The touch point does not coincide with the user's subjective aim point
Aliases: perceived input point · touch offset · subjective aim
What it is
The two sibling cards cover how the system reduces contact to a point and how that point moves with posture; this one is about the person's side of the gap. What a person subjectively believes they are pointing at — usually a visible icon or label centre — and the touch coordinate the system finally reports are two different reference frames that need not coincide. The literature calls what a user believes they tapped the perceived input point, and shows it deviates systematically from the geometric centroid the system computes from the contact patch.
Why it happens
Why does the perceived aim point drift away from the geometric centroid rather than lining up with it? Because aiming relies on visual feedback — typically whatever sliver of the target is still visible past the fingertip or above the fingernail, not "the geometric centre of the area the finger presses down on." The point vision last locked onto before contact, and the centroid computed from the contact patch after contact, are products of two different physical processes: the former is the visual target from the motor-planning stage, the latter is the mechanical result once the movement is executed. This is also why the offset correlates so strongly with the touch point's position relative to the fingernail — its underlying cause is a systematic gap between "where the eye guided the finger" and "where the finger's contact area actually landed," not undifferentiated motor noise. That is precisely why it can be modelled per user and per finger angle instead of being discarded as unpredictable error.
Studying it
Studying this requires separating three things: the target position a participant was asked to aim at, the raw contact region once the finger lands, and the coordinate the system reports. A typical design has participants aim at a series of targets at known positions, logging the reported point's offset from each target while a high-speed camera or auxiliary sensor captures finger posture, so the offset can be regressed against angle, nail position, and similar variables. The methodological point worth stressing: using a binary hit/miss outcome as the sole dependent variable hides exactly the fact that matters — that the offset is directional and predictable — and that predictability is the entire premise behind later offset-compensation techniques that correct the reported point using cues like nail position. Reporting hit rate alone erases this signal.
Where it stops holding
The offset is usually immaterial for large, low-risk targets — being off by a few millimetres does not stop you from hitting a button that covers most of a card. It matters most for small targets, dense layouts, and precise operations such as selecting an exact point on a map or placing the first stroke in a drawing app. Direction and magnitude are also strongly tied to grip: the offset pattern for one-handed thumb use differs from two-handed index-finger use, and applying a model trained under one posture to correct the other can push error in the wrong direction and create new mistakes. This card also describes a case of "aimed correctly but still missed," which is a different failure from not having seen the target clearly in the first place — the latter is fixed by improving visibility and feedback, not by hit tolerance or offset compensation.
Applying it
- When sizing hit tolerance, add more margin on the side of the target that sits above the visual aim point or is occluded by the fingertip, rather than expanding evenly in every direction.
- For precision tasks — exact point selection, the first stroke of a drawing — consider an indirect aiming aid such as a magnifier or crosshair that explicitly shows the user both reference frames ("where your eye is" and "where your finger is pressing") instead of silently correcting behind the scenes.
- Validate with participants using their natural grip rather than testers tapping vertically with an index finger, which systematically understates real aiming offset.