Building compositional tasks with shared neural subspaces

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Фев. 1, 2024

Cognition is remarkably flexible; we are able to rapidly learn and perform many different tasks

Язык: Английский

---

Dimensionality reduction beyond neural subspaces with slice tensor component analysis

Nature Neuroscience, Год журнала: 2024, Номер 27(6), С. 1199 - 1210

Опубликована: Май 6, 2024

Abstract Recent work has argued that large-scale neural recordings are often well described by patterns of coactivation across neurons. Yet the view variability is constrained to a fixed, low-dimensional subspace may overlook higher-dimensional structure, including stereotyped sequences or slowly evolving latent spaces. Here we argue task-relevant in data can also cofluctuate over trials time, defining distinct ‘covariability classes’ co-occur within same dataset. To demix these covariability classes, develop sliceTCA (slice tensor component analysis), new unsupervised dimensionality reduction method for tensors. In three example datasets, motor cortical activity during classic reaching task primates and recent multiregion mice, show capture more structure using fewer components than traditional methods. Overall, our theoretical framework extends population incorporating additional classes variables capturing structure.

Язык: Английский

---

Flexible control of sequence working memory in the macaque frontal cortex

Neuron, Год журнала: 2024, Номер 112(20), С. 3502 - 3514.e6

Опубликована: Авг. 22, 2024

Язык: Английский

---

The enigmatic HCN channels: A cellular neurophysiology perspective

Proteins Structure Function and Bioinformatics, Год журнала: 2023, Номер 93(1), С. 72 - 92

Опубликована: Ноя. 19, 2023

Abstract What physiological role does a slow hyperpolarization‐activated ion channel with mixed cation selectivity play in the fast world of neuronal action potentials that are driven by depolarization? That puzzling question has piqued curiosity physiology enthusiasts about cyclic nucleotide‐gated (HCN) channels, which widely expressed across body and especially neurons. In this review, we emphasize need to assess HCN channels from perspective how they respond time‐varying signals, while also accounting for their interactions other co‐expressing receptors. First, illustrate unique structural functional characteristics allow them mediate negative feedback loop neurons express in. We present several implications response including gain, voltage sag rebound, temporal summation, membrane potential resonance, inductive phase lead, spike triggered average, coincidence detection. Next, argue overall impact on critically relies Interactions intrinsic oscillations, earning “pacemaker channel” moniker, regulate frequency adaptation, plateau potentials, neurotransmitter release presynaptic terminals, initiation at axonal initial segment. explore spatially non‐homogeneous subcellular distributions different subtypes Finally, discuss plasticity is prevalent can encoding, homeostatic, neuroprotective functions neuron. summary, form an important class diversity owing gating kinetics made puzzle first place.

Язык: Английский

---

Learning reshapes the hippocampal representation hierarchy

Proceedings of the National Academy of Sciences, Год журнала: 2025, Номер 122(11)

Опубликована: Март 10, 2025

A key feature of biological and artificial neural networks is the progressive refinement their representations with experience. In neuroscience, this fact has inspired several recent studies in sensory motor systems. However, less known about how higher associational cortical areas, such as hippocampus, modify throughout learning complex tasks. Here, we focus on associative learning, a process that requires forming connection between different variables for appropriate behavioral response. We trained rats space-context task monitored hippocampal activity entire period, over days. This allowed us to assess changes context, movement direction, position, well relationship behavior. identified hierarchical representational structure encoding these three was preserved learning. Nevertheless, also observed at lower levels hierarchy where context encoded. These were local space restricted physical positions identification necessary correct decision-making, supporting better decoding contextual code compression. Our results demonstrate not only accommodates relationships but enables efficient through minimal space. Beyond our work reveals representation mechanism might be implemented other performing similar

Язык: Английский

---

Editorial overview: Computational neuroscience as a bridge between artificial intelligence, modeling and data

Current Opinion in Neurobiology, Год журнала: 2024, Номер 84, С. 102835 - 102835

Опубликована: Янв. 6, 2024

Язык: Английский

---

Learning reshapes the hippocampal representation hierarchy

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 21, 2024

Abstract A key feature of biological and artificial neural networks is the progressive refinement their representations with experience. In neuroscience, this fact has inspired several recent studies in sensory motor systems. However, less known about how higher associational cortical areas, such as hippocampus, modify throughout learning complex tasks. Here we focus on associative learning, a process that requires forming connection between different variables for appropriate behavioral response. We trained rats spatial-context task monitored hippocampal activity entire period, over days. This allowed us to assess changes context, movement direction position, well relationship behavior. identified hierarchical representational structure encoding these three was preserved learning. Nevertheless, also observed at lower levels hierarchy where context encoded. These were local space restricted physical positions identification necessary correct decision making, supporting better decoding contextual code compression. Our results demonstrate not only accommodates relationships but enables efficient through minimal space. Beyond our work reveals representation mechanism might be implemented other performing similar

Язык: Английский

---

Information is asymmetry: spatial relations were encoded by asymmetric mnemonic manifolds

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Март 14, 2024

Abstract Substantial evidence suggests that working memory (WM) leverages relational representations to provide flexible support for cognitive functions, a capacity likely derived from the dynamic nature of neural codes in WM. However, how these represent and maintain relations remains unclear. Here, we examined transformation geometries dorsal prefrontal cortex monkeys performing visuospatial delayed-match/nonmatch task, where were instructed hold spatial location white square WM match it with subsequent square. We found sensory manifold during square’s presence mnemonic after offset both aligned stimulus manifold. significant differences emerged between manifolds, exhibiting little correlation their geometries. Further analysis on revealed process expansion followed by flattening: asymmetric first expanded into symmetric geometry immediately onset offset, which then gradually flattened along dimensions different those initially expanded, culminating an This reconstruction not only remained its faithfulness but also gained flexibility meet task demands. In sum, this asymmetry symmetry back precisely illustrates dynamics reconstruction, shedding lights subjective generates accurate illusory representation world lived in.

Язык: Английский

---

Domain-specific Schema Reuse Supports Flexible Learning to Learn in Primate Brain

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Окт. 27, 2024

Abstract Prior knowledge accelerates subsequent learning of similarly structured problems - a phenomenon termed “learning to learn” by forming and reusing generalizable neural representations, i.e., the schemas. However, stability-plasticity dilemma, how exploit stable schemas facilitate while remaining flexible towards possible changes, is not well understood. We hypothesize that restricting specific functional, e.g., decision-making, subspace making it orthogonal other subspaces allows brain balance stability plasticity. To test it, we trained three macaques on visuomotor mapping tasks recorded activity in dorsolateral premotor cortex. By delineating decision stimulus subspaces, identified schema-like manifold within only subspace. The reuse significantly facilitated learning. In addition, exhibited trend be subspace, minimizing interference between these two domains. Our results revealed functional domains can preserve useful maintaining orthogonality with allowing for adaptation new environments, thereby resolving dilemma. This finding provides insights into mechanisms underlying brain’s capability learn both fast flexibly, which also inspire more efficient algorithms artificial intelligence systems working open, dynamic environments.

Язык: Английский

---

Policy optimization emerges from noisy representation learning

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Ноя. 3, 2024

A bstract Nervous systems learn representations of the world and policies to act within it. We present a framework that uses reward-dependent noise facilitate policy opti- mization in representation learning networks. These networks balance extracting normative features task-relevant information solve tasks. Moreover, their changes reproduce several experimentally observed shifts neural code during task learning. Our presents biologically plausible mechanism for emergent optimization amid evidence plays vital role governing dynamics. Code is available at: NeuralThermalOptimization.

Язык: Английский

---