It is found that a structured entorhinal-hippocampal memory model, which provides a content-independent metric scaffold based on grid cells for storing experience, coupled to a policy recurrent network, succeeds at all three forms of generalization.
Abstract
Flexible behavior requires generalizable memory and learning. For example, we rapidly learn to commute in new cities by reusing our knowledge of Euclidean two-dimensional space and structures like roundabouts and subway systems without forgetting how to get to a favorite restaurant back home. Yet we lack a detailed understanding of how the brain uses existing knowledge to generalize while retaining the memory of specific past experiences. To address this gap, we combine behavioral measurements, neural recordings, and computational modeling in an abstract sequential image navigation task to study three forms of generalization: mnemonic generalization, from visual to mental navigation; transitive generalization, from trained to novel routes; and structural generalization, from familiar to new environments. In contrast to monkeys and humans, recurrent neural networks failed at all generalizations. We found that a structured entorhinal-hippocampal memory model, which provides a content-independent metric scaffold based on grid cells for storing experience, coupled to a policy recurrent network, succeeds at all three. The content-independent scaffold enables mnemonic and transitive generalization through path integration and facilitates structural generalization by allowing reuse of a previously learned action policy network. Moreover, the scaffold’s high combinatorial capacity permits continual learning without catastrophic forgetting. We recorded neural activity from the entorhinal cortex and posterior parietal cortex of two monkeys performing the task and found two distinct computations across the neural population. Modularizing an entorhinal and parietal action policy network to separately track distance and initiate actions captured the distinct population dynamics and improved model performance. Finally, we added a reinforcement learning module to the network that enabled it to learn an appropriate scale factor to align the grid periodicity with the environmental temporal structure. Our findings reveal that an architecture which factorizes invariant metric representations from rapid sensory associations and a transferable policy learns, generalizes, and remembers like the brain.
Findings suggest that the mEC network generates low-dimensional dynamics more broadly, with grid-cell activity representing one manifestation and more complex manifold topologies supporting task-related representations during working memory.
Li Yuan, Jia-Rui Wang, Xiaohe Li et al.· bioRxiv· 0 citations
Animals can solve new, complex tasks by reusing and adapting what they’ve learned before. This kind of flexibility depends not just on having prior experience, but on how that experience was structured in the first place. The design of early training curriculum is especially important: poorly structured experiences can...
John C. Bowler, D. Azhar, Cambria M. Jensen et al.· bioRxiv· 5 citations
It is indicated that LC inputs to the hippocampus preferentially facilitate spatial memory updating rather than uniformly enhancing spatial learning, and a complementary framework for quantifying the organization of radial-maze trajectories is introduced.
This work investigated whether naturalistically encoded memories that were either episodic or additionally embedded in a navigational experience relied on overlapping or distinct neural networks in the hippocampus, ERC, and lateral occipital cortex.
Kim V. Nguyen, L. Skalaban, Büşra Tanrıverdi et al.· Journal of Cognitive Neurosc...· 0 citations
Associative memory encoding shifts CA3 into a distinct representational regime, decoupling the immediate precision of a new cognitive map from the experience-dependent refinement of its context-specific population embedding.
Nathalie Immerzeel, Cecilia Gallego-Carracedo, L. Mauro et al.· bioRxiv· 0 citations
Across visual recognition, vision-language understanding, ego-exo video understanding, and embodied vision-language-action learning, this method consistently improves learning under online and uncertain data streams, with gains exceeding 50 percentage points over replay-free alternatives in embodied manipulation.
Hong-Wei Yan, Kang-Lei Zhou, Qi-Hao Cheng et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.