OK, time to spell something out for you again golfy.
If you were serious about coming up with a workable test protocol to prove or disprove your claim that people can hear your thoughts you wouldn't yet have tried your cat/ship test with your polygraph because you would have been too busy doing tests to establish how reliable it was in such a test.
You would have started using questions you already know the answer to - say whether it's daytime or nightime, or whether your volunteer is male or female - and watched the output of your polygraph to see whether it gave different readings when your volunteer told the truth or lied about what they'd written down. If, after many such tests, you were convinced it did, you would then have gone on to questions to which you did not know the answer and seen if you could use those different readings to determine that answer. For example you could have got your volunteers to take a card at random from a deck, write down on a piece of paper whether it was a red card or a black card, put both paper and card in an envelope and give it to you. You would then have asked them whether they had written down 'red' or 'black, having told them to say 'No' both times, used the readings to determine which it actually was, and then opened the envelope to see if you were correct. After a few dozen such tests you would have known whether your polygraph is reliable enough for it to be worth running your cat/ship test with it. If you decided it was then you would have to accept before you ran the first cat/ship test that a negative result would prove your claim wrong just as surely as a positive result would prove it right.
But you haven't done any of that, have you golfy? You dived straight in to your cat/ship tests and are trying to use them to determine whether or not your polygraph is reliable. In other words you are assuming as fact the very claim your test is supposed to be demonstrating in order to determine whether the tool you're hoping to use is reliable enough to make your protocol workable. If you cannot grasp why this is wrong then you have no hope of ever producing an acceptable protocol.