Change blindness
We miss large changes to a scene when they happen out of focus.
What it means
Change blindness is the striking failure to detect substantial changes to a visual scene when the change coincides with some disruption — a blink, an eye movement, a film cut, or a brief blank flashed between two images. The mechanism reveals that we do not hold a complete, detailed internal copy of the visual world: perception retains far less than introspection suggests, and detecting a change normally requires attention focused on the very feature that changes, so without a localized transient to draw the eye, even large alterations go unnoticed. It is a close relative of inattentional blindness, the difference being that change blindness concerns failing to notice that something altered, whereas inattentional blindness concerns failing to see an unexpected object at all. Real-world demonstrations are dramatic, such as people in conversation failing to notice that the stranger giving them directions was swapped for a different person mid-exchange. A nuance is that people are markedly overconfident about their ability to spot such changes, a 'change blindness blindness.' It matters for eyewitness reliability, interface and warning design, driving safety, and a sober understanding of how partial our moment-to-moment awareness really is.
Why the world still looks stable
Vision samples the world sparsely. Each fixation extracts the gist and a few attended details; the rest is left in the world, to be re-sampled when needed. This normally works because a real change produces a motion transient, a sudden local signal that pulls the eye automatically to its source. A blink, an eye movement, a film cut or a flashed blank swamps that signal, so the transient no longer singles out the changed spot. Detection then falls back on effortful comparison: you must have attended the item before the change, held it in memory, and be attending the same place afterward. Miss any link in that chain and the altered scene simply reads as the current, correct one.
What the flicker task and the door study show
Two paradigms anchor the finding. In the flicker task (Rensink, O'Regan and Clark, 1997), an image and an edited copy alternate with a brief blank between them; observers hunt for the change and, for large alterations away from the scene's centre of interest, take on average around eleven seconds and seventeen cycles to locate it, though it is obvious once found. The door study (Simons and Levin, 1998) moved this into the street: when a stranger asking directions was swapped for a different person behind a passing door, only about half of pedestrians noticed, and detection tracked social group, since participants who saw the experimenters as peers caught the switch far more often than those who did not.
The confidence gap
People are systematically wrong about their own susceptibility. Levin, Momen, Drivdahl and Simons (2000) called this change blindness blindness: told about the door and flicker scenarios, roughly 83 percent of respondents predicted they would spot changes that only about 11 percent of real observers had actually detected. The error is not vanity but a faulty model of how seeing works, the assumption that perception is a rich, stable recording in which a large change ought to jump out. Because the belief feels obvious from the inside, it resists correction, which is why the phenomenon reliably surprises even people who have just been told about it. The gap matters most wherever a person's confidence is taken as evidence of accuracy.
Where the cost is real
The lab curiosity has teeth in applied settings. In staged-crime studies, a change of perpetrator between the event and a later view raises mistaken identifications, and witnesses who missed the switch can name the wrong person with inflated confidence (Fitzgerald, Oriet and Price, 2016), a dangerous pairing when jurors read confidence as reliability. Driving offers a running example: a hazard that appears while your eyes are on a mirror or a phone can be missing from the scene you resume. Interface and cockpit design work the other way, using motion, colour flashes or sound precisely to manufacture the transient that a silent state change would otherwise fail to advertise, so the update captures attention instead of slipping past it.
What it does not prove
Change blindness is genuinely robust, but its more sweeping interpretations overreach. Failing to report a change shows that detection broke down, not that the scene left no trace: primed and forced-choice measures often reveal implicit registration of a change the observer cannot consciously name. Simons and Rensink (2005) warn against the leap from change blindness to the claim that we represent almost nothing outside the current focus of attention, the so-called grand illusion. The honest reading is narrower and more useful. Comparison across a disruption is fragile and attention-limited, so any system that relies on a person noticing a difference, from safety warnings to audit trails, should not assume they will.
Examples
In studies, people giving directions don't notice the stranger they're talking to was swapped for a different person.
Continuity errors survive in finished films because of it: the wine glass is half full before the cut and brimming after, and across the cut almost nobody sees it move.
A checkout page reloads and the delivery charge is now higher. No flicker in the corner of your eye points at it, so the number simply reads as what it always was.
A close-up magician swaps a red card for a black one during a sweeping flourish; the spectator's gaze stays on the card the whole time, but the flourish's motion swamps the transient the substitution would otherwise create, and they later insist the card never changed.
A guard watching a wall of monitors looks down to log an entry; on the return glance an unattended bag has been removed from the concourse, and the empty spot reads as normal.
First described in Simons & Levin (1998).
Key references
- Fitzgerald, R. J., Oriet, C., & Price, H. L. (2016). Change blindness and eyewitness identification: Effects on accuracy and confidence. Legal and Criminological Psychology, 21(1), 189-201. doi.org/10.1111/lcrp.12044
- Simons, D. J., & Rensink, R. A. (2005). Change blindness: Past, present, and future. Trends in Cognitive Sciences, 9(1), 16-20. doi.org/10.1016/j.tics.2004.11.006
- Levin, D. T., Momen, N., Drivdahl, S. B., & Simons, D. J. (2000). Change blindness blindness: The metacognitive error of overestimating change-detection ability. Visual Cognition, 7(1-3), 397-412. doi.org/10.1080/135062800394865
- Simons, D. J., & Levin, D. T. (1998). Failure to detect changes to people during a real-world interaction. Psychonomic Bulletin & Review, 5(4), 644-649. doi.org/10.3758/BF03208840
- Rensink, R. A., O'Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceive changes in scenes. Psychological Science, 8(5), 368-373. doi.org/10.1111/j.1467-9280.1997.tb00427.x