A history of randomized task designs in fMRI

In the early days of fMRI, data were acquired using methods adapted mainly from PET imaging. Sets of similar stimuli were presented in extended blocks, with stimulus conditions changing from block to block. While this method provided optimum statistical power, it also presented a variety of potential confounds, including changes in attention, alertness, expectancy, and practice effects within and between blocks. Event-related paradigms using unpredictable, randomized stimulus sequences had been used in EEG studies for over 50 years before the development of fMRI, and provided a means to overcome these issues. However, the temporal dispersion of BOLD fMRI activity resulted in responses to successive stimuli adding together, making it difficult to perform rapid event-related paradigms using fMRI. Here we describe the background and history of methods developed to overcome this limitation, allowing rapid, randomized stimulus sequences to be used with fMRI. The advantages and disadvantages of this technique and how these methods have been applied in a variety of experimental settings are discussed.