LINKING TASK-MODULATED FUNCTIONAL CONNECTIVITY TO INDIVIDUAL DIFFERENCES IN BEHAVIOR: THE CASE OF FACE PROCESSING
Open Access
- Author:
- Elbich, Daniel Benjamin
- Graduate Program:
- Psychology
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- April 01, 2016
- Committee Members:
- Kathyrn Suzanne Scherf, Thesis Advisor/Co-Advisor
Ping Li, Committee Member
Koraly Elisa Perez-Edgar, Committee Member
Peter Cm Molenaar, Committee Member - Keywords:
- fMRI
Connectivity
Face Processing
Individual Differences
GIMME
Graph Theory - Abstract:
- The study of the brain is currently undergoing a critical transition from the study of individual regions to the study of connectivity between regions. In this newly emerging focus on how the brain functions as a network, a network is represented by a series of regions (nodes) and the connections between them (edges). One domain in which the network level approach has made great strides is in face-processing, but there still many important questions remaining with regards to how individual differences in connectivity relate to differences in face-processing behaviors. In this project, I used state-of-the art connectivity methods to address the following questions: 1) is variation in face recognition behavior related to variation in the topologic organization of the face-processing network, 2) how do the relation between network organization and behavior change when viewing faces versus other visual objects? I behavioral tested and scanned 40 adult participants, measuring face and object recognition ability and brain responses to faces. I conducted effective connectivity analyses to determine the network topology of the full face-processing network (core and extended) during face, object, and place viewing at 3 levels of analysis. Finally these topologies were quantified using Graph Theoretical Metrics and Pattern Recognition analyses. The results show that as face recognition ability increases, the PCC and bilateral Amygdala become less hub-like within the network. Additionally, the vmPFC becomes an inter-modular hub as face recognition increases. Finally, superior face recognizers exhibit more unique network topologies during face viewing compared to object viewing, while average and low recognizers do not. These results show that not only do high performers have fewer and more direct connections, resulting in more compact network topologies, but also that these topologies are specific to viewing faces and not other visual categories. Further, high performers networks when viewing faces are organized differently than when viewing objects, but such is not the case for average or low recognizers. This research is the first of its kind to merge effective connectivity, graph theory, and pattern recognition to study individual differences in network topology.