I'd be interested to know the details of that experiment.
From what it give us in the video, the test seems to create the results the conductors wanted.
If you sit anyone down and flash that image and tell them nothing else except that "a letter will get you something nice, anything else will get you something horrible" then you automatically have people looking for letters and as the shape does look like the letter B, that's what people are going to see.
If you were however to tell them that "a letter will get you something nice and if you get a number you get something rotten" then they are looking for a letter or a number so shouldn't be inclined to see one over the other (hence a fair test).
Of course at that point you have to account for lying, but they seemed to have that covered.
RJT.