Continual Knowledge Updating in LLM Systems: Learning Through Multi-Timescale Memory Dynamics

Andreas Pattichis; Constantine Dovrolis

arxiv: 2605.05097 · v2 · submitted 2026-05-06 · 💻 cs.LG · cs.AI· cs.CL

Continual Knowledge Updating in LLM Systems: Learning Through Multi-Timescale Memory Dynamics

Andreas Pattichis , Constantine Dovrolis This is my paper

Pith reviewed 2026-05-08 18:08 UTC · model grok-4.3

classification 💻 cs.LG cs.AIcs.CL

keywords continual learningLLM knowledge updatingmulti-timescale memoryknowledge graphsynaptic consolidationexternal memoryassociative memoryBenna-Fusi model

0 comments

The pith

Coupling fast and slow variables on knowledge-graph edges lets external memory adapt on its own for continual LLM updates.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

LLMs are trained once but the world keeps changing, so external memory is needed for updates. Most systems manage this memory explicitly with rules and interventions. This paper suggests letting the memory adapt on its own by using coupled multi-timescale dynamics modeled after biological synaptic consolidation. In the proposed Memini system, knowledge is stored as a directed graph where each edge has a fast and a slow variable that interact according to the Benna-Fusi model. This single coupling mechanism produces immediate usability of new info, strengthening through repetition, and fading of unused knowledge, turning memory into an active learning substrate.

Core claim

The paper establishes that from the coupling of fast and slow variables on graph edges following the Benna-Fusi model, episodic sensitivity, gradual consolidation, and selective forgetting emerge as facets of a single mechanism. This reframes external memory as a learning substrate that reorganizes through its own dynamics.

What carries the argument

Memini associative memory: a directed graph whose edges each carry two coupled internal variables, one fast and one slow, following the Benna-Fusi model of synaptic consolidation.

Load-bearing premise

The Benna-Fusi multi-timescale coupling on the edges of an LLM knowledge graph will produce stable continual learning without interference, scalability bottlenecks, or loss of previously consolidated knowledge.

What would settle it

Run Memini on a sequence of updating facts and check whether retrieval accuracy for both old and new knowledge remains high over time or whether graph dynamics produce interference or forgetting.

Figures

Figures reproduced from arXiv: 2605.05097 by Andreas Pattichis, Constantine Dovrolis.

**Figure 1.** Figure 1: Overview of Memini. (a) Write path. The LLM extracts entities and co-occurrences from arriving documents. Each co-occurrence updates an edge of the association graph through coupled fast and slow variables. Edge thickness reflects wfast. (b) Read path. Entities from the user’s query activate seed nodes (red). Activation propagates along wfast-weighted edges for a bounded number of steps, and the passages a… view at source ↗

**Figure 2.** Figure 2: shows the wfast and wslow trajectories for four entity pairs across the 13-document stream under identical parameters. The four were selected so that each illustrates one of the event-pattern groups later defined in view at source ↗

read the original abstract

LLMs are trained once, then deployed into a world that never stops changing. External memory compensates for this, but most systems manage it explicitly rather than letting it adapt on its own. Biological memory works differently: coupled multi-timescale dynamics make new associations immediately usable, strengthen what repetition confirms, and let the rest fade. We argue that external memory should follow a similar principle. In Memini, this view takes the form of an associative memory that organizes knowledge as a directed graph. Each edge carries two coupled internal variables, one fast and one slow, following the Benna-Fusi model of synaptic consolidation. From this coupling, episodic sensitivity, gradual consolidation, and selective forgetting emerge as facets of a single mechanism, reframing external memory as a learning substrate that reorganizes through its own dynamics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a high-level proposal for a graph-based memory with Benna-Fusi dynamics on edges, but it supplies no implementation, math, or tests to show the claimed behaviors actually work.

read the letter

The paper's main idea is to build an associative memory for LLMs as a directed graph where each edge holds two coupled variables, one fast and one slow, drawn from the Benna-Fusi synaptic model. The authors claim this single mechanism produces episodic sensitivity, gradual consolidation, and selective forgetting, so the memory reorganizes itself instead of needing explicit rules for updates and pruning. That is the central pitch in the abstract and title. The combination of that biological model with an LLM knowledge graph does not appear in the cited prior work, so the specific framing counts as new. The motivation section also makes a clear case that most external memory systems today are managed too rigidly and could benefit from more automatic adaptation. Beyond that, the paper stays at the level of a conceptual sketch. There are no equations showing how LLM-generated associations are turned into graph edges, no description of conflict resolution when new information contradicts old entries, and no stability or scaling analysis. The stress-test note is accurate on this point: without a fixed-point argument, Lyapunov check, or even a small simulation, there is no way to verify whether interference stays low or whether the graph blows up as new associations arrive. The paper engages honestly with the biological literature it imports, but it does not derive or test the core behaviors inside its own setup. This kind of work is mainly for researchers already interested in bio-inspired continual learning or memory-augmented LLMs. A reader who wants algorithms, code, or empirical results on knowledge retention will find little to use. I would not bring it to a reading group in its current state because there is not enough detail to discuss productively. It does not deserve peer review yet; the claims need at least a prototype and basic validation before referees could assess them.

Referee Report

3 major / 0 minor

Summary. The paper proposes Memini, a biologically inspired associative memory for continual knowledge updating in LLMs. Knowledge is represented as a directed graph in which each edge maintains two coupled internal variables (fast and slow) governed by the Benna-Fusi synaptic consolidation model. The central claim is that this per-edge coupling automatically produces episodic sensitivity, gradual consolidation, and selective forgetting, allowing the external memory to reorganize through its intrinsic dynamics rather than explicit management.

Significance. If the claimed emergence of stable continual-learning behaviors can be rigorously demonstrated, the work would offer a principled alternative to hand-engineered memory systems in LLM deployments. By importing and adapting a well-studied multi-timescale consolidation mechanism, it reframes external memory as an active learning substrate and could reduce interference and maintenance overhead in long-running LLM agents.

major comments (3)

[Model description] The manuscript asserts that 'from this coupling, episodic sensitivity, gradual consolidation, and selective forgetting emerge as facets of a single mechanism' but supplies neither the explicit update equations for the fast-slow variables on graph edges nor any fixed-point, Lyapunov, or scaling analysis. This absence is load-bearing for the central claim that the Benna-Fusi dynamics will remain stable and interference-free as new LLM-generated associations arrive and the graph grows.
[Implementation and dynamics] No pseudocode, conflict-resolution rule, or graph-expansion procedure is provided for how the system handles simultaneous or contradictory updates from the LLM. Without these details it is impossible to verify whether selective forgetting actually prevents memory bloat or whether previously consolidated knowledge is preserved, directly undermining the claim of stable continual learning.
[Evaluation] The paper contains no simulations, ablation studies, or comparisons against existing external-memory baselines. Because the soundness of the proposal rests entirely on the unexamined extrapolation of the Benna-Fusi model to knowledge-graph edges, the lack of any quantitative validation makes the central reframing of external memory untestable in its current form.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and insightful comments, which have helped us improve the clarity and rigor of the manuscript. We have revised the paper substantially to address each major point by adding the requested mathematical details, implementation specifications, and preliminary empirical results. Our responses are provided below.

read point-by-point responses

Referee: The manuscript asserts that 'from this coupling, episodic sensitivity, gradual consolidation, and selective forgetting emerge as facets of a single mechanism' but supplies neither the explicit update equations for the fast-slow variables on graph edges nor any fixed-point, Lyapunov, or scaling analysis. This absence is load-bearing for the central claim that the Benna-Fusi dynamics will remain stable and interference-free as new LLM-generated associations arrive and the graph grows.

Authors: We agree that the original version did not explicitly restate the Benna-Fusi equations in the graph-edge setting. The revised manuscript now includes the complete update rules for the fast and slow variables on each directed edge, with the fast variable responding immediately to new associations and the slow variable accumulating evidence over time. We also add a short analysis of the fixed points of the coupled system and discuss stability under incremental graph growth, building directly on the original Benna-Fusi results while noting the additional constraints imposed by the associative structure. revision: yes
Referee: No pseudocode, conflict-resolution rule, or graph-expansion procedure is provided for how the system handles simultaneous or contradictory updates from the LLM. Without these details it is impossible to verify whether selective forgetting actually prevents memory bloat or whether previously consolidated knowledge is preserved, directly undermining the claim of stable continual learning.

Authors: We acknowledge that these operational details were missing. The revised manuscript contains a new section with pseudocode for the update procedure, a conflict-resolution rule that weights incoming updates by the current strength of the slow variable (so that consolidated edges resist overwriting), and an explicit graph-expansion mechanism that adds new nodes and edges only when the fast variable exceeds a threshold. These additions make clear how selective forgetting reduces bloat while protecting strongly consolidated associations. revision: yes
Referee: The paper contains no simulations, ablation studies, or comparisons against existing external-memory baselines. Because the soundness of the proposal rests entirely on the unexamined extrapolation of the Benna-Fusi model to knowledge-graph edges, the lack of any quantitative validation makes the central reframing of external memory untestable in its current form.

Authors: The original manuscript was framed as a conceptual proposal. We have now added a dedicated evaluation section with simulation results on a synthetic continual-association task. The experiments include ablations that isolate the contribution of the fast-slow coupling and direct comparisons against a baseline associative memory without multi-timescale dynamics. These results demonstrate the claimed emergent behaviors. Full-scale LLM agent experiments remain future work owing to computational cost. revision: partial

Circularity Check

0 steps flagged

No significant circularity: external model adoption with independent application

full rationale

The paper adopts the Benna-Fusi multi-timescale synaptic consolidation model from prior biological literature as the core mechanism for edge dynamics in its knowledge graph. It asserts that episodic sensitivity, gradual consolidation, and selective forgetting emerge from this coupling but does not derive these behaviors via internal fitting, self-definition, or reduction to the paper's own fitted parameters. No equations in the provided text create a self-referential loop, no predictions are statistically forced by subset fits, and the central premise rests on an external reference rather than a self-citation chain or imported ansatz from the authors' prior work. The proposal is therefore self-contained against its inputs, with any limitations falling under validity or completeness rather than circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the direct transferability of the Benna-Fusi model to artificial memory graphs and on the assumption that the resulting dynamics will be beneficial and stable for LLM knowledge updating.

axioms (1)

domain assumption The Benna-Fusi model of synaptic consolidation can be applied to edges in an LLM associative memory graph to produce the desired learning behaviors.
Invoked in the abstract as the basis for the coupled fast-slow variables.

invented entities (1)

Memini associative memory no independent evidence
purpose: Graph-structured external memory whose edges carry coupled fast and slow variables for continual LLM knowledge updating.
New system introduced in the paper.

pith-pipeline@v0.9.0 · 5436 in / 1288 out tokens · 40343 ms · 2026-05-08T18:08:36.065136+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

Cost.FunctionalEquation / Foundation.AlphaCoordinateFixation washburn_uniqueness_aczel; J_uniquely_calibrated_via_higher_derivative unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

dw_fast/dt = -w_fast/τ_fast + C(w_slow - w_fast) + I(t); dw_slow/dt = -w_slow/τ_slow + C(w_fast - w_slow). τ_fast = 2, τ_slow = 10, C = 0.2, b = 1.

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

[1]

, title =

Anderson, John R. , title =. Cognitive Psychology , volume =. 1974 , month = oct, issn =

work page 1974
[2]

Advances in Neural Information Processing Systems 38 , publisher =

Behrouz, Ali and Razaviyayn, Meisam and Zhong, Peilin and Mirrokni, Vahab , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page
[3]

Advances in Neural Information Processing Systems 38 , publisher =

Behrouz, Ali and Zhong, Peilin and Mirrokni, Vahab , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page
[4]

and Fusi, Stefano , title =

Benna, Marcus K. and Fusi, Stefano , title =. Nature Neuroscience , volume =. 2016 , issn =

work page 2016
[5]

The Journal of Neuroscience , volume =

Bi, Guo-qiang and Poo, Mu-ming , title =. The Journal of Neuroscience , volume =. 1998 , doi =

work page 1998
[6]

and Loftus, Elizabeth F

Collins, Allan M. and Loftus, Elizabeth F. , title =. Psychological Review , volume =. 1975 , issn =

work page 1975
[7]

IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =

A continual learning survey:. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =. 2022 , issn =

work page 2022
[8]

and Eisenschlos, Julian Martin and Gillick, Daniel and Eisenstein, Jacob and Cohen, William W

Dhingra, Bhuwan and Cole, Jeremy R. and Eisenschlos, Julian Martin and Gillick, Daniel and Eisenstein, Jacob and Cohen, William W. , title =. Transactions of the Association for Computational Linguistics , volume =. 2022 , month = mar, issn =

work page 2022
[9]

2025 , month = feb, howpublished =

Edge, Darren and Trinh, Ha and Cheng, Newman and Bradley, Joshua and Chao, Alex and Mody, Apurva and Truitt, Steven and Metropolitansky, Dasha and Ness, Robert Osazuwa and Larson, Jonathan , title =. 2025 , month = feb, howpublished =

work page 2025
[10]

and Abbott, Larry F

Fusi, Stefano and Drew, Patrick J. and Abbott, Larry F. , title =. Neuron , volume =. 2005 , issn =

work page 2005
[11]

Guti. From. Proceedings of the 42nd International Conference on Machine Learning , series =

work page
[12]

Proceedings of the 13th International Conference on Human-Agent Interaction , series =

Honda, Yudai and Fujita, Yuki and Zempo, Keiichi and Fukushima, Shogo , title =. Proceedings of the 13th International Conference on Human-Agent Interaction , series =. 2026 , month = jan, publisher =

work page 2026
[13]

2026 , howpublished =

Jiang, Hanqi and Chen, Junhao and Pan, Yi and Chen, Linyi and You, Wei and Zhou, Yi and Zhang, Renxin and Sikora, Andrea and Zhao, Lu and Abate, Yohannes and Liu, Tianming , title =. 2026 , howpublished =

work page 2026
[14]

Proceedings of the 35th International Conference on Machine Learning , series =

Kaplanis, Christos and Shanahan, Murray and Clopath, Claudia , title =. Proceedings of the 35th International Conference on Machine Learning , series =

work page
[15]

2026 , howpublished =

Lau, Kwun Hang and Zhang, Fangyuan and Ruan, Boyu and Zhou, Yingli and Guo, Qintian and Zhang, Ruiyuan and Zhou, Xiaofang , title =. 2026 , howpublished =

work page 2026
[16]

Mind the gap:

Lazaridou, Angeliki and Kuncoro, Adhiguna and Gribovskaya, Elena and Agrawal, Devang and Li. Mind the gap:. Advances in Neural Information Processing Systems 34 , pages =

work page
[17]

Retrieval-

Lewis, Patrick and Perez, Ethan and Piktus, Aleksandra and Petroni, Fabio and Karpukhin, Vladimir and Goyal, Naman and K. Retrieval-. Advances in Neural Information Processing Systems 33 , pages =

work page
[18]

and McNaughton, Bruce L

McClelland, James L. and McNaughton, Bruce L. and O'Reilly, Randall C. , title =. Psychological Review , volume =. 1995 , doi =

work page 1995
[19]

Augmented

Mialon, Gr. Augmented. Transactions on Machine Learning Research , year =

work page
[20]

N. Why. Perspectives on Psychological Science , volume =. 2015 , month = sep, issn =

work page 2015
[21]

and Stoica, Ion and Gonzalez, Joseph E

Packer, Charles and Wooders, Sarah and Lin, Kevin and Fang, Vivian and Patil, Shishir G. and Stoica, Ion and Gonzalez, Joseph E. , title =. 2024 , month = feb, howpublished =

work page 2024
[22]

and Cai, Carrie J

Park, Joon Sung and O'Brien, Joseph C. and Cai, Carrie J. and Morris, Meredith Ringel and Liang, Percy and Bernstein, Michael S. , title =. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology , series =. 2023 , month = oct, publisher =

work page 2023
[23]

Leveraging

Pavlovi. Leveraging. 2025 , month = dec, howpublished =

work page 2025
[24]

ACM Computing Surveys , volume =

Shi, Haizhou and Xu, Zihao and Wang, Hengyi and Qin, Weiyi and Wang, Wenyuan and Wang, Yibin and Wang, Zifeng and Ebrahimi, Sayna and Wang, Hao , title =. ACM Computing Surveys , volume =. 2025 , issn =

work page 2025
[25]

Proceedings of the 39th International Conference on Machine Learning , series =

Sun, Tianxiang and Shao, Yunfan and Qian, Hong and Huang, Xuanjing and Qiu, Xipeng , title =. Proceedings of the 39th International Conference on Machine Learning , series =

work page
[26]

Behavioral and Brain Sciences , volume =

Tulving, Endel , title =. Behavioral and Brain Sciences , volume =. 1984 , month = jun, issn =

work page 1984
[27]

and Ebbesen, Ebbe B

Wixted, John T. and Ebbesen, Ebbe B. , title =. Memory & Cognition , volume =. 1997 , doi =

work page 1997
[28]

2024 , month = feb, howpublished =

Wu, Tongtong and Luo, Linhao and Li, Yuan-Fang and Pan, Shirui and Vu, Thuy-Trang and Haffari, Gholamreza , title =. 2024 , month = feb, howpublished =

work page 2024
[29]

Advances in Neural Information Processing Systems 38 , publisher =

Xu, Wujiang and Liang, Zujie and Mei, Kai and Gao, Hang and Tan, Juntao and Zhang, Yongfeng , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page
[30]

2024 , month = feb, howpublished =

Zaharia, Matei and Khattab, Omar and Chen, Lingjiao and Davis, Jared Quincy and Miller, Heather and Potts, Christopher and Zou, James and Carbin, Michael and Frankle, Jonathan and Rao, Naveen and Ghodsi, Ali , title =. 2024 , month = feb, howpublished =

work page 2024
[31]

Learning and Memory: A Comprehensive Reference , pages =

Zenke, Friedemann and Laborieux, Axel , title =. Learning and Memory: A Comprehensive Reference , pages =. 2025 , doi =

work page 2025
[32]

Proceedings of the AAAI Conference on Artificial Intelligence , volume =

Zhong, Wanjun and Guo, Lianghong and Gao, Qiqi and Ye, He and Wang, Yanlin , title =. Proceedings of the AAAI Conference on Artificial Intelligence , volume =. 2024 , doi =

work page 2024

[1] [1]

, title =

Anderson, John R. , title =. Cognitive Psychology , volume =. 1974 , month = oct, issn =

work page 1974

[2] [2]

Advances in Neural Information Processing Systems 38 , publisher =

Behrouz, Ali and Razaviyayn, Meisam and Zhong, Peilin and Mirrokni, Vahab , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page

[3] [3]

Advances in Neural Information Processing Systems 38 , publisher =

Behrouz, Ali and Zhong, Peilin and Mirrokni, Vahab , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page

[4] [4]

and Fusi, Stefano , title =

Benna, Marcus K. and Fusi, Stefano , title =. Nature Neuroscience , volume =. 2016 , issn =

work page 2016

[5] [5]

The Journal of Neuroscience , volume =

Bi, Guo-qiang and Poo, Mu-ming , title =. The Journal of Neuroscience , volume =. 1998 , doi =

work page 1998

[6] [6]

and Loftus, Elizabeth F

Collins, Allan M. and Loftus, Elizabeth F. , title =. Psychological Review , volume =. 1975 , issn =

work page 1975

[7] [7]

IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =

A continual learning survey:. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =. 2022 , issn =

work page 2022

[8] [8]

and Eisenschlos, Julian Martin and Gillick, Daniel and Eisenstein, Jacob and Cohen, William W

Dhingra, Bhuwan and Cole, Jeremy R. and Eisenschlos, Julian Martin and Gillick, Daniel and Eisenstein, Jacob and Cohen, William W. , title =. Transactions of the Association for Computational Linguistics , volume =. 2022 , month = mar, issn =

work page 2022

[9] [9]

2025 , month = feb, howpublished =

Edge, Darren and Trinh, Ha and Cheng, Newman and Bradley, Joshua and Chao, Alex and Mody, Apurva and Truitt, Steven and Metropolitansky, Dasha and Ness, Robert Osazuwa and Larson, Jonathan , title =. 2025 , month = feb, howpublished =

work page 2025

[10] [10]

and Abbott, Larry F

Fusi, Stefano and Drew, Patrick J. and Abbott, Larry F. , title =. Neuron , volume =. 2005 , issn =

work page 2005

[11] [11]

Guti. From. Proceedings of the 42nd International Conference on Machine Learning , series =

work page

[12] [12]

Proceedings of the 13th International Conference on Human-Agent Interaction , series =

Honda, Yudai and Fujita, Yuki and Zempo, Keiichi and Fukushima, Shogo , title =. Proceedings of the 13th International Conference on Human-Agent Interaction , series =. 2026 , month = jan, publisher =

work page 2026

[13] [13]

2026 , howpublished =

Jiang, Hanqi and Chen, Junhao and Pan, Yi and Chen, Linyi and You, Wei and Zhou, Yi and Zhang, Renxin and Sikora, Andrea and Zhao, Lu and Abate, Yohannes and Liu, Tianming , title =. 2026 , howpublished =

work page 2026

[14] [14]

Proceedings of the 35th International Conference on Machine Learning , series =

Kaplanis, Christos and Shanahan, Murray and Clopath, Claudia , title =. Proceedings of the 35th International Conference on Machine Learning , series =

work page

[15] [15]

2026 , howpublished =

Lau, Kwun Hang and Zhang, Fangyuan and Ruan, Boyu and Zhou, Yingli and Guo, Qintian and Zhang, Ruiyuan and Zhou, Xiaofang , title =. 2026 , howpublished =

work page 2026

[16] [16]

Mind the gap:

Lazaridou, Angeliki and Kuncoro, Adhiguna and Gribovskaya, Elena and Agrawal, Devang and Li. Mind the gap:. Advances in Neural Information Processing Systems 34 , pages =

work page

[17] [17]

Retrieval-

Lewis, Patrick and Perez, Ethan and Piktus, Aleksandra and Petroni, Fabio and Karpukhin, Vladimir and Goyal, Naman and K. Retrieval-. Advances in Neural Information Processing Systems 33 , pages =

work page

[18] [18]

and McNaughton, Bruce L

McClelland, James L. and McNaughton, Bruce L. and O'Reilly, Randall C. , title =. Psychological Review , volume =. 1995 , doi =

work page 1995

[19] [19]

Augmented

Mialon, Gr. Augmented. Transactions on Machine Learning Research , year =

work page

[20] [20]

N. Why. Perspectives on Psychological Science , volume =. 2015 , month = sep, issn =

work page 2015

[21] [21]

and Stoica, Ion and Gonzalez, Joseph E

Packer, Charles and Wooders, Sarah and Lin, Kevin and Fang, Vivian and Patil, Shishir G. and Stoica, Ion and Gonzalez, Joseph E. , title =. 2024 , month = feb, howpublished =

work page 2024

[22] [22]

and Cai, Carrie J

Park, Joon Sung and O'Brien, Joseph C. and Cai, Carrie J. and Morris, Meredith Ringel and Liang, Percy and Bernstein, Michael S. , title =. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology , series =. 2023 , month = oct, publisher =

work page 2023

[23] [23]

Leveraging

Pavlovi. Leveraging. 2025 , month = dec, howpublished =

work page 2025

[24] [24]

ACM Computing Surveys , volume =

Shi, Haizhou and Xu, Zihao and Wang, Hengyi and Qin, Weiyi and Wang, Wenyuan and Wang, Yibin and Wang, Zifeng and Ebrahimi, Sayna and Wang, Hao , title =. ACM Computing Surveys , volume =. 2025 , issn =

work page 2025

[25] [25]

Proceedings of the 39th International Conference on Machine Learning , series =

Sun, Tianxiang and Shao, Yunfan and Qian, Hong and Huang, Xuanjing and Qiu, Xipeng , title =. Proceedings of the 39th International Conference on Machine Learning , series =

work page

[26] [26]

Behavioral and Brain Sciences , volume =

Tulving, Endel , title =. Behavioral and Brain Sciences , volume =. 1984 , month = jun, issn =

work page 1984

[27] [27]

and Ebbesen, Ebbe B

Wixted, John T. and Ebbesen, Ebbe B. , title =. Memory & Cognition , volume =. 1997 , doi =

work page 1997

[28] [28]

2024 , month = feb, howpublished =

Wu, Tongtong and Luo, Linhao and Li, Yuan-Fang and Pan, Shirui and Vu, Thuy-Trang and Haffari, Gholamreza , title =. 2024 , month = feb, howpublished =

work page 2024

[29] [29]

Advances in Neural Information Processing Systems 38 , publisher =

Xu, Wujiang and Liang, Zujie and Mei, Kai and Gao, Hang and Tan, Juntao and Zhang, Yongfeng , title =. Advances in Neural Information Processing Systems 38 , publisher =

work page

[30] [30]

2024 , month = feb, howpublished =

Zaharia, Matei and Khattab, Omar and Chen, Lingjiao and Davis, Jared Quincy and Miller, Heather and Potts, Christopher and Zou, James and Carbin, Michael and Frankle, Jonathan and Rao, Naveen and Ghodsi, Ali , title =. 2024 , month = feb, howpublished =

work page 2024

[31] [31]

Learning and Memory: A Comprehensive Reference , pages =

Zenke, Friedemann and Laborieux, Axel , title =. Learning and Memory: A Comprehensive Reference , pages =. 2025 , doi =

work page 2025

[32] [32]

Proceedings of the AAAI Conference on Artificial Intelligence , volume =

Zhong, Wanjun and Guo, Lianghong and Gao, Qiqi and Ye, He and Wang, Yanlin , title =. Proceedings of the AAAI Conference on Artificial Intelligence , volume =. 2024 , doi =

work page 2024