H567/02 Section A - Simons and Chabris (1999): visual inattention

Visible but unnoticed

Conscious visual experience depends on attention, not simply on an image reaching the eyes. Earlier selective-looking studies asked observers to follow one basketball game while another event occurred. Unexpected objects were often missed, but superimposed transparent displays raised a question: did the unusual video format cause the effect?

This differs from change blindness, failure to detect a change between views. The new study tested a sustained event moving through a continuous scene. It asked whether the effect survived an opaque display filmed as one scene, whether a more difficult counting task increased misses, and whether similarity to the attended team improved noticing.

The study belongs to the cognitive area because it infers selective processing from performance and reports. The attention theme concerns limits and priorities in awareness. Noticing depends on what the observer is trying to do; missing an event is not equivalent to poor eyesight.

Conditions and observers

Four factors produced sixteen independent conditions in a controlled laboratory experiment. Each observer saw only one combination, preserving the unexpected event's surprise.

Four 75-second videos showed two teams of three players passing basketballs. After 44–48 seconds, the unexpected actor crossed the scene for about five seconds. Transparent versions combined separate recordings of teams and actor. Opaque versions filmed all seven actors simultaneously, so they could occlude one another naturally. The main gorilla event was a walk across the scene, not the chest-thumping sequence used in a separate follow-up.

There were 228 participants, almost all undergraduates. Participation was voluntary, with some receiving a candy bar or payment as part of a larger testing session. Thirty-six records were excluded: 14 knew about the phenomenon, 9 lost count, 7 had incompletely or inaccurately recorded counts, 5 gave unclear answers and 1 had an extreme count. The analysed sample was 192, twelve per condition: $228-36=192$ and $16\times12=192$.

The principal dependent variable was reporting the unexpected event, coded as noticed/not noticed. Pass counts also checked performance of the attended task. Excluding people who knew the demonstration helps maintain surprise, but narrows the result to suitable naive observers.

Counting and questioning

Observers were tested individually and gave informed consent. They were instructed to attend to one shirt-colour team and silently count its passes, either as one total or as two separate totals. After viewing, they wrote their counts on paper.

The researcher then used increasingly specific surprise questions: first whether anything unusual was noticed, then whether anything besides the six players appeared, and finally whether the particular unexpected actor appeared. A “yes” led to a request for details; once the event was identified, further questions were skipped. Prior knowledge was checked afterwards. Debriefing included replaying the video on request.

Twenty-one experimenters used a written protocol specifying instructions, timing, data recording and debriefing. This promotes procedural reliability across researchers. Testing sessions lasted about 5–10 minutes, but television monitors varied from 13 to 36 inches, creating some variation in stimulus size.

The surprise questioning measured reportable detection. Specific final prompts might encourage guessing, but only one observer who did not report the event after the first question reported it later. That supports consistency across prompts rather than a large prompting effect.