Event-related fMRI: comparison of conditions with varying BOLD overlap

Recently, event-related fMRI-experiments have been reported in which subsequent trials were separated by only 2 sec or less. Because the BOLD response needs 10 sec and longer to return to baseline, the event-related signal in these experiments has to be extracted from the overlapping responses elicited by successive trials. Usually it is assumed that this convolved signal is a summation of the overlapping BOLD responses. We tested this assumption by comparing event-related signals in conditions with little and with substantial BOLD overlap in two fMRI experiments of a task-switching paradigm. We analyzed the difference in the activational time course elicited by a critical task and a baseline task when trials of both tasks were separated by intertrial intervals of 15 sec or when the critical trials were embedded in a stream of baseline trials with ISI = 1 sec. The change of the BOLD signal elicited by the critical trials showed a high correspondence between both experiments in five out of six cortical ROI. Our data support the view that BOLD overlap leads to largely linear signal changes. In the present study, task-related increases in the BOLD response were detected equally well with substantial BOLD overlap as with mostly nonoverlapping BOLD responses.