population-level representations, which incorporate all active neurons, remain robust and accurately preserve these gradients. This observation led us to hypothesize that neurons recruited over time (e.g., dynamic, frequency-selective cells) also contribute to threat-value representations. Consistent with findings in the hippocampus showing that neurons can encode task contingencies through firing-rate modulation despite responding selectively to a single location (Gagliardi et al., 2024; Huxter et al., 2003; Sanders et al., 2019), we tested whether dynamic, frequency-selective clusters exhibited firing-rate differences proportional to learned threat value." (page 15)

    Regarding the reviewer's suggestion that the characteristics of the newly recruited neurons may reflect learning during repeated non-reinforced test sessions, we agree that retrieval-dependent memory updating likely contributes to the reorganization of the dynamic neuronal population, and we now explicitly acknowledge this possibility in the Discussion. However, we do not believe that our findings are fully explained by repeated non-reinforced retrieval alone. First, no-shock control animals underwent the same repeated testing but failed to develop graded neuronal representations, indicating that repeated exposure in the absence of associative learning is insufficient to account for the observed changes. Second, both behavioral discrimination and the corresponding population-level neural gradients became progressively sharper over time, consistent with refinement of learned threat representations rather than an effect of repeated testing alone, which must lead to extinction.

    In summary, we thank the reviewer for highlighting both the ambiguity of our original wording and an important alternative interpretation. In response, we have clarified the text to explicitly define what we mean by associative processes, added a GLM analysis demonstrating that the newly recruited neurons encode learned threat value beyond freezing behavior, and revised the Discussion to acknowledge that retrieval-dependent memory updating likely contributes to the reorganization of the dynamic neuronal population.

    (10) The following points all relate to the Discussion and reiterate many of the points above.

    (a) 'A subset of neurons remains consistently active across sessions, preserving core components of the memory trace and supporting inference of emotional valence for novel sounds, while neurons recruited after conditioning progressively acquire valence selectivity at remote time points.'

    'Inference of emotional valence' is unclear and unwarranted for all of the reasons provided above regarding the use of language.

    We modified the language as stated in the prior points.

    (b) '...Our data reconcile these views by demonstrating that cortical representations of emotional valence emerge rapidly after learning and persist within stable subnetworks, even as the broader population undergoes substantial turnover. This architecture preserves core mnemonic content while allowing flexibility in the surrounding ensemble.'

    These statements assume that the PL neuronal responses reflect something more than the levels of freezing behavior to the different stimuli; what are the grounds for this assumption?

    We incorporated the new GLM analysis to address this point and conclusions.

    (c) 'Importantly, these subnetworks encode both learned contingencies and the inferred valence of novel stimuli along a graded representational axis, suggesting that strong recurrent connectivity provides a stable scaffold for emotional memory representations.'

    What is a graded representational axis, and what part of the first statement suggests that 'strong recurrent connectivity provides a stable scaffold for emotional memory representations'? If the authors' goal was to make statements about emotional memory representations vis-à-vis emotional memory content, they should have used protocols that allowed them to probe such content. The auditory fear conditioning protocol used here [followed by tests for generalization to other auditory stimuli that differ in frequency from the conditioned tone] is not one that lends itself to analysis of emotional memory representations or content.

    We agree that the term "graded representational axis" was insufficiently defined and could be interpreted in multiple ways. Because this terminology was not essential to our conclusions, we have removed it and instead describe the observed phenomenon as a graded population representation of learned threat value across tone frequencies. We also removed the statement suggesting that recurrent connectivity provides a stable scaffold for these representations, as this mechanistic interpretation is not directly supported by our data.

    We also agree that some sections of the manuscript overstated the scope of our conclusions and have revised the wording accordingly. Our study uses neuronal activity recorded during memory retrieval after learning, an approach widely used in studies of systems consolidation to infer how learned information is represented within neural populations. Accordingly, we have revised the manuscript to explicitly state that our findings pertain to neural representations of learned threat value during memory retrieval rather than the broader content of emotional memories.

    Finally, we agree that our data are correlational and do not establish the causal role of the neuronal representations we identify. Throughout the manuscript, we now refer more precisely to population- and single-neuron correlates of learned threat value during memory retrieval following auditory fear conditioning.

    (d) 'Dynamic tone-selective responsive neurons emerge independently of learning, as they are present in both control and experimental mice, reflecting pre-existing PL sensory-driven properties (Hockley & Malmierca, 2024; Zikopoulos & Barbas, 2006).'

    Maybe. They are also likely to have developed as a consequence of the repeated testing on days 1, 15, and 30, which involved intermixed exposures to the tones of different frequencies. That is, rather than 'pre-existing PL sensory-driven properties', the responses of these neurons might reflect the emergence of discrimination between the various tones across testing, and greater suppression of freezing to the non-trained tones compared to the trained tone across the various test intervals.

    We thank the reviewer for this thoughtful comment. Our interpretation that these neurons reflect preexisting sensory-driven properties of PL cortex is based on two observations. First, tone-selective neuronal clusters were present in both conditioned and no-shock control animals, consistent with previous reports of sensory responsiveness in PL cortex [18, 19]. Second, these responses were already present during the first retrieval session, when the intermediate frequencies were presented for the first time. Thus, they cannot be explained by repeated exposure to those tones across subsequent test sessions.

    We therefore interpret the frequency-selective response properties as pre-existing features of PL circuitry that are present independently of conditioning. In contrast, associative learning modifies the firing activity of these neurons, allowing them to contribute to graded representations of learned threat value. This interpretation is supported by our GLM analysis, which showed that, after accounting for freezing, tone identity significantly predicted the activity of frequency-selective neurons in conditioned animals but not in no-shock controls. Thus, while the frequency-selective response properties are present independently of learning, associative learning modifies how these neurons encode learned threat value. We have revised the manuscript to clarify this distinction.

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP-positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under-examined in the field.

    We thank the reviewer for appreciating our design to track ensembles over time and the relevance of studying the neural substrates of generalization.

    Weaknesses:

    (1) Difficult to determine if responses treated as encoding stimulus valence are driven instead by the behavior that the stimulus elicits, freezing.

    We thank the reviewer for this thoughtful and constructive comment. We agree that an alternative interpretation is that the graded neuronal responses may partially reflect freezing-related activity rather than representations of learned threat value. In the revised manuscript, we acknowledge that previous studies have identified PL neurons whose activity tracks freezing independently of stimulus identity or associative content. To directly address this possibility, we implemented the reviewer's suggestion by fitting a generalized linear model (GLM) to the neuronal activity time series derived from the Ca2+ signals, using tone identity and freezing behavior as predictors. Because tone identity is fixed across trials, whereas freezing varies both during tone presentation and across trials (see below our answer to the Recommendations to Authors), this approach allowed us to dissociate their respective contributions to neuronal activity. We are grateful for this excellent suggestion, which has substantially strengthened both the manuscript and the conclusions that can be drawn from our data. The new analyses are summarized in Figures 4 and 8.

    In the points below we summarize the new findings.

    (2) The study implies that the identified ensembles are causally related to valence memory, but no experimental interventions are performed to justify this.

    We appreciate the reviewer's point. We agree that our data are correlational in nature and that establishing a causal relationship between identified ensembles and valence memory would require experimental interventions such as combinations of optogenetic and two-photon manipulations, which are beyond the scope of the present study but represent an important direction for future work.

    We examined inter-individual variability in freezing relative to the proportion of graded cells but the number of mice used in this study (CS+3= 5 and CS+15=7) did not give us enough power to reach significance.

    Therefore, we modified the manuscript terminology accordingly, replacing causal language with phrasing that accurately reflects the correlational nature of our conclusions.

    Recommendations for the authors:

    Reviewer #2 (Recommendations for the authors):

    Many sections of the paper should be rewritten along the lines that I have suggested in my public review and below.

    Minor Comments:

    (1) INTRO - 'This broad accessibility reduces spatial specificity and increases learning variability...'.

    Broad accessibility of what, exactly? And how does the 'broad accessibility' reduce spatial specificity and increase learning variability? That is, I do not understand what the terms 'reduced spatial specificity' and 'increased learning variability' refer to at this point in the first paragraph...

    We rewrote the introduction and discussion to address the points raised by the reviewer.

    (1) INTRO - 'The prelimbic cortex (PL) contributes to the expression (Burgos-Robles et al., 2009; SierraMercado et al., 2011; Sotres-Bayon & Quirk, 2010) and the proper discrimination and generalization of threat memories (Rosas-Vidal et al., 2025; Stujenske et al., 2022).'

    What is achieved by calling it 'proper' discrimination and generalization? Can't one simply say that the PL contributes to the expression, discrimination, and generalization of threat memories?

    This was corrected.

    (3) INTRO - '... and that population-level firing rate similarity across stimulus presentations determines threat generalization'.

    Or, alternatively, that generalization of conditioned freezing responses from the tone CS to variants along the dimension of Hz values is reflected in systematic changes in the firing rate of PL neuronal ensembles; when the test stimulus is similar to the conditioned stimulus, the two elicit similar behavioural responses and evoke a similar population-level firing rate in the respective PL neuronal ensembles.

    We have revised the Introduction as stated above. However, as discussed in our detailed responses, freezing behavior cannot fully account for the observed patterns of PL activity.

    (4) METHODS - 'Memory retrieval was tested on days 1, 15, and 30 after conditioning to probe early, long-term, and remote memory (Bontempi et al., 1996). During retrieval, mice were tested in a novel context with the CS+, CS1-, and two intermediate frequencies (7 and 11 kHz), presented in semirandom order, with each tone repeated three times (Fig. 1a).'

    Why was testing conducted in a different context than that of conditioning? This is likely to result in an underestimation of generalization to the different tones...

    In tone fear conditioning, it is always customary to test in a different context to dissociate conditioning to the context vs conditioning to the tones, which usually take place simultaneously in the same context [20]. Therefore, testing generalization in a novel context gives the correct estimate of generalization to the tones in the absence of contextual conditioning confounds. Please note that while overall freezing levels may be lower in a novel context due to the absence of contextual conditioning, the relative generalization gradient across tones — which is what your study measures — is unlikely to be systematically distorted by context change.

    (5) RESULTS - 'No-shock control mice showed no significant differences in freezing across frequencies on any testing day (p > 0.05; Fig. 1b, right), confirming that freezing reflected associative learning.'

    The inference doesn't follow from the result described. Was there more freezing among animals in the shocked groups compared to those in the no-shock group? I presume so - my point is that this comparison is the one that most directly speaks to the presence or absence of associative learning.

    Experimental animals exhibited not only higher overall freezing but also graded freezing responses across tone frequencies. It is important to note that no-shock controls did not display this pattern, ruling out the possibility that the different frequencies themselves elicited graded behavioral responses. To clarify this point, we revised the sentence as follows: "No-shock control mice showed no significant differences in freezing across frequencies on any testing day (p > 0.05; Fig. 1b, right), confirming that the graded freezing patterns resulted from associative learning rather than the acoustic properties of the tones." (Pg. 6)

    (6) 'Across animals and sessions, we identified distinct neuronal populations showing positive modulation, negative modulation, mixed responses, or no consistent response to sound (Fig. 2b)...'

    To be clear, do you mean to say that there were distinct neuronal populations that consistently [i.e., across all three sessions] increased their responses to the tones [positive modulation], decreased their responses to the tones [negative modulation], showed variable responses to the tones [mixed responses], and did not respond to tones [not modulated]?

    The sentence refers to neuronal populations identified within each recording session based on their responses to the tones, not to neurons that maintained the same response profile across all three sessions. We have revised the text to make this distinction explicit. The only stable patterns across sessions were observed in graded neurons that were stable across retrieval.

    “Across animals, we identified distinct neuronal subpopulations showing positive modulation, negative modulation, mixed responses, or no consistent response to sound in each session (Fig. 2b)” Pg. 7

    (7) What does 'active' mean in relation to Figure 2? Does this refer to neurons that displayed either positive responses, negative responses, and/or mixed responses? In the text, it is stated that 'Sound responder neurons were classified using a test that detected modulation based on magnitude relative to baseline variability, allowing reliable identification of both transient and sustained responses while remaining robust to noise...'

    I can't work out if this is the same classification criteria used for the determination of positive modulation, negative modulation, and mixed responding.

    We thank the reviewer for pointing out this ambiguity. In Figure 2, the term "active" referred to neurons that exhibited significant sound-evoked modulation and were subsequently classified as showing positive, negative, or mixed responses. Thus, active and sound-responsive refer to the same population of neurons. We removed the word active to avoid confusion.

    The reference to transient and sustained responses describes the temporal profile of the calcium signals rather than separate response categories. Some neurons exhibited brief calcium transients that rose and decayed rapidly, whereas others displayed sustained activity throughout the tone presentation. The sound-response detection algorithm was designed to reliably identify both temporal response profiles. We have revised the manuscript to make these definitions explicit. We modified the sentence as follows: “Sound-responsive neurons were identified using a statistical test that detected activity modulation relative to baseline variability, allowing reliable identification of responses while remaining robust to noise. This approach was effective for neurons exhibiting either brief calcium transients that rose and decayed rapidly or sustained activity throughout the tone presentation.” Pg. 7-8

    (8) 'A moderate proportion of neurons was present across all retrieval sessions, with no differences between groups (p > 0.05).'

    Do you mean to say that 'A moderate proportion of neurons was ACTIVE across all retrieval sessions, with no differences between groups (p > 0.05)'?

    We replaced the word present and replaced it with “active”. Pg. 8

    (9) In the section titled, 'Different subnetworks encode acoustic versus learned properties of sound association', it is stated that:

    'If neurons encoding graded responses carry core mnemonic information, they should exhibit enhanced stability over time. To test this hypothesis, we quantified the proportion of registered neurons that retained their cluster identity across at least two retrieval sessions and compared these values to a shuffled null distribution (10,000 iterations), with multiple comparisons controlled using the BenjaminiHochberg procedure.'

    What does the comparison to the shuffled null distribution tell us exactly? I accept that some neurons were stable positive responders across at least two sessions. The comparison to the shuffled null distribution creates a false impression about the robustness of this stability or the 'enhanced stability over time'.

    Our intention in comparing the observed stability to a shuffled null distribution was to evaluate whether the proportion of neurons retaining cluster identity exceeded chance levels expected from random assignment. The shuffled distribution therefore provides a statistical baseline against which the observed degree of stability can be evaluated. We agree, however, that the wording “enhanced stability over time” may be confusing regarding this finding. We rephrased this paragraph to clarify that a subset of neurons retained cluster identity across all retrieval sessions at levels greater than expected by chance as follows:

    “These data demonstrate that graded clusters remain consistently active at levels exceeding chance, preserving their cellular identity and providing a stable representation of learned contingencies and generalization gradients.” Pg. 15.

    (10) ABSTRACT. The abstract states that, 'Stimulus-evoked population similarity scaled precisely with behavioral generalization, and consistent population states emerged only for tones associated with shock or those eliciting strong generalized freezing, indicating that population-level similarity predicts inferred threat.'

    I believe that the sentence could be rewritten as, 'Stimulus-evoked population similarity reflected the degree of generalization, and consistent population states emerged only for tones associated with shock or those eliciting strong generalized freezing.'

    We revised the text according the reviewer’s suggestion; however, we had to shorten the sentence due to word limits. “Population similarity tracked behavioral generalization, whereas consistent population states emerged only for shock-associated or highly generalized tones.” Pg. 2

    Reviewer #3 (Recommendations for the authors):

    Major points:

    (1) The ensembles with graded activation in proportion to stimulus valence are described at various points in the manuscript as "maintaining the emotional 'gist'", "preserving core components of the memory trace", and "preserving core components of the memory trace". This conclusion is premature because there is an alternative interpretation. The graded response ensembles would also be consistent with coding for the freezing behavior itself, irrespective of the specific memory or stimulus association that drives it. An ensemble that encodes a behavior in this way would not be considered mnemonic, just as motor neurons in the spinal cord are not, even if they may fire during a conditioned response. Indeed, previous work has identified neurons in the prelimbic cortex that encode freezing independently from the stimuli that signal an aversive outcome (e.g., Kyriazi, Headley, and Pare 2020; Casanova, Pouget, ..., Vetere 2024).

    There are two ways the authors can address this point.

    (a) Fit a generalized linear model to the time series of inferred spiking activity from the Ca2+ signal and include stimuli and freezing as predictors. Since freezing behavior is inconsistent across trials, while stimulus presence is fixed, they can be disassociated. If, after accounting for freezing, responsiveness neurons still show a graded coding of stimuli that agrees with inferred aversiveness, this would strengthen their claim that they have identified an ensemble that corresponds with mnemonic or salience aspects of the stimuli.

    (b) Conduct no further analysis but cover the issue in the discussion as a limitation to their study and to dampen some of the language throughout the manuscript that implies that a memory trace has been identified.

    We thank the reviewer for this thoughtful and constructive comment. We agree that an important alternative interpretation is that graded-response ensembles could reflect freezing-related activity rather than representations of learned threat value. To directly address this possibility, we implemented a Generalized Linear Model (GLM) analysis, as suggested by the reviewer. The GLM was fitted to the activity of every sound-responsive neuron included in the population analyses and simultaneously incorporated tone identity and continuous freezing probability (derived probabilistically from miniscope acceleration) as predictors, allowing us to quantify their independent contributions to neuronal activity.

    We want to note that freezing was quantified from the miniscope's inertial measurement unit (IMU) using a two-component Gaussian mixture model applied to the log-transformed body-acceleration signal. Rather than classifying freezing with a binary threshold, we used the posterior probability of the low-movement state as a continuous freezing regressor. This approach captures graded variations in immobility and provides a more conservative test of tone encoding, because it accounts for more behaviour-related variance than a binary classifier, making it more difficult to detect an independent contribution of tone identity.

    We applied the GLM both to all sound-responsive neurons contributing to the population response curves and separately to the identified frequency-selective and graded neuronal subpopulations. Across all analyses, tone identity consistently explained neuronal activity better than freezing. Furthermore, the freezing-corrected tone β coefficients scaled with learned threat value, with graded neurons exhibiting the strongest monotonic gradients, indicating that they provide the most robust representation of learned threat value. These findings demonstrate that the graded coding of learned threat value persists after accounting for freezing behavior and therefore cannot be explained simply by the behavioral expression of fear. The new analyses are presented in Figures 4 and 8. Notably, although freezing-dominant neurons were present in both the tone-selective and graded populations, they represented only a small fraction of each group and were least prevalent among graded neurons (6–8%), further supporting the conclusion that graded neurons primarily encode learned threat value.

    In addition, we revised the manuscript to avoid language implying that these neuronal populations constitute a mnemonic trace. Instead, we consistently describe them as encoding learned threat value, a more accurate interpretation that is directly supported by the new GLM analyses.

    (2) The title makes a seemingly causal claim by using the term 'arise', "Learned and inferred valence arise from interactions between stable and dynamic subnetworks". While it is true that the authors show that both stable and dynamic ensembles encode valence, they do not demonstrate that the behavioral expression of valence depends on these codes, nor their interaction. Experimentally testing this is beyond the scope of this study (holographic two-photon stimulation of transient and stable ensembles?), but they may be able to get closer to it by examining inter-individual variability. The authors could measure the proportion of neurons in each subject that participate in the stable (graded responding) and dynamic (stimulus-specific) ensembles, and see if they predict individual differences in the expression of freezing behavior or its generalization. Indeed, this correlation may change across testing days.

    We agree that the term “arise” in the title may imply a stronger causal relationship than is directly supported by the present data. We modified the title in the resubmission as follows: “Complementary stable and dynamic prelimbic ensembles encode learned threat value underlying generalization and discrimination”

    The new LGM analysis confirms that a large proportion of neural activity can be predicted by tone threat value; therefore, we think this title fully captures our findings.

    We also appreciate the reviewer's suggestion to examine inter-individual variability. In the revised manuscript, we tested whether the proportion of graded neurons correlated with freezing behavior. However, the limited number of experimental animals in each experimental group provided insufficient statistical power to reliably assess this relationship. Accordingly, we revised the manuscript to clarify that our conclusions are based on correlational observations rather than causal inferences.

    In summary, we revised the title and related language throughout the manuscript to avoid implying causal mechanisms beyond the scope of the current experiments.

    Minor points:

    (1) I was surprised by the absence of an ensemble in the No-shock group that responded uniformly to all stimuli. Can the authors confirm this?

    Yes, we confirm this finding. It was unexpected to us as well. We would like to clarify, however, that some control neurons may have responded to more than one frequency, but these responses were too infrequent or too weak to be classified as a distinct graded neuronal population by our clustering algorithm. Thus, while broadly responsive neurons may have been present in the control group, they did not form a robust, identifiable ensemble comparable to that observed after fear conditioning.

    (2) Several different approaches were used to analyze the same Ca2+ responses to stimuli across testing days. These were the "Sound responder classification", "Average stimulus-aligned trace procedure", "Population similarity over time across tone pairs", and the construction of "Stimulus response vectors". These feature differing alignment/binning/interpolation, normalization, and response quantification procedures, and it is unclear why they cannot all be in agreement, at least when it comes to alignment and normalization.

    We thank the reviewer for this careful reading of our Methods. All analyses were performed on the same underlying calcium imaging dataset, but they were designed to address different aspects of the data and therefore required different preprocessing steps. The analyses share a common initial pipeline leading to the calcium traces (all z-scored across the session). Differences in subsequent processing (e.g., use of ΔF/F versus CASCADE-deconvolved activity, normalization, baseline correction, temporal binning, and interpolation) were introduced only when required by the specific analysis method.

    To make this clearer, we have substantially revised the Methods. We added a new overview of preprocessing section that summarizes the common preprocessing pipeline and explicitly distinguishes the shared steps from those that are analysis-specific. We also included a summary table describing the input signal (ΔF/F or CASCADE-deconvolved activity), normalization procedure, and temporal processing used for each analysis. Finally, the individual Methods sections were revised to eliminate redundancies and more clearly describe the steps to avoid confusion. We hope these revisions make the rationale for the different preprocessing procedures and the overall analytical workflow more transparent (Pg. 22-23)

    (3) In the methods section "Window-wise response quantification" the Ca2+ signal was baselinesubtracted and divided by the standard deviation in the baseline across trials, but that data was already presumably z-normalized to the baseline of each trial ("Data alignment and normalization"). This second step of normalization seems excessive. Why is it not sufficient to just take the average peri-stimulus response across the z-normalized trials from the "Data alignment and normalization" section? This is simpler and would capture the effect size of the response relative to baseline.

    We thank the reviewer for this careful observation.

    The two operations are also not the same normalization applied twice; they standardize different sources of variability. During the alignment step, each trial is z-scored relative to its own baseline by dividing by the standard deviation of that trial's baseline across time. This places all trials on a common within-trial scale before averaging. In the window-wise step, the trial-averaged, baseline-subtracted response is expressed relative to the standard deviation of the per-trial baseline levels across trials—a distinct quantity that reflects trial-to-trial baseline stability rather than within-trial fluctuations. The purpose of this second term was to down-weight windows in cells with unstable baselines across trials, and it entered the analysis only as a significance criterion; the magnitude threshold defining a sound responder was applied to the trial-averaged baseline-relative response itself. We have revised the Methods to clarify the distinct roles of these two normalization steps.

    To further address this concern, we re-ran the sound-responder classification after removing the second (between-trial) normalization step, so that responder detection depended only on the per-trial baselinerelative response magnitude and its temporal persistence. Across all cells, tones, and sessions (n = 89,504 cell–tone–session classifications), the two procedures agreed on 95.3% of labels. The small fraction of cells whose labels changed were almost exclusively those lying immediately at the detection threshold: 86.6% of changes involved cells moving into or out of the "modulated" category, whereas direct reversals between excitatory and inhibitory classification occurred in only 8 of 89,504 cases (0.009%). Consistent with the between-trial standard deviation being a less stable quantity when few trials are available, label changes were approximately twice as frequent in the three-trial retrieval sessions (5.2%) as in the ten-trial conditioning sessions (2.1%). Overall responder proportions changed only minimally (positive responders +2.2%, negative responders +4.3%), and all population-level findings—including the graded threat-value gradient across tones, its absence in no-shock controls, and its persistence after controlling for freezing in the , as analysis—were unaffected. These analyses demonstrate that our conclusions are robust to this methodological choice.

    (4) It would increase confidence in the tracking of neurons across days if the authors showed some example images of neurons tracked across days.

    We added an example in the Supplement. Additionally, we now provide measures of registration quality (Fig. S3)

    (5) Table S2 is a bit confusing. I take it that Graded A/B were only for CS15, and Graded C/D/E were only for CS3. Also, Common B1-4 were the cells with positive responses to individual stimuli, and Common C1-4 were the cells with negative responses to stimuli. If this is the case, it should be explained in the figure legend (or even better, clusters should be named and numbered consistently in all figures.

    Thank you for pointing this out, we corrected the Table to indicate which test corresponds to which figure and cluster, specifying which ones were positive or negative modulated. Please note old Table 2 is now Table 5

    (6) The term network and subnetworks implies some connectivity between neurons, but in this study, it is used to refer to the ensembles of cells activated in a similar manner. Since connectivity is never assessed, it would be better if the authors stuck to the terms ensembles or populations.

    We changed the wording and now use ensembles or populations

    (7) The legend for Figure S3 has the text 'eded', which seems to be a typo.

    We corrected this typo.

    References

    (1) Kitamura, T., et al., Engrams and circuits crucial for systems consolidation of a memory. Science, 2017. 356(6333): p. 73–78.

    (2) DeNardo, L.A., et al., Temporal evolution of cortical ensembles promoting remote memory retrieval. Nat Neurosci, 2019. 22(3): p. 460–469.

    (3) Tome, D.F., et al., Dynamic and selective engrams emerge with memory consolidation. Nat Neurosci, 2024. 27(3): p. 561–572.

    (4) Mau, W., M.E. Hasselmo, and D.J. Cai, The brain in motion: How ensemble fluidity drives memory-updating and flexibility. Elife, 2020. 9.

    (5) Gallego, J.A., et al., Long-term stability of cortical population dynamics underlying consistent behavior. Nat Neurosci, 2020. 23(2): p. 260–270.

    (6) Zaki, Y. and D.J. Cai, Memory engram stability and flexibility. Neuropsychopharmacology, 2024. 50(1): p. 285–293.

    (7) Lacagnina, A.F., et al., Distinct hippocampal engrams control extinction and relapse of fear memory. Nat Neurosci, 2019. 22(5): p. 753–761.

    (8) Sangha, S., Plasticity of Fear and Safety Neurons of the Amygdala in Response to Fear Extinction. Front Behav Neurosci, 2015. 9: p. 354.

    (9) Bouton, M.E., S. Maren, and G.P. McNally, Behavioral and Neurobiological Mechanisms of Pavlovian and Instrumental Extinction Learning. Physiol Rev, 2021. 101(2): p. 611–681.

    (10) Jenkins, H.M. and R.H. Harrison, Effect of discrimination training on auditory generalization. J Exp Psychol, 1960. 59: p. 246–53.

    (11) Dunsmoor, J.E. and K.S. LaBar, Effects of discrimination training on fear generalization gradients and perceptual classification in humans. Behav Neurosci, 2013. 127(3): p. 350–6.

    (12) Herzog, K., et al., Reducing Generalization of Conditioned Fear: Beneficial Impact of Fear Relevance and Feedback in Discrimination Training. Front Psychol, 2021. 12: p. 665711.

    (13) Lommen, M.J.J., et al., Training discrimination diminishes maladaptive avoidance of innocuous stimuli in a fear conditioning paradigm. PLoS One, 2017. 12(10): p. e0184485.

    (14) Casanova, J.P., et al., Threat-dependent scaling of prelimbic dynamics to enhance fear representation. Neuron, 2024. 112(14): p. 2304–2314 e6.

    (15) Kyriazi, P., D.B. Headley, and D. Pare, Different Multidimensional Representations across the Amygdalo-Prefrontal Network during an Approach-Avoidance Task. Neuron, 2020. 107(4): p. 717–730 e5.

    (16) McKenzie, S. and H. Eichenbaum, Consolidation and reconsolidation: two lives of memories? Neuron, 2011. 71(2): p. 224–33.

    (17) Winocur, G. and M. Moscovitch, Memory transformation and systems consolidation. J Int Neuropsychol Soc, 2011. 17(5): p. 766–80.

    (18) Hockley, A. and M.S. Malmierca, Auditory processing control by the medial prefrontal cortex: A review of the rodent functional organisation. Hear Res, 2024. 443: p. 108954.

    (19) Zikopoulos, B. and H. Barbas, Prefrontal projections to the thalamic reticular nucleus form a unique circuit for attentional mechanisms. J Neurosci, 2006. 26(28): p. 7348–61.

    (20) Phillips, R.G. and J.E. LeDoux, Differential contribution of amygdala and hippocampus to cued and contextual fear conditioning. Behav Neurosci, 1992. 106(2): p. 274–85.

  14. Version published to 10.7554/elife.111177.2 on eLife
  15. Mentioned by @bertogliolab.bsky.social on Bluesky

    Complementary stable and dynamic prelimbic ensembles encode learned threat value underlying generalization and discrimination www.biorxiv.org/content/10.6...

  16. Version published to 10.64898/2026.03.08.710406v3 on bioRxiv
  17. eLife

    eLife Assessment

    This potentially valuable study investigates the anti-senescence effects of red light exposure, proposing that reduced SIRT4 levels enhance fatty acid metabolism and H3K9ac, thereby attenuating ageing-related phenotypes. The authors use multiple approaches, including cultured cells, animal models, and molecular analyses, to support their conclusions. Following revision, many of the concerns previously raised by the reviewers have been addressed. The evidence is solid, whereas additional controls and stronger mechanistic data are still needed to fully substantiate the proposed pathway, particularly the mechanism by which red light exposure leads to reduced SIRT4 levels.

  18. eLife

    Reviewer #1 (Public review):

    Summary:

    Deng and colleagues pursue the possibility that red light exposure can provide some benefits and anti-senescence effects in aged mouse models. In addition, they show how red light influence metabolism in cultured keratinocytes. The authors provide a long dissection of the potential paths involved in the changes promoted by red light exposure, identifying CytC oxidase, SIRT4, PPARa and MCD as key players.

    Strengths:

    The authors did a thorough exploration of the multiple potential avenues by which red light exposure influence metabolism. The in vitro and in vivo evidence nicely complement each other.

    Weaknesses:

    This is a challenging hypothesis that would require some additional experimental controls. The pathway dissection, while extensive, sometimes is approach in unconvincing ways and the results are not always evident to judge or interpret. Technically, the western blots and transcriptomic analyses require notable improvements.

    Comments on revised version.

    The revised version of the manuscript provides some improvements. However, I feel that many aspects remain poorly addressed. In the authors' favour, many of these limitations are now acknowledged in their rebuttal, as well as in the discussion section.

  19. eLife

    Reviewer #2 (Public review):

    Summary:

    This work identifies a previously unknown way that red light can slow ageing. The authors show that red light lowers the level of a protein called SIRT4 in skin cells. Reducing SIRT4 boosts fatty acid use and increases a type of histone modification that keeps genes active. These changes help cells clear away signs of ageing, reduce inflammation, and restore normal metabolism. The findings open the possibility of developing new treatments that target SIRT4 to reverse age‑related decline.

    Strengths:

    The evidence is solid because the authors use several complementary methods. They test red light in both cultured cells and naturally aged mice, and they confirm the key role of SIRT4 by silencing its gene. Measurements of metabolism, protein changes, and ageing markers all point in the same direction. However, the exact way red light lowers SIRT4 levels is not fully explained, which leaves a minor gap. Overall, the conclusions are well supported and convincing.

    Weaknesses:

    The paper does not evolve to use the mechanistic discoveries of the manuscript to help our community to identify the mechanism of photobiomodulation, which is not known so far.

    I would like to draw your attention to a recently published paper by Herrera et al. (FEBS Letters 2025, doi:10.1002/1873-3468.70195), which shows that red light (660 nm) stimulates mitochondrial fatty acid oxidation in keratinocytes via AMPK‑dependent phosphorylation of ACC, without altering expression of electron transport chain complexes. I believe this paper is highly complementary to current study.

    Herrera et al. demonstrate that red light increases basal, ATP‑linked, and maximal oxygen consumption rates in keratinocytes specifically through enhanced fatty acid oxidation (inhibited by etomoxir). This independently validates the central finding of the current manuscript ,i.e., red light boosts lipid metabolism, strengthening the robustness of this concept.

    While the current manuscript focusses on the SIRT4‑MCD axis, Herrera et al. identify AMPK phosphorylation and ACC inhibition as key effectors. Authors can integrate and expand their discussion, since SIRT4 downregulation may converge on AMPK activation, or they may represent parallel, reinforcing mechanisms. This would enrich the mechanistic model and open new hypotheses.

    The mechanism of photobiomodulation: Herrera et al. explicitly challenge the prevailing paradigm that red light acts solely via cytochrome c oxidase (by showing long‑lasting effects, unchanged OXPHOS protein levels, and no difference in permeabilized cells). The current finding (red light acts through SIRT4 downregulation, i.e., not direct enzymatic activation, aligns perfectly with Herrera´s critique.

    Long‑term metabolic effects - Herrera et al. show that a single red light exposure elevates oxygen consumption for up to 2 days. The current study focuses on changes at 12‑24 h. Their data extend the time window and suggest that the metabolic reprogramming you describe may persist longer than currently discussed, which is clinically relevant.

    Discussing Herrera et al. results would not only acknowledge independent, corroborating evidence but also allow the authors to position your SIRT4‑centric mechanism within a broader, emerging understanding of red‑light photobiomodulation.

    Comments on the latest version:

    The authors have made a terrific work in answering the reviewers and modifying the manuscript.

  20. eLife

    Author response:

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

    We sincerely thank the editors and reviewers for your careful evaluation of our manuscript and for the constructive recommendations that have helped us improve the rigor, clarity, and balance of the study. We are pleased that the reviewers recognized the potential value of linking red light exposure to SIRT4 downregulation, fatty acid metabolism, H3K9 acetylation, and attenuation of ageing-related phenotypes. We have revised the manuscript extensively in response to the reviewers’ comments.

    In particular, we have clarified the wavelength specificity of the red-light response, reanalyzed and more cautiously interpreted the omics data, improved the presentation and quantification of semi-quantitative experiments, revised statistical reporting, corrected gene/pathway annotations, toned down mechanistic claims where direct evidence was insufficient, and expanded the Discussion to integrate recent evidence on red-light-induced fatty acid oxidation and AMPK/ACC signaling. We also added a dedicated limitations paragraph addressing the use of female mice, the absence of a complete in vivo wavelength-control and source-blocked sham cohort, and the need for future direct metabolic flux and isolated mitochondria studies.

    Public Reviews:

    Reviewer #1 (Public review):

    Weaknesses:

    This is a challenging hypothesis that would require some additional experimental controls. The pathway dissection, while extensive, is sometimes approached in unconvincing ways, and the results are not always evident to judge or interpret. Technically, the western blots and transcriptomic analyses require notable improvements.

    We would like to thank the reviewer for the careful and patient examination of the issues identified in our manuscript. The poor quality of some of the Western blot bands in Figure 4 may have been caused by inappropriate electrophoresis conditions during the Western blot experiments. In the revised manuscript, we will optimize the electrophoresis conditions to obtain higher-quality protein bands and update the quantitative data. Regarding the quantification format, we believe that heatmaps provide a more intuitive representation of trends in protein expression across different treatment groups. This approach more accurately reflects the results of our biological replicates than simply analyzing the significance of differences in the grayscale values of protein bands. For the analysis of transcriptomic data, we will conduct a more detailed analysis of signal pathway enrichment and the identified differentially expressed genes to ensure that predicted genes are excluded from our current results and redundant data presentation is removed.

    Regarding additional experimental controls, such as incorporating experimental data under blue light treatment conditions as a control for red light. While exploring the optimal red light irradiation dose at the cellular level, we simultaneously conducted experiments on the effects of blue light irradiation at the same dose on keratinocyte activity. The results indicated that as the blue light irradiation dose increased (0–160 J/cm2), the keratinocyte activity exhibited a dose-dependent decline. This indicates that blue light is phototoxic to keratinocytes. The relevant experimental results have already been published in our previous study (Communications Biology 2024, doi: 10.1038/s42003-024-06973-1). Taken together with the data from our study, this demonstrates that the anti-ageing effects of red light reported in the current manuscript are indeed driven by red light.

    Reviewer #2 (Public review):

    Weaknesses:

    The paper does not evolve to use the mechanistic discoveries of the manuscript to help our community to identify the mechanism of photobiomodulation, which is not known so far.

    I would like to draw attention to a recently published paper by Herrera et al. (FEBS Letters 2025, doi:10.1002/1873-3468.70195), which shows that red light (660 nm) stimulates mitochondrial fatty acid oxidation in keratinocytes via AMPK‑dependent phosphorylation of ACC, without altering expression of electron transport chain complexes. I believe this paper is highly complementary to the current study.

    Herrera et al. demonstrate that red light increases basal, ATP-linked, and maximal oxygen consumption rates in keratinocytes specifically through enhanced fatty acid oxidation (inhibited by etomoxir). This independently validates the central finding of the current manuscript, i.e., red light boosts lipid metabolism, strengthening the robustness of this concept.

    While the current manuscript focuses on the SIRT4-MCD axis, Herrera et al. identify AMPK phosphorylation and ACC inhibition as key effectors. The authors can integrate and expand their discussion, since SIRT4 downregulation may converge on AMPK activation, or they may represent parallel, reinforcing mechanisms. This would enrich the mechanistic model and open new hypotheses.

    The mechanism of photobiomodulation: Herrera et al. explicitly challenge the prevailing paradigm that red light acts solely via cytochrome c oxidase (by showing long-lasting effects, unchanged OXPHOS protein levels, and no difference in permeabilised cells). The current finding (red light acts through SIRT4 downregulation, i.e., not direct enzymatic activation) aligns perfectly with Herrera´s critique.

    Long-term metabolic effects-Herrera et al. show that a single red light exposure elevates oxygen consumption for up to 2 days. The current study focuses on changes at 12-24 h. Their data extend the time window and suggest that the metabolic reprogramming you describe may persist longer than currently discussed, which is clinically relevant.

    Discussing Herrera et al.'s results would not only acknowledge independent, corroborating evidence but would also allow the authors to position their SIRT4-centric mechanism within a broader, emerging understanding of red-light photobiomodulation.

    We would like to thank the reviewer for providing us with constructive suggestions for discussion. Our results showed that under red light conditions, both glycolipid and lipid metabolism were activated in keratinocytes, and cellular metabolic flux increased. The activation of lipid metabolism directly led to an increase in metabolism-associated H3K9ac and drove the upregulation of anti-ageing-related genes; we believe this is key to the anti-ageing effects of red light. Mechanistic analysis combining proteomics and acetylation proteomics revealed that red light significantly downregulated SIRT4 expression and increased the acetylation of MCD, a protein regulated by SIRT4 that governs cellular fatty acid oxidation rates. Through validation using cell-level knockdown and inhibitors, we confirmed that SIRT4 inhibition exerts anti-ageing effects in vitro and that inhibiting MCD function under red light conditions suppresses H3K9ac. These results establish the role of the SIRT4-MCD signalling axis in mediating the anti-ageing effects of red light.

    The study by Herrera et al. included a substantial body of validation data confirming the role of red light in promoting fatty acid oxidation, providing robust empirical support for our research. Furthermore, Herrera et al. revealed that red light-induced fatty acid oxidation depends on AMPK and ACC phosphorylation. This mechanism of red-light photobiomodulation may refute the notion that its bio-regulatory effects rely solely on the action of mitochondrial cytochrome c oxidase. Furthermore, together with our study revealing that red light exerts anti-ageing photobiomodulatory effects via the SIRT4-MCD signalling axis, these findings independently confirm that red light regulates cellular fatty acid oxidation, thereby demonstrating the pivotal role of activated fatty acid oxidation in the bio-regulatory effects of red light. In the revised manuscript, we will include a discussion on the potential link between the red light-driven downregulation of SIRT4 and the phosphorylation of AMPK/ACC. This will be of positive value in elucidating how SIRT4 exerts its anti-ageing effects by regulating lipid metabolism, as well as in explaining the possible mechanisms by which red light downregulates SIRT4.

    Recommendations for the authors:

    Summary of Major Revisions

    Changes made in the revised manuscript:

    (1) Added a clearer explanation of why the 625-635 nm red-light regimen was considered the active intervention and how the available blue-light data from our previous work support wavelength-dependent effects on keratinocytes.

    (2) Revised the language describing inflammatory regulation. We now avoid presenting red light as producing a uniform anti-inflammatory effect and instead describe selective remodeling of ageing-associated inflammatory and SASP signatures.

    (3) Improved figure presentation and quantification for immunofluorescence, metabolite, and western blot assays; clarified image-analysis regions, replicate numbers, and normalization procedures.

    (4) Reanalyzed transcriptomic, proteomic, and acetyl-proteomic datasets with appropriate multiple-testing correction and corrected erroneous pathway/gene annotations in metabolic gene panels.

    (5) Replaced overly strong causal wording with more conservative language, especially regarding PI3K/Akt/mTOR, cytochrome c oxidase, SIRT4 localization, PPARα immunofluorescence, and direct fatty acid oxidation flux.

    (6) Expanded the Discussion to incorporate Herrera et al. (FEBS Letters 2025, doi:10.1002/1873-3468.70195), highlighting convergence between the SIRT4-MCD model and AMPK/ACC-dependent fatty acid oxidation.

    (7) Corrected typographical, nomenclature, and figure-legend inconsistencies throughout the manuscript.

    Reviewer #1 (Recommendations for the authors):

    (1) Wavelength specificity and need for a non-red-light control

    As a reader, one is left wondering whether the effects are due to red light specifically. An important control would have been to irradiate mice and cells with another light wavelength, such as blue light.

    We agree that wavelength specificity is a critical issue for interpreting photobiomodulation studies. In the revised manuscript, we have clarified that the anti-ageing and metabolic effects described here apply specifically to our 625-635 nm red-light regimen, rather than to visible light in general. We have also added a discussion of our previously published blue-light experiments, in which keratinocyte viability decreased in a dose-dependent manner across the same 0-160 J/cm2 dose range (Communications Biology 2024, doi: 10.1038/s42003-024-06973-1). These data indicate that blue light and red light produce distinct biological outcomes in keratinocytes. Because high-dose blue light was cytotoxic under comparable cellular conditions and because the present study was designed to investigate the long-term effects of red light in aged mice, we did not perform prolonged in vivo blue-light irradiation as an ageing intervention.

    Changes made in the revised manuscript:

    Clarified in the revised Introduction and Discussion that the conclusions are specific to 625-635 nm red light under the irradiation parameters used in this study. (Lines 92 to 94, Lines 1146-1149)

    Added text summarizing the published blue-light comparison data from our previous study (Communications Biology 2024, doi: 10.1038/s42003-024-06973-1), including the dose-dependent decline in keratinocyte activity after blue-light irradiation. (Lines 90 to 92, Lines 1146-1149)

    Added data on the wavelength range of the red light used in this study. (Lines 529 to 531, Fig S1a)

    (2) Complexity of inflammatory effects

    The manuscript repeatedly emphasizes anti-inflammatory effects, yet some cytokines such as IL-18, Ccl2, TNF-α, Ccl2, and IL-8 appear increased. This suggests that the effects may be more complex than presented and may require additional readouts or stronger statistical power.

    We unanimously agree that the inflammatory response to red light should not be described as a simple, uniform suppression of all cytokines. We have demonstrated that changes in the levels of the senescence-associated secretory phenotype (SASP) at the cellular level and in skin tissue following red light treatment not only indicate that red light-induced metabolic activation can reduce the age-related inflammatory baseline, but also reveal red light-driven short-term reparative effects or stress-related cytokine responses. We consider this to be consistent with the findings, and the downregulation of NF-κB-related signalling observed in skin tissue following periodic red light irradiation of aged mice further supports the conclusion that red light alleviates the age-related inflammatory baseline. In fact, in our previous study, we did observe that red light treatment promoted increased levels of the cytokine Ccl2, which plays an important positive role in rapid wound healing (Communications Biology 2024, doi: 10.1038/s42003-024-06973-1). In the revised manuscript, we have reworded the relevant Results and Discussion sections to indicate that red light remodels ageing-associated inflammatory signalling rather than globally reducing every inflammatory mediator. We have also toned down statements suggesting that red light ‘reverses’ or ‘suppresses’ inflammation where the underlying data support a more selective effect.

    Changes made in the revised manuscript:

    Replaced broad terms such as “anti-inflammatory effects” with more precise wording such as “remodeling of ageing-associated inflammatory signaling” where appropriate. (Lines 541 to 543, Lines 573 to 574, Lines 893 to 895)

    Expanded the Discussion to explain that red-light-induced metabolic activation may simultaneously reduce senescence-associated inflammatory tone while allowing transient reparative or stress-related cytokine responses. (Lines 1135 to 1142)

    (3) Figure clarity, semi-quantitative methods, western blot quality, and inconsistent band patterns

    Many differences are difficult to see or require orthogonal validation. Some tissue-specific signals and western blots are difficult to judge. Several western blots are of poor quality, and multiple markers show inconsistent band profiles across experiments, including SIRT4 in Figure 5.

    We thank the reviewer for highlighting these technical and presentation issues. We have reviewed the semi-quantitative data and revised the presentation of the figures to improve their interpretability. For Western blot experiments, we optimised the electrophoresis and transfer conditions, replaced low-quality representative images where possible, and updated the semi-quantitative results. Furthermore, regarding the lack of clarity in the SIRT4 protein band, we have conducted repeat experiments and updated the main text to include a clearer image of the band. The issue with the annotation of the protein location was in fact due to an oversight during the data analysis process; we have carried out a detailed review and provided the uncropped full-length Western blot images for all experiments in the Supplementary Materials for the reviewers’ scrutiny. Finally, we would also like to point out that factors such as sample origin, protein extraction, electrophoresis conditions, antibody exposure time, and potential non-specific detection may all contribute to differences in band patterns. At present, the core conclusions regarding SIRT4 are supported by multiple lines of evidence, including mRNA analysis, immunofluorescence, Western blotting of bands at the expected sizes, and SIRT4 knockdown experiments, rather than being based solely on any single semi-quantitative Western blot result. We therefore believe that the conclusions drawn from the data presented in the revised manuscript are equally convincing.

    Changes made in the revised manuscript:

    Replaced or improved low-quality Western blot panels and updated quantitative analyses in revised Figures 3-5 and associated supplementary material. (Fig 3m, Fig 4, Fig 5f)

    Clarified the normalization approach for H3K9ac/H3 and target/loading-control comparisons, and the use of Actin or H3 as appropriate loading controls. (Lines 283 to 293)

    Bands with nonspecific profiles were excluded from quantitative conclusions and the manuscript conclusions no longer depend on those ambiguous signals.

    Revised the Results (Repeat the experiment to update the low-quality Bands) to avoid overstating changes that are not clearly visible or not supported by statistical analysis. (Fig 4n and r)

    (4) Choice of pharmacological agents and need for genetic strategies

    The choice of drugs in Figure 4 is puzzling. More specific and widely used inhibitors could be used to block PI3K/Akt or mTOR, and natural agonists such as insulin or EGF could be used. Genetic strategies should complement these observations.

    We agree that pharmacological perturbation experiments should be interpreted with caution. In this study, our criteria for selecting inhibitors were based on transcriptomic and proteomic analyses; we sought to determine how the most direct inhibition of red light-activated signalling pathways would affect H3K9ac levels. In the revised manuscript, we have clarified the rationale for the compounds used and have reduced the causal weight assigned to these inhibitor/agonist experiments. These data are now presented as supportive evidence that red light is associated with metabolism-related signalling changes, rather than as definitive proof that PI3K/Akt/mTOR is the primary upstream mechanism. We have also emphasised the genetic SIRT4 knockdown experiments as a more direct mechanistic test for the SIRT4-centred part of the model. We acknowledge that additional experiments using more selective inhibitors, physiological agonists such as insulin or EGF, and genetic perturbation of PI3K/Akt/mTOR components would be valuable for future studies.

    Changes made in the revised manuscript:

    Revised the text describing pharmacological experiments to distinguish supportive pathway modulation from direct causal evidence. (Lines 787 to 789)

    Added a limitation and future direction noting that genetic perturbation of PI3K/Akt/mTOR and physiological pathway activation with insulin or EGF would strengthen the model. (Lines 1190 to 1194)

    (5) Serum NADH measurement

    In Figure 1t, the authors measure serum NADH. NADH is poorly detectable in serum or plasma, and changes may reflect blood-cell lysis during collection rather than circulating NADH.

    We appreciate this technical concern. We have revised the manuscript so that serum NADH is no longer used as a central mechanistic readout. We now treat this measurement only as an exploratory indicator of systemic redox-related changes and explicitly acknowledge that serum or plasma NADH is vulnerable to artifacts from blood-cell disruption during sampling. The mechanistic interpretation has been shifted toward cellular and tissue measurements, including intracellular NADH/NADPH/GSH, ATP, acetyl-CoA, fatty acid uptake, and H3K9ac, which are more directly relevant to keratinocyte metabolic remodeling.

    Changes made in the revised manuscript:

    Removed serum NADH from the main causal argument linking red light to metabolic flux and H3K9ac.

    Placed greater emphasis on cell-based metabolite assays, tissue acetyl-CoA, and H3K9ac measurements as the main metabolic-epigenetic evidence. (Lines 582 to 587, Fig 1s)

    (6) Direct assessment of glycolysis and fatty acid oxidation

    The authors propose that red light increases glycolysis and fatty acid oxidation, but this could be assessed directly rather than through surrogate measures.

    We agree. Our current data include multiple metabolic readouts, including glucose and fatty acid uptake, ATP, NADH/NADPH/GSH, triglycerides, fatty acids, pyruvate, lactate, acetyl-CoA, and MCD-dependent changes; however, these assays are not equivalent to direct flux measurements such as Seahorse extracellular flux analysis, isotope tracing, or etomoxir-sensitive respiration. We have therefore revised the wording throughout the manuscript to distinguish metabolic remodeling and fatty-acid-oxidation-related signatures from direct measurements of fatty acid oxidation flux. We also incorporated the recent independent work by Herrera et al., which directly measured oxygen consumption and demonstrated red-light-induced fatty acid oxidation in keratinocytes. This external evidence supports the biological plausibility of our SIRT4-MCD model while making clear which aspects are directly measured in our study and which are inferred.

    Changes made in the revised manuscript:

    Replaced overstrong language such as “red light increases fatty acid oxidation” with “red light promotes PPAR-α-related fatty acid metabolism pathway” where direct flux data were not measured in our experiments. (Lines 882 to 883)

    Expanded the Discussion to integrate direct FAO evidence from Herrera et al. and to place the SIRT4-MCD axis within a broader red-light metabolic framework. (Lines 1169 to 1189)

    (7) Incorrect annotation of metabolic genes in Figure 3e

    Acss2, Aldh3b1 and Aldh3a1 are not glycolytic enzymes, Aldh3a3 does not appear to exist, and several enzymes classified as FAO are fatty acid synthesis enzymes. This questions the interpretation of the data.

    We thank the reviewer for identifying these annotation errors. We have rechecked the gene names and pathway assignments in the transcriptomic analysis and corrected the metabolic gene panels. We have confirmed that Acss2 is an acetyl-CoA synthase involved in the metabolism of acetate to acetyl-CoA. The Aldh family genes, meanwhile, are associated with aldehyde metabolism and detoxification. We have made the corresponding adjustments in the manuscript. The incorrectly listed Aldh3a3 entry has been removed. Furthermore, we have categorised genes involved in fatty acid metabolism as ‘fatty acid metabolism-related genes’, rather than grouping them all under the FAO category. These revisions have significantly improved the accuracy of the metabolic interpretation.

    Changes made in the revised manuscript:

    Reannotated Figure 3e and the corresponding Results text to correct glycolysis, TCA cycle, pentose phosphate pathway, and fatty acid metabolism categories. (Fig 3d)

    Removed the erroneous Acss2 and Aldh family genes. (Fig 3d)

    Revised the metabolic model to avoid using incorrectly grouped genes as evidence for direct fatty acid oxidation. (Lines 707 to 709)

    (8) Transcriptomic analysis and implausible volcano-plot p-values

    The transcriptomic analysis raises concerns. For example, the volcano plot in Figure 3d appears incorrect, with -log10(P-value) around 300 despite n=3 biological replicates.

    We thank the reviewer for pointing out this important issue. We have reopened the transcriptomic data and found that the extremely high -log10(P-value) in the original volcano plot were caused by the automatic replacement of very small P-values—generated during the differential expression analysis—with zero in the tabular data. To avoid misleading visualisations, we have regenerated the volcano plot using Q-values in place of the original P-values. Differentially expressed genes were defined as those with a Q-value < 0.05 and |log2 fold change| > 1. Furthermore, for visualisation purposes only, the upper limit for q-values was set to 1 × 10-50 for values below 1 × 10-50. This adjustment does not affect the statistical classification of differentially expressed genes but prevents over-interpretation of extremely small values. The revised volcano plots and legends have been updated accordingly. To avoid any potential misinterpretation arising from these updates, the updated volcano plots are presented in the supplementary materials.

    Changes made in the revised manuscript:

    Reanalyzed transcriptomic data using appropriate multiple-testing correction and revised the volcano plot. (Fig S3a)

    Corrected the y-axis transformation and removed implausible -log10(P-value) presentation. (Fig S3a)

    Updated Methods to specify the statistical workflow for transcriptomic differential expression and pathway enrichment. (Supplementary materials Lines 46 to 52)

    Moved the analysis of metabolic pathways based on transcriptomic data to the supplementary material, thereby reducing the reliance of the conclusions on transcriptomic data (Fig S3b).

    (9) Need to tone down mechanistic claims regarding PI3K/Akt/mTOR, cytochrome c oxidase, SIRT4, and PPARα

    The mechanisms proposed must be toned down. PI3K/Akt/mTOR should not be called glycolytic pathways, the link to red light or cytochrome c oxidase is vague, SIRT4 reduction requires mitochondrial counterstaining, and PPARα appears cytosolic after SIRT4 knockdown.

    We agree and have substantially revised the mechanistic language. PI3K/Akt/mTOR is no longer referred to as a ‘glycolytic pathway’; instead, it is described as a metabolism-related signalling axis that may influence glucose uptake, growth and nutrient-responsive metabolism. We have also toned down statements attributing red-light effects directly to cytochrome c oxidase, as our study primarily examines downstream metabolic and epigenetic remodelling rather than direct photoreceptor activation. With regard to SIRT4, we have revised the text to avoid interpreting changes in SIRT4 immunofluorescence alone as evidence of altered mitochondrial abundance or mitochondrial localisation. The conclusion is now based on a combination of SIRT4 mRNA levels, western blot bands of the expected size, immunofluorescence trends, and SIRT4 knockdown phenotypes. With regard to PPARα, we have re-examined the PPARα antibody used for the cellular immunofluorescence experiments. In the original Figure 5p, we mistakenly used a PPARα antibody (PPARα, Abclonal, A25296) that is only suitable for Western blot (WB) experiments; we believe this was the cause of the mislocalisation of the fluorescent signal; Consequently, we conducted new experiments using a PPARα antibody (PPARα, Abclonal, A22887) specifically designed for cellular immunofluorescence. The relevant experimental data have been corrected in the manuscript.

    Changes made in the revised manuscript:

    Replaced “PI3K/Akt/mTOR glycolytic pathway” with “The PI3K-AKT signalling pathway is involved in the regulation of glucose metabolism” throughout the revised manuscript. (Lines 701 to 705, Lines 809 to 810, Lines 813, Lines 1193)

    Reduced mechanistic certainty around cytochrome c oxidase and framed it as a possible upstream photoreceptor rather than an experimentally proven mechanism in this study. (Lines 827 to 830, Lines 842 to 848)

    Repeat the PPARα immunofluorescence staining experiment. (Fig 5p)

    (10) Need for isolated mitochondria experiments and red/blue light comparison of mitochondrial respiration

    If the effect of red light relies on mitochondrial cytochromes, additional proof would be needed, potentially using isolated mitochondria and comparing how red and blue light influence respiration capacity.

    We agree that isolated mitochondria experiments would be an important way to test direct mitochondrial photoreception. Because the present study was designed around cellular and in vivo metabolic-epigenetic remodeling, we did not perform isolated mitochondria irradiation experiments. To address this concern, we have toned down statements implying direct cytochrome activation and revised the Discussion to distinguish between direct mitochondrial photoreceptor models and downstream metabolic reprogramming. We also added a future direction proposing isolated mitochondria or permeabilized-cell experiments comparing red and blue light effects on respiration, ATP-linked OCR, maximal respiration, and FAO-dependent respiration. The revised manuscript now emphasizes that our data support a downstream SIRT4-MCD-H3K9ac mechanism after red-light exposure, while the proximal photophysical event remains to be fully defined.

    Changes made in the revised manuscript:

    Added discussion of the need for isolated mitochondria, permeabilized-cell, and wavelength-comparison respiration experiments. (Lines 1194 to 1197)

    Reviewer #2 (Recommendations for the authors):

    (1) Statistical reporting, post-hoc tests, normality/equal-variance tests, exact p-values, and FDR control

    The manuscript states that one-way ANOVA followed by Tukey or Dunnett tests was used, but it does not consistently specify the post-hoc correction for each figure. Normality and equal-variance tests are not reported, p-values are shown only as asterisks, and FDR control is not mentioned for transcriptomics and proteomics.

    We agree that the statistical reporting needed to be more complete. We have revised the Statistics and reproducibility section and the figure legends to specify the statistical test used for each experiment, the post-hoc correction applied after ANOVA, the number of independent biological replicates, and the definition of error bars. Where multiple comparisons were performed, we now state whether Tukey’s or Dunnett’s correction was used. Regarding P-value presentation, we have retained the use of asterisks in the figures as visual indicators of statistical significance to maintain figure readability. For transcriptomic, proteomic, and acetyl-proteomic analyses, we have revised the Methods section to state that multiple-testing correction was performed using the Benjamini–Hochberg false-discovery-rate procedure. Adjusted P values or Q values were used for differential-expression and pathway-enrichment analyses. These revisions clarify the statistical workflow and strengthen the reproducibility of the study.

    Changes made in the revised manuscript:

    Revised the Statistics and reproducibility section to define statistical tests, post-hoc corrections, assumption checks, and multiple-testing correction. (Lines 509 to 522)

    Updated relevant figure legends to include n values, statistical tests, post-hoc corrections, and definitions of significance symbols. (Lines 515 to 516)

    Added FDR control details for RNA-seq, proteomics, acetyl-proteomics, and pathway-enrichment analyses. (Lines 450 to 455)

    (2) Figure clarity and quantitative analysis of fluorescence, JC-1, metabolite, and western blot data

    Several figures lack clarity or appropriate quantification. Figure 1i-j H3K9ac quantification should be based on whole-image or multiple fields; Figure 2e JC-1 should include red/green ratio quantification; Figure 2k-p metabolite data should include absolute concentrations; Figure 3j needs appropriate loading controls.

    We appreciate these specific suggestions and have revised the figure presentation accordingly. For H3K9ac immunofluorescence in skin sections, we have clarified the anatomical region quantified and performed a more objective quantification using multiple fields/regions per section rather than relying on a visually selected dashed area. The dashed regions in the representative images were made clearer and the quantification criteria were added to the Methods and legend. For JC-1 staining, the bar chart on the right-hand side of the mitochondrial membrane potential fluorescence image in Figure 2e shows the quantitative data for the red/green fluorescence ratio obtained from independent experiments; compared with providing only a representative image, these data offer a more easily interpretable quantitative measure of mitochondrial membrane potential. To avoid any potential misunderstanding, we have corrected the vertical axis. For metabolite assays, we clarified normalization to cell number or protein content and revised the data presentation to include absolute or normalized concentrations where available, rather than relying solely on fold changes with variable y-axis scaling.

    Changes made in the revised manuscript:

    Revised Figure 1i-j quantification using multiple fields/regions per mouse section and improved dashed-region visibility. (Lines 431 to 440, Fig 1i and j)

    Corrected the vertical axis of the quantitative data for the JC-1 red/green fluorescence ratio. (Fig 2e and Fig S2a)

    Updated metabolite panels and/or source data to include absolute or protein-normalized values where available, and standardized y-axis interpretation. (Fig 2k-p and Fig 4s and v, Given the diversity of intracellular fatty acid and triglyceride species, absolute quantification based solely on absorbance measurements would be technically challenging and may not accurately reflect the content of each molecular component. Therefore, we presented the changes in fatty acid and glycerol levels as percentage-normalized relative absorbance values, which allowed consistent comparison among the experimental groups.)

    (3) Experimental design limitations: sex of mice and sham control

    Only female C57BL/6 mice were used, although aging and metabolic responses can be sex-dependent. The thermal-control argument lacks a true sham control in which mice are placed in the same apparatus with the light blocked at the source.

    We agree with these points. We have added a section on limitations stating that all aged mice used in this study were female, and that sex-dependent responses to red light, SIRT4 regulation, metabolism and skin ageing should be investigated in future studies using both male and female cohorts. Furthermore, regarding the design of the non-irradiated control group: although the control mice underwent the same depilation and routine procedures, they did not receive red light irradiation. However, we also acknowledge that establishing a sham-irradiated control group with light shielding would allow for stricter control of factors such as restraint, contact with equipment and procedural stress. However, given that this experiment involved a continuous cyclic photoperiodic treatment lasting two years, we were unable to supplement the study with a control experiment involving only red light shielding. Nevertheless, based on the fact that we observed only minimal changes in the mice’s skin temperature following red light irradiation, we believe that the primary factor driving the alleviation of the skin ageing phenotype in the mice remains red light-induced.

    Changes made in the revised manuscript:

    Added a limitation noting that the study used female C57BL/6 mice only and that sex as a biological variable should be addressed in future studies. (Lines 1198 to 1201)

    (4) Textual errors, nomenclature inconsistencies, and ChIP-qPCR normalization

    Several textual errors and inconsistencies should be corrected, including Pparg1a/Ppargc1a, Ricotr/Rictor, Pi3k/PI3K, Sirt4/SIRT4 protein nomenclature, and the use of RPL30 normalization in ChIP-qPCR without showing that RPL30 is unchanged.

    We thank the reviewer for their careful reading. We have corrected the typographical errors and standardised gene and protein nomenclature throughout the manuscript and figure legends. Specifically, Ppargc1α has been corrected to Ppargc1a, Ricotr to Rictor, and the capitalisation of PI3K has been standardised. We now use Sirt4 for the mouse gene and SIRT4 for the protein, applying the same convention to other genes and proteins. For ChIP-qPCR, we have revised the Methods and Results sections to describe normalisation against input and IgG controls more clearly, and to specify the role of the RPL30 locus as an internal control. We have also included data in the Supplementary Materials showing relative enrichment of H3K9ac in the RPL30 promoter region in PAM212 cells before and after red light irradiation; the results indicate that H3K9ac enrichment at the RPL30 locus remained stable across treatment groups after normalization to input DNA and correction against IgG background. This result indicates that the use of RPL30 as an internal control in ChIP-qPCR experiments is feasible.

    Changes made in the revised manuscript:

    Corrected Ppargc1a, Rictor, PI3K, Sirt4/SIRT4, and related nomenclature throughout the manuscript.

    Supplement the experimental results on the effect of red-light irradiation on the level of H3K9ac enrichment at the RPL30 locus in keratinocytes. (Lines 339-352, Lines 539 to 541, Fig S1d)

    (5) Additional Revision Addressing the Public Review and Herrera et al.

    The reviewer suggested integrating the recent study by Herrera et al. showing that 660 nm red light stimulates mitochondrial fatty acid oxidation in keratinocytes through AMPK-dependent phosphorylation of ACC, without changing electron transport chain complex expression. The reviewer also noted that these findings may complement the SIRT4-MCD axis and challenge a cytochrome-c-oxidase-only model of photobiomodulation.

    We are grateful for this constructive suggestion. We have expanded the Discussion to incorporate Herrera et al. and to place our SIRT4-MCD-centered mechanism within the broader emerging model of red-light-driven metabolic remodeling. Herrera et al. provide direct oxygen-consumption evidence that red light enhances fatty acid oxidation in keratinocytes and that this effect involves AMPK/ACC signaling. This is highly complementary to our data, in which red light decreases SIRT4, increases acetylation of MCD, promotes fatty-acid-metabolism-related signatures, elevates acetyl-CoA, and increases H3K9ac. In the revised Discussion, we propose two nonexclusive models: red light-induced SIRT4 downregulation may converge with AMPK/ACC-dependent relief of fatty acid oxidation, or the two pathways may represent parallel reinforcing mechanisms that together enhance lipid metabolic flux.

    Changes made in the revised manuscript:

    Added a paragraph discussing Herrera et al. in the revised Discussion. (Lines A1169 to 1189, Lines 1206 and 1208)

    Revised the conceptual model of red-light photobiomodulation to emphasize downstream metabolic reprogramming rather than direct cytochrome c oxidase activation alone. (Lines 827 to 830, Lines 843 to 844)

    Added future directions to test whether red-light-induced SIRT4 downregulation causally affects AMPK/ACC phosphorylation and FAO-dependent respiration. (Lines 1169 to 1189)

    We again thank the editors and reviewers for their thoughtful and constructive comments. The revised manuscript now provides a more rigorous and balanced presentation of the evidence, distinguishes direct measurements from inferred metabolic flux, corrects pathway annotations, improves figure quantification and statistical transparency, and places the SIRT4-MCD-H3K9ac mechanism within a broader framework of red-light-induced fatty acid metabolic remodeling. We believe these revisions substantially strengthen the manuscript and clarify both the significance and the limitations of our findings.

  21. Version published to 10.64898/2026.03.08.710406v2 on bioRxiv
  22. eLife

    eLife Assessment

    This valuable study examines how the prelimbic cortex represents learned and generalized threat over time and identifies potentially distinct stable and dynamic subnetworks that may support these functions. The work is conceptually interesting and is strengthened by the longitudinal calcium imaging approach and the inclusion of key control groups. However, the evidence supporting the claims is incomplete, particularly because the interpretations regarding inference, time-dependent representational change, and the dissociation of neural activity from freezing behavior extend beyond what is currently established by the data.

  23. eLife

    Reviewer #1 (Public review):

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a no-shock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure.

    To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences. However, in its current form, the manuscript does not yet fully support the strength of the conceptual claims. Several issues limit confidence in the interpretation, including the possibility that repeated testing itself contributes to changes across days, uncertainty about the relationship between neural activity and freezing behavior, limited quantitative documentation of longitudinal cell registration, and a number of problems in figure clarity and statistical framing. Overall, the study contains promising observations, but the claims should be narrowed, and several analyses or controls would be needed to fully support the proposed framework.

    Detailed Comments

    (1) A general concern is that the repeated test procedure itself may contribute to extinction. Because the animals are exposed to multiple CS frequencies across multiple test days, and each tone is presented three times per session, some of the reported changes in behavior and neural activity across days could reflect extinction or repeated nonreinforced retrieval rather than the passage of time per se. This is especially relevant given that the manuscript makes claims about recent versus remote representations and representational drift over 30 days. At a minimum, the authors should discuss this limitation explicitly and temper claims about time-dependent changes. Ideally, they would include a control group in which animals are tested only once or twice (e.g., at an early and later time point with fewer CS frequencies), or a reduced-frequency testing design that minimizes extinction while still allowing evaluation of recent versus remote memory.

    (2) More generally, some of the reported learning-related neural differences may be driven by behavioral differences, particularly freezing, rather than by learning or generalization per se. For example, animals that freeze more to certain frequencies may show corresponding neural response differences simply because freezing alters PL activity. The authors should examine this possibility more directly. Analyses testing whether recorded cells encode freezing behavior, or whether tone frequency-related neural differences remain robust when comparing high- and low-freezing epochs, would help determine whether the reported effects reflect learned stimulus value rather than behavioral state differences.

    (3) A central feature of the manuscript is the analysis of neural response properties over an extended period of time, up to 30 days after learning. However, aside from a brief mention in the Methods that spatial registration was used, the manuscript provides very little quantitative information about this critical aspect of the study. The paper would be strengthened by including explicit metrics describing longitudinal cell tracking, such as the number and proportion of ROIs retained across all sessions, distributions of spatial-footprint correlations or centroid distances across days, and representative examples of matched imaging fields over time. Without this information, it is difficult to assess how strongly the longitudinal claims are supported.

    (4) The text states that "Figs. 1c and 1d show GCaMP6f expression in PL, representative calcium footprints, and activity traces". However, the figure as presented does not clearly show all of these elements, at least not in a way that matches the description in the Results. The correspondence between text and figure should be corrected.

    (5) The labeling of Figure 2a is insufficient for interpretation. The legend states that the panel shows raster plots of sound responsiveness, but the axes and scaling are not clearly defined. It is not clear from the figure what the x-axis represents, whether the y-axis corresponds to individual neurons, where the CS period occurs, or what the activity scale at the right denotes. Also, the term 'rasters' implies that spikes were analyzed. It seems that the spike inference approach (CASCADE) was only used for later analyses. Perhaps 'heat-plot' would be more accurate here? Generally, this figure should be annotated more clearly so that the reader can understand it without referring back to the Methods.

    (6) In relation to Figure 3, the analysis of population-averaged responses across tone frequencies is useful, but the manuscript would be stronger with additional statistical analyses across time and across groups. For example, if the authors want to argue that learning induces graded changes in neural responses and that these evolve across time, they should directly compare within-group responses across days and also compare matched frequencies between the conditioned groups and the no-shock controls. These analyses would help establish whether the observed differences are genuinely learning dependent and whether they change significantly over time.

    (7) The inclusion of two different CS+ frequencies and a no-shock control is a strength of the study and substantially improves the interpretation that graded neural responses are related to learning and generalization rather than to simple sensory processing or passage of time. That said, I am not entirely comfortable with the use of the term "inference" throughout the manuscript. What is being measured here appears closer to sensory generalization than inference in a stronger cognitive sense. The current task does not clearly require that animals infer hidden structure or stimulus value through abstract reasoning; rather, the generalized stimulus may simply be treated as similar to the conditioned cue. The terminology should therefore be reconsidered or softened.

    (8) I also found the use of the term "valence" somewhat problematic. The manuscript appears to use valence to refer to graded responding across tones with different aversive significance, but valence typically refers more broadly to distinctions between appetitive and aversive value. Here, terms such as "threat value," "aversive value," may be more precise. The authors should consider revising this language throughout.

  24. eLife

    Reviewer #2 (Public review):

    Summary:

    The following points are those that occurred to me across readings of the paper. They are listed in what I take to be the order of their significance. Many of the points relate to the loose use of language and invocation of concepts that are not warranted, given the study design and results obtained.

    Major Comments:

    (1) The concept of ensemble turnover is interesting - the way it is introduced and discussed implies some type of spontaneous change in the neural underpinnings of fear discrimination and generalization in the PL. But, of course, every trial involves an opportunity to learn about the threat CS or the generalization test stimuli, and I am troubled by the thought that stability in the neural underpinnings of fear discrimination and generalization will actually reflect the level of defensive behaviours evoked on different trial types and/or the discrepancy between those behaviours and the outcome of a given trial in the generalization test. That is, stability in the neural underpinnings may be related to an animal's certainty or uncertainty in the contingency between a stimulus and danger; or, put another way, an animal's confidence that danger will or won't occur given the presence of some stimulus. This is not uninteresting. It is, however, not considered anywhere in the paper, which is overloaded with references to inferred threat values and integration of information across different types of stimuli. The protocol is not one that requires inference about anything or integration across anything.

    (2) I appreciate the link to Gu and Johansen in paragraph 3 of the Introduction, but the type of generalization under investigation here is not the same as the type of 'generalization' studied by Gu and Johansen [who used a sensory preconditioning protocol]. Nonetheless, the authors have forced the language used by Gu and Johansen into their paper, and this has created tension [at least for this reader] as the concepts introduced by Gu and Johansen [inference, integration] are simply not relevant given the generalization protocol used here. Here are a few examples of points where the tension might interfere with a reader's understanding:

    a. 'We hypothesized that generalization to novel stimuli depends on stable subnetwork organization that enables comparisons between learned and inferred valence, as well as population-level features that reduce variability across related representations.'

    I understand the words in the hypothesis, but can't form a representation of what is being said because of the reference to terms that stand in need of clarification [inferred valence, variability across related representations], but, ultimately, won't be clarified. This needs to be re-expressed so that the reader can appreciate what is being said.

    b. 'Our results show that stable cortical subnetworks integrate the emotional "gist" of memory and inferred valence for novel cues over time, despite ongoing ensemble reorganization, and that population-level firing rate similarity across stimulus presentations determines threat generalization.'

    Again, what does this mean? How is the gist of a memory integrated with inferred valence for novel cues over time? The statement simply doesn't make sense. This needs to be rewritten for clarity.

    c. 'In CS⁺15 mice, positively modulated sound-responsive neurons exhibited graded tone activity reflecting the contingency learned valence as well as the inferred valence of novel tones across testing days...'.

    Can this be rewritten as 'In CS⁺15 mice, positively modulated sound-responsive neurons exhibited graded activity to the tone CS and its variants that were used to assess generalization.'? The overloading of the text with references to 'contingency learned valence' and 'inferred valence' is unnecessary and makes it much harder to understand what has been shown in the results.

    (3) Re the same passage of text as in 2c:

    Is it the case that these neurons are simply tracking the expression of freezing to the various tones? The same question applies to the results obtained for the CS+3 mice. If this is the case, then why should the results be taken to support the banner statement that 'Sound-modulated PL population responses encode learned and inferred valence' - these analyses do not support that statement. And, as indicated, I don't believe that the language of learned and inferred valence is appropriate to such statements, given the nature of the protocol used and results obtained. It is a study looking at how populations of neurons in the PL respond during presentations of auditory stimuli that were subject to discriminative conditioning, and during tests of generalized freezing to other [intermediate] auditory stimuli.

    (4) It is stated that:

    'In no-shock controls, although both positive and negative responses were present, population activity was not modulated by tone frequency or valence'.

    What does this mean? I can understand that population activity was not modulated by tone frequency. But what does it mean to say that it was not modulated by valence? Why should it have been when none of the tones were conditioned in this group and, hence, mice were responding to all the tones equally? And given that this is true, I don't understand the use of 'valence' here, or the subsequent statements in this paragraph that 'graded responses require associative learning' and that 'PL population responses encode graded sound-valence associations that reflect both learning and inference, closely matching behavioral generalization.' The latter statement is particularly unwarranted and, again, highlights a major issue with the paper. It could and should be rewritten as 'PL population responses reflect behavioral generalization.' There is nothing in the additional language that adds to the reader's understanding of what has been shown. The reference to 'graded sound-valence associations that reflect both learning and inference' is completely unwarranted, given the nature of this study. It is anathema to the vast literature on stimulus generalization. If the authors wished to make statements of this sort, they should have taken a different approach, perhaps using protocols like those featured in Gu and Johansen.

    (5) The section titled, 'Consistently active neurons preserve valence representations as newly recruited neurons sharpen remote memory traces' ends with the following summary:

    'Together, these results indicate that consistently active neurons maintain stable representations of learned and inferred sound associations across time, whereas neurons recruited after conditioning progressively acquire graded tuning at later retrieval stages. This dynamic refinement suggests that cortical memory representations become increasingly selective during systems consolidation, while a stable neuronal subpopulation preserves the core emotional content of the memory.'

    Once again, the summary is not in keeping with the results obtained. The 'dynamic refinement' of representations is far more likely to reflect the repeated testing across days 1, 15, and 30 rather than anything to do with systems consolidation - at the very least, it is the simplest interpretation of the results. The impact of repeated testing is evident in the sharpening of generalization gradients over time, which is contrary to what is otherwise observed in the literature - the incredibly well -documented broadening of generalization gradients with time. Given this impact of repeated testing, surely the changes in the neuronal population that underlie performance are more likely to reflect the learning that occurs on days 1, 15, and 30, which is reflected in reduced freezing to the non-conditioned tones. If this is a reasonable take on the results, then I don't see the basis for invoking systems consolidation at all, and I don't see the basis for inferring a stable neuronal subpopulation that preserves the emotional content of the memory. Rather, non-reinforced presentations of 'never-reinforced' tones result in recruitment of additional neurons that result in suppression of freezing responses to those stimuli.

    (6) In the section titled, 'Population vector similarity at stimulus onset determines degree of generalization', it is stated that:

    'Because population similarity peaked shortly after stimulus onset, we quantified similarity during the first 5 s after tone onset relative to the CS⁺. In CS⁺15 mice, population similarity was highest for 15/15 and 15/11 tone pairs with no differences between them.'

    Isn't this consistent with the view that the population response in the PL simply reflects the level of freezing? Freezing to the 15-15 and 15-11 tones is most likely to be similar on their first presentation prior to the effects of extinction on the 11 Hz tone; hence the results obtained. That is, these results appear to clearly indicate that neuronal responses in the PL reflect the degree of stimulus generalization, as evidenced in freezing behavior. Given all that we know about the involvement of the PL in expressing fear responses, it is not appropriate to claim that 'population vector similarity at stimulus onset *determines* the degree of generalization. The PL responses simply reflect the varying levels of performance displayed to the different types of tones. What have I missed that could be taken to support additional statements?

    Later in the same section, it is stated that 'population-level similarity at stimulus onset scales with behavioral threat generalization and is maximal for tones associated with robust threat responses.' For simplicity and, therefore, clarity, this should be rewritten as 'population-level similarity at stimulus onset reflects behavioral threat generalization.'

    (7) In the section titled, 'Different subnetworks encode acoustic versus learned properties of sound association', it is stated that:

    'Our previous analyses show that learned and inferred associations are represented at the population level. However, these results do not resolve whether graded responses arise from pooled activity of frequency-selective neurons or from subnetworks encoding integrated learned valence across tones.'

    What does it mean to say 'integrated learned valence across tones'? As it presently stands, the meaning of the phrase is unclear. It only makes sense if one supposes that generalized freezing responses to the 11 and 7 kHZ tones reflect separate associations between those tones and the aversive foot shock US. This supposition is inconsistent with the rich literature on generalization of Pavlovian conditioned fear responses. Specifically, it is inconsistent with the many theories of fear generalization, which attribute the reduction in fear as one moves away from the specific conditioned stimulus to a decrement in the ability of the test stimulus to activate the trained CS-US association. My strong impression is that the authors would do well to ground their findings in theories of stimulus/fear generalization, of which there are many. This would better serve the results obtained [and the reader's appreciation of them] - at present, the unnecessary invocation of concepts does very little to enhance the reader's appreciation or understanding of what has been found in the study.

    (8) Another example of what has been a common theme in this review :

    '...we hypothesized that the PL active ensemble segregates into functionally distinct subnetworks: one encoding tone-specific sensory features with dynamic characteristics, and another responding to all frequencies encoding stable core memory content and inferred emotional valence.'

    What does it mean to say 'all frequencies encoding stable core memory content and inferred emotional valence'? Do the authors mean to say '...and another that tracks freezing/defensive responses regardless of whether they were elicited by the trained CS or one of the generalization test stimuli'?

    (9) It is stated that - 'Graded clusters encode emotional valence but constitute only a fraction of the active population; yet valence coding at the population level remains accurate and precise. This indicates that neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.'

    What does this mean? Are the authors trying to say that - 'Some clusters of PL neurons track freezing responses. In spite of the fact that these are only a fraction of the total active neuronal population, the population-level response of PL neurons also tracks the levels of fear to the trained tone and its variants used in the test for generalization.' If this is what one wants to say, then the final statement in the reproduced section does not follow. That is, there is no indication that 'neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.' As noted, the characteristics of other ensembles that become active across the repeated tests on days 1, 15, and 30 are more likely to reflect learning from non-reinforcement that occurs within and across those sessions. Perhaps this is what is meant by the phrase, 'shaped by associative processes'? If so, it should be stated explicitly instead of left to the reader to work out.

    (10) The following points all relate to the Discussion and reiterate many of the points above.

    a. 'A subset of neurons remains consistently active across sessions, preserving core components of the memory trace and supporting inference of emotional valence for novel sounds, while neurons recruited after conditioning progressively acquire valence selectivity at remote time points.'

    'Inference of emotional valence' is unclear and unwarranted for all of the reasons provided above regarding the use of language.

    b. '...Our data reconcile these views by demonstrating that cortical representations of emotional valence emerge rapidly after learning and persist within stable subnetworks, even as the broader population undergoes substantial turnover. This architecture preserves core mnemonic content while allowing flexibility in the surrounding ensemble.'

    These statements assume that the PL neuronal responses reflect something more than the levels of freezing behavior to the different stimuli; what are the grounds for this assumption?

    c. 'Importantly, these subnetworks encode both learned contingencies and the inferred valence of novel stimuli along a graded representational axis, suggesting that strong recurrent connectivity provides a stable scaffold for emotional memory representations.'

    What is a graded representational axis, and what part of the first statement suggests that 'strong recurrent connectivity provides a stable scaffold for emotional memory representations'? If the authors' goal was to make statements about emotional memory representations vis-à-vis emotional memory content, they should have used protocols that allowed them to probe such content. The auditory fear conditioning protocol used here [followed by tests for generalization to other auditory stimuli that differ in frequency from the conditioned tone] is not one that lends itself to analysis of emotional memory representations or content.

    d. 'Dynamic tone-selective responsive neurons emerge independently of learning, as they are present in both control and experimental mice, reflecting pre-existing PL sensory-driven properties (Hockley & Malmierca, 2024; Zikopoulos & Barbas, 2006).'

    Maybe. They are also likely to have developed as a consequence of the repeated testing on days 1, 15, and 30, which involved intermixed exposures to the tones of different frequencies. That is, rather than 'pre-existing PL sensory-driven properties', the responses of these neurons might reflect the emergence of discrimination between the various tones across testing, and greater suppression of freezing to the non-trained tones compared to the trained tone across the various test intervals.

  25. eLife

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP-positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under-examined in the field.

    Weaknesses:

    (1) Difficult to determine if responses treated as encoding stimulus valence are driven instead by the behavior that the stimulus elicits, freezing.

    (2) The study implies that the identified ensembles are causally related to valence memory, but no experimental interventions are performed to justify this.

  26. eLife

    Author response:

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    The authors combine discriminative auditory fear conditioning with longitudinal in vivo calcium imaging to ask how prelimbic (PL) representations of learned and generalized threat evolve across recent and remote memory time points. Using two different CS+ frequencies and a no-shock control group, they report that PL population activity tracks graded behavioral generalization, that population similarity is highest for tones eliciting strong threat responding, and that distinct subnetworks can be identified that appear to encode tone-specific sensory features versus learned threat-related response structure.

    To my knowledge, this may be the first study to comprehensively examine neural encoding of fear generalization in prelimbic cortex (PL). The manuscript is ambitious and technically interesting, and several aspects are potentially important. In particular, the suggestion that neurons showing graded, learning-related response patterns become selectively stabilized over time is intriguing. The inclusion of two CS+ training conditions and a no-shock control also strengthens the case that at least some of the reported effects are related to associative learning rather than simple sensory differences. However, in its current form, the manuscript does not yet fully support the strength of the conceptual claims. Several issues limit confidence in the interpretation, including the possibility that repeated testing itself contributes to changes across days, uncertainty about the relationship between neural activity and freezing behavior, limited quantitative documentation of longitudinal cell registration, and a number of problems in figure clarity and statistical framing. Overall, the study contains promising observations, but the claims should be narrowed, and several analyses or controls would be needed to fully support the proposed framework.

    Detailed Comments

    (1) A general concern is that the repeated test procedure itself may contribute to extinction. Because the animals are exposed to multiple CS frequencies across multiple test days, and each tone is presented three times per session, some of the reported changes in behavior and neural activity across days could reflect extinction or repeated nonreinforced retrieval rather than the passage of time per se. This is especially relevant given that the manuscript makes claims about recent versus remote representations and representational drift over 30 days. At a minimum, the authors should discuss this limitation explicitly and temper claims about time-dependent changes. Ideally, they would include a control group in which animals are tested only once or twice (e.g., at an early and later time point with fewer CS frequencies), or a reduced-frequency testing design that minimizes extinction while still allowing evaluation of recent versus remote memory.

    We agree with the reviewer that repeated testing is an inherent limitation of longitudinal memory studies and may itself contribute to some neural changes across sessions. However, several aspects of our behavioral design and results argue against extinction or repeated nonreinforced retrieval as the primary drivers of the observed effects. Importantly, discrimination ratios remained stable or increased across time rather than progressively diminishing as would be expected under extinction (this new analysis will be added to the resubmission). Nevertheless, we will address this important point in the Discussion and explicitly acknowledge that repeated retrieval may contribute to some component of the observed representational changes.

    (2) More generally, some of the reported learning-related neural differences may be driven by behavioral differences, particularly freezing, rather than by learning or generalization per se. For example, animals that freeze more to certain frequencies may show corresponding neural response differences simply because freezing alters PL activity. The authors should examine this possibility more directly. Analyses testing whether recorded cells encode freezing behavior, or whether tone frequency-related neural differences remain robust when comparing high- and low-freezing epochs, would help determine whether the reported effects reflect learned stimulus value rather than behavioral state differences.

    We thank the reviewer for raising this important point, which was also noted by the other reviewers. To address this issue, we will implement Reviewer 3’s suggested Generalized Linear Model (GLM) analysis using inferred spiking activity derived from the Ca2+ signals, with both tone identity and freezing behavior included as predictors. Because freezing behavior varies across trials whereas stimulus identity is fixed, this approach will allow us to dissociate their respective contributions to neuronal activity. If, after accounting for freezing behavior, responsive neurons continue to exhibit graded coding consistent with inferred threat value, this would strengthen the interpretation that the identified ensembles reflect generalization gradients related to aversive value rather than freezing behavior alone. Otherwise, we will adjust the conclusions according to the interpretation that freezing itself drives the generalization gradients.

    (3) A central feature of the manuscript is the analysis of neural response properties over an extended period of time, up to 30 days after learning. However, aside from a brief mention in the Methods that spatial registration was used, the manuscript provides very little quantitative information about this critical aspect of the study. The paper would be strengthened by including explicit metrics describing longitudinal cell tracking, such as the number and proportion of ROIs retained across all sessions, distributions of spatial-footprint correlations or centroid distances across days, and representative examples of matched imaging fields over time. Without this information, it is difficult to assess how strongly the longitudinal claims are supported.

    We thank the reviewer for this suggestion. We will include measures of registration quality in the resubmission.

    (4) The text states that "Figs. 1c and 1d show GCaMP6f expression in PL, representative calcium footprints, and activity traces". However, the figure as presented does not clearly show all of these elements, at least not in a way that matches the description in the Results. The correspondence between text and figure should be corrected.

    We will correct correspondence between text and Figure.

    (5) The labeling of Figure 2a is insufficient for interpretation. The legend states that the panel shows raster plots of sound responsiveness, but the axes and scaling are not clearly defined. It is not clear from the figure what the x-axis represents, whether the y-axis corresponds to individual neurons, where the CS period occurs, or what the activity scale at the right denotes. Also, the term 'rasters' implies that spikes were analyzed. It seems that the spike inference approach (CASCADE) was only used for later analyses. Perhaps 'heat-plot' would be more accurate here? Generally, this figure should be annotated more clearly so that the reader can understand it without referring back to the Methods.

    Thank you for this suggestion. We will clarify the labelling of the Figure 2a and call the graphs “activity-plots”.

    (6) In relation to Figure 3, the analysis of population-averaged responses across tone frequencies is useful, but the manuscript would be stronger with additional statistical analyses across time and across groups. For example, if the authors want to argue that learning induces graded changes in neural responses and that these evolve across time, they should directly compare within-group responses across days and also compare matched frequencies between the conditioned groups and the no-shock controls. These analyses would help establish whether the observed differences are genuinely learning dependent and whether they change significantly over time.

    We will redo the Statistics of Figure 3 to take into account the following variables: group (CS15, CS3, no shocks), frequency (3, 7, 11, 15), and day of testing (2, 15, 30).

    (7) The inclusion of two different CS+ frequencies and a no-shock control is a strength of the study and substantially improves the interpretation that graded neural responses are related to learning and generalization rather than to simple sensory processing or passage of time. That said, I am not entirely comfortable with the use of the term "inference" throughout the manuscript. What is being measured here appears closer to sensory generalization than inference in a stronger cognitive sense. The current task does not clearly require that animals infer hidden structure or stimulus value through abstract reasoning; rather, the generalized stimulus may simply be treated as similar to the conditioned cue. The terminology should therefore be reconsidered or softened.

    We thank the reviewer for appreciating the strengths of the experimental design and for this thoughtful suggestion regarding terminology. We agree that the term “inference” may overstate the cognitive processes engaged by the current task. Accordingly, we will revise the terminology throughout the manuscript to describe these effects as graded generalization of threat value across stimuli.

    (8) I also found the use of the term "valence" somewhat problematic. The manuscript appears to use valence to refer to graded responding across tones with different aversive significance, but valence typically refers more broadly to distinctions between appetitive and aversive value. Here, terms such as "threat value," "aversive value," may be more precise. The authors should consider revising this language throughout.

    We will correct the language and use “threat value”.

    Reviewer #2 (Public review):

    Summary:

    The following points are those that occurred to me across readings of the paper. They are listed in what I take to be the order of their significance. Many of the points relate to the loose use of language and invocation of concepts that are not warranted, given the study design and results obtained.

    Major Comments:

    (1) The concept of ensemble turnover is interesting - the way it is introduced and discussed implies some type of spontaneous change in the neural underpinnings of fear discrimination and generalization in the PL. But, of course, every trial involves an opportunity to learn about the threat CS or the generalization test stimuli, and I am troubled by the thought that stability in the neural underpinnings of fear discrimination and generalization will actually reflect the level of defensive behaviours evoked on different trial types and/or the discrepancy between those behaviours and the outcome of a given trial in the generalization test. That is, stability in the neural underpinnings may be related to an animal's certainty or uncertainty in the contingency between a stimulus and danger; or, put another way, an animal's confidence that danger will or won't occur given the presence of some stimulus. This is not uninteresting. It is, however, not considered anywhere in the paper, which is overloaded with references to inferred threat values and integration of information across different types of stimuli. The protocol is not one that requires inference about anything or integration across anything.

    We thank the reviewer for these important points, which we address in further detail below.

    Ongoing learning during test sessions: The reviewer correctly notes that unreinforced test presentations may constitute extinction-learning trials and that some neural changes across days could therefore reflect ongoing learning rather than spontaneous ensemble reorganization. However, new analyses indicate that extinction is unlikely to be the primary driver of our findings. Discrimination ratios do not decay over time; instead, they either sharpen or remain stable across sessions (new analyses to be included in the resubmission). These results argue against robust extinction as the primary source of the neural changes observed across sessions. This interpretation is also consistent with the strength of our conditioning protocol, which used 10 CS+ shock pairings and 10 CS− no-shock pairings specifically to minimize extinction across repeated testing sessions. Nevertheless, we acknowledge that the current design cannot fully dissociate time-dependent consolidation from retrieval-induced plasticity, and we will explicitly discuss this limitation in the revised Discussion.

    Stability reflecting behavioral consistency: We agree this alternative cannot be fully excluded. However, the cluster stability analyses assess identity at the level of response profile across all four frequencies, not response magnitude alone. Tone-selective clusters, which also show consistent behavioral correlates (firing rate correlates with threat-value, Fig. S8), do not show equivalent profile stability, suggesting that the stability of graded clusters is not simply a consequence of behavioral consistency. This point will be added to the Discussion in the resubmission.

    Language of "inference" and "integration": The reviewer is correct that responses to novel tones are consistent with graded stimulus generalization. We will substantially revise the manuscript to replace "inference" and "integration" with more precise language describing graded frequency generalization gradients.

    (2) I appreciate the link to Gu and Johansen in paragraph 3 of the Introduction, but the type of generalization under investigation here is not the same as the type of 'generalization' studied by Gu and Johansen [who used a sensory preconditioning protocol]. Nonetheless, the authors have forced the language used by Gu and Johansen into their paper, and this has created tension [at least for this reader] as the concepts introduced by Gu and Johansen [inference, integration] are simply not relevant given the generalization protocol used here. Here are a few examples of points where the tension might interfere with a reader's understanding:

    We thank the reviewer for these specific and constructive criticisms. We will revise the manuscript throughout to remove or redefine terms like "inferred valence" and "integration," replacing them with clearer, more accurate descriptions of gradient generalization of threat value. Below we address each point raised by the reviewer regarding terminology clarifications.

    (a) 'We hypothesized that generalization to novel stimuli depends on stable subnetwork organization that enables comparisons between learned and inferred valence, as well as population-level features that reduce variability across related representations.'

    I understand the words in the hypothesis, but can't form a representation of what is being said because of the reference to terms that stand in need of clarification [inferred valence, variability across related representations], but, ultimately, won't be clarified. This needs to be re-expressed so that the reader can appreciate what is being said.

    The hypothesis will be rewritten as: "We hypothesized that generalization to tones acoustically similar to the CS+ and CS− depends on the emergence of stable ensembles encoding threat value, and that population-level response similarity across stimuli would correlate with the degree of behavioral fear generalization, consistent with prior work in auditory cortex [1]."

    (b) 'Our results show that stable cortical subnetworks integrate the emotional "gist" of memory and inferred valence for novel cues over time, despite ongoing ensemble reorganization, and that population-level firing rate similarity across stimulus presentations determines threat generalization.'

    Again, what does this mean? How is the gist of a memory integrated with inferred valence for novel cues over time? The statement simply doesn't make sense. This needs to be rewritten for clarity.

    The summary statement will be rewritten: "Our results show that stable cortical sub-ensembles preserve the emotional content of the fear memory over time, despite ongoing ensemble reorganization, and that population-level firing rate similarity in response to tones associated with threat correlates with the degree of behavioral threat generalization."

    (c) 'In CS⁺15 mice, positively modulated sound-responsive neurons exhibited graded tone activity reflecting the contingency learned valence as well as the inferred valence of novel tones across testing days...'.

    Can this be rewritten as 'In CS⁺15 mice, positively modulated sound-responsive neurons exhibited graded activity to the tone CS and its variants that were used to assess generalization.'? The overloading of the text with references to 'contingency learned valence' and 'inferred valence' is unnecessary and makes it much harder to understand what has been shown in the results.

    We will adopt the reviewer's suggested rewording: "In CS+15 mice, positively modulated sound-responsive neurons exhibited graded activity to the tone CS and its variants that were used to assess generalization."

    We will systematically review the entire manuscript to ensure consistency with this revised framing.

    (3) Re the same passage of text as in 2c:

    Is it the case that these neurons are simply tracking the expression of freezing to the various tones? The same question applies to the results obtained for the CS+3 mice. If this is the case, then why should the results be taken to support the banner statement that 'Sound-modulated PL population responses encode learned and inferred valence' - these analyses do not support that statement. And, as indicated, I don't believe that the language of learned and inferred valence is appropriate to such statements, given the nature of the protocol used and results obtained. It is a study looking at how populations of neurons in the PL respond during presentations of auditory stimuli that were subject to discriminative conditioning, and during tests of generalized freezing to other [intermediate] auditory stimuli.

    The reviewer is correct that the graded population responses observed in PL could reflect freezing behavior across tone frequencies rather than encoding an abstract threat-value representation. This important concern was also raised by other reviewers. To address it directly, we will follow Reviewer 3’s suggestion and implement a Generalized Linear Model (GLM) using inferred spiking activity derived from the Ca2+ signals, with both tone identity and freezing behavior included as predictors. This analysis will allow us to dissociate the respective contributions of tone frequency and freezing to the graded neural responses. Based on the outcome of this analysis, we will revise and appropriately adjust our conclusions.

    In addition, we will revise the section heading and surrounding text to remove the terminology of “learned and inferred valence.” Instead, the findings will be described more conservatively as: “PL population responses reflect behavioral generalization to auditory stimuli following discriminative fear conditioning.”

    (4) It is stated that:

    'In no-shock controls, although both positive and negative responses were present, population activity was not modulated by tone frequency or valence'.

    What does this mean? I can understand that population activity was not modulated by tone frequency. But what does it mean to say that it was not modulated by valence? Why should it have been when none of the tones were conditioned in this group and, hence, mice were responding to all the tones equally? And given that this is true, I don't understand the use of 'valence' here, or the subsequent statements in this paragraph that 'graded responses require associative learning' and that 'PL population responses encode graded sound-valence associations that reflect both learning and inference, closely matching behavioral generalization.' The latter statement is particularly unwarranted and, again, highlights a major issue with the paper. It could and should be rewritten as 'PL population responses reflect behavioral generalization.' There is nothing in the additional language that adds to the reader's understanding of what has been shown. The reference to 'graded sound-valence associations that reflect both learning and inference' is completely unwarranted, given the nature of this study. It is anathema to the vast literature on stimulus generalization. If the authors wished to make statements of this sort, they should have taken a different approach, perhaps using protocols like those featured in Gu and Johansen.

    The reviewer is correct that controls do not form threat associations; however, these animals still could respond differentially to distinct frequencies, something that is not reflected in the data. We will correct the section indicating that distinct neutral frequencies do not produce graded responses: "graded responses require associative learning" will be retained but reframed simply as: "graded frequency-dependent population responses were absent in animals that did not receive fear conditioning." The concluding statement of the paragraph will be rewritten as: "PL population responses reflect behavioral generalization to acoustically similar stimuli following discriminative conditioning," in line with the reviewer's suggestion.

    (5) The section titled, 'Consistently active neurons preserve valence representations as newly recruited neurons sharpen remote memory traces' ends with the following summary:

    'Together, these results indicate that consistently active neurons maintain stable representations of learned and inferred sound associations across time, whereas neurons recruited after conditioning progressively acquire graded tuning at later retrieval stages. This dynamic refinement suggests that cortical memory representations become increasingly selective during systems consolidation, while a stable neuronal subpopulation preserves the core emotional content of the memory.'

    Once again, the summary is not in keeping with the results obtained. The 'dynamic refinement' of representations is far more likely to reflect the repeated testing across days 1, 15, and 30 rather than anything to do with systems consolidation - at the very least, it is the simplest interpretation of the results. The impact of repeated testing is evident in the sharpening of generalization gradients over time, which is contrary to what is otherwise observed in the literature - the incredibly well -documented broadening of generalization gradients with time. Given this impact of repeated testing, surely the changes in the neuronal population that underlie performance are more likely to reflect the learning that occurs on days 1, 15, and 30, which is reflected in reduced freezing to the non-conditioned tones. If this is a reasonable take on the results, then I don't see the basis for invoking systems consolidation at all, and I don't see the basis for inferring a stable neuronal subpopulation that preserves the emotional content of the memory. Rather, non-reinforced presentations of 'never-reinforced' tones result in recruitment of additional neurons that result in suppression of freezing responses to those stimuli.

    We respectfully disagree with the reviewer’s interpretation. While repeated testing cannot be entirely excluded as a contributing factor, several lines of evidence suggest that it cannot fully account for our observations.

    Regarding extinction: discrimination ratios between CS+ and all other frequencies either remained stable or increased over time (new analysis included in resubmission), indicating that animals continued to discriminate threat value across the testing period rather than showing the progressive suppression expected under extinction — the opposite of what we observe.

    Regarding the recruitment of new neurons: repeated non-reinforced tone exposure would be expected to produce stimulus-specific adaptation — characterized by reduced, less discriminative neural responsiveness and flatter tuning profiles [2]— not the progressive sharpening we observe. The same would be expected if these neurons represent or are associated with new extinction learning.

    Finally, sharpening of generalization gradients during repeated within-subjects testing has been reported previously [3], suggesting that successive exposures may promote more precise discrimination in some cases. Consistent with this, discrimination learning has also been shown to narrow or sharpen fear generalization gradients rather than broaden them [4], supporting the idea that discriminative conditioning enhances stimulus specificity during testing. Although we cannot exclude the possibility that more extended training could eventually broaden the generalization gradient, under the training parameters and temporal window used in our study, the data support a progressive sharpening of the gradient over time. In the revised Discussion, we will present systems consolidation as the primary interpretive framework and further elaborate on why repeated testing is unlikely to account for the full pattern of behavioral and neural findings reported here.

    (6) In the section titled, 'Population vector similarity at stimulus onset determines degree of generalization', it is stated that:

    'Because population similarity peaked shortly after stimulus onset, we quantified similarity during the first 5 s after tone onset relative to the CS⁺. In CS⁺15 mice, population similarity was highest for 15/15 and 15/11 tone pairs with no differences between them.'

    Isn't this consistent with the view that the population response in the PL simply reflects the level of freezing? Freezing to the 15-15 and 15-11 tones is most likely to be similar on their first presentation prior to the effects of extinction on the 11 Hz tone; hence the results obtained. That is, these results appear to clearly indicate that neuronal responses in the PL reflect the degree of stimulus generalization, as evidenced in freezing behavior. Given all that we know about the involvement of the PL in expressing fear responses, it is not appropriate to claim that 'population vector similarity at stimulus onset *determines* the degree of generalization. The PL responses simply reflect the varying levels of performance displayed to the different types of tones. What have I missed that could be taken to support additional statements?

    The GLM analysis described in our response to reviewers 1 and 3 will directly address the contribution of freezing. We will report these results in the resubmission and revise the interpretive language in the manuscript accordingly.

    However, regarding the analysis of population vector similarity, we need to clarify a point of confusion. The reviewer states “Freezing to the 15-15 and 15-11 tones is most likely to be similar on their first presentation prior to the effects of extinction on the 11 Hz tone; hence the results obtained”. The similarity vectors were calculated by correlating activity across all tone presentations within each testing day, not only the first two presentations. In Fig. 4, “Early” and “Late” refer to the order of a tone within a trial, which we will clarify more explicitly in the resubmission. Notably, repeated-measures analyses did not reveal any effect of the time variable (Fig. 4e,f), indicating that similarity across tone presentations remained high for tones associated with high threat value. Importantly, our data showed no evidence that responses to 11 kHz or 15 kHz in the CS15 group, or to 3 kHz in the CS3 group, exhibited extinction-like patterns at either the behavioral or neural level. Therefore, the persistence of high population similarity across time provides additional evidence against extinction as the primary explanation for our findings.

    We will remove the word "determines" from the manuscript, as our data cannot conclusively establish a causal relationship.

    Later in the same section, it is stated that 'population-level similarity at stimulus onset scales with behavioral threat generalization and is maximal for tones associated with robust threat responses.' For simplicity and, therefore, clarity, this should be rewritten as 'population-level similarity at stimulus onset reflects behavioral threat generalization.'

    We will make this correction.

    (7) In the section titled, 'Different subnetworks encode acoustic versus learned properties of sound association', it is stated that:

    'Our previous analyses show that learned and inferred associations are represented at the population level. However, these results do not resolve whether graded responses arise from pooled activity of frequency-selective neurons or from subnetworks encoding integrated learned valence across tones.'

    What does it mean to say 'integrated learned valence across tones'? As it presently stands, the meaning of the phrase is unclear. It only makes sense if one supposes that generalized freezing responses to the 11 and 7 kHZ tones reflect separate associations between those tones and the aversive foot shock US. This supposition is inconsistent with the rich literature on generalization of Pavlovian conditioned fear responses. Specifically, it is inconsistent with the many theories of fear generalization, which attribute the reduction in fear as one moves away from the specific conditioned stimulus to a decrement in the ability of the test stimulus to activate the trained CS-US association. My strong impression is that the authors would do well to ground their findings in theories of stimulus/fear generalization, of which there are many. This would better serve the results obtained [and the reader's appreciation of them] - at present, the unnecessary invocation of concepts does very little to enhance the reader's appreciation or understanding of what has been found in the study.

    We thank the reviewer for raising this point. The phrase "integrated learned valence across tones" refers specifically to a subpopulation of neurons that respond to all four frequencies in a graded manner, with response magnitude scaling according to threat value. This is distinct from tone-selective neurons, which respond preferentially to a single frequency. The neurons responding to all tones in a graded manner are present only in conditioned animals and not in no-shock controls, demonstrating that their graded response profile is shaped by associative learning.

    We agree, however, that the phrase "integrated learned valence" is unnecessarily opaque and we will replace it with more precise language: these neurons will be described as showing graded frequency-dependent responses whose magnitude scales with threat value. We believe this subpopulation represents a genuinely novel finding that complements the behavioral generalization literature by identifying a specific neural substrate for the generalization gradient within PL.

    (8) Another example of what has been a common theme in this review:

    '...we hypothesized that the PL active ensemble segregates into functionally distinct subnetworks: one encoding tone-specific sensory features with dynamic characteristics, and another responding to all frequencies encoding stable core memory content and inferred emotional valence.'

    What does it mean to say 'all frequencies encoding stable core memory content and inferred emotional valence'? Do the authors mean to say '...and another that tracks freezing/defensive responses regardless of whether they were elicited by the trained CS or one of the generalization test stimuli'?

    As stated in our previous responses, in the resubmission we will determine the contribution of freezing. If we find that freezing predicts graded neural responses, we will adjust the language of the manuscript.

    (9) It is stated that - 'Graded clusters encode emotional valence but constitute only a fraction of the active population; yet valence coding at the population level remains accurate and precise. This indicates that neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.'

    What does this mean? Are the authors trying to say that - 'Some clusters of PL neurons track freezing responses. In spite of the fact that these are only a fraction of the total active neuronal population, the population-level response of PL neurons also tracks the levels of fear to the trained tone and its variants used in the test for generalization.' If this is what one wants to say, then the final statement in the reproduced section does not follow. That is, there is no indication that 'neurons newly recruited into the population-likely frequency-selective and organized within learning-independent clusters-can be shaped by associative processes through modulation of firing activity.' As noted, the characteristics of other ensembles that become active across the repeated tests on days 1, 15, and 30 are more likely to reflect learning from non-reinforcement that occurs within and across those sessions. Perhaps this is what is meant by the phrase, 'shaped by associative processes'? If so, it should be stated explicitly instead of left to the reader to work out.

    We thank the reviewer for highlighting the lack of clarity in this passage and agree that the original phrasing was insufficiently precise. What we intended to convey is that only a subset of PL neurons displays graded tuning that tracks behavioral generalization across tones. Nevertheless, despite constituting only a fraction of the total active population, this graded coding is also reflected at the population level. Therefore, we suggest that neurons recruited into the active population after conditioning — likely frequency-selective neurons — contribute to the graded population responses through changes in their firing-rate activity, which is modulated by threat value (Fig. S8). We will rewrite this passage in the resubmission to make this interpretation explicit rather than leaving it to the reader to infer.

    Regarding the reviewer's suggestion that the characteristics of newly recruited neurons more likely reflect learning from non-reinforced exposures during repeated test sessions, we respectfully maintain that this interpretation is difficult to reconcile with two aspects of our data. First, graded-response neurons are absent in no-shock controls that are exposed to nonreinforced repeated testing. Second, as detailed in our responses to previous points, the progressive sharpening of population responses over time is inconsistent with what would be expected from repeated non-reinforced exposure, which would more plausibly produce broader or flatter tuning profiles.

    We agree that the phrase "shaped by associative processes" was ambiguous and will replace it with explicit language clarifying that we refer to fear conditioning as the associative process driving the emergence of graded responses, rather than any learning occurring during the test sessions themselves.

    (10) The following points all relate to the Discussion and reiterate many of the points above.

    (a) 'A subset of neurons remains consistently active across sessions, preserving core components of the memory trace and supporting inference of emotional valence for novel sounds, while neurons recruited after conditioning progressively acquire valence selectivity at remote time points.'

    'Inference of emotional valence' is unclear and unwarranted for all of the reasons provided above regarding the use of language.

    We will modify the language as stated in the prior points.

    (b) '...Our data reconcile these views by demonstrating that cortical representations of emotional valence emerge rapidly after learning and persist within stable subnetworks, even as the broader population undergoes substantial turnover. This architecture preserves core mnemonic content while allowing flexibility in the surrounding ensemble.'

    These statements assume that the PL neuronal responses reflect something more than the levels of freezing behavior to the different stimuli; what are the grounds for this assumption?

    We will incorporate new analysis (GLM) to better address this point and conclusions.

    (c) 'Importantly, these subnetworks encode both learned contingencies and the inferred valence of novel stimuli along a graded representational axis, suggesting that strong recurrent connectivity provides a stable scaffold for emotional memory representations.'

    What is a graded representational axis, and what part of the first statement suggests that 'strong recurrent connectivity provides a stable scaffold for emotional memory representations'? If the authors' goal was to make statements about emotional memory representations vis-à-vis emotional memory content, they should have used protocols that allowed them to probe such content. The auditory fear conditioning protocol used here [followed by tests for generalization to other auditory stimuli that differ in frequency from the conditioned tone] is not one that lends itself to analysis of emotional memory representations or content.

    We thank the reviewer for this comment and agree that both phrases require clarification or revision.

    By "graded representational axis" we intended to convey that PL population activity varies systematically as a function of stimulus similarity to the conditioned tone — that is, population responses are not categorical but scale continuously with spectral proximity to the CS+. We agree this was not clearly stated and will revise the manuscript accordingly.

    Regarding recurrent connectivity, we agree with the reviewer that nothing in our data directly measures or manipulates connectivity between neurons. This statement was intended as a speculative interpretive hypothesis in the Discussion, motivated by the established literature linking strong recurrent connectivity in prefrontal circuits to stable population-level representations [5]. However, we acknowledge that invoking it in this context, without direct evidence, risks overstating our conclusions. We will revise this sentence to make its speculative nature explicit and ground it more carefully in the cited literature rather than presenting it as an inference from our own data.

    In summary, we will ensure our conclusions will be restricted to population-level coding of learned threat value and its generalization across auditory frequencies. We will revise the relevant passages in the Discussion to ensure that speculative interpretations regarding emotional memory content are either removed or clearly flagged as speculative hypotheses.

    (d) 'Dynamic tone-selective responsive neurons emerge independently of learning, as they are present in both control and experimental mice, reflecting pre-existing PL sensory-driven properties (Hockley & Malmierca, 2024; Zikopoulos & Barbas, 2006).'

    Maybe. They are also likely to have developed as a consequence of the repeated testing on days 1, 15, and 30, which involved intermixed exposures to the tones of different frequencies. That is, rather than 'pre-existing PL sensory-driven properties', the responses of these neurons might reflect the emergence of discrimination between the various tones across testing, and greater suppression of freezing to the non-trained tones compared to the trained tone across the various test intervals.

    We thank the reviewer for this point. Our interpretation that these neurons reflect pre-existing PL sensory-driven properties was based on the observation that tone-selective responses were present in control animals that never received conditioning, consistent with prior reports of sensory responsiveness in PL cortex ([6, 7]. Because these responses emerge from the first time we expose mice to the intermediate frequencies, they cannot be explained by repeated exposure. Moreover, we did not observe progressive refinement, emergence of discrimination-like changes, or suppression of responding to non-reinforced tones in control mice. This difference between conditioned and control animals indicates that repeated tone exposure alone is not sufficient to produce the observed dynamics — associative learning is necessary. We therefore maintain that the tone-selective responses of these neurons reflect pre-existing sensory-driven properties of PL cortex that are present independently of conditioning history.

    In summary, we thank the reviewer for suggesting clarifications to our interpretation, for raising the possibility that freezing behavior may contribute to graded neural responses, and for raising the question of whether repeated tone exposure may contribute to the properties of neurons recruited after conditioning. In the revised manuscript, we will include additional analyses to better dissociate the contributions of freezing behavior and tone identity, clarify passages that were insufficiently precise, and include a paragraph in the Discussion addressing potential alternative explanations alongside our own interpretation of the data.

    Reviewer #3 (Public review):

    Summary:

    Normandin et al. explore the coding of stimuli predicting an aversive event in the prelimbic cortex. Stimuli could either be explicitly paired, explicitly unpaired, or novel but with an inferred association with the aversive event (generalization). Long-term tracking of GCaMP-positive neurons allowed them to examine how coding evolves out to a month following training. In general, they found two types of ensemble codes. One was ensembles coding for each stimulus independently, but with enhanced responding to the one eliciting a freezing response. The other was ensembles that responded to all stimuli in proportion to their similarity to the stimulus paired with the aversive event, either increasing or decreasing their activation with the degree of freezing elicited by a stimulus. Importantly, this second set of ensembles was more stable across days, potentially providing a memory trace.

    Strengths:

    (1) The authors track ensembles in prelimbic cortex over long time scales, providing valuable information on the consolidation of neural codes.

    (2) Neural coding of generalization is examined, which is under-examined in the field.

    We thank the reviewer for appreciating our design to track ensembles over time and the relevance of studying the neural substrates of generalization.

    Weaknesses:

    (1) Difficult to determine if responses treated as encoding stimulus valence are driven instead by the behavior that the stimulus elicits, freezing.

    We thank the reviewer for this thoughtful and constructive comment. We agree that an alternative interpretation is that the graded-response ensembles may partially reflect freezing-related activity rather than mnemonic or salience-related representations of the conditioned stimuli themselves. In the revision, we will acknowledge that prior work has identified PL neurons that encode freezing independently of stimulus identity or associative content. Furthermore, we will implement the reviewer’s suggested generalized linear model (GLM) approach using inferred spiking activity derived from the Ca2+ signals. Specifically, we will include both stimulus identity and freezing behavior as predictors. Because freezing varies across trials whereas stimulus presentation is fixed, this analysis will allow us to dissociate the relative contributions of stimulus-related versus freezing-related activity to the graded neuronal responses. We thank the reviewer for this excellent suggestion.

    If graded stimulus coding remains significant after accounting for freezing behavior, this would strengthen the interpretation that these ensembles encode learned salience or associative properties of the stimuli rather than behavioral output alone. Conversely, if freezing explains a substantial proportion of the variance, we will revise our interpretation accordingly.

    (2) The study implies that the identified ensembles are causally related to valence memory, but no experimental interventions are performed to justify this.

    We appreciate the reviewer's point. We agree that our data are correlational in nature and that establishing a causal relationship between identified ensembles and valence memory would require experimental interventions such holographic two-photon manipulations, which are beyond the scope of the present study but represent an important direction for future work.

    To provide an indirect link between ensemble organization and behavior within the constraints of the current dataset, we will examine inter-individual variability in the revised manuscript. Specifically, we will test whether the proportion of neurons participating in stable graded-response ensembles versus dynamic stimulus-specific ensembles predicts individual differences in freezing behavior and fear generalization across retrieval sessions. If animals with a higher proportion of stable graded-response neurons show stronger discrimination and less generalization to non-conditioned tones, this would strengthen the association between ensemble organization and behavioral outcome, while remaining correlational in interpretation.

    We will modify the manuscript terminology accordingly, replacing causal language with phrasing that accurately reflects the associative nature of our conclusions.

    References

    (1) Aschauer, D.F., et al., Learning-induced biases in the ongoing dynamics of sensory representations predict stimulus generalization. Cell Rep, 2022. 38(6): p. 110340.

    (2) Kato, H.K., S.N. Gillet, and J.S. Isaacson, Flexible Sensory Representations in Auditory Cortex Driven by Behavioral Relevance. Neuron, 2015. 88(5): p. 1027–1039.

    (3) Vervliet, B., et al., Generalization gradients in human predictive learning: Effects of discrimination training and within-subjects testing. Learning and Motivation, 2011. 42(3): p. 210–220.

    (4) Dunsmoor, J.E. and K.S. LaBar, Effects of discrimination training on fear generalization gradients and perceptual classification in humans. Behav Neurosci, 2013. 127(3): p. 350–6.

    (5) Mante, V., et al., Context-dependent computation by recurrent dynamics in prefrontal cortex. Nature, 2013. 503(7474): p. 78–84.

    (6) Hockley, A. and M.S. Malmierca, Auditory processing control by the medial prefrontal cortex: A review of the rodent functional organisation. Hear Res, 2024. 443: p. 108954.

    (7) Zikopoulos, B. and H. Barbas, Prefrontal projections to the thalamic reticular nucleus form a unique circuit for attentional mechanisms. J Neurosci, 2006. 26(28): p. 7348–61.

  27. Version published to 10.7554/elife.111177.1 on eLife
  28. Mentioned by @rridrobot.bsky.social on Bluesky

    RRIDs were included in this in bioRxiv paper. We value the author's support of reproducibility. #OpenScience #OpenResearch #reproducibility

  29. Mentioned by @biorxivpreprint.bsky.social on Bluesky

    Learned and inferred valence arise from interactions between stable and dynamic subnetworks https://www.biorxiv.org/content/10.64898/2026.03.08.710406v1

  30. Mentioned by @biorxiv-neursci.bsky.social on Bluesky

    Learned and inferred valence arise from interactions between stable and dynamic subnetworks https://www.biorxiv.org/content/10.64898/2026.03.08.710406v1

  31. Version published to 10.64898/2026.03.08.710406v1 on bioRxiv