Perspectives on the estimation of effective connectivity from neuroimaging data

In the late 19th century, the early investigations of brain function were dominated by the concept of functional segregation. This focus was driven largely by the data available to scientists of that era. Patients with circumscribed lesions were found who were impaired in one particular ability while other abilities remained largely intact. In the first part of the 20th century, the idea of functional segregation fell into disrepute and the doctrine of “mass action” held sway, proposing that higher abilities depended on the function of the brain “as a whole.” Although seen as opposing concepts, functional integration and functional segregation are not mutually exclusive, but exist only in relation to each other. For the last decade, functional neuroimaging techniques have been used to infer functional and effective connectivity in the human brain. However, the advent of event-related fMRI experiments have apparently complicated analyses of effective connectivity because variance components induced by different events cannot be easily separated at the hemodynamic level. Only recently, a new technique, dynamic causal modeling (DCM) was introduced, which provides a general framework for the analysis of effective connectivity and also allows the estimation of effective connectivity analyses in rapid event-related designs.