Motor Variability in Timing: Why No Two Taps Are the Same
A plain-language guide to the noise inside your timing — and how researchers split it into two separate sources.
Short answer: when you try to tap a steady beat, every interval comes out slightly different, because two separate sources of noise sit inside every response. One is the central timekeeper that decides when to move; the other is the motor delay between that decision and the finger actually landing. The Wing–Kristofferson model shows how the wobble you can measure splits into these two parts — and that split is why timing errors carry a very particular fingerprint: a tap that lands late tends to be followed by one that lands early.
The wobble is real, and it is not sloppiness
Even a trained drummer, asked to tap once per 500 ms in silence, produces intervals that scatter around the target — some 480 ms, some 520 ms, rarely exactly 500. This scatter is called motor timing variability, and the important point is that it is not a sign of carelessness or a lack of skill. It is a built-in property of biological timing. Nervous systems generate intervals with noise, and that noise never fully disappears; even the best performers reduce it rather than remove it. Understanding where the wobble comes from is the first step toward thinking clearly about both practice and measurement, because the two halves of the problem — deciding when to move and actually moving — behave quite differently.
The Wing–Kristofferson model: two sources, not one
In 1973, Alan Wing and A. B. Kristofferson proposed an idea that still anchors the field. When you tap without an external cue — self-paced, from your own memory of the beat — your performance runs through two stages. First, an internal timekeeper (a central “clock”) emits trigger signals at roughly the target interval. Second, each trigger has to travel through the motor system — nerves, muscles, the mechanics of the hand — before the tap actually happens, and that journey takes a slightly different amount of time on every repetition.
The measurable interval between two taps is therefore not the clock interval on its own. It is the clock interval, plus the motor delay on the second tap, minus the motor delay on the first. Because the same motor delay appears with a plus sign in one interval and a minus sign in the next, the two sources leave different traces in the data — which is what makes the model testable rather than just a tidy story.
The tell-tale sign: late tap, then early tap
The model’s best-known prediction is a negative correlation between neighbouring intervals (technically, a negative lag-one autocorrelation). If a single motor delay happens to be unusually long, it stretches the interval that ends with that tap and shortens the very next interval that begins from it. So a long interval tends to be followed by a short one, and the reverse — an alternating, see-saw pattern riding on top of any slower drift.
This is genuinely useful, because it lets researchers pull the two sources apart from a single stream of taps. Clock noise and motor noise contribute differently to the total variance and to that lag-one relationship, so with enough taps you can estimate how much of a person’s wobble is “clock” and how much is “implementation.” Wing and Kristofferson’s decomposition, and the many refinements since, turned motor variability into something you can quantify rather than merely feel.
Where the variance comes from
It helps to lay the sources side by side. The table below separates the two internal components the model describes from a third, thoroughly practical one that lives in any measuring device — including a rhythm game running in a browser.
| Source of variance | What it is | How it shows up | What may reduce it |
|---|---|---|---|
| Central clock (timekeeper) | The stage that decides when to move and generates the interval | Sets the baseline scatter; feeds slow drift in tempo | Steady reference practice, subdividing, internalising the beat |
| Motor implementation | The variable delay from trigger to the tap physically landing | Creates the see-saw (a long interval, then a short one) | Relaxed, efficient movement; less muscular tension |
| Device / input latency | Time the hardware and software add before registering the tap | Adds jitter that can masquerade as motor noise | Low-latency input, treating it as a known offset, averaging many taps |
Faster intervals wobble less — but not proportionally
Timing noise is not a fixed number of milliseconds. Broadly, the variability of an interval grows with its length: reproducing a 1000 ms beat scatters more, in absolute terms, than reproducing a 300 ms one. This scaling is often described as roughly Weber-like — the standard deviation rises with the interval, so the relative error stays steadier than the absolute error. The relationship is not perfectly proportional across every duration, and the clock and motor components appear to scale differently, but the general rule holds: shorter intervals are tighter in absolute milliseconds, longer intervals looser. This is closely tied to the same just-noticeable-difference logic that governs how we hear tempo, which we unpack in our piece on the tempo JND and Weber’s law.
What this means for practice and measurement
Two lessons follow, and both are cautious ones. First, some residual wobble is normal — chasing perfectly identical taps is chasing a target biology does not offer. A realistic aim is to narrow the scatter, not to erase it. Second, because the total variability blends a clock component and a motor component, practice may act on the two in different ways. Steady work with a reference beat — subdividing, tapping along, then holding the beat once the cue drops away — is generally thought to sharpen the internal timekeeper, while relaxed, economical movement may trim the motor contribution. We say “may” and “generally thought” on purpose: the evidence points this way, but individual results vary, and the model itself is more a description of the noise than a training recipe. For a fuller, hands-on take, see our guide to improving rhythm and timing.
If you want to measure someone’s timing, a single tap tells you almost nothing — the noise guarantees it will be off. What carries information is the distribution over many taps: its spread (how variable) and its pattern (that see-saw signature). This is exactly why timing tasks ask for repeated responses rather than one heroic attempt. It is also why honest measurement has to account for the device itself: input latency and its jitter add to whatever the body contributes, and if that added noise is large or uneven it can be mistaken for motor variability. Well-built timing tools try to keep device jitter small and, where they can, treat it as a known offset rather than pretending it isn’t there.
Put your own variability to the test
Curious how tight your own timing is? In Chronfork, two intervals are compared and you judge which is faster, so the game homes in on the smallest tempo difference you can reliably tell apart. Because it repeats the judgement many times, it reads your distribution — the spread the Wing–Kristofferson picture is all about — not a single lucky tap.
Frequently asked questions
Why are no two of my taps ever the same?
Because two independent sources of noise sit inside every response: the central timekeeper that decides when to move, and the variable motor delay before the tap actually lands. Both fluctuate slightly on every repetition, so the measured interval is never identical twice. This is a normal property of biological timing, not a flaw in your technique.
What is the Wing–Kristofferson model?
It is a 1973 account of self-paced timing that splits the variability of tapping into two parts — a central clock component and a motor-implementation component. Its key testable claim is that motor noise produces a negative correlation between neighbouring intervals, which lets researchers estimate the two components from a single stream of taps.
What is the difference between clock variance and motor variance?
Clock variance comes from the timekeeper that sets when a movement should happen; it feeds the baseline scatter and slow drift. Motor variance comes from the changeable delay between that command and the tap landing; it produces the alternating long-then-short pattern. The model separates them because they leave different fingerprints in the data.
Why does a late tap tend to be followed by an early one?
A single long motor delay stretches the interval that ends with that tap and, at the same time, shortens the next interval that starts from it. The one delay pushes two neighbouring intervals in opposite directions, which is why the timing shows a see-saw, negative correlation between successive taps.
Does more practice remove timing variability?
Not entirely. Some residual wobble appears to be built into biological timing, so a realistic goal is to narrow the scatter rather than eliminate it. Practice may reduce the clock and motor components in different ways, but individual results vary and the effect is best treated cautiously.
Do longer intervals carry more timing noise?
Generally yes, in absolute terms: the variability of a reproduced interval tends to grow with its length, in a roughly Weber-like way. Shorter intervals are usually tighter in absolute milliseconds; longer ones looser. The scaling is not perfectly proportional, and the clock and motor parts may scale differently.
Keep reading
- Sensorimotor synchronization: how the same timing machinery locks on to an external beat instead of running free.
- Your internal clock for tempo: the central timekeeper that sits at one end of the Wing–Kristofferson split.
- Tempo memory and recall: how well you can store and reproduce a beat once the cue is gone.
Sources: Alan M. Wing & A. B. Kristofferson, Response delays and the timing of discrete motor responses, Perception & Psychophysics (1973); Bruno H. Repp, Sensorimotor synchronization: A review of the tapping literature, Psychonomic Bulletin & Review (2005); Bruno H. Repp & Yi-Huang Su, Sensorimotor synchronization: A review of recent research (2006–2012), Psychonomic Bulletin & Review (2013); Alan M. Wing, Voluntary timing and brain function: an information processing approach, Brain and Cognition (2002).
Understand it, now train it. The daily games stay free forever. A Personal Pass adds progress tracking (trends + export); Teacher / School tools let you assign it and follow a whole class. For teachers & self-learners →