Research

Our research (Nair Lab – MU College of Engineering) involves interdisciplinary collaborations with neuroscience labs and research groups, focusing on three areas - computational neural engineering, machine learning/AI, and cyberinfrastructures (CI):

• Headley Lab and Pare Lab – Rutgers University

• Sah Lab – Queensland University, Australia

• Kundu Lab - MU School of Medicine, Neurology

• Schulz Lab – MU Biological Sciences

Nair Lab

Systems and Computational Neuroscience

The Nair lab projects involve reverse engineering the brain circuits, at intracellular, cellular, systems/networks, cognitive, and behavior/clinical levels, in close association with neuro- and physician-scientists. The projects require expertise and tools from several areas - computational modeling (biophysical, etc.), machine learning/AI, and cyberinfrastructure (CI), and combinations of these. 

We model a neuron as a nonlinear electrical circuit and combine many neurons to form networks at the 'biophysical' level. We also developed reduced order versions of such models, and black box (statistical, machine learning), etc. versions. Using a variety of such computational models, we use system theoretic concepts to investigate how neurons/network circuits implement functions, including learning. Some of the neurocomputational properties include structure-function as well as system level issues such as adaptation and learning (LTP/LTD, etc.), robustness, control and related ones for these nonlinear dynamic circuits, at cellular, systems/networks, cognitive, and behavior/clinical levels.

Some of our focus areas:

• Neural oscillations - Automated identification of oscillations using in vivo LFP data for detect oscillations online in experiments; Modeling gamma and its modulation in the amygdala; Beta-gamma coupling in network models of M1 cortex; Mechanisms underlying the generation of spontaneous ripples in amygdala and hippocampus; Automated prediction of oscillatory bursts in vivo using LFP data

• Single neuron and network models - Integration of dendritic spikes in a morphologically detailed single cell model; Machine learning models for automated identification of neuron types from neuropixel probe data; How do neurons and networks maintain steady output in the presence of variability in both intrinsic and synaptic properties?

• Amygdala and fear - Role of NMDARs in generating network activity during the processing of auditory tones and inducing plasticity during fear conditioning and extinction;  

• Machine learning/AI models for mechanistic understanding of brain function and for identifying potential targets for neuromodulation for pathophysiology - what neural circuits/regions supporting working memory (WM) in humans?  what circuits/regions support substance-abuse disorder pathophysiology?

Computational models of brain will provide a fundamental understanding of the underlying pathophysiology in both health and disease (e.g., working memory, substance-abuse, anxiety disorders, PTSD). Such models will also help lay the groundwork for innovative interventions including pharmacological, psychotherapeutic, and also neuromodulation by permitting rapid in-computer experimentation.

NIH Research Training (BRAIN Initiative)

Computational Neuroscience: Models & Neurobiology (for pre-doc, post-docs, medical students, residents and faculty)

Undergraduate Research

• NSF Neuro REU Project

Research Experiences for Teachers

• NSF Research Experiences for Teachers in Neural Engineering

  • Six-week research experience for K-12 teachers.

Modeling Resources

• See CyNeuro.org for open-source research and teaching software hosted by the Lab