---
name: Perceptual duration
slug: perceptual-duration
category: motion
status: published
created: 2026-08-21T00:00:00.000Z
modified: 2026-08-21T00:00:00.000Z
definition: How long a spring appears to take, measured to the point the
  movement stops being visible rather than to its mathematical rest, the figure
  to quote against a timed animation.
aliases:
  - name: perceived duration
    source: community
  - name: settling time
    source: community
  - name: visual duration
    source: community
tags:
  - perceived-performance
  - perception
relations:
  contrastWith:
    - duration
    - frame-rate
  variantOf: []
  partOf: []
  seeAlso:
    - damping
implementations: []
sources:
  - title: Animation Vocabulary
    url: https://animations.dev/vocabulary
  - title: The Vocabulary of Motion
    url: https://motion-vocabulary.vercel.app/
demo: inline
exhibit: false
useWhen: a spring must be compared with a duration
---

A tween has a number written on it. A [spring](/spring-animation) does not: it is
described by mass, stiffness and damping, it approaches its target
asymptotically, and formally it never arrives. That is fine as physics and
useless as a specification, because the moment anyone asks how the spring
compares with the 400 millisecond tween beside it, a number is needed. The
perceptual duration is that number: the time from the start to the last movement a
person could actually see, with everything after it counted as arrived.

The measurement needs a threshold, and the threshold is a choice rather than a
constant. Common ones are the point where the remaining distance falls under one
percent of the travel, or under half a pixel on the screen it will run on, or
where velocity drops below some small figure and stays there. Different tools
answer differently, which is why two libraries can report different durations for
the same three parameters and both be right. State the threshold when the number
matters, treat it as approximate when it does not, and remember that a springier
setting can look slower than a stiffer one even when its measured time is shorter,
because the wobble at the end keeps drawing the eye.

The word perceptual is doing a second job too, and it is worth keeping the two
apart. Felt time is not clock time: a well eased move reads as quicker than a
linear one of identical length, an interruptible move reads as quicker than one
that has to finish, and motion that starts the instant a control is pressed reads
as quicker than motion of the same length that begins after a delay. That is craft
lore with attention research behind it rather than a law, and the honest anchors
are the response-time thresholds: under roughly 100 milliseconds a result feels
instantaneous, and past about a second a person's attention has started to wander
off the task.

Use the figure the way any other duration is used. Quote it in a review, compare
it against the timed transitions it will sit beside, and once a family of springs
has been chosen, write the settled numbers down where the rest of the system can
spend them, which is what a [motion token](/motion-token) is for. A spring that
measures 500 milliseconds and a tween of 400 belong to the same conversation. A
spring nobody has measured belongs to no conversation at all.