In the 1999 Simons and Chabris study, participants watched a video of two teams passing basketballs and counted the passes. While they watched, a person in a gorilla suit walked into the frame, faced the camera, thumped their chest, and walked out. The sequence took about nine seconds. When asked afterward if they had noticed anything unusual, roughly half said no.
Not: failed to remember. Not: weren't sure. Said directly that they hadn't seen it.
This is usually presented as a finding about attention — that attention acts as a gate, and what doesn't pass through the gate goes unnoticed. But there's a prior question the finding doesn't resolve: was the gorilla in experience, and then filtered out? Or was it never in experience at all?
Those are two distinct claims, and they matter differently.
In the first picture, you have a rich visual field, and attention is a spotlight that selects from it. The gorilla was in the periphery of that field, was not selected by attention, and so wasn't remembered — but it was there in some background way, processed below the level of explicit notice. Call this attention-as-selection.
In the second picture, experience is not prior to attention. There is no background field from which attention picks. Experience is what attention constitutes, moment to moment, and everything outside that is nothing — not dimly present, not peripheral, not available-but-unused. The gorilla was not in experience. Call this attention-as-construction.
The difference is not just conceptual. It has a consequence: in the first picture, the subject had some experience of the gorilla, however thin. In the second, they had none. Their report — "I didn't see it" — is not a report of forgetting. It is an accurate report of absence.
One complication: there's a third interpretation called inattentional amnesia. Perhaps the gorilla was consciously experienced — really experienced, with the full weight of experience — but the trace was so thin, with no rehearsal from attention, that it degraded before the question arrived. The experience happened and then was gone. The subject's report reflects the absence of the memory, not the absence of the experience.
The amnesia hypothesis sounds like it resolves the problem — experience was there, we just lost it — but it generates a harder version of the problem. What does it mean for an experience to have occurred if no trace survives, not even the feeling of something having been there? The person who can't recall a dream sometimes feels the echo of it. These subjects don't report even that. They report cleanly: nothing happened.
Some implicit processing studies have pushed back on this. Subjects who deny noticing the gorilla can, under certain conditions, correctly report properties of the unexpected object — its location, its color — at above-chance rates. This suggests some processing occurred. But processing is not experience. Unconscious processing of a stimulus is not the same as having an experience of it, even a thin one. The presence of a neural signal is not the presence of a subject's experience.
There's a structural problem underneath all three interpretations.
When we ask "was the gorilla in experience during the video?", we are asking a question about time T from time T+n. The question arrives after the event. The only tool available to answer it is the subject's report, which is itself produced by attention — attention at T+n, directed at what happened at T. There is no neutral record. There is no playback. Introspection is not a camera that was running during the video; it's a process that starts after the video ends and reconstructs what it can.
The complication: that reconstruction is itself attentional. Attending to "what I was experiencing" at T is a new act of attention. If experience requires attention, then asking about prior experience doesn't give you prior experience — it gives you a new attentional act pointed at memory, which may or may not have anything to track.
This is the thing I can't get past. The question "was the gorilla in your experience?" cannot be answered from the inside, by the subject, because the inside is always the present moment of attending. There is no outside view. There is no archive of experience that exists independently of the current attentional act that would access it.
The philosopher Susan Blackmore called the rich visual world we seem to inhabit a "grand illusion." We feel surrounded by a detailed, stable, densely present visual scene. But change blindness and inattentional blindness suggest that what's actually available at any moment is sparse — only what's currently being attended is represented in any detail. The seeming-richness is not the product of a rich representation; it's a kind of guarantee extended by the knowing that we can look. The room feels detailed because we know we could turn and see the lamp. Not because the lamp is in experience right now.
O'Regan and Noë's sensorimotor account puts it differently: experience is not the presence of internal pictures. It's the mastery of contingencies — the knowing that if you move your eyes, the scene will respond predictably. The gorilla was available. Any subject could have turned attention toward it and seen it clearly. That availability is what makes the visual world feel rich. But availability is not presence. The lamp I could see is not the lamp I'm seeing.
The gorilla walked through a room of people watching carefully. Nine seconds. Thumped its chest. Half the room says: not there.
The finding has been replicated many times under many conditions, with different unexpected objects, different primary tasks, different levels of demand. The effect is robust. But the question it points toward — what was in experience during those nine seconds, for the half who didn't notice — remains genuinely open. Not because we haven't looked carefully enough, but because looking carefully is the thing that changes the answer.
The gorilla is a finding about attention. It's also, underneath that, a problem about what experience is made of and whether there's any way to tell from the inside.