Multimodal subspace identification for modeling discrete-continuous spiking and field potential population activity DOI Creative Commons
Parima Ahmadipour, Omid G. Sani, Bijan Pesaran

et al.

Journal of Neural Engineering, Journal Year: 2023, Volume and Issue: 21(2), P. 026001 - 026001

Published: Nov. 29, 2023

Abstract Objective. Learning dynamical latent state models for multimodal spiking and field potential activity can reveal their collective low-dimensional dynamics enable better decoding of behavior through fusion. Toward this goal, developing unsupervised learning methods that are computationally efficient is important, especially real-time applications such as brain–machine interfaces (BMIs). However, remains elusive spike-field data due to heterogeneous discrete-continuous distributions different timescales. Approach. Here, we develop a multiscale subspace identification (multiscale SID) algorithm enables modeling dimensionality reduction data. We describe the combined Poisson Gaussian observations, which derive new analytical SID method. Importantly, also introduce novel constrained optimization approach learn valid noise statistics, critical statistical inference state, neural activity, behavior. validate method using numerical simulations with local population recorded during naturalistic reach grasp Main results. find accurately learned signals extracted from these signals. Further, it fused information, thus identifying modes predicting compared single modality. Finally, existing expectation-maximization Poisson–Gaussian had much lower training time while being in having or similar accuracy Significance. Overall, an accurate particularly beneficial when interest, online adaptive BMIs track non-stationary reducing offline neuroscience investigations.

Language: Английский

Stability from subspace rotations and traveling waves DOI Open Access
Tamal Batabyal, Scott L. Brincat, Jacob Donoghue

et al.

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

Published: Feb. 19, 2024

Abstract Cortical activity shows the ability to recover from distractions. We analyzed neural prefrontal cortex (PFC) of monkeys performing working memory tasks with mid-memory-delay distractions (a cued gaze shift or an irrelevant visual input). After distraction there were state-space rotational dynamics that returned spiking population patterns similar those pre-disruption. In fact, rotations fuller when task was performed correctly versus errors made. found a correspondence between and traveling waves across surface PFC. This suggests role for emergent like in recovery

Language: Английский

Citations

1
Stability from subspace rotations and traveling waves DOI Creative Commons
Earl K. Miller, Tamal Batabyal, Scott L. Brincat

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: March 15, 2024

Abstract Cortical activity shows stability, including the ability to recover from disruptions. We analyzed spiking prefrontal cortex (PFC) of monkeys performing working memory tasks with mid-memory-delay distractions. Perturbation by events (a gaze shift or visual inputs) caused rotational dynamics in subspace that could return patterns similar those before perturbation. In fact, after a distraction, rotations were fuller when task was correctly performed vs errors made. found direct correspondence between state-space and traveling waves rotating across surface PFC. This suggests role for cortical stability trajectories waves.

Language: Английский

Citations

1
Integrity of the circadian clock determines regularity of high-frequency and diurnal LFP rhythms within and between brain areas DOI Creative Commons
Paul Volkmann, A. Geiger,

Anisja Hühne-Landgraf

et al.

Molecular Psychiatry, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 29, 2024

Abstract Circadian clocks control most physiological processes of many species. We specifically wanted to investigate the influence environmental and endogenous rhythms their interplay on electrophysiological dynamics neuronal populations. Therefore, we measured local field potential (LFP) time series in wild-type Cryptochrome 1 2 deficient ( Cry1/2 −/− ) mice suprachiasmatic nucleus accumbens under regular light conditions constant darkness. Using refined descriptive statistical analyses, systematically profiled LFP activity. show that both strongly rhythmicity signals frequency components, but also shape patterns much smaller scales, as activity is significantly less at each more synchronous within between brain areas than mice. These results functional circadian are integral for non-circadian coordination ensemble dynamics.

Language: Английский

Citations

1
Unsupervised Neural Manifold Alignment for Stable Decoding of Movement from Cortical Signals DOI

Mohammadali Ganjali,

Alireza Mehridehnavi,

Sajed Rakhshani

et al.

International Journal of Neural Systems, Journal Year: 2023, Volume and Issue: 34(01)

Published: Nov. 15, 2023

The stable decoding of movement parameters using neural activity is crucial for the success brain-machine interfaces (BMIs). However, can be unstable over time, leading to changes in used movement, which hinder accurate decoding. To tackle this issue, one approach transfer a stable, low-dimensional manifold dimensionality reduction techniques and align manifolds across sessions by maximizing correlations manifolds. practical use stabilization requires knowledge true subject intentions such as target direction or behavioral state. overcome limitation, an automatic unsupervised algorithm proposed that determines intention before alignment presence rotation scaling sessions. This combined with method decoder instabilities. effectiveness BMI stabilizer represented two-dimensional (2D) hand velocity two rhesus macaque monkeys during center-out-reaching task. performance evaluated correlation coefficient R-squared measures, demonstrating higher compared state-of-the-art stabilizer. results offer benefits determination intents long-term Overall, offers promising solution achieving applications.

Language: Английский

Citations

3
Multimodal subspace identification for modeling discrete-continuous spiking and field potential population activity DOI Creative Commons
Parima Ahmadipour, Omid G. Sani, Bijan Pesaran

et al.

Journal of Neural Engineering, Journal Year: 2023, Volume and Issue: 21(2), P. 026001 - 026001

Published: Nov. 29, 2023

Abstract Objective. Learning dynamical latent state models for multimodal spiking and field potential activity can reveal their collective low-dimensional dynamics enable better decoding of behavior through fusion. Toward this goal, developing unsupervised learning methods that are computationally efficient is important, especially real-time applications such as brain–machine interfaces (BMIs). However, remains elusive spike-field data due to heterogeneous discrete-continuous distributions different timescales. Approach. Here, we develop a multiscale subspace identification (multiscale SID) algorithm enables modeling dimensionality reduction data. We describe the combined Poisson Gaussian observations, which derive new analytical SID method. Importantly, also introduce novel constrained optimization approach learn valid noise statistics, critical statistical inference state, neural activity, behavior. validate method using numerical simulations with local population recorded during naturalistic reach grasp Main results. find accurately learned signals extracted from these signals. Further, it fused information, thus identifying modes predicting compared single modality. Finally, existing expectation-maximization Poisson–Gaussian had much lower training time while being in having or similar accuracy Significance. Overall, an accurate particularly beneficial when interest, online adaptive BMIs track non-stationary reducing offline neuroscience investigations.

Language: Английский

Citations

3