Date of Award

Spring 1-1-2016

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Psychology

First Advisor

Marie T. Banich

Second Advisor

Tor D. Wager

Third Advisor

Rafael M. Frongillo

Fourth Advisor

Ronald M. Carter

Fifth Advisor

Scott I. Vrieze

Abstract

The field of human brain mapping has made immense progress in recent years by making tens of thousand associations between the brain and psychological states using functional magnetic resonance imaging (fMRI). However, there is a growing appreciation of the limited ability to determine the specificity between brain-cognition mappings in individual studies. Without surveying a diverse range of psychological states, it is difficult to know if a brain region is preferentially recruited by a given state, or a more domain-general process that underlies it. In a related issue, several recent efforts have attempted to find the fundamental computational units of the brain by using statistical learning techniques to identify discrete regions on the basis of properties that constrain information processing, such as connectivity. However, it’s not clear how well these brain atlases describe the high-level functional organization of the brain.

In this dissertation, I apply relatively unbiased data-driven methods to a database of nearly 12,000 fMRI studies to comprehensively map psychological states to discrete regions in human frontal cortex– a complex, high-level association area of the brain. On the basis of activation patterns across studies, I identify functionally distinct whole-brain networks composed of spatially contiguous subregions. While each network exhibits distinct functional associations, subregions within each network show much more similar, yet dissociable profiles. In contrast with strong localizationist accounts, we find distributed associations between psychological states and brain anatomy, suggesting moderate functional selectivity in many parts of frontal cortex.

In the last section, I quantitatively assess various approaches for clustering the brain into discrete regions by comparing novel meta-analytic atlases to existing brain atlases from other brain modalities. Across a variety of metrics, I find evidence that meta-analytic atlases are robust and may provide a better account of the task-dependent organization of the brain than atlases from other brain modalities. I conclude by discussing future approaches for using large-scale meta-analysis to better understand how the brain gives rise to psychological function.

Share

COinS