Causal Inference for fMRI Time Series Data with Systematic Errors of Measurement in a Balanced On/Off Study of Social Evaluative Threat

• Activity Number: 493
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Imaging
• Author(s): Michael Sobel*+ and Martin Lindquist
• Companies: Columbia University and Johns Hopkins Bloomberg School of Public Health
• Keywords: Balanced design ; Bold contrast ; Causal inference ; fMRI ; Longitudinal data ; Measurement error

Abstract

Functional magnetic resonance imaging (fMRI) has facilitated advances in understanding human brain function. Neuroscientists are interested in the effects of external stimuli on brain activity and causal relationships among brain regions. However, they have not stated what they mean by causation or defined the effects they purport to estimate. Using potential outcomes notation, we construct a framework for causal inference on blood oxygenation level dependent (BOLD) outcomes in fMRI time series data. In the usual causal inference literature,outcomes are assumed to be measured without error and treatment effects are estimated by comparing exchangeable subjects receiving different treatment regimens. In contrast, the BOLD outcomes are measured with systematic error, typically task related,and in the most widely used experimental design in fMRI research, all subjects are assigned to the same regimen. Thus, we define effects free of systematic error and develop alternative conditions for identifying these. Our results are of general interest to researchers working with outcomes measured with error and in fields where large amounts of data are collected on relatively few subjects.