Inattentional Blindness

What this demonstrates

In 1999, Daniel Simons and Christopher Chabris showed 228 observers a short video of people passing a basketball. Observers were asked to count passes — either by the white-shirted team or the black-shirted team. About half failed to notice a person in a full gorilla costume walk into the scene, beat their chest, and walk out. The gorilla was on screen for roughly nine seconds.

The effect is not a failure of visual processing. Neuroimaging evidence suggests the unexpected stimulus reaches low-level processing even when unreported. What fails is the connection from processing to reportable awareness — the gorilla doesn't assemble a trace strong enough to survive until the question is asked. Jeremy Wolfe's inattentional amnesia hypothesis holds that there may be a brief flicker of experience that dissolves before it can be accessed at T+25s. The alternative account: the gorilla was never experienced at all, only processed below the threshold of awareness.

The behavioral signatures of these two accounts are identical. You cannot distinguish them from outside, and you cannot distinguish them from inside: asking "did you see it?" at T+25s gives you an attentional act at T+25s, not access to experience at T+12s. The question of what was "in experience" during the trial may not be answerable by any available method.

What it can't show: the simulation uses bouncing geometric shapes, not a real scene with human motion — the original effect relies partly on the gorilla's salience relative to a rich social scene. This version cannot screen for prior knowledge of the 1999 study: if you know about the gorilla experiment, you may have watched for unexpected elements rather than counting, which reduces the blindness effect and typically degrades counting accuracy. The simulation also cannot resolve the core debate it demonstrates: whether inattentional blindness involves no experience of the missed stimulus, or a thin flicker of experience that leaves no traceable residue. Both accounts look identical from the outside and, per the evidence, from the inside too.

Related: entry-502 · Not Seen