A generative model of memory construction and consolidation

Nature Human Behaviour, Journal Year: 2024, Volume and Issue: 8(3), P. 526 - 543

Published: Jan. 19, 2024

Abstract Episodic memories are (re)constructed, share neural substrates with imagination, combine unique features schema-based predictions and show distortions that increase consolidation. Here we present a computational model in which hippocampal replay (from an autoassociative network) trains generative models (variational autoencoders) to (re)create sensory experiences from latent variable representations entorhinal, medial prefrontal anterolateral temporal cortices via the formation. Simulations effects of memory age lesions agreement previous models, but also provide mechanisms for semantic memory, episodic future thinking, relational inference including boundary extension. The explains how predictable conceptual elements stored reconstructed by efficiently combining both neocortical systems, optimizing use limited storage new unusual information. Overall, believe training provides comprehensive account construction, imagination

Language: Английский

---

A model of autonomous interactions between hippocampus and neocortex driving sleep-dependent memory consolidation

Proceedings of the National Academy of Sciences, Journal Year: 2022, Volume and Issue: 119(44)

Published: Oct. 24, 2022

How do we build up our knowledge of the world over time? Many theories memory formation and consolidation have posited that hippocampus stores new information, then “teaches” this information to neocortex time, especially during sleep. But it is unclear, mechanistically, how actually works—How are these systems able interact periods with virtually no environmental input accomplish useful learning shifts in representation? We provide a framework for thinking about question, neural network model simulations serving as demonstrations. The composed neocortical areas, which replay memories one another completely autonomously simulated Oscillations leveraged support error-driven leads changes representation behavior. has non–rapid eye movement (NREM) sleep stage, where dynamics between tightly coupled, helping reinstate high-fidelity versions attractors, REM more freely explore existing attractors. find alternating NREM stages, alternately focuses model’s on recent remote facilitates graceful continual learning. thus an account can without any external drive cortical protect old integrated.

Language: Английский

---

Boosting generalization of fine-tuning BERT for fake news detection

Information Processing & Management, Journal Year: 2024, Volume and Issue: 61(4), P. 103745 - 103745

Published: April 13, 2024

Language: Английский

---

Structure transfer and consolidation in visual implicit learning

Published: Jan. 24, 2025

Transfer learning, the re-application of previously learned higher-level regularities to novel input, is a key challenge in cognition. While previous empirical studies investigated human transfer learning supervised or reinforcement for explicit knowledge, it unknown whether such occurs during naturally more common implicit and unsupervised and, if so, how related memory consolidation. We compared newly acquired abstract knowledge by extending visual statistical paradigm context. found but with important differences depending on explicitness/implicitness knowledge. Observers acquiring initial could structures immediately. In contrast, observers same amount showed opposite effect, structural interference transfer. However, sleep between phases, observers, while still remaining implicit, switched their behaviour pattern as did. This effect was specific not after non-sleep Our results highlight similarities generalizable relying consolidation restructuring internal representations.

Language: Английский

---

Beyond hippocampus: Thalamic and prefrontal contributions to an evolving memory

Neuron, Journal Year: 2024, Volume and Issue: 112(7), P. 1045 - 1059

Published: Jan. 24, 2024

Language: Английский

---

The relational bottleneck as an inductive bias for efficient abstraction

Trends in Cognitive Sciences, Journal Year: 2024, Volume and Issue: 28(9), P. 829 - 843

Published: May 9, 2024

Language: Английский

---

Prefrontal cortical ripples mediate top-down suppression of hippocampal reactivation during sleep memory consolidation

Current Biology, Journal Year: 2024, Volume and Issue: 34(13), P. 2801 - 2811.e9

Published: June 3, 2024

Language: Английский

---

A comparison of hippocampal and retrosplenial cortical spatial and contextual firing patterns

Hippocampus, Journal Year: 2024, Volume and Issue: 34(7), P. 357 - 377

Published: May 21, 2024

The hippocampus (HPC) and retrosplenial cortex (RSC) are key components of the brain's memory navigation systems. Lesions either region produce profound deficits in spatial cognition HPC neurons exhibit well-known firing patterns (place fields). Recent studies have also identified an array navigation-related RSC. However, there has been little work comparing response properties information coding mechanisms these two brain regions. In present study, we examined RSC tasks which commonly used to study rodents, open field foraging with environmental context manipulation continuous T-maze alternation. We found striking similarities kinds contextual encoded by Neurons both regions carried about rat's current location, trajectories goal locations, reliably differentiated contexts. several differences. For example, head direction was a prominent component representations but only weakly HPC. different schemes, even when they same kind information. As expected, employed sparse scheme characterized compact, high contrast place fields, location dominant representations. were more consistent distributed scheme. Instead compact exhibited broad, reliable, directional tuning, typically multiple navigational variables. observed highlight closely related functions RSC, whereas differences types schemes suggest that likely make somewhat contributions cognition.

Language: Английский

---

Interactive Continual Learning: Fast and Slow Thinking

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Journal Year: 2024, Volume and Issue: 6, P. 12882 - 12892

Published: June 16, 2024

Language: Английский

---

Computational Models of Hippocampal Cognitive Function

Oxford University Press eBooks, Journal Year: 2025, Volume and Issue: unknown, P. 809 - 856

Published: Jan. 1, 2025

Abstract Computational models of the hippocampus are invaluable in exploring link between neurons and behavior, enabling hypothetical mechanisms to be defined precisely examined quantitatively. This chapter reviews many these, including spatial mnemonic function, that stress feedforward processing through hippocampal system, those stressing recurrent within it. It first, then associative or episodic memory function. Finally, it recent proposals unify functions hippocampus. These based on two related considerations: correspondence place index-like representations low dimensional latent variables efficiently describe underlying causes sensory input; benefit finding capture common relational structure across multiple contexts for prediction generalization. Overall, these instantiate cognitive maps states concepts relationships them.

Language: Английский

---