2-D Neural Geometry Underpins Hierarchical Organization of Sequence in Human Working Memory DOI Creative Commons
Ying Fan,

Muzhi Wang,

Nai Ding

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Feb. 21, 2024

Abstract Working memory (WM) is constructive in nature. Instead of passively retaining information, WM reorganizes complex sequences into hierarchically embedded chunks to overcome capacity limits and facilitate flexible behavior. To investigate the neural mechanisms underlying hierarchical reorganization WM, we performed two electroencephalography (EEG) one magnetoencephalography (MEG) experiments, wherein humans retain a temporal sequence items, i.e., syllables, which are organized chunks, multisyllabic words. We demonstrate that 1-D represented by 2-D representational geometry arising from parietal-frontal regions, with separate dimensions encoding item position within chunk sequence. Critically, this observed consistently different experimental settings, even during tasks discouraging correlates Overall, these findings strongly support reorganized factorized multi-dimensional also speaks general structure-based organizational principles given WM’s involvement many cognitive functions.

Language: Английский

Neural population geometry: An approach for understanding biological and artificial neural networks DOI Creative Commons
SueYeon Chung,

L. F. Abbott

Current Opinion in Neurobiology, Journal Year: 2021, Volume and Issue: 70, P. 137 - 144

Published: Oct. 1, 2021

Advances in experimental neuroscience have transformed our ability to explore the structure and function of neural circuits. At same time, advances machine learning unleashed remarkable computational power artificial networks (ANNs). While these two fields different tools applications, they present a similar challenge: namely, understanding how information is embedded processed through high-dimensional representations solve complex tasks. One approach addressing this challenge utilize mathematical analyze geometry representations, i.e., population geometry. We review examples geometrical approaches providing insight into biological networks: representation untangling perception, geometric theory classification capacity, disentanglement abstraction cognitive systems, topological underlying maps, dynamic motor dynamical cognition. Together, findings illustrate an exciting trend at intersection learning, neuroscience, geometry, which provides useful population-level mechanistic descriptor task implementation. Importantly, descriptions are applicable across sensory modalities, brain regions, network architectures timescales. Thus, has potential unify networks, bridging gap between single neurons, populations behavior.

Language: Английский

Citations

200
The computational foundations of dynamic coding in working memory DOI Creative Commons
Jake P. Stroud, John Duncan, Máté Lengyel

et al.

Trends in Cognitive Sciences, Journal Year: 2024, Volume and Issue: 28(7), P. 614 - 627

Published: April 4, 2024

Working memory (WM) is a fundamental aspect of cognition. WM maintenance classically thought to rely on stable patterns neural activities. However, recent evidence shows that population activities during undergo dynamic variations before settling into pattern. Although this has been difficult explain theoretically, network models optimized for typically also exhibit such dynamics. Here, we examine versus coding in data, classical models, and task-optimized networks. We review principled mathematical reasons why do not, while naturally coding. suggest an update our understanding maintenance, which computational feature rather than epiphenomenon.

Language: Английский

Citations

17
50 years of mnemonic persistent activity: quo vadis? DOI Creative Commons
Xiao‐Jing Wang

Trends in Neurosciences, Journal Year: 2021, Volume and Issue: 44(11), P. 888 - 902

Published: Oct. 13, 2021

Language: Английский

Citations

79
Constructing neural network models from brain data reveals representational transformations linked to adaptive behavior DOI Creative Commons
Takuya Ito, Guangyu Robert Yang,

Patryk A. Laurent

et al.

Nature Communications, Journal Year: 2022, Volume and Issue: 13(1)

Published: Feb. 3, 2022

Abstract The human ability to adaptively implement a wide variety of tasks is thought emerge from the dynamic transformation cognitive information. We hypothesized that these transformations are implemented via conjunctive activations in “conjunction hubs”—brain regions selectively integrate sensory, cognitive, and motor activations. used recent advances using functional connectivity map flow activity between brain construct task-performing neural network model fMRI data during control task. verified importance conjunction hubs computations by simulating over this empirically-estimated model. These empirically-specified simulations produced above-chance task performance (motor responses) integrating sensory rule hubs. findings reveal role supporting flexible computations, while demonstrating feasibility models gain insight into brain.

Language: Английский

Citations

43
Ketamine induces multiple individually distinct whole-brain functional connectivity signatures DOI Creative Commons
Flora Moujaes, Jie Lisa Ji, Masih Rahmati

et al.

eLife, Journal Year: 2024, Volume and Issue: 13

Published: April 17, 2024

Background: Ketamine has emerged as one of the most promising therapies for treatment-resistant depression. However, inter-individual variability in response to ketamine is still not well understood and it unclear how ketamine’s molecular mechanisms connect its neural behavioral effects. Methods: We conducted a single-blind placebo-controlled study, with participants blinded their treatment condition. 40 healthy received acute (initial bolus 0.23 mg/kg, continuous infusion 0.58 mg/kg/hr). quantified resting-state functional connectivity via data-driven global brain related individual ketamine-induced symptom variation cortical gene expression targets. Results: found that: (i) both effects are multi-dimensional, reflecting robust variability; (ii) principal gradient effect matched somatostatin (SST) parvalbumin (PVALB) patterns humans, while mean did not; (iii) mapped onto distinct gradients ketamine, which were resolvable at single-subject level. Conclusions: These results highlight importance considering ketamine. They also have implications development individually precise pharmacological biomarkers selection psychiatry. Funding: This study was supported by NIH grants DP5OD012109-01 (A.A.), 1U01MH121766 R01MH112746 (J.D.M.), 5R01MH112189 5R01MH108590 NIAAA grant 2P50AA012870-11 (A.A.); NSF NeuroNex 2015276 (J.D.M.); Brain Behavior Research Foundation Young Investigator Award SFARI Pilot (J.D.M., A.A.); Heffter Institute (Grant No. 1–190420) (FXV, KHP); Swiss Neuromatrix 2016–0111) National Science under framework Neuron Cofund 01EW1908) (KHP); Usona (2015 – 2056) (FXV). Clinical trial number: NCT03842800

Language: Английский

Citations

9
Maintenance and transformation of representational formats during working memory prioritization DOI Creative Commons
Daniel Pacheco, Marie-Christin Fellner, Lukas Kunz

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Sept. 19, 2024

Language: Английский

Citations

5
A transient high-dimensional geometry affords stable conjunctive subspaces for efficient action selection DOI Creative Commons
Atsushi Kikumoto, Apoorva Bhandari, Kazuhisa Shibata

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Oct. 1, 2024

Language: Английский

Citations

4
Intermittent rate coding and cue-specific ensembles support working memory DOI Creative Commons
Matthew F. Panichello, Donatas Jonikaitis,

Yu jin Oh

et al.

Nature, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 6, 2024

Abstract Persistent, memorandum-specific neuronal spiking activity has long been hypothesized to underlie working memory 1,2 . However, emerging evidence suggests a potential role for ‘activity-silent’ synaptic mechanisms 3–5 This issue remains controversial because either view largely relied on datasets that fail capture single-trial population dynamics or indirect measures of spiking. We addressed this controversy by examining the mnemonic information single trials obtained from large, local populations lateral prefrontal neurons recorded simultaneously in monkeys performing task. Here we show does not persist during delays, but instead alternates between coordinated ‘On’ and ‘Off’ states. At level neurons, Off states are driven both loss selectivity memoranda return firing rates spontaneous levels. Further exploiting large-scale recordings used here, is available patterns functional connections among ensembles Our results suggest intermittent periods coexist with support memory.

Language: Английский

Citations

4
Comparing representations and computations in single neurons versus neural networks DOI
Camilo Libedinsky

Trends in Cognitive Sciences, Journal Year: 2023, Volume and Issue: 27(6), P. 517 - 527

Published: April 1, 2023

Language: Английский

Citations

10
The topology and geometry of neural representations DOI Creative Commons
Baihan Lin, Nikolaus Kriegeskorte

Proceedings of the National Academy of Sciences, Journal Year: 2024, Volume and Issue: 121(42)

Published: Oct. 7, 2024

A central question for neuroscience is how to characterize brain representations of perceptual and cognitive content. An ideal characterization should distinguish different functional regions with robustness noise idiosyncrasies individual brains that do not correspond computational differences. Previous studies have characterized by their representational geometry, which defined the dissimilarity matrix (RDM), a summary statistic abstracts from roles neurons (or responses channels) characterizes discriminability stimuli. Here, we explore further step abstraction: geometry topology representations. We propose topological similarity analysis, an extension analysis uses family geotopological statistics generalizes RDM while de-emphasizing geometry. evaluate this in terms sensitivity specificity model selection using both simulations MRI (fMRI) data. In simulations, ground truth data-generating layer representation neural network models are same other layers instances (trained random seeds). fMRI, visual area areas measured subjects. Results show topology-sensitive characterizations population codes robust interindividual variability maintain excellent unique signatures regions.

Language: Английский

Citations

3