Figure 1: Scheme for different binocular rivalry paradigms.

A) Scheme of typical binocular rivalry setting. When an observer views dissimilar stimuli, e.g., achromatic sinusoidal gratings tilted −45° and +45° viewed dichoptically, perceptual competition arises in which experiences gradually change across time, known as binocular rivalry. The traditional task for the observer is then to continuously report what is seen via key presses assigned to categories by an experimenter, here a 3-Alternative-Forced-Choice task. B) Schematic overview of the InFoRM: Rivalry paradigm. During Indicate-Me (Phase 1), participants explore the stimulus-space, moving a joystick to modify binocular-non-rivaling stimuli in real-time that generate corresponding changes of the physical image. The participants were then asked to move the joystick to highlight images that they consider representative of six canonical rivalry states (‘exclusive left-tilted’, ‘exclusive right-tilted’, ‘piecemeal’, ‘equal superimposition’, ‘superimposition with left-tilted predominance’, and ‘superimposition with right-tilted predominance’), that have been reported in previous rivalry literature. During Follow-Me (Phase 2), participants moved the joystick to match perceptual reports for physically changing binocular-non-rivaling-stimuli to confirm their understanding of the relationship between the joystick position and stimulus appearance. Participants followed four trials that reproduced the rivalry experiences of author JS and four trials that reproduced the six rivalry states the participant had generated themselves during phase 1 - Indicate-Me. This trained participants to track their changing experiences during perceptual rivalry while also capturing the participant’s joystick position for each of six canonical perceptual categories. These data used to build estimates of introspection for each category, indicated with different colors in the classification figure. During Rival-Me (Phase 3), participants reported their perception with the same instruction as for Phase 2. The resulting data were then analyzed with various techniques, including the illustrated Hidden Markov Models. During Replay-Me (Phase 4), participants’ responses during the Phase 3-Rival-Me dichoptic-trials were used to generate physically changing binocular stimuli, that the participant again tracked which validated their individual perceptual-state-space. These data from Phase 3 and Phase 4 were then analyzed for similarity illustrated by the plot for one representative participant.