Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Master of Science

Program

Physics

Supervisor

Dr. Andrea Soddu

Abstract

Integrated Information Theory (IIT) is a framework developed to explain consciousness, arguing that conscious systems consist of interacting elements that are integrated through their causal properties. In this study, we present the first application of IIT to functional magnetic resonance imaging (fMRI) data and investigate whether its principal metric, Phi, can meaningfully quantify resting-state cortical activity patterns. Data was acquired from 17 healthy subjects who underwent sedation with propofol, a short acting anesthetic. Using PyPhi, a software package developed for IIT, we thoroughly analyze how Phi varies across different networks and throughout sedation. Our findings indicate that variations in Phi closely reflect the conscious state of patients in the frontoparietal and dorsal attention networks, which are responsible for higher-order cognitive functions. Despite existing limitations in applying IIT to empirical data, we obtained promising results that merit further applications of this framework to fMRI.

Summary for Lay Audience

Consciousness is a highly debated subject in neuroscience, and many different theories have been proposed to explain it. In this area of research, Integrated Information Theory (IIT) has emerged as a leading framework. IIT argues that a system, such as the brain, generates “consciousness” through the integration of its elements, which can be computed using the causes and effects associated with the system’s interactions. The principal metric of this framework is integrated information, or Phi, which measures the extent of a system’s integration. Although promising developments have been made so far, studies applying IIT to empirical neuroimaging data are limited. In this work, we provide a seminal application of this framework to resting-state functional magnetic resonance imaging data (fMRI). fMRI is a technique that allows for analysis of brain activity patterns over space and time. “Resting-state” (also known as “task-negative”) acquisition describes measurements of spontaneous brain activity patterns when subjects are not involved in a particular task. Studies employing this technique have identified a series of resting-state networks (RSNs), which are collections of correlated regions associated with the brain’s functions at rest. We apply our analysis to several RSNs acquired from subjects who underwent sedation with propofol, a short acting anesthetic. To test whether Phi is a valid marker of consciousness, we thoroughly analyzed how it varies throughout the different states of awareness induced by the sedative. In our discussions, we relate our results to existing literature on these networks and how they are affected by propofol. Most notably, we found that Phi corresponds to the conscious state of subjects in higher-order networks of the brain that are responsible for awareness and perception. Altogether, our implementation presents a promising procedure for computing integrated information from fMRI data. Ultimately, our goal is to provide a foundation for future neuroimaging studies that apply IIT to better understand neurological disorders and other states of altered consciousness.

Available for download on Thursday, August 31, 2023

Share

COinS