And that's the difficult part, to catch every confound you can think of, or at least minimize them. For sensory leaks if you catch anything it's enough for you to go off of previous experience and make connections, with better accuracy than pure guessing or going off 'nlp' (or whatever sense of interest, could be something 'common' like prioprioception v hit detection in sports) alone since its more data to process, but not too much to overwhelm, enough to the subjects to 'cheat' having more sensory tools at their disposal and generate noise. Things in the world tend to have certain sensory attachments to them that are obvious in some ways depending the sense in question and symbolic/subconscious in others.

Parapsych seems at this point a temporary catch for these things before a particular focus gets shuffled into one of the older areas of study, I find it easier to catch it on that end rather than at the beginning, especially as removed from all things remotely academic as I am.