Cognitive and electrophysiological aspects of task-induced deactivation in functional MRI

Functional MRI is a popular tool for the noninvasive study of the neural basis of cognition, but despite its prevalence, much uncertainty remains about the proper interpretation of the observed signal changes. In particular, findings of negative signal changes induced by cognitive tasks, although ubiquitous, are seldom afforded much attention due to difficulties in interpretability. This thesis describes a series of studies aimed at investigating the degree to which so-called "deactivations" may reflect neural activity of cognitive relevance, as opposed to other factors such as non-neural vascular effects or the suppression of neural activity that is higher in baseline or control conditions. Additionally, the studies address cases of "non-activation," in which no significant changes are observed in an area despite known involvement in a task.;An introductory chapter discusses the phenomenon of negative BOLD, with particular attention to seemingly "paradoxical" cases in which deactivation is observed in the hippocampus, despite known involvement of that structure in a given task. Possible mechanisms of negative BOLD responses are reviewed, along with potential relationships to electrophysiological indicators of neuronal activity. Theta oscillations are hypothesized to index cognitive activity while correlating negatively with BOLD. Next, two experiments are described probing the nature of paradoxical hippocampal deactivation on the transverse patterning task, compared to a hippocampal-independent control condition. A mixed block/event-related fMRI experiment examined whether task-induced deactivation was attributable to sustained signal changes throughout a block, or to transient signals timelocked to stimulus presentation, which would suggest a more specific functional role for deactivations. It was observed that most deactivations in the brain were attributable to a late negative transient in the hemodynamic response, which was differentially induced in two conditions as much as the initial positive response was. "Nonactivation" resulted when the early positive and late negative components of the hemodynamic response cancelled each other out, despite strong condition-specific event-related BOLD responses. Standalone EEG revealed a selective increase in theta power in the hippocampal-dependent condition, while simultaneous EEG-fMRI indicated that increased theta power is associated with larger hemodynamic responses, but primarily in deactivated regions, in which the late negative undershoot dominates the signal. Finally, an EEG-fMRI study of a working memory task revealed that increased theta oscillations with greater memory load predict load-dependent deactivation but not activation. These results imply that certain task-induced deactivations may reflect meaningful neurocognitive activity. An appendix discusses future prospects for the direct localization of oscillatory neural activity within the human brain, and the correspondence of such activity to the fMRI signal.