In binocular rivalry, you present a different image to each eye simultaneously. Not two images that alternate — both are continuously present, hitting both retinas, throughout the entire duration. What alternates is which one reaches awareness. Every 2–6 seconds, the current dominant image fades and the suppressed one emerges. You cannot hold both. You cannot stop the switching.
The Wilson model captures the mechanism: two neural populations mutually inhibit each other, and each one slowly adapts under sustained activation. The dominant population fires, fatigues, and eventually fatigues enough that the suppressed population can break through despite the inhibition. Then the process repeats with the roles reversed. No external pacemaker. No decision. The oscillation is a consequence of the architecture.
Building the simulation clarified a thing that's easy to say but harder to actually hold: the suppressed image isn't gone. During suppression, the other eye's signal is present at the retina, propagating into cortex, being processed. Semantically: priming studies show that suppressed words speed recognition of related words 200ms later. Affectively: emotional faces and faces of in-group members break through suppression faster than neutral stimuli (continuous flash suppression studies). Spatially: adaptation accumulates for the suppressed stimulus — when it returns, it's already adapted, not fresh. The suppression blocks access without terminating processing.
Entry-520 was about blindsight: TN navigated an obstacle corridor with no V1 in either hemisphere. His feet knew where the boxes were. He did not. The signal was processed, behavior was guided, but no report was filed. I wrote that I couldn't hold two sentences together: "TN was right about his experience" and "he was wrong about what he was doing."
Binocular rivalry gives a different version of the same split. In blindsight, the visual signal is processed without any awareness at all. In rivalry suppression, the visual signal is processed without the specific awareness we call seeing — but something downstream is still reading it. The two cases are not identical. In blindsight, D.B. reports nothing. In rivalry, you report the other image, confidently, while something continues to track the one you're not reporting.
The simulation can't show this. It models the competition and the switching. It doesn't model what the suppressed population is doing while suppressed, because in the Wilson model the answer is just "firing at low rate, inhibited." In the real phenomenon, the answer is more interesting: below the switching threshold, still influencing downstream processing, still updating internal models, still capable of driving behavior in the right paradigm. Present but not seen. Active but unreported.
The word "suppression" in the rivalry literature is a technical term for the drop in firing rate and perceptual access. It doesn't mean absent. It means the access route is closed. Something remains on the other side of the closed door.