Motion draws attention to where change happened
Aliases: change transient · local motion signal · orienting to change
What it is
A state can change on screen and still be missed, unless the change itself produces a local motion transient. A brief shift in luminance or position pulls attention to that patch, without the viewer already knowing where to look.
The job is locating the change, not explaining where an object came from and not filling a wait. Without that local signal, a large refresh can still register as nothing having happened.
Why it happens
The visual system does not keep a full previous frame. Change is usually caught because the change emits a local motion signal: it travels quickly in the magnocellular pathway and the periphery is sensitive to it, so conscious search is not required first.
If old and new states cut, the transient is drowned by a whole-display rewrite and only memory comparison remains — memory that holds a handful of attended objects. Motion remakes “something is changing here” as a salient local event: a path of travel, a fade confined to the target, a brief highlight on the target. Moving the entire screen at once behaves like a blank flash; the local event no longer stands out.
Studying it
The usual contrast is “local transient present” versus “transient drowned by an interruption”:
- Flicker paradigm: a blank between two frames; the measure is cycles to detection. The blank erases the local transient.
- Transition contrast: the same change with displacement or a fade as the treatment, a hard cut as control.
- Centre-of-interest manipulation: change on the currently fixated object versus in the periphery.
Independent variables: presence of a transient, spatial extent (local versus full display), whether the change sits at the centre of interest. Dependent variables: detection rate, time to detection, time for the first fixation to land on the changed region.
In interface work this tests whether notifications, list insertions and filter refreshes are actually seen. Lab observers know a change exists and hunt for it; field users do not, so detection is lower.
Where it stops holding
- If the user is already looking at the control, the transient is redundant; a cut is visible.
- Several flashes at once cancel orienting: competing transients do not guarantee the task-relevant one wins.
- Large optic flow can cause discomfort; using a full-page scale as a “reminder” turns orienting into a vestibular conflict.
- This is about seeing where the change is, not understanding what it means.
Applying it
- Give a local transition only to the task-relevant change; do not move the whole page.
- For in-place changes (row insert, filter refresh, badge count), use a short displacement, opacity, or a brief marker so the transient sits on the object.
- Check: record the same flow with and without a transition, then immediately ask what changed and where. If the no-transition group cannot name the location, the transient was never built.
Related
- Same group: F7.01.2 Motion shows where an element came from and belongs · F7.01.3 Motion fills a wait and changes how long it feels · F7.01.4 Motion that does none of those jobs is pure cost
- Nearby: A5.04 Change blindness · A1.14 Motion perception · A5.07 Attentional capture
- Search terms:
motion transient·change blindness·attentional capture