I just tried the binocular rivalry test and I think the way it is described on this site isn’t sufficiently controlled. It asks one to try to prime your expectations by thinking of red prior to looking at the image. The way I was attempting this was to 1. hold my red/cyan filters in front of my eyes, 2. get the figure centred on the screen, 3. close my eyes and attempt to imagine red, then 4. open my eyes and report on which image showed. The difference between the primed and unprimed version of the test is whether step is left out. A problem with the approach I took is that I think closing the eyes is not a good idea because ambient light still filters through the filters and the thin tissue of the eyelid, which allows red light through. This is an external stimulus “leaking” into the experiment that I think is difficult to control for. I could imagine, but I didn’t try doing the following instead: only allow yourself, say 10 seconds for the visualization phase so that you only have your eyes closed for exactly 10 s. Closing the eyes is not sufficient; I think you need a very good method for blocking out all light with an opaque material of some kind, and use this for 10 s for both the primed and unprimed versions of the test.