Maryam Modir Shanechi

Decoding continuous goal-directed movement from human brain-wide intracranial recordings.

This study decodes continuous goal-directed movement from brain-wide intracranial recordings using PSID, showing widespread neural correlates across multiple brain regions.

Event Detection and Classification from Multimodal Time Series with Application to Neural Data

This paper develops a method to detect and classify events in multimodal time series data, demonstrating its effectiveness on neural spike-LFP data for eye movement detection and classification.

Modeling and dissociation of intrinsic and input-driven neural population dynamics underlying behavior

This paper develops a method to dissociate intrinsic neural dynamics from input-driven dynamics in neural-behavioral data, showing it better captures behaviorally relevant dynamics across tasks and subjects.

Multimodal subspace identification for modeling discrete-continuous spiking and field potential population activity

This paper introduces a multiscale subspace identification algorithm for modeling multimodal spiking and field potential data, showing improved efficiency and accuracy in capturing low-dimensional dynamics and predicting behavior.

Post-Stimulus Encoding of Decision Confidence in EEG: Toward a Brain-Computer Interface for Decision Making

This paper investigates whether decision confidence can be decoded from EEG signals post-stimulus and pre-response to enable a brain-computer interface for improving decision making.

Dynamical flexible inference of nonlinear latent structures in neural population activity

DFINE infers nonlinear latent structures in neural population activity, achieving flexible inference and better prediction of behavior and neural activity.

Modeling and Inference Methods for Switching Regime-Dependent Dynamical Systems with Multiscale Neural Observations

This paper develops a model and inference methods for regime-dependent multiscale neural dynamics, showing improved state estimation compared to single-scale or stationary methods.

Where is all the nonlinearity: flexible nonlinear modeling of behaviorally relevant neural dynamics using recurrent neural networks

This paper introduces RNN PSID, a nonlinear dynamic modeling method for analyzing neural dynamics, and shows it outperforms linear and non-preferential methods in behavior decoding.

Modeling behaviorally relevant neural dynamics enabled by preferential subspace identification (PSID)

This paper introduces PSID, a method to dissociate behaviorally relevant neural dynamics from irrelevant ones, revealing lower-dimensional and more predictive dynamics in monkey motor tasks.