| |
Abstract:
In many experimental paradigms, the constraints on timing
imposed by fMRI appreciably change the nature of the task. Spacing
trials close together in time, however, decreases the ability to
detect differences between conditions due to the slow hemodynamic
response. We present a genetic algorithm and simulations designed
to enable fMRI designs with very short inter-trial intervals
(around 1 s). The genetic algorithm finds an order of trial
presentation across experimental conditions that simultaneously
maximizes the signal in the design at detectable frequencies,
minimizes the colinearity between predictors in the design matrix,
and minimizes deviations from a perfectly counterbalanced design.
Simulation results are applied to an event-related fMRI study of
task switching. The imaging results show that trial spacing of less
than 1.5 seconds apart is viable for at least some cognitive
paradigms.
|
