Abstract
Bipolar disorder is a serious chronic condition, and there is great interest in understanding brain-based
mechanisms that contribute to disorder symptoms. In this proposal, we focus on one promising candidate:
efficiency of evidence accumulation (EEA). EEA is measured in specialized models from computational
psychiatry, and it quantifies a basic neurocognitive ability to accumulate information from a stimulus in noisy
conditions in order to select appropriate responses. Substantial reductions in EEA are found in bipolar
disorder, as well as other major psychiatric disorders, and they contribute to impulsivity and disease severity.
There is a critical gap in knowledge, however: At the current time, we know little about the brain mechanism
that produce reduced EEA in bipolar disorder, or in any other psychiatric disorder.
In this project, we address this gap using the methods of network neuroscience. Substantial evidence from
large datasets strongly supports a flexible network reconfiguration model of EEA. This model says EEA
depends on the brain’s ability to adaptively reconfigure connectivity patterns of brain networks across cognitive
demands and task contexts. The model suggests the novel hypothesis that reduced EEA in bipolar disorder
arises from deficits in flexible network reconfiguration. We test this hypothesis with U. of Michigan’s unique
Prechter Longitudinal Study of Bipolar Disorder (headed by Co-I McInnis). We study 130 healthy adults and
130 adults with bipolar disorder, who complete a battery of behavioral tasks to measure EEA and a battery of
neuroimaging tasks optimized to measure flexibility of brain network reconfiguration.
A centerpiece of our approach is the use of brain basis set (BBS), a multivariate predictive modeling
framework. This method lets us “summarize” tens of thousands of changes in connectivity patterns across the
brain in terms of a modest number of basic reconfiguration components. BBS lets us identify what networks
reconfigure as well as how much they reconfigure. Using BBS, we will quantify brain network reconfiguration
deficits in bipolar disorder. We in addition link deficits in EEA and reduced brain network reconfiguration
specifically to an impulsive/affectively-unstable subtype of bipolar disorder and to impulsivity factor scores.
Finally, we elucidate the etiology of deficits in task-evoked brain network reconfiguration. We use multivariate
methods to delineate how reduced task-evoked network flexibility arises from alterations in the brain’s task-free
functional and structural architecture.
EEA is a computational metric that rigorously quantifies core neurocognitive deficits in bipolar disorder. This
project leverages computational psychiatry, network neuroscience, and multi-modal imaging to delineate brain
network mechanisms that underpin EEA. Success here lays the foundation for a broader network neuroscience
research program examining impairments in reconfiguration/flexibility of brain networks across multiple
disorders, with the aim of pinpointing etiology and identifying potential interventions.