This item is under embargo and not available online per the author's request. For access information, please visit


Date of Award

Spring 5-15-2019

Author's School

School of Engineering & Applied Science

Author's Department

Biomedical Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Traditionally neuroscience research has focused on characterizing the topography and patterns of brain activation evoked by specific cognitive or behavioral tasks to understand human brain functions. This activation-based paradigm treated underlying spontaneous brain activity, a.k.a. intrinsic activity, as noise hence irrelevant to cognitive or behavioral functions. This view, however, has been profoundly modified by the discovery that intrinsic activity is not random, but temporally correlated at rest in widely distributed spatiotemporal patterns, so called resting state networks (RSN). Studies of temporal correlation of spontaneous activity among brain regions, or functional connectivity (FC), have yielded important insights into the network organization of the human brain. However, the underlying fundamental relationship between intrinsic and task-evoked brain activity has remained unclear, becoming an increasingly important topic in neuroscience. An emerging view is that neural activity evoked by a task and the associated behavior is influenced and constrained by intrinsic activity. Additionally, intrinsic activity may be shaped in the course of development or adult life by neural activity evoked by a task through a Hebbian learning process. This thesis aims to reveal correspondences between intrinsic activity and task-evoked activity to better understand the nature and function of intrinsic brain activity. We measured in human visual cortex the blood oxygen level dependent (BOLD) signal with fMRI to analyze the multivoxel activity patterns and FC structures of intrinsic activity, and compare them to those evoked by natural and synthetic visual stimuli.

In chapter 1, we review previous evidence of an association between intrinsic and task-evoked activity across studies using different experimental methods. Two experimental strategies from the literature were adapted to our own experiments. First, from anesthetized animal studies of intrinsic activity in visual cortex, we set out to measure macro-scale multi-voxel patterns of spontaneous activity fluctuations as they relate to visually driven patterns of activity (Chapters 2 and 4). Second, from inter-subject correlation studies of visual activity driven by natural stimuli, we measure relationships between intrinsic and evoked activity, specifically in relation to their topographic similarity at the network level (Chapter 5).

In Chapter 2 to 4, we establish a multivariate-pattern analysis (MVPA) approach to evaluate patterns of intrinsic and task-evoked activity. The main idea is that patterns of activity induced by behaviorally relevant stimuli over long periods of time would be represented in spontaneous activity patterns within the same areas. To test the idea, in Chapter 2, we compare the overall degree of pattern similarities between resting-state activity patterns, frame-by-frame (framewise), and visual-stimulus evoked activity patterns for natural (face, body, scenes, man-made objects) and synthetic (phase and position scrambled) object images during low-level detection task. We found that the variability, not the mean, of pattern similarity was significantly higher for natural than synthetic stimuli in visual occipital regions that preferred particular stimulus categories. Chapter 3 extends the static categorical pattern similarity measure of Chapter 2 into a temporal correlation measure. We built pattern-based FC matrices for different stimulus categories (e.g. a face specific multivoxel pattern) in regions that preferred particular stimulus categories (e.g. FFA, STG), and showed that the occurrence of a specific categorical pattern generalizes across category specific regions. These pattern-based FCs resemble that of resting-state FC of the same regions supporting that resting state patterns are related to category-specific stimulus-evoked multivoxel activity patterns. In Chapter 4, we repeat the analysis used in Chapter 2 with language stimuli. Language stimuli (alphabetic letters and English words) are interesting as they are learned through intensive training as kids learn to read. Therefore, they represent a non-natural category of stimuli that is, however, highly trained in literate individuals.

The visual stimuli used in Chapter 2 to 4 are designed specifically for a laboratory environment that does not correspond to realistic ecological environments. In Chapter 5, to overcome this limitation, we use the more naturalistic visual experience of movie-watching and compare the whole-brain FC network structure of movie-watching and of resting-state. We show the whole-brain FC structure evoked by movie-watching is partly constrained by the resting network structure.

In conclusion, our experiments show that the link between intrinsic activity and task-evoked activity is not only limited to inter-regional interactions (as in regular resting-state FC), hence potentially reflecting anatomical connectivity or modulations of excitability between cortical regions, but extends to multivoxel patterns that carry information about specific stimulus categories. This result supports the notion that intrinsic activity constrains task-evoked, not only in terms of topography or activation levels, but also in terms of the information states that are represented in cortex.


English (en)


Maurizio Corbetta

Committee Members

Dennis Barbour, Brendan Juba, Eric C. Leuthardt, Steven Petersen,


Permanent URL:

Available for download on Thursday, April 15, 2021