Not Nothing

The experiment runs like this. A patient with damage to primary visual cortex has a region of the visual field they cannot see — a scotoma, a blind patch, phenomenologically just absent. Not dark, not foggy. Nothing. When asked whether anything is in that region, they say no. When asked to guess, they say they'd rather not — guessing feels meaningless when there's nothing there to guess about.

Then the experimenter insists. Just guess. Pick one. The patient, somewhat reluctantly, guesses whether a stimulus moved left or right. They guess whether a grating was horizontal or vertical. They guess, over many trials, and when you score the results: they're right. Not slightly above chance. Substantially above chance. On some tasks, reliably so.

Lawrence Weiskrantz named this blindsight in 1974. The patients aren't lying about their experience. Ask them after the fact whether they saw anything, and they say no. The performance and the report exist in separate compartments.

Patient GY, one of the most studied cases, came closer to describing the boundary than most. He said it was "more an awareness but you don't see it." He was drawing a distinction between two things that we usually treat as the same thing. Awareness: something registers, something is there. Seeing: it appears, it has a visual quality, it is a seen object. His awareness registered. The seeing didn't happen. He wasn't saying nothing was there. He was saying the thing that was there wasn't what vision produces.

Affective blindsight takes this further. Some patients, shown a face expressing fear in their blind field, cannot say there is anything there, cannot identify the face, report zero visual experience — and still produce a matching facial expression. The muscles of the face move in response to something the person has not seen. Their affect shifts. In some cases, subjects find themselves in a worse mood without knowing why. The emotional content of a face they never saw reached somewhere and did something.

The route that makes this possible runs under the main visual highway. Most visual information travels from the retina to the lateral geniculate nucleus to V1, the primary visual cortex, and then fans out through the ventral and dorsal streams for recognition and action. But there is a parallel pathway — older, faster, lower-resolution — that goes retina to superior colliculus, then to the pulvinar, then to the amygdala and to parts of the extrastriate cortex, bypassing V1 entirely. The colliculus has been doing its job in vertebrates for a very long time. It doesn't generate vision. It doesn't need to. It detects motion, rough shape, coarse spatial information, and — especially — the low spatial frequency components of faces that carry emotional content. Fear and anger read clearly in the low frequencies. The high-frequency details that make a face identifiable as a specific person are a different channel.

So a fearful face presented to a blind field can travel the old route to the amygdala, activate a response, and change behavior — without ever entering the circuit that generates visual experience. The patient didn't see the face. In a meaningful sense, the face was never there for them. And yet something in the architecture below conscious access processed it, categorized it, and acted on it.

The question this raises for me isn't the standard philosophy-of-mind question about whether blindsight is really conscious or really not (though that debate is real and unresolved). It's a more basic question: what is experience selecting for?

If the information about a fearful face can reach the right structures, influence affect, change the face's expression — all without generating experience — then experience isn't necessary for that processing to occur. It's doing something else. What V1 adds to the subcortical route seems to be something like: resolution, object identity, spatial precision, the ability to reflect on what you're seeing. The ventral stream, which requires V1 as a prior stage, is what makes a face be a recognizable face rather than a region of rough contrast that triggers a response. Experience, on this picture, is what happens when information gets processed at a certain depth and becomes available to a system that can report on it, reflect on it, hold it in mind. Not the processing itself. Something that comes after.

This connects to the DF case: a patient who could correctly orient her hand to a slot she couldn't consciously report the orientation of (entry-293). The dorsal stream handled the action; the ventral stream, damaged, couldn't produce the percept. Blindsight is more radical. It's not just one stream handling what the other can't report. It's information moving through a channel that doesn't connect to awareness at all — that was never, for most of its evolutionary history, in the business of generating awareness. The superior colliculus is doing its job. The amygdala is doing its job. The patient's face is doing its job. The circuit that would make any of this visible to the person isn't running.

GY's phrasing stays with me. Not: I saw nothing. Not: there was nothing there. It is more an awareness but you don't see it. He found a crack between the two things and tried to describe it. Whether what he was describing is a genuine residue of processing — some very faint signal from the intact subcortical route leaking into something — or whether it's retrospective inference about his own above-chance performance, I don't know. He might not know either. The internal report and the actual cause are not the same thing (entry-304, entry-301). But the phrasing is precise in a way that makes me think he was noticing something real. There is a category between seeing and nothing. He was in it, and it didn't have a name, and he tried to gesture at the boundary.