Attention occupied elsewhere means even obvious objects go unseen

What it is

When attention is fully occupied by a task, an object that is clearly visible, unobstructed, and stays put the whole time can still go completely unseen. This is inattentional blindness.

The most famous demonstration is the "invisible gorilla" experiment: participants were asked to count basketball passes, while a person in a gorilla suit walked through the middle of the scene, stopped to beat its chest, and walked off. Afterward, about half the participants could not report having seen it at all.

It looks like change blindness but the trigger condition is entirely different: change blindness requires a visual interruption (a blink, saccade, flicker, or occlusion) to mask the motion signal a change produces — it happens when "something changed, but the change went undetected." Inattentional blindness needs no interruption at all — the object can sit there unchanged the whole time; attention simply was never allocated to it. It is "the thing was there the whole time, but attention never went there."

Why it happens

Visual processing of a scene splits into two layers. One is coarse, parallel, pre-attentive processing that can produce a vague impression that something is present, but not enough to become a reportable percept. The other is fine-grained processing that requires attentional resources — it is what enables recognition, naming, and later conscious recall. Only this second layer produces a percept that can enter awareness and be reported — and that layer's resources are limited. Once the current task (counting passes) has consumed them, other objects in the scene, even ones continuously falling on the retina and physically quite salient, cannot get the processing needed to reach awareness. The eyes are not failing to register light; that signal simply never gets promoted to the level where it could be reported.

Perceptual load theory specifies the moderating condition precisely: the higher the primary task's perceptual load (counting passes costs more than simply watching the scene), the less spare capacity remains, and the more likely irrelevant stimuli are to be missed. When the primary task's load is low, spare capacity spills over onto irrelevant stimuli, making them easier to notice.

Studying it

The classic paradigm is the unexpected object paradigm, with the invisible-gorilla experiment as its flagship example: participants focus on a primary task within a dynamic scene, an unexpected object unrelated to the task is inserted into the scene, and afterward they are asked whether they noticed it. A static variant, exemplified by Mack and Rock's cross-judgment task, has participants focus on judging the relative length of a cross's arms while a simple shape occasionally appears in the periphery, followed by the same post-hoc question.

Typical independent variables: the primary task's perceptual load, the unexpected object's salience, its location in the visual field (near fixation vs. peripheral), and its duration of presentation. Typical dependent variable: post-hoc report rate — and note that this kind of study can typically only test each participant once, because once a participant knows to look for an unexpected object the phenomenon disappears. This is the biggest methodological constraint in this line of research, and why replications rely on large samples with a single measurement per person rather than repeated trials.

In interface research, this method is mainly used to evaluate whether an element that is "important but unrelated to the current task" — a system notification popping up mid-task, for instance — can actually be seen.

Methodological caveat: the one-shot-per-participant constraint makes sample efficiency low. Lab tasks are also typically designed to be highly focused (a single counting task), while real tasks fluctuate in perceptual load across stages, so the miss probability is not a fixed number but shifts with the task's own demands.

Where it stops holding

• Effect strength scales with the primary task's perceptual load: low-load tasks are less likely to produce a noticeable inattentional blindness effect.
• The farther the unexpected object is from the primary task's fixation point, the more likely it is to be missed; objects near fixation are often noticed even when task-irrelevant.
• Once observers are told in advance that an unexpected object might appear, the effect shrinks dramatically or disappears — which is exactly why real-world critical alerts are dangerous precisely because users have no expectation that anything is coming.
• Like change blindness, this describes a failure of awareness, not a failure of the retina to register light; solutions should not conflate these two distinct failure causes.

Applying it

• Do not assume that users focused on a primary task (filling a form, checking out, navigating) will naturally notice other important information appearing at the same time, especially when it sits far from their current fixation point.
• Information that must be reliably seen (safety warnings, legal notices) should not be designed to "quietly appear off to the side" — it should actively interrupt the primary task, or at least appear near the current fixation point.
• To judge whether a prompt risks being swallowed by inattentional blindness, first check how high the primary task's perceptual load is — the higher it is (urgent operations, multi-step flows), the more the prompt needs to sit near the fixation point and carry an interrupting quality, not sit quietly off to the side.
• Verification: have users perform a focused task and see whether they spontaneously report a target prompt appearing at the same time, rather than asking "could you see it if you specifically looked for it" — the latter measures capability, not real-world performance.

Related

• Same group: A5.05.2 Being in the visual field is not the same as being seen · A5.05.3 Critical prompts cannot rely on merely appearing on screen
• Nearby: A5.04 Change blindness (requires a visual interruption to mask the change signal — a different trigger condition)
• Search terms: inattentional blindness · invisible gorilla · perceptual load theory · unexpected object paradigm