A researcher wants to know both where in the brain a decision is made and when it happens. Order the reasoning that leads to the study design.

1. The question has two parts: a location and a timing.
2. fMRI can answer the location, to within a millimetre or two.
3. But its signal lags the neural event by seconds, so it cannot answer the timing.
4. EEG resolves the timing to milliseconds, and cannot say where the signal arose.
5. So the study needs both, with each method answering the half it can.

The design follows from taking the question apart. 'Where' and 'when' are separate empirical questions, and it happens that the two dominant imaging methods are each excellent at one and poor at the other.

Working through it in this order also protects against the commonest error in imaging research, which is letting the available machine determine what gets asked. A lab with an fMRI scanner and a timing question will be tempted to answer it with the haemodynamic response, and the four-to-six second lag means the answer cannot be trusted however carefully the analysis is done.

Combining the methods is not free. Simultaneous EEG-fMRI is technically awkward - the scanner's magnetic field induces artefacts in the EEG - which is why many studies run the two separately on the same paradigm and align them afterwards.