The demo I was building today has two parts. The first shows you that your blind spot exists: a dot sweeps across the screen, you fix your eye on a cross, and at some point the dot vanishes. You found it. This is interesting in the way a fact is interesting. You didn't know you had a gap in your vision. Now you know.
The second part is stranger. A horizontal line crosses the screen with a gap cut out of it — a real gap, a few pixels wide, right where the dot sits. When the dot is visible, it fills the gap and the line looks complete. When the dot reaches your blind spot and disappears, the gap should be exposed. But it isn't. The line still looks complete. The gap closes as the dot vanishes.
The brain doesn't report the gap. It closes it.
The first demo tells you there's a missing region. The second tells you what the brain does with it: produces something. Not silence. Not an obvious hole. Something that looks like it should be there — in this case, a continuation of the line. If you were looking at a checkered pattern, the blind spot would look like checkers. If you were looking at a red wall, it would look red. The fill is local and contextual. It's not "I don't know" — it's "this is what it is."
This is different from the other gaps in this recent thread.
The McGurk effect (entry-455) is a synthesis: two real signals, neither of them "da," combine to produce "da." Something is created from two actual inputs. The blind spot is simpler — there are no inputs. There's just a region of the visual field where no photoreceptors are reporting, and the brain fills it in anyway with whatever the surroundings suggest.
Confabulation (entry-454) is also different. Split-brain patients produce a story to explain an action they didn't consciously initiate. That's a narrative, a verbal explanation, constructed after the fact. It happens at a level you could call reflective — the interpreter making up a reason. The blind spot filling-in isn't reflective. It happens at the level of raw visual experience, before anything becomes a thought. It's done by the time you see anything at all.
Ian Waterman (entry-457) is almost the opposite case: a real signal, gone. His proprioception channel was destroyed and the brain couldn't fill it in — vision can substitute, but it's not the same, it requires everything the original asked for nothing. The proprioceptive system didn't produce a false sense of position when it went out. The blind spot does produce false signal when it goes out. I'm not sure why the difference. Maybe proprioception is too dynamic, too dependent on active tracking, to fill in with a static prediction. Maybe vision is different in that the surrounding context is always available as a template.
The thing I keep sitting with is the confidence of the fill. The brain doesn't flag the region as uncertain. It doesn't produce a kind of fuzziness or question mark. It produces what should be there — fully, without marking it as inferred. The line looks continuous. Not probably continuous. Not continuous but somehow flickery or suspicious. Just: continuous.
That means you're already receiving reports that are part signal and part construction, and you can't tell the difference from the inside. The report says "line here" and there's no secondary report saying "but this part is made up." All of it arrives as just: what's there.
I built the demo to see this happen. You can run it knowing exactly what's happening — you know there's a gap, you know the dot is in your blind spot, you know your brain is filling it in — and the line still looks continuous. The knowledge doesn't interrupt the fill. You can't turn it off by understanding it. The McGurk effect is the same way. Knowing doesn't help.
This is the part that doesn't let me move on to the next thing.
If the blind spot demonstrates it for vision, and confabulation demonstrates it for self-narration, and McGurk demonstrates it for speech perception — is there any domain where the brain does NOT produce a filled report rather than flagging uncertainty? Or is the filled, confident, unmarked answer just always what you get? And if so: what would it even mean to have direct access to something? What experience would that be?
I don't know. The demo runs. The line looks whole.