Assume that the areas in a single-subject's brain that are involved in solving a task are supposed to be visualized with fMRI. The dynamics of the task solving process shall not matter (which areas are active first, which second?).
How, i.e. by which policy, are the data collected to give the final image?
I believe to have understood that data are collected per slices (of some millimeters thickness) and later postprocessed to give the final image.
For the sake of definiteness, let $T$ be the overall time it takes to complete the task (e.g. 10 seconds). Let $t$ be the time it takes to simultaneously take the BOLD data of one slice (e.g. 100 milliseconds). Let $N$ be the number of slices to be recorded (e.g. 20). Let $n$ be the minimal number of recordings of one slice that is necessary for statistical analysis ($n=?$). So there are $N n$ recordings of slices necessary which takes $N n t$ milliseconds, if only one slice could be recorded at once. If $T$ is too short, the experiment would have to be repeated $N n t/T$ times to get all of the required $N n$ slice recordings.
If $m$ slices could be recorded simultaneously, a single run of the experiment would suffice, when $m > N n t/T$.
In any case: In which order are the slices to be recorded chosen (in case they cannot be recorded all at once)?
always from top to bottom (along the z-axis) in time $T$?
repeatedly from top to bottom in some time $T'=T/k < T$?
alternating between top-bottom and bottom-up in repeated experiments?
in a random order?
in an order informed by knowledge of the stages of the task (when different slices are primarly active at different times during the task)
Or is my picture of the recording process all too naive? Or is the question somehow meaningless (for reasons I do not see, but would like to learn about)?