> wasn't it the case that if the overall trial length was a multiple of the TR, the
> bold response was always sampled in the same points, leading to a poor
> estimation of the HRF?
It is the jittering of the inter-stimulus intervals that would help you sample different phases of the hemodynamic response curve. In other words, compared to a fixed inter-stimulus interval, your regressors would be less correlated with each other, and you'd gain higher statistical power (more successfully detect what you are looking for). The range of jitter values depends on overall consideration of multiple factor such as total scanning time, number of trials per condition, etc.
Gang
