Complexity
Order/Disorder We characterize the regularity and irregularity of local field potential and spiking time series across different sleep-wake states using entropy measures like sample entropy or permutation entropy. This is done at various timescales using multiscale measures on neuropixel electrophysiological data.
Scale We examine the self-organization and criticality of neural events, exploring phase transitions in driven vs. control regimes and intrinsic activity. This project uses calcium imaging data to compare cortical regions in early visual areas, focusing on both deep and superficial layers.
Wave Our team is building quantification tools to characterize traveling waves measured with high-density miniature electrocorticograms. Using AI, we analyze various features of these waves to study the interaction of different rhythms through traveling waves.
Pattern
Synthesis We construct latent spaces from multiarea neuropixel recordings and use generative models to create synthetic data that resembles observed spike-time measurements. These models help us generate data-driven governing equations coordinating activity among recorded regions.
Randomness We examine the random vs. structured nature of spiking activity measured with neuropixels. Our objective is to map this data to a spectrum of random to highly structured coordinate spaces, aiding in the development of metrics for quantifying measurements.
Multiscale Correlation Inspired by simulated annealing and the Ising model, we build multiscale correlation maps between pairs of neurons and higher-order activities. This dynamic mapping uses CA imaging and neuropixel data to create a representation of neural coordination.
Dynamics/Computation/Control
System Identification Using AI-driven data modeling, we examine the interactions between inhibitory and excitatory neurons through multielectrode recording. Our objective is to derive governing equations that control these interactions.
Causal Mapping We use empirical dynamic modeling to build causal mappings of interareal communication in neurophysiological elements during the wake-sleep cycle. This involves analyzing LFP and spike data to understand causal relationships.
Koopman Autoencoders We develop Koopman autoencoders to identify constrained latent spaces, examining system dimensionality and regularity/irregularity in different states. This is applied to LFP and spike data.
Topological Data Analysis Through AI and manifold learning, we study perceptual representation in early cortical areas using neuropixel and CA imaging data. Predictive coding tasks help us explore the emergence of these representations.
Laminar Physiology/Cell Type
State-Dependent Changes We examine state-dependent changes in laminar physiology by decomposing current sink and sources into spatial modes. Dynamic mode decomposition and machine learning techniques help us profile state-dependent modes and their dynamics using multielectrode arrays.
Morphological Characteristics Our research focuses on the morphological characteristics of inhibitory neurons across different cortical layers, using patch-seq data to explore their physiological profiles and morpho-characteristics.
Cell-Type Contributions We analyze state-dependent contributions of excitatory vs. inhibitory neurons to spatiotemporal motifs, aiming to build a biophysically constrained model of laminar dynamics.
Network Motifs
Higher Order Interactions We investigate higher-order interactions among neurons using CA imaging data in early visual areas. Information theory and motif extraction using machine learning help us understand population dynamics in response to visual stimuli.
Causal Dependence We use reservoir computing and CA imaging data to infer short-term causal dependencies between neuronal populations from time-series measurements, examining the causal structure of population activity.
Information Spaces Employing tools from topology and information geometry, we analyze neuropixel recordings to build information spaces. These topological structures help us understand multiscale attributes of single neuron spiking and their contribution to neural manifolds.