A launched action has a non-cancelable window

What it is

Between "intending to act" and "already acting," there's a definite boundary: cross it, and even an instruction to stop right now can't halt the limb mid-course — it finishes the movement as planned. This is the onset of the non-cancelable window (the point of no return). It isn't a fuzzy continuum where cancellation gets gradually harder the later it comes; there's a comparatively sharp transition zone — almost always stoppable before it, almost never stoppable after.

Why it happens

Once a motor program starts releasing, the muscle activation commands are already issued and moving down the execution pipeline. Stopping the limb requires a separate inhibitory signal to catch up with and override that pipeline before it finishes. If the stop signal arrives early enough, the inhibitory command can still intervene before the muscles actually activate. But once the motor program has advanced into the middle or later part of execution, even a stop signal that gets issued can't complete the full chain — generate the signal, conduct it to the muscle, override activation already underway — within the time remaining, so the movement runs to completion as planned. This is essentially a race between two processes: continuing the already-launched action, and generating and conducting a stop command. Whichever finishes first determines whether the action can be intercepted.

Studying it

The standard paradigm has participants respond to a go signal while a "stop" signal is randomly inserted at varying delays after the action has begun preparing or just started, instructing them to try their best not to produce the movement. Systematically varying that delay traces a curve of successful-stop rate dropping as delay increases; the delay where the curve falls sharply toward zero marks the onset of the non-cancelable window for that specific task. Such experiments are typically explained with a race model — treating continued execution and stop-signal generation as two independent processes racing each other, with whichever finishes first determining the outcome.

Where it stops holding

Exactly where the window's onset falls depends on the type of action, individual differences, and how well-rehearsed the task is — there's no single number that applies across all movements. Simpler, faster, more practiced movements have an earlier window onset and less room to be stopped. Also, lab "stop" signals are typically a clear external cue, while in real settings a user often changes their mind mid-course on their own; self-initiated stopping and externally cued stopping may not share the same latency, so applying a lab-measured window width directly to spontaneous cancellation needs caution.

Applying it

• Any design that expects a user to change their mind mid-action — undo, a second confirmation, blocking a hazardous operation — should place its intervention point before the window closes, not hope to intercept the action while it's already underway; if a confirmation dialog or warning shows up too late, the user may see it and still be unable to stop.
• For irreversible or costly operations, put the confirmation step as a separate action before the movement actually launches, rather than relying on a cancel gesture during the movement itself (like "release to cancel") — the latter requires the user to complete a fresh stop decision before the window closes, which may itself come too late.
• How to check: log the proportion of cases where a user still completes the original action after an interception prompt appears, measured from trigger to completion. If that proportion is near one hundred percent, the prompt is landing inside the non-cancelable window and the intervention point needs to move earlier.

Related

• Same group: A8.05.1 An action is programmed in full before it is launched · A8.05.2 Movement complexity increases the preparation time needed to program it · A8.05.4 Cancelling an already-programmed action itself takes time and carries a cost · A8.05.5 Changing the target inside that window guarantees a mis-hit
• Nearby: A8.04 Psychological Refractory Period · A10.14 Forcing Functions and Interlocks
• Search terms: point of no return · stop-signal paradigm · response inhibition