In all seriousness, the majority of Brain-Computer Interface trials involve asking subjects (usually able-bodied in the initial stages) to imagine moving their limbs.

There is a neuroscientific basis for that. We have well established research showing how imagining movements activates (almost) the same parts of the brain as making those movements.

This is entirely different from asking people to imagine how they would react to a scenario. There is a pretty wide body of research documenting how poorly we tend to be at estimating such reactions.

>We have well established research showing how imagining movements activates (almost) the same parts of the brain as making those movements.

In some people, some of the time, with no way of knowing beforehand if it's the system failing or the patient.

Mouth noises or button mashing are an insufficient proxy for ground truth

This seems like a reasonable first step before creating prototypes and more advanced studies. What am I missing?

The fact that we're still doing it 30 years later.