Date of Award

Fall 2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Interdepartmental Neuroscience Program

First Advisor

Scheinost, Dustin

Abstract

Data-driven functional connectivity analyses in the chemical senses are relatively uncommon, with current literature largely predominated by Psychophysiological Interaction (PPI), Dynamic Causal Modeling (DCM), or Structural Equation Modeling (SEM) techniques. The aim of this dissertation is to expand data-driven analyses in the chemical senses (and using chemosensory stimuli) to promote the use of techniques not based on a priori hypotheses or derived from potentially irreproducible literature.Chapter One details a systematic review of functional connectivity analysis in the chemosenses, which demonstrates that, out of 103 studies examined, the vast majority use hypothesis-driven techniques. Moreover, these studies comprise topics of olfaction, gustation, chemesthesis, flavor, and multimodal "other" sense, with common subtopics including pleasantness, attention, intensity, and olfactory-visual integration, among others. In Chapter Two, we take our motivation a step further, and apply data- driven functional connectivity analysis to two datasets to examine neural correlates of taste intensity perception and anosmia, respectively. Both datasets were originally collected for hypothesis-driven analysis and reanalyzed using a data-driven approach, Connectome-based Predictive Modeling (CPM), to yield new insights. Regarding taste intensity, I demonstrate that connectivity among several canonical networks, including ventral-attention, somato-motor, default mode, and fronto-parietal, among others, can account for approximately 16% of the variance in sucrose and citric acid intensity ratings. Regarding anosmia, I yield additional support for previously conducted voxel-wise analyses, which suggests that a network consisting of ventromedial prefrontal cortex, frontal operculum, and posterior cingulate cortex can discriminate between those with anosmia or normosmia with a positive and negative predictive value of 64% and 63%, respectively. Chapter Three begins the discussion on the neuroimaging of obesity, first detailing a review conducted to explore the effects of adiposity and metabolic dysfunction on the brain. I observed evidence suggesting that both adiposity and metabolic dysfunction can affect cognition independently of one another, with memory, executive function, and medial temporal lobe structures particularly impacted. These effects may have both distinct and overlapping etiologies, perhaps resulting from oxidative stress, sleep apnea, central inflammation, decreased neurogenesis, and silent infarcts, among other causes. Chapter Four describes a human fMRI study in which an appetitive, complex chemosensory stimulus (i.e., milkshake) was delivered to subjects with overweight and obesity. I conducted a secondary data analysis to uncover neural predictors of adiposity and metabolic dysfunction, using this dataset to take advantage of Greene et al. (2018)’s finding that task-based paradigms, in comparison to resting-state, better amplify prediction of individual traits. I found largely separable functional connectivity networks predicting both adiposity (as indexed by waist circumference) and fasting insulin, suggesting dissociable pathophysiological phenotypes for obesity and diabetes. Chapter Five describes a data-driven analysis conducted on collated data collected in the laboratory over the course of several years. This analysis makes use of an easily acquired metric derived from the T2* MR signal, called eR2*, which has been experimentally related to both non-heme iron in the brain and presynaptic dopamine tone by association. I found a significant negative association between BMI and striatal eR2* across subjects in both our in-house dataset, as well as in subject data acquired from a freely available open dataset. Data analyzed included both resting-state data as well as task-based data acquired during the administration of chemosensory stimuli; though we note that the T2* signal is agnostic to data type. This dissertation demonstrates several key insights, namely that 1) there is a dearth of data-driven functional connectivity studies in the chemosenses, 2) data-driven analyses using shared and collated data yield novel insights into functional networks integral to taste intensity perception and post-traumatic anosmia, 3) metabolic dysfunction and adiposity can influence cognition independent of one another, 4) discrete functional networks can predict both adiposity (via waist circumference) and metabolic markers (via fasting insulin), suggesting separable pathophysiologies, and finally, that 5) an easily acquired MR metric indicative of non-heme iron and presynaptic dopamine tone is associated with BMI, suggesting aberrant dopaminergic function in overweight and obesity. Each empirical finding was completed using data-driven procedures and either collated or open data, or some combination. This thesis is therefore a motivation for those employing chemosensory neuroimaging procedures to adopt data-driven and “big data” approaches.

Download

Share

COinS