Research

Granger causality and its application in neuroscience

Cognitive brain functions are achieved through cooperative neural computing. Multisensor recording and functional imaging provide the multivariate data for studying the brain mechanisms of cognition from a network perspective. Conventionally, patterns of neural interactions are assessed by such techniques as cross correlation and coherence, which do not yield information on the direction of signal transmission. Since neural interactions are inherently directional, e.g., action potentials travel from one neuron to another, being able to assess the directionality of neuronal interactions is thus a highly desired capability. Development over the last decade has shown that Granger causality is a key technique to furnish this capability. Our lab has played a key role in this development and published some of the classical papers in the area. Today, Granger causality, together with other multivariate time series methods, are routinely applied by members of the lab to unravel the cooperative nature of neural computation.

Single-trial analysis of event-related potentials

A common experimental method to study cognitive brain functions is to record electric potentials from the brain over repeated performance of the same task (called trials) and average them over all the trials. The event-related potential (ERP) obtained this way, while useful, is incapable of capturing trial-to-trial variability of the brain responses. Our lab has developed methods to cope with this problem. Our latest technique, referred to as ASEO, is an iterative parameter estimation method, which estimates the ERPs on a single trial basis and has been successfully applied to a number of neuroscience problems.

Simultaneous EEG-fMRI recordings

EEG and fMRI are the two major imaging modalities for the noninvasive imaging of human brain functions. EEG, the older method of the two, has exquisite temporal resolution, and is capable of tracking neural events with millisecond precision. It has, however, poor spatial resolution. FMRI, on the other hand, has good spatial resolution but poor temporal resolution. The advent of simultaneous EEG and fMRI recording technology holds the promise to overcome these shortcomings. We are using this technology to address questions in a number of neuroscience areas, including control of attention, novelty detection, emotional control, and fear learning.

Characterization and function of neuronal oscillations

The functional significance of oscillatory neuronal in normal brain function and as a marker of brain pathology is being increasingly recognized. This is a major area of interest for our lab. We are among the first to apply advanced signal processing methods such as Granger causality to characterizing neuronal oscillations in nonhuman primates. Current topics of interest include alpha oscillations in humans and nonhuman primates, theta oscillations in humans and rodents, phase-amplitude coupling of neuronal oscillations in emotional processing, and brain-to-brain synchrony in human subjects engaged in social interactions.

Areas of current research activity with translational significance

1. Cognitive decline in orthopedic surgery and in Parkinson’s disease.
2. Neural mechanisms of pain and treatment strategies.
3. Neural mechanisms of cognitive impairments in MCI and in HIV infections.
4. Neural basis of fear and anxiety.
5. Animal models of schizophrenia.
6. Adverse impact of antiepileptic drugs on brain function.
7. Physiology and treatment of central fatigue in aging and in Parkinson’s disease.